Datasets:

---

bigcode

/

jupyter-parsed

like 4

Follow

BigCode 1.35k

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "[](http://introml.analyticsdojo.com)\n<center><h1>Introduction to Python - Groupby and Pivot Tables</h1></center>\n<center><h3><a href = 'http://introml.analyticsdojo.com'>introml.analyticsdojo.com</a></h3></center>\n\n\n", "_____no_output_____" ], [ "# Groupby and Pivot Tables", "_____no_output_____" ] ], [ [ "!wget https://raw.githubusercontent.com/rpi-techfundamentals/spring2019-materials/master/input/train.csv\n!wget https://raw.githubusercontent.com/rpi-techfundamentals/spring2019-materials/master/input/test.csv", "_____no_output_____" ], [ "import numpy as np \nimport pandas as pd \n\n# Input data files are available in the \"../input/\" directory.\n# Let's input them into a Pandas DataFrame\ntrain = pd.read_csv(\"train.csv\")\ntest = pd.read_csv(\"test.csv\")", "_____no_output_____" ] ], [ [ "### Groupby\n- Often it is useful to see statistics by different classes.\n- Can be used to examine different subpopulations", "_____no_output_____" ] ], [ [ "train.head()", "_____no_output_____" ], [ "print(train.dtypes)", "_____no_output_____" ], [ "#What does this tell us? \ntrain.groupby(['Sex']).Survived.mean()", "_____no_output_____" ], [ "#What does this tell us? \ntrain.groupby(['Sex','Pclass']).Survived.mean()", "_____no_output_____" ], [ "#What does this tell us? Here it doesn't look so clear. We could separate by set age ranges.\ntrain.groupby(['Sex','Age']).Survived.mean()", "_____no_output_____" ] ], [ [ "### Combining Multiple Operations\n- *Splitting* the data into groups based on some criteria\n- *Applying* a function to each group independently\n- *Combining* the results into a data structure", "_____no_output_____" ] ], [ [ "s = train.groupby(['Sex','Pclass'], as_index=False).Survived.sum()\ns['PerSurv'] = train.groupby(['Sex','Pclass'], as_index=False).Survived.mean().Survived\ns['PerSurv']=s['PerSurv']*100\ns['Count'] = train.groupby(['Sex','Pclass'], as_index=False).Survived.count().Survived\nsurvived =s.Survived\ns", "_____no_output_____" ], [ "#What does this tell us? \nspmean=train.groupby(['Sex','Pclass']).Survived.mean()\nspcount=train.groupby(['Sex','Pclass']).Survived.sum()\nspsum=train.groupby(['Sex','Pclass']).Survived.count()\nspsum", "_____no_output_____" ] ], [ [ "### Pivot Tables\n- A pivot table is a data summarization tool, much easier than the syntax of groupBy. \n- It can be used to that sum, sort, averge, count, over a pandas dataframe. \n- Download and open data in excel to appreciate the ways that you can use Pivot Tables. ", "_____no_output_____" ] ], [ [ "#Load it and create a pivot table.\nfrom google.colab import files\nfiles.download('train.csv')", "_____no_output_____" ], [ "#List the index and the functions you want to aggregage by. \npd.pivot_table(train,index=[\"Sex\",\"Pclass\"],values=[\"Survived\"],aggfunc=['count','sum','mean',])", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "body = client_cred.get_object(Bucket=bucket,Key='311_Service_Requests_from_2010_to_Present_min.csv')['Body']\n# add missing __iter__ method, so pandas accepts body as file-like object\nif not hasattr(body, \"__iter__\"): body.__iter__ = types.MethodType( __iter__, body )\n\ndf_data_1 = pd.read_csv(body)\ndf_data_1.head()\n", "_____no_output_____" ], [ "df_data_1.columns", "_____no_output_____" ], [ "df_data_1['Complaint Type'].value_counts()", "_____no_output_____" ], [ "df_data_1['Incident Zip'].value_counts()", "_____no_output_____" ], [ "df1 = df_data_1[ df_data_1['Complaint Type'].isin(['HEATING', 'HEAT/HOT WATER'])]\ndf1 = df1[df1['Borough'] == 'BRONX']\ndf1['complaint'] = 1\ndf1.rename(columns={'Incident Zip': 'ZipCode', 'Incident Address': 'Address'}, inplace = True)\ndf1.head(5)", "_____no_output_____" ], [ "body = client_cred.get_object(Bucket=bucket,Key='BX_18v1.csv')['Body']\n# add missing __iter__ method, so pandas accepts body as file-like object\nif not hasattr(body, \"__iter__\"): body.__iter__ = types.MethodType( __iter__, body )\n\ndf_data_2 = pd.read_csv(body, usecols = ['Address', 'BldgArea', 'BldgDepth', 'BuiltFAR', 'CommFAR', 'FacilFAR', 'Lot', 'LotArea', 'LotDepth', 'NumBldgs', 'NumFloors', 'OfficeArea', 'ResArea', 'ResidFAR', 'RetailArea', 'YearBuilt', 'YearAlter1', 'ZipCode', 'YCoord', 'XCoord'])\ndf_data_2.head()", "_____no_output_____" ], [ "df_data_2.columns", "_____no_output_____" ], [ "df1.columns", "_____no_output_____" ], [ "df = pd.merge(df_data_2, df1, on=['ZipCode', 'Address'], how='outer')\ndf.head(5)", "_____no_output_____" ], [ "df.columns", "_____no_output_____" ], [ "df = df.drop(['XCoord', 'YCoord', 'Unnamed: 0', 'Unique Key','Created Date', 'Closed Date', 'Complaint Type', 'Borough', 'Latitude', 'Longitude'], axis=1)", "_____no_output_____" ], [ "df.columns", "_____no_output_____" ], [ "df.fillna(0, inplace=True)\ndf['complaint'].value_counts()", "_____no_output_____" ], [ "df.info()", "<class 'pandas.core.frame.DataFrame'>\nInt64Index: 679322 entries, 0 to 679321\nData columns (total 25 columns):\nLot 679322 non-null float64\nZipCode 679322 non-null float64\nAddress 679322 non-null object\nLotArea 679322 non-null float64\nBldgArea 679322 non-null float64\nResArea 679322 non-null float64\nOfficeArea 679322 non-null float64\nRetailArea 679322 non-null float64\nNumBldgs 679322 non-null float64\nNumFloors 679322 non-null float64\nLotDepth 679322 non-null float64\nBldgDepth 679322 non-null float64\nYearBuilt 679322 non-null float64\nYearAlter1 679322 non-null float64\nBuiltFAR 679322 non-null float64\nResidFAR 679322 non-null float64\nCommFAR 679322 non-null float64\nFacilFAR 679322 non-null float64\nLocation Type 679322 non-null object\nStreet Name 679322 non-null object\nAddress Type 679322 non-null object\nCity 679322 non-null object\nStatus 679322 non-null object\nResolution Description 679322 non-null object\ncomplaint 679322 non-null float64\ndtypes: float64(18), object(7)\nmemory usage: 134.8+ MB\n" ], [ "from scipy.stats import spearmanr\ndf.corr(method='pearson')['complaint']", "_____no_output_____" ], [ "from sklearn.linear_model import LogisticRegression\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import mean_squared_error, r2_score\nfrom sklearn.preprocessing import StandardScaler\n\nscaler = StandardScaler()\ny = df['complaint']\nx = df[['BldgArea', 'ResArea', 'NumFloors', 'BldgDepth', 'YearBuilt', 'YearAlter1', 'BuiltFAR', 'ResidFAR', 'FacilFAR']]\nx = scaler.fit_transform(x)\nX_train, X_test, y_train, y_test = train_test_split(x, y, test_size=0.3, random_state=17)", "_____no_output_____" ], [ "logisticRegr = LogisticRegression(solver = 'liblinear')\nlogisticRegr.fit(X_train, y_train)\ny_pred = logisticRegr.predict(X_test)\n\nscore = logisticRegr.score(X_test, y_test)\nprint(score)", "0.9311618914900612\n" ], [ "import matplotlib.pyplot as plt\nimport seaborn as sns\nfrom sklearn import metrics\n\ncm = metrics.confusion_matrix(y_test, y_pred)\nplt.figure(figsize=(8,8))\nsns.heatmap(cm, annot=True, fmt=\".3f\", linewidths=.5, square = True, cmap = 'Blues_r');\nplt.ylabel('Actual label');\nplt.xlabel('Predicted label');\nall_sample_title = 'Accuracy Score: {0}'.format(score)\nplt.title(all_sample_title, size = 15);", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Notebook #2\n\nIn this notebook, we worked with the result dataset from Notebook #1 and computed rolling statistics (mean, difference, std, max, min) for a list of features over various time windows. \nThis was the most time consuming and computational expensive part of the entire tutorial. We encountered some roadblocks and found some workarounds. Please see below for more details.\n\n## Outline\n\n- [Define Rolling Features and Window Sizes](#Define-list-of-features-for-rolling-compute,-window-sizes)\n- [Issues and Solutions](#What-issues-we-encountered-using-Pyspark-and-how-we-solved-them?)\n- [Rolling Compute](#Rolling-Compute)\n - [Rolling Mean](#Rolling-Mean)\n - [Rolling Difference](#Rolling-Difference)\n - [Rolling Std](#Rolling-Std)\n - [Rolling Max](#Rolling-Max)\n - [Rolling Min](#Rolling-Min)\n- [Join Results](#Join-result-dataset-from-the-five-rolling-compute-cells:)\n \n", "_____no_output_____" ] ], [ [ "import pyspark.sql.functions as F\nimport time\nimport subprocess\nimport sys\nimport os\nimport re\n\nfrom pyspark import SparkConf\nfrom pyspark import SparkContext\nfrom pyspark import SQLContext\nfrom pyspark.sql.types import *\nfrom pyspark.sql.functions import col,udf,lag,date_add,explode,lit,concat,unix_timestamp\nfrom pyspark.sql.dataframe import *\nfrom pyspark.sql.window import Window\nfrom pyspark.sql.types import DateType\nfrom datetime import datetime, timedelta\nfrom pyspark.sql import Row\n\nstart_time = time.time()\n", "_____no_output_____" ] ], [ [ "## Define list of features for rolling compute, window sizes", "_____no_output_____" ] ], [ [ "rolling_features = [\n 'warn_type1_total', 'warn_type2_total', \n 'pca_1_warn','pca_2_warn', 'pca_3_warn', 'pca_4_warn', 'pca_5_warn',\n 'pca_6_warn','pca_7_warn', 'pca_8_warn', 'pca_9_warn', 'pca_10_warn',\n 'pca_11_warn','pca_12_warn', 'pca_13_warn', 'pca_14_warn', 'pca_15_warn',\n 'pca_16_warn','pca_17_warn', 'pca_18_warn', 'pca_19_warn', 'pca_20_warn',\n 'problem_type_1', 'problem_type_2', 'problem_type_3','problem_type_4',\n 'problem_type_1_per_usage1','problem_type_2_per_usage1',\n 'problem_type_3_per_usage1','problem_type_4_per_usage1',\n 'problem_type_1_per_usage2','problem_type_2_per_usage2',\n 'problem_type_3_per_usage2','problem_type_4_per_usage2', \n 'fault_code_type_1_count', 'fault_code_type_2_count', 'fault_code_type_3_count', 'fault_code_type_4_count', \n 'fault_code_type_1_count_per_usage1','fault_code_type_2_count_per_usage1',\n 'fault_code_type_3_count_per_usage1', 'fault_code_type_4_count_per_usage1',\n 'fault_code_type_1_count_per_usage2','fault_code_type_2_count_per_usage2',\n 'fault_code_type_3_count_per_usage2', 'fault_code_type_4_count_per_usage2']\n \n# lag window 3, 7, 14, 30, 90 days\nlags = [3, 7, 14, 30, 90]\n\nprint(len(rolling_features))\n", "46\n" ] ], [ [ "## What issues we encountered using Pyspark and how we solved them?\n\n- If the entire list of **46 features** and **5 time windows** were computed for **5 different types of rolling** (mean, difference, std, max, min) all in one go, we always ran into \"StackOverFlow\" error. \n- It was because the lineage was too long and Spark could not handle it.\n- We could either create checkPoint and materialize it throughout the process.\n- OR break the workload into chunks and save the result from each chunk as parquet file.\n\n## A few things we found helpful:\n- Before the rolling compute, save the upstream work as a parquet file in Notebook_1 (\"Notebook_1_DataCleansing_FeatureEngineering\"). It will speed up the whole process because we no need to repeat all the previous steps. It will also help reduce the lineage.\n- Print out the lag and feature name to track progress.\n- Use \"htop\" command from the terminal to keep track how many CPUs are running for a particular task. For rolling compute, we were considering two potential approaches: 1) Use Spark clusters on HDInsight to perform rolling compute in parallel; 2) Use single node Spark on a powerful VM. By looking at htop dashboard, we saw all the 32 cores were running at the same time for a single task (for example compute rolling mean). So if say we divide the workload onto multiple nodes and each node runs a type of rolling compute, the amount of time taken will be comparable with running everything in a sequential manner on a single node Spark on a powerful machine.\n- Use \"%%time\" for each cell to get an estimate of the total run time, we will then have a better idea where and what to optimze the process.\n- Materialize the intermediate results by either caching in memory or writing as parquet files. We chose to save as parquet files because we did not want to repeat the compute again in case cache() did not work or any part of the rolling compute did not work.\n- Why parquet? There are many reasons, just to name a few: parquet not only saves the data but also the schema, it is a preferred file format by Spark, you are allowed to read only the data you need, etc..\n\n<br>\n", "_____no_output_____" ], [ "## Rolling Compute\n### Rolling Mean", "_____no_output_____" ] ], [ [ "%%time\n\n# Load result dataset from Notebook #1\ndf = sqlContext.read.parquet('/mnt/resource/PysparkExample/notebook1_result.parquet')\n\nfor lag_n in lags:\n wSpec = Window.partitionBy('deviceid').orderBy('date').rowsBetween(1-lag_n, 0)\n for col_name in rolling_features:\n df = df.withColumn(col_name+'_rollingmean_'+str(lag_n), F.avg(col(col_name)).over(wSpec))\n print(\"Lag = %d, Column = %s\" % (lag_n, col_name))\n\n# Save the intermediate result for downstream work\ndf.write.mode('overwrite').parquet('/mnt/resource/PysparkExample/data_rollingmean.parquet')\n", "Lag = 3, Column = warn_type1_total\nLag = 3, Column = warn_type2_total\nLag = 3, Column = pca_1_warn\nLag = 3, Column = pca_2_warn\nLag = 3, Column = pca_3_warn\nLag = 3, Column = pca_4_warn\nLag = 3, Column = pca_5_warn\nLag = 3, Column = pca_6_warn\nLag = 3, Column = pca_7_warn\nLag = 3, Column = pca_8_warn\nLag = 3, Column = pca_9_warn\nLag = 3, Column = pca_10_warn\nLag = 3, Column = pca_11_warn\nLag = 3, Column = pca_12_warn\nLag = 3, Column = pca_13_warn\nLag = 3, Column = pca_14_warn\nLag = 3, Column = pca_15_warn\nLag = 3, Column = pca_16_warn\nLag = 3, Column = pca_17_warn\nLag = 3, Column = pca_18_warn\nLag = 3, Column = pca_19_warn\nLag = 3, Column = pca_20_warn\nLag = 3, Column = problem_type_1\nLag = 3, Column = problem_type_2\nLag = 3, Column = problem_type_3\nLag = 3, Column = problem_type_4\nLag = 3, Column = problem_type_1_per_usage1\nLag = 3, Column = problem_type_2_per_usage1\nLag = 3, Column = problem_type_3_per_usage1\nLag = 3, Column = problem_type_4_per_usage1\nLag = 3, Column = problem_type_1_per_usage2\nLag = 3, Column = problem_type_2_per_usage2\nLag = 3, Column = problem_type_3_per_usage2\nLag = 3, Column = problem_type_4_per_usage2\nLag = 3, Column = fault_code_type_1_count\nLag = 3, Column = fault_code_type_2_count\nLag = 3, Column = fault_code_type_3_count\nLag = 3, Column = fault_code_type_4_count\nLag = 3, Column = fault_code_type_1_count_per_usage1\nLag = 3, Column = fault_code_type_2_count_per_usage1\nLag = 3, Column = fault_code_type_3_count_per_usage1\nLag = 3, Column = fault_code_type_4_count_per_usage1\nLag = 3, Column = fault_code_type_1_count_per_usage2\nLag = 3, Column = fault_code_type_2_count_per_usage2\nLag = 3, Column = fault_code_type_3_count_per_usage2\nLag = 3, Column = fault_code_type_4_count_per_usage2\nLag = 7, Column = warn_type1_total\nLag = 7, Column = warn_type2_total\nLag = 7, Column = pca_1_warn\nLag = 7, Column = pca_2_warn\nLag = 7, Column = pca_3_warn\nLag = 7, Column = pca_4_warn\nLag = 7, Column = pca_5_warn\nLag = 7, Column = pca_6_warn\nLag = 7, Column = pca_7_warn\nLag = 7, Column = pca_8_warn\nLag = 7, Column = pca_9_warn\nLag = 7, Column = pca_10_warn\nLag = 7, Column = pca_11_warn\nLag = 7, Column = pca_12_warn\nLag = 7, Column = pca_13_warn\nLag = 7, Column = pca_14_warn\nLag = 7, Column = pca_15_warn\nLag = 7, Column = pca_16_warn\nLag = 7, Column = pca_17_warn\nLag = 7, Column = pca_18_warn\nLag = 7, Column = pca_19_warn\nLag = 7, Column = pca_20_warn\nLag = 7, Column = problem_type_1\nLag = 7, Column = problem_type_2\nLag = 7, Column = problem_type_3\nLag = 7, Column = problem_type_4\nLag = 7, Column = problem_type_1_per_usage1\nLag = 7, Column = problem_type_2_per_usage1\nLag = 7, Column = problem_type_3_per_usage1\nLag = 7, Column = problem_type_4_per_usage1\nLag = 7, Column = problem_type_1_per_usage2\nLag = 7, Column = problem_type_2_per_usage2\nLag = 7, Column = problem_type_3_per_usage2\nLag = 7, Column = problem_type_4_per_usage2\nLag = 7, Column = fault_code_type_1_count\nLag = 7, Column = fault_code_type_2_count\nLag = 7, Column = fault_code_type_3_count\nLag = 7, Column = fault_code_type_4_count\nLag = 7, Column = fault_code_type_1_count_per_usage1\nLag = 7, Column = fault_code_type_2_count_per_usage1\nLag = 7, Column = fault_code_type_3_count_per_usage1\nLag = 7, Column = fault_code_type_4_count_per_usage1\nLag = 7, Column = fault_code_type_1_count_per_usage2\nLag = 7, Column = fault_code_type_2_count_per_usage2\nLag = 7, Column = fault_code_type_3_count_per_usage2\nLag = 7, Column = fault_code_type_4_count_per_usage2\nLag = 14, Column = warn_type1_total\nLag = 14, Column = warn_type2_total\nLag = 14, Column = pca_1_warn\nLag = 14, Column = pca_2_warn\nLag = 14, Column = pca_3_warn\nLag = 14, Column = pca_4_warn\nLag = 14, Column = pca_5_warn\nLag = 14, Column = pca_6_warn\nLag = 14, Column = pca_7_warn\nLag = 14, Column = pca_8_warn\nLag = 14, Column = pca_9_warn\nLag = 14, Column = pca_10_warn\nLag = 14, Column = pca_11_warn\nLag = 14, Column = pca_12_warn\nLag = 14, Column = pca_13_warn\nLag = 14, Column = pca_14_warn\nLag = 14, Column = pca_15_warn\nLag = 14, Column = pca_16_warn\nLag = 14, Column = pca_17_warn\nLag = 14, Column = pca_18_warn\nLag = 14, Column = pca_19_warn\nLag = 14, Column = pca_20_warn\nLag = 14, Column = problem_type_1\nLag = 14, Column = problem_type_2\nLag = 14, Column = problem_type_3\nLag = 14, Column = problem_type_4\nLag = 14, Column = problem_type_1_per_usage1\nLag = 14, Column = problem_type_2_per_usage1\nLag = 14, Column = problem_type_3_per_usage1\nLag = 14, Column = problem_type_4_per_usage1\nLag = 14, Column = problem_type_1_per_usage2\nLag = 14, Column = problem_type_2_per_usage2\nLag = 14, Column = problem_type_3_per_usage2\nLag = 14, Column = problem_type_4_per_usage2\nLag = 14, Column = fault_code_type_1_count\nLag = 14, Column = fault_code_type_2_count\nLag = 14, Column = fault_code_type_3_count\nLag = 14, Column = fault_code_type_4_count\nLag = 14, Column = fault_code_type_1_count_per_usage1\nLag = 14, Column = fault_code_type_2_count_per_usage1\nLag = 14, Column = fault_code_type_3_count_per_usage1\nLag = 14, Column = fault_code_type_4_count_per_usage1\nLag = 14, Column = fault_code_type_1_count_per_usage2\nLag = 14, Column = fault_code_type_2_count_per_usage2\nLag = 14, Column = fault_code_type_3_count_per_usage2\nLag = 14, Column = fault_code_type_4_count_per_usage2\nLag = 30, Column = warn_type1_total\nLag = 30, Column = warn_type2_total\nLag = 30, Column = pca_1_warn\nLag = 30, Column = pca_2_warn\nLag = 30, Column = pca_3_warn\nLag = 30, Column = pca_4_warn\nLag = 30, Column = pca_5_warn\nLag = 30, Column = pca_6_warn\nLag = 30, Column = pca_7_warn\nLag = 30, Column = pca_8_warn\nLag = 30, Column = pca_9_warn\nLag = 30, Column = pca_10_warn\nLag = 30, Column = pca_11_warn\nLag = 30, Column = pca_12_warn\nLag = 30, Column = pca_13_warn\nLag = 30, Column = pca_14_warn\nLag = 30, Column = pca_15_warn\nLag = 30, Column = pca_16_warn\nLag = 30, Column = pca_17_warn\nLag = 30, Column = pca_18_warn\nLag = 30, Column = pca_19_warn\nLag = 30, Column = pca_20_warn\nLag = 30, Column = problem_type_1\nLag = 30, Column = problem_type_2\nLag = 30, Column = problem_type_3\nLag = 30, Column = problem_type_4\nLag = 30, Column = problem_type_1_per_usage1\nLag = 30, Column = problem_type_2_per_usage1\nLag = 30, Column = problem_type_3_per_usage1\nLag = 30, Column = problem_type_4_per_usage1\nLag = 30, Column = problem_type_1_per_usage2\nLag = 30, Column = problem_type_2_per_usage2\nLag = 30, Column = problem_type_3_per_usage2\nLag = 30, Column = problem_type_4_per_usage2\nLag = 30, Column = fault_code_type_1_count\nLag = 30, Column = fault_code_type_2_count\nLag = 30, Column = fault_code_type_3_count\nLag = 30, Column = fault_code_type_4_count\nLag = 30, Column = fault_code_type_1_count_per_usage1\nLag = 30, Column = fault_code_type_2_count_per_usage1\nLag = 30, Column = fault_code_type_3_count_per_usage1\nLag = 30, Column = fault_code_type_4_count_per_usage1\nLag = 30, Column = fault_code_type_1_count_per_usage2\nLag = 30, Column = fault_code_type_2_count_per_usage2\nLag = 30, Column = fault_code_type_3_count_per_usage2\nLag = 30, Column = fault_code_type_4_count_per_usage2\nLag = 90, Column = warn_type1_total\nLag = 90, Column = warn_type2_total\nLag = 90, Column = pca_1_warn\nLag = 90, Column = pca_2_warn\nLag = 90, Column = pca_3_warn\nLag = 90, Column = pca_4_warn\nLag = 90, Column = pca_5_warn\nLag = 90, Column = pca_6_warn\nLag = 90, Column = pca_7_warn\nLag = 90, Column = pca_8_warn\nLag = 90, Column = pca_9_warn\nLag = 90, Column = pca_10_warn\nLag = 90, Column = pca_11_warn\nLag = 90, Column = pca_12_warn\nLag = 90, Column = pca_13_warn\nLag = 90, Column = pca_14_warn\nLag = 90, Column = pca_15_warn\nLag = 90, Column = pca_16_warn\nLag = 90, Column = pca_17_warn\nLag = 90, Column = pca_18_warn\nLag = 90, Column = pca_19_warn\nLag = 90, Column = pca_20_warn\nLag = 90, Column = problem_type_1\nLag = 90, Column = problem_type_2\nLag = 90, Column = problem_type_3\nLag = 90, Column = problem_type_4\nLag = 90, Column = problem_type_1_per_usage1\nLag = 90, Column = problem_type_2_per_usage1\nLag = 90, Column = problem_type_3_per_usage1\nLag = 90, Column = problem_type_4_per_usage1\nLag = 90, Column = problem_type_1_per_usage2\nLag = 90, Column = problem_type_2_per_usage2\nLag = 90, Column = problem_type_3_per_usage2\nLag = 90, Column = problem_type_4_per_usage2\nLag = 90, Column = fault_code_type_1_count\nLag = 90, Column = fault_code_type_2_count\nLag = 90, Column = fault_code_type_3_count\nLag = 90, Column = fault_code_type_4_count\nLag = 90, Column = fault_code_type_1_count_per_usage1\nLag = 90, Column = fault_code_type_2_count_per_usage1\nLag = 90, Column = fault_code_type_3_count_per_usage1\nLag = 90, Column = fault_code_type_4_count_per_usage1\nLag = 90, Column = fault_code_type_1_count_per_usage2\nLag = 90, Column = fault_code_type_2_count_per_usage2\nLag = 90, Column = fault_code_type_3_count_per_usage2\nLag = 90, Column = fault_code_type_4_count_per_usage2\nCPU times: user 848 ms, sys: 279 ms, total: 1.13 s\nWall time: 28min 6s\n" ] ], [ [ "### Rolling Difference", "_____no_output_____" ] ], [ [ "%%time\n\n# Load result dataset from Notebook #1\ndf = sqlContext.read.parquet('/mnt/resource/PysparkExample/notebook1_result.parquet')\n\nfor lag_n in lags:\n wSpec = Window.partitionBy('deviceid').orderBy('date').rowsBetween(1-lag_n, 0)\n for col_name in rolling_features:\n df = df.withColumn(col_name+'_rollingdiff_'+str(lag_n), col(col_name)-F.avg(col(col_name)).over(wSpec))\n print(\"Lag = %d, Column = %s\" % (lag_n, col_name))\n\nrollingdiff = df.select(['key'] + list(s for s in df.columns if \"rollingdiff\" in s))\n\n# Save the intermediate result for downstream work\nrollingdiff.write.mode('overwrite').parquet('/mnt/resource/PysparkExample/rollingdiff.parquet')\n", "Lag = 3, Column = warn_type1_total\nLag = 3, Column = warn_type2_total\nLag = 3, Column = pca_1_warn\nLag = 3, Column = pca_2_warn\nLag = 3, Column = pca_3_warn\nLag = 3, Column = pca_4_warn\nLag = 3, Column = pca_5_warn\nLag = 3, Column = pca_6_warn\nLag = 3, Column = pca_7_warn\nLag = 3, Column = pca_8_warn\nLag = 3, Column = pca_9_warn\nLag = 3, Column = pca_10_warn\nLag = 3, Column = pca_11_warn\nLag = 3, Column = pca_12_warn\nLag = 3, Column = pca_13_warn\nLag = 3, Column = pca_14_warn\nLag = 3, Column = pca_15_warn\nLag = 3, Column = pca_16_warn\nLag = 3, Column = pca_17_warn\nLag = 3, Column = pca_18_warn\nLag = 3, Column = pca_19_warn\nLag = 3, Column = pca_20_warn\nLag = 3, Column = problem_type_1\nLag = 3, Column = problem_type_2\nLag = 3, Column = problem_type_3\nLag = 3, Column = problem_type_4\nLag = 3, Column = problem_type_1_per_usage1\nLag = 3, Column = problem_type_2_per_usage1\nLag = 3, Column = problem_type_3_per_usage1\nLag = 3, Column = problem_type_4_per_usage1\nLag = 3, Column = problem_type_1_per_usage2\nLag = 3, Column = problem_type_2_per_usage2\nLag = 3, Column = problem_type_3_per_usage2\nLag = 3, Column = problem_type_4_per_usage2\nLag = 3, Column = fault_code_type_1_count\nLag = 3, Column = fault_code_type_2_count\nLag = 3, Column = fault_code_type_3_count\nLag = 3, Column = fault_code_type_4_count\nLag = 3, Column = fault_code_type_1_count_per_usage1\nLag = 3, Column = fault_code_type_2_count_per_usage1\nLag = 3, Column = fault_code_type_3_count_per_usage1\nLag = 3, Column = fault_code_type_4_count_per_usage1\nLag = 3, Column = fault_code_type_1_count_per_usage2\nLag = 3, Column = fault_code_type_2_count_per_usage2\nLag = 3, Column = fault_code_type_3_count_per_usage2\nLag = 3, Column = fault_code_type_4_count_per_usage2\nLag = 7, Column = warn_type1_total\nLag = 7, Column = warn_type2_total\nLag = 7, Column = pca_1_warn\nLag = 7, Column = pca_2_warn\nLag = 7, Column = pca_3_warn\nLag = 7, Column = pca_4_warn\nLag = 7, Column = pca_5_warn\nLag = 7, Column = pca_6_warn\nLag = 7, Column = pca_7_warn\nLag = 7, Column = pca_8_warn\nLag = 7, Column = pca_9_warn\nLag = 7, Column = pca_10_warn\nLag = 7, Column = pca_11_warn\nLag = 7, Column = pca_12_warn\nLag = 7, Column = pca_13_warn\nLag = 7, Column = pca_14_warn\nLag = 7, Column = pca_15_warn\nLag = 7, Column = pca_16_warn\nLag = 7, Column = pca_17_warn\nLag = 7, Column = pca_18_warn\nLag = 7, Column = pca_19_warn\nLag = 7, Column = pca_20_warn\nLag = 7, Column = problem_type_1\nLag = 7, Column = problem_type_2\nLag = 7, Column = problem_type_3\nLag = 7, Column = problem_type_4\nLag = 7, Column = problem_type_1_per_usage1\nLag = 7, Column = problem_type_2_per_usage1\nLag = 7, Column = problem_type_3_per_usage1\nLag = 7, Column = problem_type_4_per_usage1\nLag = 7, Column = problem_type_1_per_usage2\nLag = 7, Column = problem_type_2_per_usage2\nLag = 7, Column = problem_type_3_per_usage2\nLag = 7, Column = problem_type_4_per_usage2\nLag = 7, Column = fault_code_type_1_count\nLag = 7, Column = fault_code_type_2_count\nLag = 7, Column = fault_code_type_3_count\nLag = 7, Column = fault_code_type_4_count\nLag = 7, Column = fault_code_type_1_count_per_usage1\nLag = 7, Column = fault_code_type_2_count_per_usage1\nLag = 7, Column = fault_code_type_3_count_per_usage1\nLag = 7, Column = fault_code_type_4_count_per_usage1\nLag = 7, Column = fault_code_type_1_count_per_usage2\nLag = 7, Column = fault_code_type_2_count_per_usage2\nLag = 7, Column = fault_code_type_3_count_per_usage2\nLag = 7, Column = fault_code_type_4_count_per_usage2\nLag = 14, Column = warn_type1_total\nLag = 14, Column = warn_type2_total\nLag = 14, Column = pca_1_warn\nLag = 14, Column = pca_2_warn\nLag = 14, Column = pca_3_warn\nLag = 14, Column = pca_4_warn\nLag = 14, Column = pca_5_warn\nLag = 14, Column = pca_6_warn\nLag = 14, Column = pca_7_warn\nLag = 14, Column = pca_8_warn\nLag = 14, Column = pca_9_warn\nLag = 14, Column = pca_10_warn\nLag = 14, Column = pca_11_warn\nLag = 14, Column = pca_12_warn\nLag = 14, Column = pca_13_warn\nLag = 14, Column = pca_14_warn\nLag = 14, Column = pca_15_warn\nLag = 14, Column = pca_16_warn\nLag = 14, Column = pca_17_warn\nLag = 14, Column = pca_18_warn\nLag = 14, Column = pca_19_warn\nLag = 14, Column = pca_20_warn\nLag = 14, Column = problem_type_1\nLag = 14, Column = problem_type_2\nLag = 14, Column = problem_type_3\nLag = 14, Column = problem_type_4\nLag = 14, Column = problem_type_1_per_usage1\nLag = 14, Column = problem_type_2_per_usage1\nLag = 14, Column = problem_type_3_per_usage1\nLag = 14, Column = problem_type_4_per_usage1\nLag = 14, Column = problem_type_1_per_usage2\nLag = 14, Column = problem_type_2_per_usage2\nLag = 14, Column = problem_type_3_per_usage2\nLag = 14, Column = problem_type_4_per_usage2\nLag = 14, Column = fault_code_type_1_count\nLag = 14, Column = fault_code_type_2_count\nLag = 14, Column = fault_code_type_3_count\nLag = 14, Column = fault_code_type_4_count\nLag = 14, Column = fault_code_type_1_count_per_usage1\nLag = 14, Column = fault_code_type_2_count_per_usage1\nLag = 14, Column = fault_code_type_3_count_per_usage1\nLag = 14, Column = fault_code_type_4_count_per_usage1\nLag = 14, Column = fault_code_type_1_count_per_usage2\nLag = 14, Column = fault_code_type_2_count_per_usage2\nLag = 14, Column = fault_code_type_3_count_per_usage2\nLag = 14, Column = fault_code_type_4_count_per_usage2\nLag = 30, Column = warn_type1_total\nLag = 30, Column = warn_type2_total\nLag = 30, Column = pca_1_warn\nLag = 30, Column = pca_2_warn\nLag = 30, Column = pca_3_warn\nLag = 30, Column = pca_4_warn\nLag = 30, Column = pca_5_warn\nLag = 30, Column = pca_6_warn\nLag = 30, Column = pca_7_warn\nLag = 30, Column = pca_8_warn\nLag = 30, Column = pca_9_warn\nLag = 30, Column = pca_10_warn\nLag = 30, Column = pca_11_warn\nLag = 30, Column = pca_12_warn\nLag = 30, Column = pca_13_warn\nLag = 30, Column = pca_14_warn\nLag = 30, Column = pca_15_warn\nLag = 30, Column = pca_16_warn\nLag = 30, Column = pca_17_warn\nLag = 30, Column = pca_18_warn\nLag = 30, Column = pca_19_warn\nLag = 30, Column = pca_20_warn\nLag = 30, Column = problem_type_1\nLag = 30, Column = problem_type_2\nLag = 30, Column = problem_type_3\nLag = 30, Column = problem_type_4\nLag = 30, Column = problem_type_1_per_usage1\nLag = 30, Column = problem_type_2_per_usage1\nLag = 30, Column = problem_type_3_per_usage1\nLag = 30, Column = problem_type_4_per_usage1\nLag = 30, Column = problem_type_1_per_usage2\nLag = 30, Column = problem_type_2_per_usage2\nLag = 30, Column = problem_type_3_per_usage2\nLag = 30, Column = problem_type_4_per_usage2\nLag = 30, Column = fault_code_type_1_count\nLag = 30, Column = fault_code_type_2_count\nLag = 30, Column = fault_code_type_3_count\nLag = 30, Column = fault_code_type_4_count\nLag = 30, Column = fault_code_type_1_count_per_usage1\nLag = 30, Column = fault_code_type_2_count_per_usage1\nLag = 30, Column = fault_code_type_3_count_per_usage1\nLag = 30, Column = fault_code_type_4_count_per_usage1\nLag = 30, Column = fault_code_type_1_count_per_usage2\nLag = 30, Column = fault_code_type_2_count_per_usage2\nLag = 30, Column = fault_code_type_3_count_per_usage2\nLag = 30, Column = fault_code_type_4_count_per_usage2\nLag = 90, Column = warn_type1_total\nLag = 90, Column = warn_type2_total\nLag = 90, Column = pca_1_warn\nLag = 90, Column = pca_2_warn\nLag = 90, Column = pca_3_warn\nLag = 90, Column = pca_4_warn\nLag = 90, Column = pca_5_warn\nLag = 90, Column = pca_6_warn\nLag = 90, Column = pca_7_warn\nLag = 90, Column = pca_8_warn\nLag = 90, Column = pca_9_warn\nLag = 90, Column = pca_10_warn\nLag = 90, Column = pca_11_warn\nLag = 90, Column = pca_12_warn\nLag = 90, Column = pca_13_warn\nLag = 90, Column = pca_14_warn\nLag = 90, Column = pca_15_warn\nLag = 90, Column = pca_16_warn\nLag = 90, Column = pca_17_warn\nLag = 90, Column = pca_18_warn\nLag = 90, Column = pca_19_warn\nLag = 90, Column = pca_20_warn\nLag = 90, Column = problem_type_1\nLag = 90, Column = problem_type_2\nLag = 90, Column = problem_type_3\nLag = 90, Column = problem_type_4\nLag = 90, Column = problem_type_1_per_usage1\nLag = 90, Column = problem_type_2_per_usage1\nLag = 90, Column = problem_type_3_per_usage1\nLag = 90, Column = problem_type_4_per_usage1\nLag = 90, Column = problem_type_1_per_usage2\nLag = 90, Column = problem_type_2_per_usage2\nLag = 90, Column = problem_type_3_per_usage2\nLag = 90, Column = problem_type_4_per_usage2\nLag = 90, Column = fault_code_type_1_count\nLag = 90, Column = fault_code_type_2_count\nLag = 90, Column = fault_code_type_3_count\nLag = 90, Column = fault_code_type_4_count\nLag = 90, Column = fault_code_type_1_count_per_usage1\nLag = 90, Column = fault_code_type_2_count_per_usage1\nLag = 90, Column = fault_code_type_3_count_per_usage1\nLag = 90, Column = fault_code_type_4_count_per_usage1\nLag = 90, Column = fault_code_type_1_count_per_usage2\nLag = 90, Column = fault_code_type_2_count_per_usage2\nLag = 90, Column = fault_code_type_3_count_per_usage2\nLag = 90, Column = fault_code_type_4_count_per_usage2\nCPU times: user 1.18 s, sys: 383 ms, total: 1.56 s\nWall time: 45min 12s\n" ] ], [ [ "### Rolling Std", "_____no_output_____" ] ], [ [ "%%time\n\n# Load result dataset from Notebook #1\ndf = sqlContext.read.parquet('/mnt/resource/PysparkExample/notebook1_result.parquet')\n\nfor lag_n in lags:\n wSpec = Window.partitionBy('deviceid').orderBy('date').rowsBetween(1-lag_n, 0)\n for col_name in rolling_features:\n df = df.withColumn(col_name+'_rollingstd_'+str(lag_n), F.stddev(col(col_name)).over(wSpec))\n print(\"Lag = %d, Column = %s\" % (lag_n, col_name))\n\n# There are some missing values for rollingstd features\nrollingstd_features = list(s for s in df.columns if \"rollingstd\" in s)\ndf = df.fillna(0, subset=rollingstd_features)\nrollingstd = df.select(['key'] + list(s for s in df.columns if \"rollingstd\" in s))\n\n# Save the intermediate result for downstream work\nrollingstd.write.mode('overwrite').parquet('/mnt/resource/PysparkExample/rollingstd.parquet')\n", "Lag = 3, Column = warn_type1_total\nLag = 3, Column = warn_type2_total\nLag = 3, Column = pca_1_warn\nLag = 3, Column = pca_2_warn\nLag = 3, Column = pca_3_warn\nLag = 3, Column = pca_4_warn\nLag = 3, Column = pca_5_warn\nLag = 3, Column = pca_6_warn\nLag = 3, Column = pca_7_warn\nLag = 3, Column = pca_8_warn\nLag = 3, Column = pca_9_warn\nLag = 3, Column = pca_10_warn\nLag = 3, Column = pca_11_warn\nLag = 3, Column = pca_12_warn\nLag = 3, Column = pca_13_warn\nLag = 3, Column = pca_14_warn\nLag = 3, Column = pca_15_warn\nLag = 3, Column = pca_16_warn\nLag = 3, Column = pca_17_warn\nLag = 3, Column = pca_18_warn\nLag = 3, Column = pca_19_warn\nLag = 3, Column = pca_20_warn\nLag = 3, Column = problem_type_1\nLag = 3, Column = problem_type_2\nLag = 3, Column = problem_type_3\nLag = 3, Column = problem_type_4\nLag = 3, Column = problem_type_1_per_usage1\nLag = 3, Column = problem_type_2_per_usage1\nLag = 3, Column = problem_type_3_per_usage1\nLag = 3, Column = problem_type_4_per_usage1\nLag = 3, Column = problem_type_1_per_usage2\nLag = 3, Column = problem_type_2_per_usage2\nLag = 3, Column = problem_type_3_per_usage2\nLag = 3, Column = problem_type_4_per_usage2\nLag = 3, Column = fault_code_type_1_count\nLag = 3, Column = fault_code_type_2_count\nLag = 3, Column = fault_code_type_3_count\nLag = 3, Column = fault_code_type_4_count\nLag = 3, Column = fault_code_type_1_count_per_usage1\nLag = 3, Column = fault_code_type_2_count_per_usage1\nLag = 3, Column = fault_code_type_3_count_per_usage1\nLag = 3, Column = fault_code_type_4_count_per_usage1\nLag = 3, Column = fault_code_type_1_count_per_usage2\nLag = 3, Column = fault_code_type_2_count_per_usage2\nLag = 3, Column = fault_code_type_3_count_per_usage2\nLag = 3, Column = fault_code_type_4_count_per_usage2\nLag = 7, Column = warn_type1_total\nLag = 7, Column = warn_type2_total\nLag = 7, Column = pca_1_warn\nLag = 7, Column = pca_2_warn\nLag = 7, Column = pca_3_warn\nLag = 7, Column = pca_4_warn\nLag = 7, Column = pca_5_warn\nLag = 7, Column = pca_6_warn\nLag = 7, Column = pca_7_warn\nLag = 7, Column = pca_8_warn\nLag = 7, Column = pca_9_warn\nLag = 7, Column = pca_10_warn\nLag = 7, Column = pca_11_warn\nLag = 7, Column = pca_12_warn\nLag = 7, Column = pca_13_warn\nLag = 7, Column = pca_14_warn\nLag = 7, Column = pca_15_warn\nLag = 7, Column = pca_16_warn\nLag = 7, Column = pca_17_warn\nLag = 7, Column = pca_18_warn\nLag = 7, Column = pca_19_warn\nLag = 7, Column = pca_20_warn\nLag = 7, Column = problem_type_1\nLag = 7, Column = problem_type_2\nLag = 7, Column = problem_type_3\nLag = 7, Column = problem_type_4\nLag = 7, Column = problem_type_1_per_usage1\nLag = 7, Column = problem_type_2_per_usage1\nLag = 7, Column = problem_type_3_per_usage1\nLag = 7, Column = problem_type_4_per_usage1\nLag = 7, Column = problem_type_1_per_usage2\nLag = 7, Column = problem_type_2_per_usage2\nLag = 7, Column = problem_type_3_per_usage2\nLag = 7, Column = problem_type_4_per_usage2\nLag = 7, Column = fault_code_type_1_count\nLag = 7, Column = fault_code_type_2_count\nLag = 7, Column = fault_code_type_3_count\nLag = 7, Column = fault_code_type_4_count\nLag = 7, Column = fault_code_type_1_count_per_usage1\nLag = 7, Column = fault_code_type_2_count_per_usage1\nLag = 7, Column = fault_code_type_3_count_per_usage1\nLag = 7, Column = fault_code_type_4_count_per_usage1\nLag = 7, Column = fault_code_type_1_count_per_usage2\nLag = 7, Column = fault_code_type_2_count_per_usage2\nLag = 7, Column = fault_code_type_3_count_per_usage2\nLag = 7, Column = fault_code_type_4_count_per_usage2\nLag = 14, Column = warn_type1_total\nLag = 14, Column = warn_type2_total\nLag = 14, Column = pca_1_warn\nLag = 14, Column = pca_2_warn\nLag = 14, Column = pca_3_warn\nLag = 14, Column = pca_4_warn\nLag = 14, Column = pca_5_warn\nLag = 14, Column = pca_6_warn\nLag = 14, Column = pca_7_warn\nLag = 14, Column = pca_8_warn\nLag = 14, Column = pca_9_warn\nLag = 14, Column = pca_10_warn\nLag = 14, Column = pca_11_warn\nLag = 14, Column = pca_12_warn\nLag = 14, Column = pca_13_warn\nLag = 14, Column = pca_14_warn\nLag = 14, Column = pca_15_warn\nLag = 14, Column = pca_16_warn\nLag = 14, Column = pca_17_warn\nLag = 14, Column = pca_18_warn\nLag = 14, Column = pca_19_warn\nLag = 14, Column = pca_20_warn\nLag = 14, Column = problem_type_1\nLag = 14, Column = problem_type_2\nLag = 14, Column = problem_type_3\nLag = 14, Column = problem_type_4\nLag = 14, Column = problem_type_1_per_usage1\nLag = 14, Column = problem_type_2_per_usage1\nLag = 14, Column = problem_type_3_per_usage1\nLag = 14, Column = problem_type_4_per_usage1\nLag = 14, Column = problem_type_1_per_usage2\nLag = 14, Column = problem_type_2_per_usage2\nLag = 14, Column = problem_type_3_per_usage2\nLag = 14, Column = problem_type_4_per_usage2\nLag = 14, Column = fault_code_type_1_count\nLag = 14, Column = fault_code_type_2_count\nLag = 14, Column = fault_code_type_3_count\nLag = 14, Column = fault_code_type_4_count\nLag = 14, Column = fault_code_type_1_count_per_usage1\nLag = 14, Column = fault_code_type_2_count_per_usage1\nLag = 14, Column = fault_code_type_3_count_per_usage1\nLag = 14, Column = fault_code_type_4_count_per_usage1\nLag = 14, Column = fault_code_type_1_count_per_usage2\nLag = 14, Column = fault_code_type_2_count_per_usage2\nLag = 14, Column = fault_code_type_3_count_per_usage2\nLag = 14, Column = fault_code_type_4_count_per_usage2\nLag = 30, Column = warn_type1_total\nLag = 30, Column = warn_type2_total\nLag = 30, Column = pca_1_warn\nLag = 30, Column = pca_2_warn\nLag = 30, Column = pca_3_warn\nLag = 30, Column = pca_4_warn\nLag = 30, Column = pca_5_warn\nLag = 30, Column = pca_6_warn\nLag = 30, Column = pca_7_warn\nLag = 30, Column = pca_8_warn\nLag = 30, Column = pca_9_warn\nLag = 30, Column = pca_10_warn\nLag = 30, Column = pca_11_warn\nLag = 30, Column = pca_12_warn\nLag = 30, Column = pca_13_warn\nLag = 30, Column = pca_14_warn\nLag = 30, Column = pca_15_warn\nLag = 30, Column = pca_16_warn\nLag = 30, Column = pca_17_warn\nLag = 30, Column = pca_18_warn\nLag = 30, Column = pca_19_warn\nLag = 30, Column = pca_20_warn\nLag = 30, Column = problem_type_1\nLag = 30, Column = problem_type_2\nLag = 30, Column = problem_type_3\nLag = 30, Column = problem_type_4\nLag = 30, Column = problem_type_1_per_usage1\nLag = 30, Column = problem_type_2_per_usage1\nLag = 30, Column = problem_type_3_per_usage1\nLag = 30, Column = problem_type_4_per_usage1\nLag = 30, Column = problem_type_1_per_usage2\nLag = 30, Column = problem_type_2_per_usage2\nLag = 30, Column = problem_type_3_per_usage2\nLag = 30, Column = problem_type_4_per_usage2\nLag = 30, Column = fault_code_type_1_count\nLag = 30, Column = fault_code_type_2_count\nLag = 30, Column = fault_code_type_3_count\nLag = 30, Column = fault_code_type_4_count\nLag = 30, Column = fault_code_type_1_count_per_usage1\nLag = 30, Column = fault_code_type_2_count_per_usage1\nLag = 30, Column = fault_code_type_3_count_per_usage1\nLag = 30, Column = fault_code_type_4_count_per_usage1\nLag = 30, Column = fault_code_type_1_count_per_usage2\nLag = 30, Column = fault_code_type_2_count_per_usage2\nLag = 30, Column = fault_code_type_3_count_per_usage2\nLag = 30, Column = fault_code_type_4_count_per_usage2\nLag = 90, Column = warn_type1_total\nLag = 90, Column = warn_type2_total\nLag = 90, Column = pca_1_warn\nLag = 90, Column = pca_2_warn\nLag = 90, Column = pca_3_warn\nLag = 90, Column = pca_4_warn\nLag = 90, Column = pca_5_warn\nLag = 90, Column = pca_6_warn\nLag = 90, Column = pca_7_warn\nLag = 90, Column = pca_8_warn\nLag = 90, Column = pca_9_warn\nLag = 90, Column = pca_10_warn\nLag = 90, Column = pca_11_warn\nLag = 90, Column = pca_12_warn\nLag = 90, Column = pca_13_warn\nLag = 90, Column = pca_14_warn\nLag = 90, Column = pca_15_warn\nLag = 90, Column = pca_16_warn\nLag = 90, Column = pca_17_warn\nLag = 90, Column = pca_18_warn\nLag = 90, Column = pca_19_warn\nLag = 90, Column = pca_20_warn\nLag = 90, Column = problem_type_1\nLag = 90, Column = problem_type_2\nLag = 90, Column = problem_type_3\nLag = 90, Column = problem_type_4\nLag = 90, Column = problem_type_1_per_usage1\nLag = 90, Column = problem_type_2_per_usage1\nLag = 90, Column = problem_type_3_per_usage1\nLag = 90, Column = problem_type_4_per_usage1\nLag = 90, Column = problem_type_1_per_usage2\nLag = 90, Column = problem_type_2_per_usage2\nLag = 90, Column = problem_type_3_per_usage2\nLag = 90, Column = problem_type_4_per_usage2\nLag = 90, Column = fault_code_type_1_count\nLag = 90, Column = fault_code_type_2_count\nLag = 90, Column = fault_code_type_3_count\nLag = 90, Column = fault_code_type_4_count\nLag = 90, Column = fault_code_type_1_count_per_usage1\nLag = 90, Column = fault_code_type_2_count_per_usage1\nLag = 90, Column = fault_code_type_3_count_per_usage1\nLag = 90, Column = fault_code_type_4_count_per_usage1\nLag = 90, Column = fault_code_type_1_count_per_usage2\nLag = 90, Column = fault_code_type_2_count_per_usage2\nLag = 90, Column = fault_code_type_3_count_per_usage2\nLag = 90, Column = fault_code_type_4_count_per_usage2\nCPU times: user 1.03 s, sys: 411 ms, total: 1.44 s\nWall time: 30min 16s\n" ] ], [ [ "### Rolling Max", "_____no_output_____" ] ], [ [ "%%time\n\n# Load result dataset from Notebook #1\ndf = sqlContext.read.parquet('/mnt/resource/PysparkExample/notebook1_result.parquet')\n\nfor lag_n in lags:\n wSpec = Window.partitionBy('deviceid').orderBy('date').rowsBetween(1-lag_n, 0)\n for col_name in rolling_features:\n df = df.withColumn(col_name+'_rollingmax_'+str(lag_n), F.max(col(col_name)).over(wSpec))\n print(\"Lag = %d, Column = %s\" % (lag_n, col_name))\n\nrollingmax = df.select(['key'] + list(s for s in df.columns if \"rollingmax\" in s))\n\n# Save the intermediate result for downstream work\nrollingmax.write.mode('overwrite').parquet('/mnt/resource/PysparkExample/rollingmax.parquet')\n", "Lag = 3, Column = warn_type1_total\nLag = 3, Column = warn_type2_total\nLag = 3, Column = pca_1_warn\nLag = 3, Column = pca_2_warn\nLag = 3, Column = pca_3_warn\nLag = 3, Column = pca_4_warn\nLag = 3, Column = pca_5_warn\nLag = 3, Column = pca_6_warn\nLag = 3, Column = pca_7_warn\nLag = 3, Column = pca_8_warn\nLag = 3, Column = pca_9_warn\nLag = 3, Column = pca_10_warn\nLag = 3, Column = pca_11_warn\nLag = 3, Column = pca_12_warn\nLag = 3, Column = pca_13_warn\nLag = 3, Column = pca_14_warn\nLag = 3, Column = pca_15_warn\nLag = 3, Column = pca_16_warn\nLag = 3, Column = pca_17_warn\nLag = 3, Column = pca_18_warn\nLag = 3, Column = pca_19_warn\nLag = 3, Column = pca_20_warn\nLag = 3, Column = problem_type_1\nLag = 3, Column = problem_type_2\nLag = 3, Column = problem_type_3\nLag = 3, Column = problem_type_4\nLag = 3, Column = problem_type_1_per_usage1\nLag = 3, Column = problem_type_2_per_usage1\nLag = 3, Column = problem_type_3_per_usage1\nLag = 3, Column = problem_type_4_per_usage1\nLag = 3, Column = problem_type_1_per_usage2\nLag = 3, Column = problem_type_2_per_usage2\nLag = 3, Column = problem_type_3_per_usage2\nLag = 3, Column = problem_type_4_per_usage2\nLag = 3, Column = fault_code_type_1_count\nLag = 3, Column = fault_code_type_2_count\nLag = 3, Column = fault_code_type_3_count\nLag = 3, Column = fault_code_type_4_count\nLag = 3, Column = fault_code_type_1_count_per_usage1\nLag = 3, Column = fault_code_type_2_count_per_usage1\nLag = 3, Column = fault_code_type_3_count_per_usage1\nLag = 3, Column = fault_code_type_4_count_per_usage1\nLag = 3, Column = fault_code_type_1_count_per_usage2\nLag = 3, Column = fault_code_type_2_count_per_usage2\nLag = 3, Column = fault_code_type_3_count_per_usage2\nLag = 3, Column = fault_code_type_4_count_per_usage2\nLag = 7, Column = warn_type1_total\nLag = 7, Column = warn_type2_total\nLag = 7, Column = pca_1_warn\nLag = 7, Column = pca_2_warn\nLag = 7, Column = pca_3_warn\nLag = 7, Column = pca_4_warn\nLag = 7, Column = pca_5_warn\nLag = 7, Column = pca_6_warn\nLag = 7, Column = pca_7_warn\nLag = 7, Column = pca_8_warn\nLag = 7, Column = pca_9_warn\nLag = 7, Column = pca_10_warn\nLag = 7, Column = pca_11_warn\nLag = 7, Column = pca_12_warn\nLag = 7, Column = pca_13_warn\nLag = 7, Column = pca_14_warn\nLag = 7, Column = pca_15_warn\nLag = 7, Column = pca_16_warn\nLag = 7, Column = pca_17_warn\nLag = 7, Column = pca_18_warn\nLag = 7, Column = pca_19_warn\nLag = 7, Column = pca_20_warn\nLag = 7, Column = problem_type_1\nLag = 7, Column = problem_type_2\nLag = 7, Column = problem_type_3\nLag = 7, Column = problem_type_4\nLag = 7, Column = problem_type_1_per_usage1\nLag = 7, Column = problem_type_2_per_usage1\nLag = 7, Column = problem_type_3_per_usage1\nLag = 7, Column = problem_type_4_per_usage1\nLag = 7, Column = problem_type_1_per_usage2\nLag = 7, Column = problem_type_2_per_usage2\nLag = 7, Column = problem_type_3_per_usage2\nLag = 7, Column = problem_type_4_per_usage2\nLag = 7, Column = fault_code_type_1_count\nLag = 7, Column = fault_code_type_2_count\nLag = 7, Column = fault_code_type_3_count\nLag = 7, Column = fault_code_type_4_count\nLag = 7, Column = fault_code_type_1_count_per_usage1\nLag = 7, Column = fault_code_type_2_count_per_usage1\nLag = 7, Column = fault_code_type_3_count_per_usage1\nLag = 7, Column = fault_code_type_4_count_per_usage1\nLag = 7, Column = fault_code_type_1_count_per_usage2\nLag = 7, Column = fault_code_type_2_count_per_usage2\nLag = 7, Column = fault_code_type_3_count_per_usage2\nLag = 7, Column = fault_code_type_4_count_per_usage2\nLag = 14, Column = warn_type1_total\nLag = 14, Column = warn_type2_total\nLag = 14, Column = pca_1_warn\nLag = 14, Column = pca_2_warn\nLag = 14, Column = pca_3_warn\nLag = 14, Column = pca_4_warn\nLag = 14, Column = pca_5_warn\nLag = 14, Column = pca_6_warn\nLag = 14, Column = pca_7_warn\nLag = 14, Column = pca_8_warn\nLag = 14, Column = pca_9_warn\nLag = 14, Column = pca_10_warn\nLag = 14, Column = pca_11_warn\nLag = 14, Column = pca_12_warn\nLag = 14, Column = pca_13_warn\nLag = 14, Column = pca_14_warn\nLag = 14, Column = pca_15_warn\nLag = 14, Column = pca_16_warn\nLag = 14, Column = pca_17_warn\nLag = 14, Column = pca_18_warn\nLag = 14, Column = pca_19_warn\nLag = 14, Column = pca_20_warn\nLag = 14, Column = problem_type_1\nLag = 14, Column = problem_type_2\nLag = 14, Column = problem_type_3\nLag = 14, Column = problem_type_4\nLag = 14, Column = problem_type_1_per_usage1\nLag = 14, Column = problem_type_2_per_usage1\nLag = 14, Column = problem_type_3_per_usage1\nLag = 14, Column = problem_type_4_per_usage1\nLag = 14, Column = problem_type_1_per_usage2\nLag = 14, Column = problem_type_2_per_usage2\nLag = 14, Column = problem_type_3_per_usage2\nLag = 14, Column = problem_type_4_per_usage2\nLag = 14, Column = fault_code_type_1_count\nLag = 14, Column = fault_code_type_2_count\nLag = 14, Column = fault_code_type_3_count\nLag = 14, Column = fault_code_type_4_count\nLag = 14, Column = fault_code_type_1_count_per_usage1\nLag = 14, Column = fault_code_type_2_count_per_usage1\nLag = 14, Column = fault_code_type_3_count_per_usage1\nLag = 14, Column = fault_code_type_4_count_per_usage1\nLag = 14, Column = fault_code_type_1_count_per_usage2\nLag = 14, Column = fault_code_type_2_count_per_usage2\nLag = 14, Column = fault_code_type_3_count_per_usage2\nLag = 14, Column = fault_code_type_4_count_per_usage2\nLag = 30, Column = warn_type1_total\nLag = 30, Column = warn_type2_total\nLag = 30, Column = pca_1_warn\nLag = 30, Column = pca_2_warn\nLag = 30, Column = pca_3_warn\nLag = 30, Column = pca_4_warn\nLag = 30, Column = pca_5_warn\nLag = 30, Column = pca_6_warn\nLag = 30, Column = pca_7_warn\nLag = 30, Column = pca_8_warn\nLag = 30, Column = pca_9_warn\nLag = 30, Column = pca_10_warn\nLag = 30, Column = pca_11_warn\nLag = 30, Column = pca_12_warn\nLag = 30, Column = pca_13_warn\nLag = 30, Column = pca_14_warn\nLag = 30, Column = pca_15_warn\nLag = 30, Column = pca_16_warn\nLag = 30, Column = pca_17_warn\nLag = 30, Column = pca_18_warn\nLag = 30, Column = pca_19_warn\nLag = 30, Column = pca_20_warn\nLag = 30, Column = problem_type_1\nLag = 30, Column = problem_type_2\nLag = 30, Column = problem_type_3\nLag = 30, Column = problem_type_4\nLag = 30, Column = problem_type_1_per_usage1\nLag = 30, Column = problem_type_2_per_usage1\nLag = 30, Column = problem_type_3_per_usage1\nLag = 30, Column = problem_type_4_per_usage1\nLag = 30, Column = problem_type_1_per_usage2\nLag = 30, Column = problem_type_2_per_usage2\nLag = 30, Column = problem_type_3_per_usage2\nLag = 30, Column = problem_type_4_per_usage2\nLag = 30, Column = fault_code_type_1_count\nLag = 30, Column = fault_code_type_2_count\nLag = 30, Column = fault_code_type_3_count\nLag = 30, Column = fault_code_type_4_count\nLag = 30, Column = fault_code_type_1_count_per_usage1\nLag = 30, Column = fault_code_type_2_count_per_usage1\nLag = 30, Column = fault_code_type_3_count_per_usage1\nLag = 30, Column = fault_code_type_4_count_per_usage1\nLag = 30, Column = fault_code_type_1_count_per_usage2\nLag = 30, Column = fault_code_type_2_count_per_usage2\nLag = 30, Column = fault_code_type_3_count_per_usage2\nLag = 30, Column = fault_code_type_4_count_per_usage2\nLag = 90, Column = warn_type1_total\nLag = 90, Column = warn_type2_total\nLag = 90, Column = pca_1_warn\nLag = 90, Column = pca_2_warn\nLag = 90, Column = pca_3_warn\nLag = 90, Column = pca_4_warn\nLag = 90, Column = pca_5_warn\nLag = 90, Column = pca_6_warn\nLag = 90, Column = pca_7_warn\nLag = 90, Column = pca_8_warn\nLag = 90, Column = pca_9_warn\nLag = 90, Column = pca_10_warn\nLag = 90, Column = pca_11_warn\nLag = 90, Column = pca_12_warn\nLag = 90, Column = pca_13_warn\nLag = 90, Column = pca_14_warn\nLag = 90, Column = pca_15_warn\nLag = 90, Column = pca_16_warn\nLag = 90, Column = pca_17_warn\nLag = 90, Column = pca_18_warn\nLag = 90, Column = pca_19_warn\nLag = 90, Column = pca_20_warn\nLag = 90, Column = problem_type_1\nLag = 90, Column = problem_type_2\nLag = 90, Column = problem_type_3\nLag = 90, Column = problem_type_4\nLag = 90, Column = problem_type_1_per_usage1\nLag = 90, Column = problem_type_2_per_usage1\nLag = 90, Column = problem_type_3_per_usage1\nLag = 90, Column = problem_type_4_per_usage1\nLag = 90, Column = problem_type_1_per_usage2\nLag = 90, Column = problem_type_2_per_usage2\nLag = 90, Column = problem_type_3_per_usage2\nLag = 90, Column = problem_type_4_per_usage2\nLag = 90, Column = fault_code_type_1_count\nLag = 90, Column = fault_code_type_2_count\nLag = 90, Column = fault_code_type_3_count\nLag = 90, Column = fault_code_type_4_count\nLag = 90, Column = fault_code_type_1_count_per_usage1\nLag = 90, Column = fault_code_type_2_count_per_usage1\nLag = 90, Column = fault_code_type_3_count_per_usage1\nLag = 90, Column = fault_code_type_4_count_per_usage1\nLag = 90, Column = fault_code_type_1_count_per_usage2\nLag = 90, Column = fault_code_type_2_count_per_usage2\nLag = 90, Column = fault_code_type_3_count_per_usage2\nLag = 90, Column = fault_code_type_4_count_per_usage2\nCPU times: user 860 ms, sys: 316 ms, total: 1.18 s\nWall time: 24min 41s\n" ] ], [ [ "### Rolling Min", "_____no_output_____" ] ], [ [ "%%time\n\n# Load result dataset from Notebook #1\ndf = sqlContext.read.parquet('/mnt/resource/PysparkExample/notebook1_result.parquet')\n\nfor lag_n in lags:\n wSpec = Window.partitionBy('deviceid').orderBy('date').rowsBetween(1-lag_n, 0)\n for col_name in rolling_features:\n df = df.withColumn(col_name+'_rollingmin_'+str(lag_n), F.min(col(col_name)).over(wSpec))\n print(\"Lag = %d, Column = %s\" % (lag_n, col_name))\n\nrollingmin = df.select(['key'] + list(s for s in df.columns if \"rollingmin\" in s))\n\n# Save the intermediate result for downstream work\nrollingmin.write.mode('overwrite').parquet('/mnt/resource/PysparkExample/rollingmin.parquet')\n", "Lag = 3, Column = warn_type1_total\nLag = 3, Column = warn_type2_total\nLag = 3, Column = pca_1_warn\nLag = 3, Column = pca_2_warn\nLag = 3, Column = pca_3_warn\nLag = 3, Column = pca_4_warn\nLag = 3, Column = pca_5_warn\nLag = 3, Column = pca_6_warn\nLag = 3, Column = pca_7_warn\nLag = 3, Column = pca_8_warn\nLag = 3, Column = pca_9_warn\nLag = 3, Column = pca_10_warn\nLag = 3, Column = pca_11_warn\nLag = 3, Column = pca_12_warn\nLag = 3, Column = pca_13_warn\nLag = 3, Column = pca_14_warn\nLag = 3, Column = pca_15_warn\nLag = 3, Column = pca_16_warn\nLag = 3, Column = pca_17_warn\nLag = 3, Column = pca_18_warn\nLag = 3, Column = pca_19_warn\nLag = 3, Column = pca_20_warn\nLag = 3, Column = problem_type_1\nLag = 3, Column = problem_type_2\nLag = 3, Column = problem_type_3\nLag = 3, Column = problem_type_4\nLag = 3, Column = problem_type_1_per_usage1\nLag = 3, Column = problem_type_2_per_usage1\nLag = 3, Column = problem_type_3_per_usage1\nLag = 3, Column = problem_type_4_per_usage1\nLag = 3, Column = problem_type_1_per_usage2\nLag = 3, Column = problem_type_2_per_usage2\nLag = 3, Column = problem_type_3_per_usage2\nLag = 3, Column = problem_type_4_per_usage2\nLag = 3, Column = fault_code_type_1_count\nLag = 3, Column = fault_code_type_2_count\nLag = 3, Column = fault_code_type_3_count\nLag = 3, Column = fault_code_type_4_count\nLag = 3, Column = fault_code_type_1_count_per_usage1\nLag = 3, Column = fault_code_type_2_count_per_usage1\nLag = 3, Column = fault_code_type_3_count_per_usage1\nLag = 3, Column = fault_code_type_4_count_per_usage1\nLag = 3, Column = fault_code_type_1_count_per_usage2\nLag = 3, Column = fault_code_type_2_count_per_usage2\nLag = 3, Column = fault_code_type_3_count_per_usage2\nLag = 3, Column = fault_code_type_4_count_per_usage2\nLag = 7, Column = warn_type1_total\nLag = 7, Column = warn_type2_total\nLag = 7, Column = pca_1_warn\nLag = 7, Column = pca_2_warn\nLag = 7, Column = pca_3_warn\nLag = 7, Column = pca_4_warn\nLag = 7, Column = pca_5_warn\nLag = 7, Column = pca_6_warn\nLag = 7, Column = pca_7_warn\nLag = 7, Column = pca_8_warn\nLag = 7, Column = pca_9_warn\nLag = 7, Column = pca_10_warn\nLag = 7, Column = pca_11_warn\nLag = 7, Column = pca_12_warn\nLag = 7, Column = pca_13_warn\nLag = 7, Column = pca_14_warn\nLag = 7, Column = pca_15_warn\nLag = 7, Column = pca_16_warn\nLag = 7, Column = pca_17_warn\nLag = 7, Column = pca_18_warn\nLag = 7, Column = pca_19_warn\nLag = 7, Column = pca_20_warn\nLag = 7, Column = problem_type_1\nLag = 7, Column = problem_type_2\nLag = 7, Column = problem_type_3\nLag = 7, Column = problem_type_4\nLag = 7, Column = problem_type_1_per_usage1\nLag = 7, Column = problem_type_2_per_usage1\nLag = 7, Column = problem_type_3_per_usage1\nLag = 7, Column = problem_type_4_per_usage1\nLag = 7, Column = problem_type_1_per_usage2\nLag = 7, Column = problem_type_2_per_usage2\nLag = 7, Column = problem_type_3_per_usage2\nLag = 7, Column = problem_type_4_per_usage2\nLag = 7, Column = fault_code_type_1_count\nLag = 7, Column = fault_code_type_2_count\nLag = 7, Column = fault_code_type_3_count\nLag = 7, Column = fault_code_type_4_count\nLag = 7, Column = fault_code_type_1_count_per_usage1\nLag = 7, Column = fault_code_type_2_count_per_usage1\nLag = 7, Column = fault_code_type_3_count_per_usage1\nLag = 7, Column = fault_code_type_4_count_per_usage1\nLag = 7, Column = fault_code_type_1_count_per_usage2\nLag = 7, Column = fault_code_type_2_count_per_usage2\nLag = 7, Column = fault_code_type_3_count_per_usage2\nLag = 7, Column = fault_code_type_4_count_per_usage2\nLag = 14, Column = warn_type1_total\nLag = 14, Column = warn_type2_total\nLag = 14, Column = pca_1_warn\nLag = 14, Column = pca_2_warn\nLag = 14, Column = pca_3_warn\nLag = 14, Column = pca_4_warn\nLag = 14, Column = pca_5_warn\nLag = 14, Column = pca_6_warn\nLag = 14, Column = pca_7_warn\nLag = 14, Column = pca_8_warn\nLag = 14, Column = pca_9_warn\nLag = 14, Column = pca_10_warn\nLag = 14, Column = pca_11_warn\nLag = 14, Column = pca_12_warn\nLag = 14, Column = pca_13_warn\nLag = 14, Column = pca_14_warn\nLag = 14, Column = pca_15_warn\nLag = 14, Column = pca_16_warn\nLag = 14, Column = pca_17_warn\nLag = 14, Column = pca_18_warn\nLag = 14, Column = pca_19_warn\nLag = 14, Column = pca_20_warn\nLag = 14, Column = problem_type_1\nLag = 14, Column = problem_type_2\nLag = 14, Column = problem_type_3\nLag = 14, Column = problem_type_4\nLag = 14, Column = problem_type_1_per_usage1\nLag = 14, Column = problem_type_2_per_usage1\nLag = 14, Column = problem_type_3_per_usage1\nLag = 14, Column = problem_type_4_per_usage1\nLag = 14, Column = problem_type_1_per_usage2\nLag = 14, Column = problem_type_2_per_usage2\nLag = 14, Column = problem_type_3_per_usage2\nLag = 14, Column = problem_type_4_per_usage2\nLag = 14, Column = fault_code_type_1_count\nLag = 14, Column = fault_code_type_2_count\nLag = 14, Column = fault_code_type_3_count\nLag = 14, Column = fault_code_type_4_count\nLag = 14, Column = fault_code_type_1_count_per_usage1\nLag = 14, Column = fault_code_type_2_count_per_usage1\nLag = 14, Column = fault_code_type_3_count_per_usage1\nLag = 14, Column = fault_code_type_4_count_per_usage1\nLag = 14, Column = fault_code_type_1_count_per_usage2\nLag = 14, Column = fault_code_type_2_count_per_usage2\nLag = 14, Column = fault_code_type_3_count_per_usage2\nLag = 14, Column = fault_code_type_4_count_per_usage2\nLag = 30, Column = warn_type1_total\nLag = 30, Column = warn_type2_total\nLag = 30, Column = pca_1_warn\nLag = 30, Column = pca_2_warn\nLag = 30, Column = pca_3_warn\nLag = 30, Column = pca_4_warn\nLag = 30, Column = pca_5_warn\nLag = 30, Column = pca_6_warn\nLag = 30, Column = pca_7_warn\nLag = 30, Column = pca_8_warn\nLag = 30, Column = pca_9_warn\nLag = 30, Column = pca_10_warn\nLag = 30, Column = pca_11_warn\nLag = 30, Column = pca_12_warn\nLag = 30, Column = pca_13_warn\nLag = 30, Column = pca_14_warn\nLag = 30, Column = pca_15_warn\nLag = 30, Column = pca_16_warn\nLag = 30, Column = pca_17_warn\nLag = 30, Column = pca_18_warn\nLag = 30, Column = pca_19_warn\nLag = 30, Column = pca_20_warn\nLag = 30, Column = problem_type_1\nLag = 30, Column = problem_type_2\nLag = 30, Column = problem_type_3\nLag = 30, Column = problem_type_4\nLag = 30, Column = problem_type_1_per_usage1\nLag = 30, Column = problem_type_2_per_usage1\nLag = 30, Column = problem_type_3_per_usage1\nLag = 30, Column = problem_type_4_per_usage1\nLag = 30, Column = problem_type_1_per_usage2\nLag = 30, Column = problem_type_2_per_usage2\nLag = 30, Column = problem_type_3_per_usage2\nLag = 30, Column = problem_type_4_per_usage2\nLag = 30, Column = fault_code_type_1_count\nLag = 30, Column = fault_code_type_2_count\nLag = 30, Column = fault_code_type_3_count\nLag = 30, Column = fault_code_type_4_count\nLag = 30, Column = fault_code_type_1_count_per_usage1\nLag = 30, Column = fault_code_type_2_count_per_usage1\nLag = 30, Column = fault_code_type_3_count_per_usage1\nLag = 30, Column = fault_code_type_4_count_per_usage1\nLag = 30, Column = fault_code_type_1_count_per_usage2\nLag = 30, Column = fault_code_type_2_count_per_usage2\nLag = 30, Column = fault_code_type_3_count_per_usage2\nLag = 30, Column = fault_code_type_4_count_per_usage2\nLag = 90, Column = warn_type1_total\nLag = 90, Column = warn_type2_total\nLag = 90, Column = pca_1_warn\nLag = 90, Column = pca_2_warn\nLag = 90, Column = pca_3_warn\nLag = 90, Column = pca_4_warn\nLag = 90, Column = pca_5_warn\nLag = 90, Column = pca_6_warn\nLag = 90, Column = pca_7_warn\nLag = 90, Column = pca_8_warn\nLag = 90, Column = pca_9_warn\nLag = 90, Column = pca_10_warn\nLag = 90, Column = pca_11_warn\nLag = 90, Column = pca_12_warn\nLag = 90, Column = pca_13_warn\nLag = 90, Column = pca_14_warn\nLag = 90, Column = pca_15_warn\nLag = 90, Column = pca_16_warn\nLag = 90, Column = pca_17_warn\nLag = 90, Column = pca_18_warn\nLag = 90, Column = pca_19_warn\nLag = 90, Column = pca_20_warn\nLag = 90, Column = problem_type_1\nLag = 90, Column = problem_type_2\nLag = 90, Column = problem_type_3\nLag = 90, Column = problem_type_4\nLag = 90, Column = problem_type_1_per_usage1\nLag = 90, Column = problem_type_2_per_usage1\nLag = 90, Column = problem_type_3_per_usage1\nLag = 90, Column = problem_type_4_per_usage1\nLag = 90, Column = problem_type_1_per_usage2\nLag = 90, Column = problem_type_2_per_usage2\nLag = 90, Column = problem_type_3_per_usage2\nLag = 90, Column = problem_type_4_per_usage2\nLag = 90, Column = fault_code_type_1_count\nLag = 90, Column = fault_code_type_2_count\nLag = 90, Column = fault_code_type_3_count\nLag = 90, Column = fault_code_type_4_count\nLag = 90, Column = fault_code_type_1_count_per_usage1\nLag = 90, Column = fault_code_type_2_count_per_usage1\nLag = 90, Column = fault_code_type_3_count_per_usage1\nLag = 90, Column = fault_code_type_4_count_per_usage1\nLag = 90, Column = fault_code_type_1_count_per_usage2\nLag = 90, Column = fault_code_type_2_count_per_usage2\nLag = 90, Column = fault_code_type_3_count_per_usage2\nLag = 90, Column = fault_code_type_4_count_per_usage2\nCPU times: user 870 ms, sys: 306 ms, total: 1.18 s\nWall time: 23min 27s\n" ] ], [ [ "## Join result dataset from the five rolling compute cells:\n- Join in Spark is usually very slow, it is better to reduce the number of partitions before the join.\n- Check the number of partitions of the pyspark dataframe.\n- **repartition vs coalesce**. If we only want to reduce the number of partitions, it is better to use coalesce because repartition involves reshuffling which is computational more expensive and takes more time.\n<br>\n", "_____no_output_____" ] ], [ [ "# Import result dataset \nrollingmean = sqlContext.read.parquet('/mnt/resource/PysparkExample/data_rollingmean.parquet')\nrollingdiff = sqlContext.read.parquet('/mnt/resource/PysparkExample/rollingdiff.parquet')\nrollingstd = sqlContext.read.parquet('/mnt/resource/PysparkExample/rollingstd.parquet')\nrollingmax = sqlContext.read.parquet('/mnt/resource/PysparkExample/rollingmax.parquet')\nrollingmin = sqlContext.read.parquet('/mnt/resource/PysparkExample/rollingmin.parquet')\n\n# Check the number of partitions for each dataset\nprint(rollingmean.rdd.getNumPartitions())\nprint(rollingdiff.rdd.getNumPartitions())\nprint(rollingstd.rdd.getNumPartitions())\nprint(rollingmax.rdd.getNumPartitions())\nprint(rollingmin.rdd.getNumPartitions())\n", "33\n33\n33\n31\n31\n" ], [ "%%time\n\n# To make join faster, reduce the number of partitions (not necessarily to \"1\")\nrollingmean = rollingmean.coalesce(1)\nrollingdiff = rollingdiff.coalesce(1)\nrollingstd = rollingstd.coalesce(1)\nrollingmax = rollingmax.coalesce(1)\nrollingmin = rollingmin.coalesce(1)\n\nrolling_result = rollingmean.join(rollingdiff, 'key', 'inner')\\\n .join(rollingstd, 'key', 'inner')\\\n .join(rollingmax, 'key', 'inner')\\\n .join(rollingmin, 'key', 'inner')\n \n\n## Write the final result as parquet file for downstream work in Notebook_3\nrolling_result.write.mode('overwrite').parquet('/mnt/resource/PysparkExample/notebook2_result.parquet')\n", "CPU times: user 901 ms, sys: 303 ms, total: 1.2 s\nWall time: 1h 50min 38s\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Are Graphs Unique?\nThis notebook shows how to determine if each entry in the HydroNet dataset represents a unique graph. ", "_____no_output_____" ] ], [ [ "%matplotlib inline\nfrom matplotlib import pyplot as plt\nfrom hydronet.data import graph_from_dict\nfrom multiprocessing import Pool\nfrom functools import partial\nfrom tqdm import tqdm\nimport networkx as nx\nimport pandas as pd\nimport numpy as np", "_____no_output_____" ] ], [ [ "Configuration", "_____no_output_____" ] ], [ [ "cluster_size = 18", "_____no_output_____" ] ], [ [ "## Load in the Data\nLoad in a small dataset from disk", "_____no_output_____" ] ], [ [ "%%time\ndata = pd.read_json('../data/output/atomic_valid.json.gz', lines=True)\nprint(f'Loaded {len(data)} records')", "Loaded 224018 records\nCPU times: user 33.1 s, sys: 5.7 s, total: 38.8 s\nWall time: 38.8 s\n" ] ], [ [ "## Find Pairs of Isomorphic Graphs\nAssess how many training records are isomorphic", "_____no_output_____" ] ], [ [ "data.query(f'n_waters=={cluster_size}', inplace=True)\nprint(f'Downselected to {len(data)} graphs')", "Downselected to 5714 graphs\n" ] ], [ [ "Generate networkx objects for each", "_____no_output_____" ] ], [ [ "%%time\ndata['nx'] = data.apply(graph_from_dict, axis=1)", "CPU times: user 1.66 s, sys: 170 ms, total: 1.83 s\nWall time: 1.83 s\n" ] ], [ [ "Compute which graphs are isomorphic", "_____no_output_____" ] ], [ [ "data.reset_index(inplace=True)", "_____no_output_____" ], [ "matches = [[] for _ in range(len(data))]\nn_matches = 0\nwith Pool() as p:\n for i, g in tqdm(enumerate(data['nx']), total=len(data)):\n f = partial(nx.algorithms.is_isomorphic, g, node_match=dict.__eq__, edge_match=dict.__eq__)\n is_match = p.map(f, data['nx'].iloc[i+1:])\n for j, hit in enumerate(is_match):\n if hit:\n n_matches += 1\n j_real = i + j + 1\n matches[i].append(j_real)\n matches[j_real].append(i)\nprint(f'Found {n_matches} pairs of isomorphic graphs')", "100%|██████████| 5714/5714 [26:46<00:00, 3.56it/s] \n" ] ], [ [ "Add to the dataframe for safe keeping", "_____no_output_____" ] ], [ [ "data['matches'] = matches", "_____no_output_____" ], [ "data['n_matches'] = data['matches'].apply(len)", "_____no_output_____" ] ], [ [ "## Assess Energy Differences between Isomorphic Graphs\nWe want to know how large they are. Does each graph represent a local minimum, or they actually very different in energy?", "_____no_output_____" ] ], [ [ "energy_diffs = []\nfor rid, row in data.query('n_matches>0').iterrows():\n for m in row['matches']:\n if m > rid:\n energy_diffs.append(abs(row['energy'] - data.iloc[m]['energy']))", "_____no_output_____" ], [ "print(f'Maximum: {np.max(energy_diffs):.2e} kcal/mol')\nprint(f'Median: {np.percentile(energy_diffs, 50):.2e} kcal/mol')\nprint(f'Minimum: {np.min(energy_diffs):.2e} kcal/mol')", "Maximum: 1.50e-01 kcal/mol\nMedian: 5.48e-02 kcal/mol\nMinimum: 1.04e-02 kcal/mol\n" ], [ "fig, ax = plt.subplots(figsize=(3.5, 2.5))\n\nbins = np.logspace(-4, 1, 32)\nax.hist(energy_diffs, bins=bins)\nax.set_xscale('log')\n\nax.set_xlabel('$\\Delta E$ (kcal/mol)')\nax.set_ylabel('Frequency')\nfig.tight_layout()\nfig.savefig(f'figures/energy-difference-isomorphic-graphs-size-{cluster_size}.png', dpi=320)", "_____no_output_____" ] ], [ [ "For comparision, print out the range of energies for clusters", "_____no_output_____" ] ], [ [ "(data['energy'] - data['energy'].min()).describe()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Interactive question answering with OpenVINO\n\nThis demo shows interactive question answering with OpenVINO. We use [small BERT-large-like model](https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/bert-small-uncased-whole-word-masking-squad-int8-0002) distilled and quantized to INT8 on SQuAD v1.1 training set from larger BERT-large model. The model comes from [Open Model Zoo](https://github.com/openvinotoolkit/open_model_zoo/). At the bottom of this notebook, you will see live inference results from your inputs.", "_____no_output_____" ], [ "## Imports", "_____no_output_____" ] ], [ [ "import time\nfrom urllib import parse\n\nimport numpy as np\nfrom openvino.runtime import Core, Dimension\n\nimport html_reader as reader\nimport tokens_bert as tokens", "_____no_output_____" ] ], [ [ "## The model\n\n### Download the model\n\nWe use `omz_downloader`, which is a command-line tool from the `openvino-dev` package. `omz_downloader` automatically creates a directory structure and downloads the selected model. If the model is already downloaded, this step is skipped.\n\nYou can download and use any of the following models: `bert-large-uncased-whole-word-masking-squad-0001`, `bert-large-uncased-whole-word-masking-squad-int8-0001`, `bert-small-uncased-whole-word-masking-squad-0001`, `bert-small-uncased-whole-word-masking-squad-0002`, `bert-small-uncased-whole-word-masking-squad-int8-0002`, just change the model name below. Any of these models are already converted to OpenVINO Intermediate Representation (IR), so there is no need to use `omz_converter`.", "_____no_output_____" ] ], [ [ "# directory where model will be downloaded\nbase_model_dir = \"model\"\n\n# desired precision\nprecision = \"FP16-INT8\"\n\n# model name as named in Open Model Zoo\nmodel_name = \"bert-small-uncased-whole-word-masking-squad-int8-0002\"\n\nmodel_path = f\"model/intel/{model_name}/{precision}/{model_name}.xml\"\nmodel_weights_path = f\"model/intel/{model_name}/{precision}/{model_name}.bin\"\n\ndownload_command = f\"omz_downloader \" \\\n f\"--name {model_name} \" \\\n f\"--precision {precision} \" \\\n f\"--output_dir {base_model_dir} \" \\\n f\"--cache_dir {base_model_dir}\"\n! $download_command", "_____no_output_____" ] ], [ [ "### Load the model\n\nDownloaded models are located in a fixed structure, which indicates vendor, model name and precision. Only a few lines of code are required to run the model. First, we create an Inference Engine object. Then we read the network architecture and model weights from the .xml and .bin files. Finally, we compile the network for the desired device. Because of using dynamic shapes we can run code only on `CPU`.", "_____no_output_____" ] ], [ [ "# initialize inference engine\nie_core = Core()\n\n# read the model and corresponding weights from file\nmodel = ie_core.read_model(model=model_path, weights=model_weights_path)\n\n# assign dynamic shapes to every input layer\nfor input_layer in model.inputs:\n input_shape = input_layer.partial_shape\n input_shape[1] = Dimension()\n model.reshape({input_layer: input_shape})\n\n# compile the model for the CPU\ncompiled_model = ie_core.compile_model(model=model, device_name=\"CPU\")\n\n# get input and output names of nodes\ninput_keys = list(compiled_model.inputs)\noutput_keys = list(compiled_model.outputs)", "_____no_output_____" ] ], [ [ "Input keys are the names of the input nodes and output keys contain names of output nodes of the network. In the case of the BERT-large-like model, we have four inputs and two outputs.", "_____no_output_____" ] ], [ [ "[i.any_name for i in input_keys], [o.any_name for o in output_keys]", "_____no_output_____" ] ], [ [ "## Processing\n\nNLP models usually take a list of tokens as standard input. A token is a single word converted to some integer. To provide the proper input, we need the vocabulary for such mapping. We also define some special tokens like separators and a function to load the content from provided URLs.", "_____no_output_____" ] ], [ [ "# path to vocabulary file\nvocab_file_path = \"data/vocab.txt\"\n\n# create dictionary with words and their indices\nvocab = tokens.load_vocab_file(vocab_file_path)\n\n# define special tokens\ncls_token = vocab[\"[CLS]\"]\nsep_token = vocab[\"[SEP]\"]\n\n\n# function to load text from given urls\ndef load_context(sources):\n input_urls = []\n paragraphs = []\n for source in sources:\n result = parse.urlparse(source)\n if all([result.scheme, result.netloc]):\n input_urls.append(source)\n else:\n paragraphs.append(source)\n\n paragraphs.extend(reader.get_paragraphs(input_urls))\n # produce one big context string\n return \"\\n\".join(paragraphs)", "_____no_output_____" ] ], [ [ "\n### Preprocessing\n\nThe main input (`input_ids`) to used BERT model consist of two parts: question tokens and context tokens separated by some special tokens. We also need to provide: `attention_mask`, which is a sequence of integer values representing the mask of valid values in the input; `token_type_ids`, which is a sequence of integer values representing the segmentation of the `input_ids` into question and context; `position_ids`, which is a sequence of integer values from 0 to length of input, extended by separation tokens, representing the position index for each input token. To know more about input, please read [this](https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/bert-small-uncased-whole-word-masking-squad-int8-0002#input).", "_____no_output_____" ] ], [ [ "# generator of a sequence of inputs\ndef prepare_input(question_tokens, context_tokens, input_keys):\n input_ids = [cls_token] + question_tokens + [sep_token] + context_tokens + [sep_token]\n # 1 for any index\n attention_mask = [1] * len(input_ids)\n # 0 for question tokens, 1 for context part\n token_type_ids = [0] * (len(question_tokens) + 2) + [1] * (len(context_tokens) + 1)\n\n # create input to feed the model\n input_dict = {\n \"input_ids\": np.array([input_ids], dtype=np.int32),\n \"attention_mask\": np.array([attention_mask], dtype=np.int32),\n \"token_type_ids\": np.array([token_type_ids], dtype=np.int32),\n }\n\n # some models require additional position_ids\n if \"position_ids\" in [i_key.any_name for i_key in input_keys]:\n position_ids = np.arange(len(input_ids))\n input_dict[\"position_ids\"] = np.array([position_ids], dtype=np.int32)\n\n return input_dict", "_____no_output_____" ] ], [ [ "### Postprocessing\n\nThe results from the network are raw (logits). We need to use the softmax function to get the probability distribution. Then, we are looking for the best answer in the current part of the context (the highest score) and we return the score and the context range for the answer.", "_____no_output_____" ] ], [ [ "# based on https://github.com/openvinotoolkit/open_model_zoo/blob/bf03f505a650bafe8da03d2747a8b55c5cb2ef16/demos/common/python/openvino/model_zoo/model_api/models/bert.py#L163\ndef postprocess(output_start, output_end, question_tokens, context_tokens_start_end, input_size):\n\n def get_score(logits):\n out = np.exp(logits)\n return out / out.sum(axis=-1)\n\n # get start-end scores for context\n score_start = get_score(output_start)\n score_end = get_score(output_end)\n\n # index of first context token in tensor\n context_start_idx = len(question_tokens) + 2\n # index of last+1 context token in tensor\n context_end_idx = input_size - 1\n\n # find product of all start-end combinations to find the best one\n max_score, max_start, max_end = find_best_answer_window(start_score=score_start,\n end_score=score_end,\n context_start_idx=context_start_idx,\n context_end_idx=context_end_idx)\n\n # convert to context text start-end index\n max_start = context_tokens_start_end[max_start][0]\n max_end = context_tokens_start_end[max_end][1]\n\n return max_score, max_start, max_end\n\n\n# based on https://github.com/openvinotoolkit/open_model_zoo/blob/bf03f505a650bafe8da03d2747a8b55c5cb2ef16/demos/common/python/openvino/model_zoo/model_api/models/bert.py#L188\ndef find_best_answer_window(start_score, end_score, context_start_idx, context_end_idx):\n context_len = context_end_idx - context_start_idx\n\n score_mat = np.matmul(\n start_score[context_start_idx:context_end_idx].reshape((context_len, 1)),\n end_score[context_start_idx:context_end_idx].reshape((1, context_len)),\n )\n \n # reset candidates with end before start\n score_mat = np.triu(score_mat)\n # reset long candidates (>16 words)\n score_mat = np.tril(score_mat, 16)\n # find the best start-end pair\n max_s, max_e = divmod(score_mat.flatten().argmax(), score_mat.shape[1])\n max_score = score_mat[max_s, max_e]\n\n return max_score, max_s, max_e", "_____no_output_____" ] ], [ [ " Firstly, we need to create a list of tokens from the context and the question. Then, we are looking for the best answer in the context. The best answer should come with the highest score.", "_____no_output_____" ] ], [ [ "def get_best_answer(question, context, vocab, input_keys):\n # convert context string to tokens\n context_tokens, context_tokens_start_end = tokens.text_to_tokens(text=context.lower(),\n vocab=vocab)\n # convert question string to tokens\n question_tokens, _ = tokens.text_to_tokens(text=question.lower(), vocab=vocab)\n\n network_input = prepare_input(question_tokens, context_tokens, input_keys)\n input_size = len(context_tokens) + len(question_tokens) + 3\n\n # openvino inference\n request = compiled_model.create_infer_request()\n request.infer(inputs=network_input)\n\n # postprocess the result getting the score and context range for the answer\n score_start_end = postprocess(output_start=request.get_tensor(name=\"output_s\").data[0],\n output_end=request.get_tensor(name=\"output_e\").data[0],\n question_tokens=question_tokens,\n context_tokens_start_end=context_tokens_start_end,\n input_size=input_size)\n\n # return the part of the context, which is already an answer\n return context[score_start_end[1]:score_start_end[2]], score_start_end[0]", "_____no_output_____" ] ], [ [ "### Main Processing Function\n\nRun question answering on specific knowledge base and iterate through the questions.", "_____no_output_____" ] ], [ [ "def run_question_answering(sources):\n print(f\"Context: {sources}\", flush=True)\n context = load_context(sources)\n\n if len(context) == 0:\n print(\"Error: Empty context or outside paragraphs\")\n return\n\n while True:\n question = input()\n # if no question - break\n if question == \"\":\n break\n\n # measure processing time\n start_time = time.perf_counter()\n answer, score = get_best_answer(question=question, context=context, vocab=vocab, input_keys=input_keys)\n end_time = time.perf_counter()\n\n print(f\"Question: {question}\")\n print(f\"Answer: {answer}\")\n print(f\"Score: {score:.2f}\")\n print(f\"Time: {end_time - start_time:.2f}s\")", "_____no_output_____" ] ], [ [ "## Run\n\n### Run on local paragraphs\n\nChange sources to your own to answer your questions. You can use as many sources as you want. Usually, you need to wait a few seconds for the answer, but the longer context the longer the waiting time. The model is very limited and sensitive for the input. The answer can depend on whether there is a question mark at the end. The model will try to answer any of your questions even there is no good answer in the context, so in that case, you can see random results.\n\nSample source: Computational complexity theory paragraph (from [here](https://rajpurkar.github.io/SQuAD-explorer/explore/v2.0/dev/Computational_complexity_theory.html))\n\nSample questions:\n- What is the term for a task that generally lends itself to being solved by a computer?\n- By what main attribute are computational problems classified utilizing computational complexity theory?\n- What branch of theoretical computer science deals with broadly classifying computational problems by difficulty and class of relationship?\n\nIf you want to stop the processing just put an empty string.\n\n*Note: Firstly, run the code below and then put your questions in the box.*", "_____no_output_____" ] ], [ [ "sources = [\"Computational complexity theory is a branch of the theory of computation in theoretical computer \"\n \"science that focuses on classifying computational problems according to their inherent difficulty, \"\n \"and relating those classes to each other. A computational problem is understood to be a task that \"\n \"is in principle amenable to being solved by a computer, which is equivalent to stating that the \"\n \"problem may be solved by mechanical application of mathematical steps, such as an algorithm.\"]\n\nrun_question_answering(sources)", "_____no_output_____" ] ], [ [ "### Run on websites\n\nYou can also provide urls. Note that the context (knowledge base) is built from website paragraphs. If some information is outside the paragraphs, the algorithm won't able to find it.\n\nSample source: [OpenVINO wiki](https://en.wikipedia.org/wiki/OpenVINO)\n\nSample questions:\n- What does OpenVINO mean?\n- What is the license for OpenVINO?\n- Where can you deploy OpenVINO code?\n\nIf you want to stop the processing just put an empty string.\n\n*Note: Firstly, run the code below and then put your questions in the box.*", "_____no_output_____" ] ], [ [ "sources = [\"https://en.wikipedia.org/wiki/OpenVINO\"]\n\nrun_question_answering(sources)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Laboratorio 10", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\nimport numpy as np\nimport pandas as pd\nfrom sklearn.datasets import load_breast_cancer\nfrom sklearn.model_selection import train_test_split, GridSearchCV\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom sklearn.metrics import plot_confusion_matrix, classification_report, accuracy_score, recall_score, f1_score\n\n%matplotlib inline", "_____no_output_____" ], [ "breast_cancer = load_breast_cancer()\nX, y = breast_cancer.data, breast_cancer.target\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)\ntarget_names = breast_cancer.target_names", "_____no_output_____" ] ], [ [ "## Ejercicio 1\n\n(1 pto.)\n\nAjusta una regresión logística a los datos de entrenamiento y obtén el _accuracy_ con los datos de test. Utiliza el argumento `n_jobs` igual a $-1$, si aún así no converge aumenta el valor de `max_iter`.\n\nHint: Recuerda que el _accuracy_ es el _score_ por defecto en los modelos de clasificación de scikit-learn.", "_____no_output_____" ] ], [ [ "# No hay que hace validacion cruzada pq la regresion logistica no tiene hiperparametros\nlr = LogisticRegression(max_iter=2100, n_jobs=-1) # ya con 2100 iteraciones aproximadamente converge aproximadamente al mismo valor que con mas iteraciones\nlr.fit(X_train, y_train)\ny_pred = lr.predict(X_test)\nprint(f\"Logistic Regression accuracy: {accuracy_score(y_test, y_pred):0.2f}\") #lr.score(X_test, y_test)", "Logistic Regression accuracy: 0.98\n" ] ], [ [ "## Ejercicio 2\n\n(1 pto.)\n\nUtiliza `GridSearchCV` con 5 _folds_ para encontrar el mejor valor de `n_neighbors` de un modelo KNN.", "_____no_output_____" ] ], [ [ "knn =KNeighborsClassifier() # Defino el modelo de KNN\nknn_grid = {\"n_neighbors\": np.arange(2, 31)} # Defino una grilla\n\nknn_cv = GridSearchCV( # Hago validacion cruzada\n KNeighborsClassifier(), # modelo a hiperparametrizar\n param_grid=knn_grid, #defino la grilla a utilizar\n cv=5, # Defino los 5 fold\n n_jobs=-1 # Uso todos los nucleos\n)\nknn_cv.fit(X_train, y_train) # Entrenarla\nknn_cv.best_params_\ny_pred1 = knn_cv.predict(X_test) ", "_____no_output_____" ], [ "#knn_cv.best_score_ \n#knn_cv.best_estimator_\nknn_cv.best_params_", "_____no_output_____" ], [ "print(f\"KNN accuracy: {accuracy_score(y_test, y_pred1):0.2f}\") #Imorimir el acurracy del moejor modelo de knn con el mejor n_neighbors", "KNN accuracy: 0.96\n" ] ], [ [ "## Ejercicio 3\n\n(1 pto.)\n\n¿Cuál modelo escogerías basándote en los resultados anteriores? Justifica", "_____no_output_____" ], [ "__Respuesta:__ El mejor modelo de los dos utilizados es el de regresion logistica ya que tiene un accuracy de 0.98 con 2100 iteraciones en comparacion con el de KNN que tiene un accuracy de 0.96 utilizando 5 fold.", "_____no_output_____" ], [ "## Ejercicio 4\n\n(1 pto.)\n\nPara el modelo seleccionado en el ejercicio anterior.\n\n* Grafica la matriz de confusión (no olvides colocar los nombres originales en los _labels_).\n* Imprime el reporte de clasificación.", "_____no_output_____" ] ], [ [ "# Me quedo con el modelo de regresion logistica \nplot_confusion_matrix(lr, X_test, y_test, display_labels=target_names) # en vez de tirar 0 y 1 te salagan valores reales malignt bening\nplt.show()", "_____no_output_____" ], [ "# Uso el y_pred que definí en el modelo de regresion logistica\nprint(classification_report(y_test, y_pred, target_names=breast_cancer.target_names)) ", " precision recall f1-score support\n\n malignant 0.97 0.97 0.97 63\n benign 0.98 0.98 0.98 108\n\n accuracy 0.98 171\n macro avg 0.97 0.97 0.97 171\nweighted avg 0.98 0.98 0.98 171\n\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Chapter 11 - Machine Learning\n\nWoo! #machinelearning #ml #AI", "_____no_output_____" ], [ "Data science is a lot of reformatting of business problems into data problems, and then collecting, cleaning, formatting, and restructuring data. ML is almost an afterthought. But it is an essential afterthought. Intro to this chapter is a good one. ", "_____no_output_____" ] ], [ [ "# supervised models \n# need to create some learnin data\n\ndef split_data(data, prob):\n results = [], []\n for row in data:\n results[0 if random.random() < prob else 1].append(row)\n return results\n\ndef train_test_split(x, y, test_pct):\n data = zip(x,y)\n train, test = split_data(data, 1 - test_pct)\n x_train, y_train = zip(*train)\n x_test, y_test = zip(*test)\n return x_train, x_test, y_train, y_test\n\n# then you can create your little model\nmodel = SomeKindOfModel()\nx_train, x_test, y_train, y_test = train_test_split(xs, ys, 0.33)\nmodel.train(x_train, y_train)\n\nperformance = model.test(x_test, y_test)\n\n", "_____no_output_____" ] ], [ [ "Models aren't necessarily graded on accuracy. If we said every person named Luke will not develop leukemia, we'd be right 98% of the time.\n\nSee the book (or a google search) for the confusion matrix, which describes true positives, true negatives, false positives (Type I error), and false negatives (Type 2 error)", "_____no_output_____" ] ], [ [ "def accuracy(tp, fp, fn, tn):\n correct = tp + tn\n total = tp+tn+fp+fn\n return correct / total\n\nprint(accuracy(70, 4930, 13930, 981070))", "0.98114\n" ], [ "# precision is accuracy of positive predictions\ndef precision(tp, fp, fn, tn):\n return tp / (tp+fp)\nprint(precision(70,4930,13930,981070))\n\n# recall is the fraction of the posinives identified\ndef recall(tp, fp, fn, tn):\n return tp / (tp + fn)\nprint(recall(70,4930,13930,981070))\n\n# both terrible = terrible model", "0.014\n0.005\n" ], [ "# sometimes these are combined into an f1 score\n\ndef f1_score(tp, fp, fn, tn):\n p = precision(tp, fp, fn, tn)\n r = recall(tp, fp, fn, tn)\n \n return 2 * p * r / (p+r)\n\nprint(f1_score(70,4930,13930,981070))", "0.00736842105263158\n" ] ], [ [ "## Bias vs Variance Trades-off\n\nFamously. Worth reading the chapter here to the end. The next chapters go into a bunch of different machine learing models we can build. I love this book so far and I like where it is headed.", "_____no_output_____" ], [ "## This concludes Chapter 10. \n\nA myriad of plethora of shittons of ML stuff exists on the web. After reading this book, I realize I should go finish my self-driving car nanodegree rather than doing the ML one. That I can learn if I have a source of data. ", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Annotating pathway into mouse Single-Cell clusters\n\nThis tutorial shows how to use the descartes_rpa module with scanpy formated data outside of Descartes. Data from the [Trajectory inference for hematopoiesis in mouse](https://scanpy-tutorials.readthedocs.io/en/latest/paga-paul15.html) tutorial will be used.", "_____no_output_____" ] ], [ [ "import scanpy as sc", "_____no_output_____" ], [ "adata = sc.datasets.paul15()", "WARNING: In Scanpy 0.*, this returned logarithmized data. Now it returns non-logarithmized data.\n... storing 'paul15_clusters' as categorical\nTrying to set attribute `.uns` of view, copying.\n" ], [ "adata.X = adata.X.astype('float64') # this is not required and results will be comparable without it", "_____no_output_____" ], [ "sc.pp.recipe_zheng17(adata)", "_____no_output_____" ], [ "sc.tl.pca(adata, svd_solver='arpack')", "_____no_output_____" ], [ "sc.pp.neighbors(adata, n_neighbors=4, n_pcs=20)", "_____no_output_____" ], [ "sc.tl.leiden(adata)", "_____no_output_____" ] ], [ [ "### Since this dataset is from mouse (Mus musculus), we pass its species as input", "_____no_output_____" ] ], [ [ "from descartes_rpa import get_pathways_for_group", "_____no_output_____" ], [ "get_pathways_for_group(adata, groupby=\"paul15_clusters\", species=\"Mus musculus\")", "/home/joao/miniconda3/envs/descartes-rpa/lib/python3.9/site-packages/scanpy/tools/_rank_genes_groups.py:419: RuntimeWarning: invalid value encountered in log2\n self.stats[group_name, 'logfoldchanges'] = np.log2(\n" ] ], [ [ "### We can look at the top 2 marker genes for each cluster", "_____no_output_____" ] ], [ [ "from descartes_rpa.pl import marker_genes", "_____no_output_____" ], [ "marker_genes(adata, n_genes=2)", "WARNING: dendrogram data not found (using key=dendrogram_paul15_clusters). Running `sc.tl.dendrogram` with default parameters. For fine tuning it is recommended to run `sc.tl.dendrogram` independently.\nWARNING: saving figure to file dotplot_marker_genes.pdf\n" ] ], [ [ "### Also, we can look at the shared pathways between clusters", "_____no_output_____" ] ], [ [ "from descartes_rpa.pl import shared_pathways", "_____no_output_____" ], [ "shared_pathways(adata, clusters=[\"9GMP\", \"1Ery\", \"17Neu\", \"4Ery\"])", "_____no_output_____" ], [ "from descartes_rpa import get_shared", "_____no_output_____" ], [ "get_shared(adata, clusters=[\"1Ery\", \"4Ery\"])", "_____no_output_____" ], [ "from descartes_rpa.pl import pathways", "_____no_output_____" ], [ "adata.uns[\"pathways\"].keys()", "_____no_output_____" ], [ "pathways(adata, \"18Eos\")", "_____no_output_____" ], [ "pathways(adata, \"3Ery\")", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ [ [ "# Djibouti\n\n* Homepage of project: https://oscovida.github.io\n* Plots are explained at http://oscovida.github.io/plots.html\n* [Execute this Jupyter Notebook using myBinder](https://mybinder.org/v2/gh/oscovida/binder/master?filepath=ipynb/Djibouti.ipynb)", "_____no_output_____" ] ], [ [ "import datetime\nimport time\n\nstart = datetime.datetime.now()\nprint(f\"Notebook executed on: {start.strftime('%d/%m/%Y %H:%M:%S%Z')} {time.tzname[time.daylight]}\")", "_____no_output_____" ], [ "%config InlineBackend.figure_formats = ['svg']\nfrom oscovida import *", "_____no_output_____" ], [ "overview(\"Djibouti\", weeks=5);", "_____no_output_____" ], [ "overview(\"Djibouti\");", "_____no_output_____" ], [ "compare_plot(\"Djibouti\", normalise=True);\n", "_____no_output_____" ], [ "# load the data\ncases, deaths = get_country_data(\"Djibouti\")\n\n# get population of the region for future normalisation:\ninhabitants = population(\"Djibouti\")\nprint(f'Population of \"Djibouti\": {inhabitants} people')\n\n# compose into one table\ntable = compose_dataframe_summary(cases, deaths)\n\n# show tables with up to 1000 rows\npd.set_option(\"max_rows\", 1000)\n\n# display the table\ntable", "_____no_output_____" ] ], [ [ "# Explore the data in your web browser\n\n- If you want to execute this notebook, [click here to use myBinder](https://mybinder.org/v2/gh/oscovida/binder/master?filepath=ipynb/Djibouti.ipynb)\n- and wait (~1 to 2 minutes)\n- Then press SHIFT+RETURN to advance code cell to code cell\n- See http://jupyter.org for more details on how to use Jupyter Notebook", "_____no_output_____" ], [ "# Acknowledgements:\n\n- Johns Hopkins University provides data for countries\n- Robert Koch Institute provides data for within Germany\n- Atlo Team for gathering and providing data from Hungary (https://atlo.team/koronamonitor/)\n- Open source and scientific computing community for the data tools\n- Github for hosting repository and html files\n- Project Jupyter for the Notebook and binder service\n- The H2020 project Photon and Neutron Open Science Cloud ([PaNOSC](https://www.panosc.eu/))\n\n--------------------", "_____no_output_____" ] ], [ [ "print(f\"Download of data from Johns Hopkins university: cases at {fetch_cases_last_execution()} and \"\n f\"deaths at {fetch_deaths_last_execution()}.\")", "_____no_output_____" ], [ "# to force a fresh download of data, run \"clear_cache()\"", "_____no_output_____" ], [ "print(f\"Notebook execution took: {datetime.datetime.now()-start}\")\n", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code" ] ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ [ [ "# Linear regression without scikit-learn\n\nIn this notebook, we introduce linear regression. Before presenting the\navailable scikit-learn classes, we will provide some insights with a simple\nexample. We will use a dataset that contains information about penguins.", "_____no_output_____" ], [ "<div class=\"admonition note alert alert-info\">\n<p class=\"first admonition-title\" style=\"font-weight: bold;\">Note</p>\n<p class=\"last\">If you want a deeper overview regarding this dataset, you can refer to the\nAppendix - Datasets description section at the end of this MOOC.</p>\n</div>", "_____no_output_____" ] ], [ [ "import pandas as pd\n\npenguins = pd.read_csv(\"../datasets/penguins_regression.csv\")\npenguins.head()", "_____no_output_____" ] ], [ [ "This dataset contains measurements taken on penguins. We will formulate the\nfollowing problem: using the flipper length of a penguin, we would like\nto infer its mass.", "_____no_output_____" ] ], [ [ "import seaborn as sns\n\nfeature_names = \"Flipper Length (mm)\"\ntarget_name = \"Body Mass (g)\"\ndata, target = penguins[[feature_names]], penguins[target_name]\n\nax = sns.scatterplot(data=penguins, x=feature_names, y=target_name)\nax.set_title(\"Flipper length in function of the body mass\")", "_____no_output_____" ] ], [ [ "<div class=\"admonition tip alert alert-warning\">\n<p class=\"first admonition-title\" style=\"font-weight: bold;\">Tip</p>\n<p class=\"last\">The function <tt class=\"docutils literal\">scatterplot</tt> from searborn take as input the full dataframe\nand the parameter <tt class=\"docutils literal\">x</tt> and <tt class=\"docutils literal\">y</tt> allows to specify the name of the columns to\nbe plotted. Note that this function returns a matplotlib axis\n(named <tt class=\"docutils literal\">ax</tt> in the example above) that can be further used to add element on\nthe same matplotlib axis (such as a title).</p>\n</div>", "_____no_output_____" ], [ "<div class=\"admonition caution alert alert-warning\">\n<p class=\"first admonition-title\" style=\"font-weight: bold;\">Caution!</p>\n<p class=\"last\">Here and later, we use the name <tt class=\"docutils literal\">data</tt> and <tt class=\"docutils literal\">target</tt> to be explicit. In\nscikit-learn, documentation <tt class=\"docutils literal\">data</tt> is commonly named <tt class=\"docutils literal\">X</tt> and <tt class=\"docutils literal\">target</tt> is\ncommonly called <tt class=\"docutils literal\">y</tt>.</p>\n</div>", "_____no_output_____" ], [ "In this problem, penguin mass is our target. It is a continuous\nvariable that roughly varies between 2700 g and 6300 g. Thus, this is a\nregression problem (in contrast to classification). We also see that there is\nalmost a linear relationship between the body mass of the penguin and its\nflipper length. The longer the flipper, the heavier the penguin.\n\nThus, we could come up with a simple formula, where given a flipper length\nwe could compute the body mass of a penguin using a linear relationship\nof the form `y = a * x + b` where `a` and `b` are the 2 parameters of our\nmodel.", "_____no_output_____" ] ], [ [ "def linear_model_flipper_mass(flipper_length, weight_flipper_length,\n intercept_body_mass):\n \"\"\"Linear model of the form y = a * x + b\"\"\"\n body_mass = weight_flipper_length * flipper_length + intercept_body_mass\n return body_mass", "_____no_output_____" ] ], [ [ "Using the model we defined above, we can check the body mass values\npredicted for a range of flipper lengths. We will set `weight_flipper_length`\nto be 45 and `intercept_body_mass` to be -5000.", "_____no_output_____" ] ], [ [ "import numpy as np\n\nweight_flipper_length = 45\nintercept_body_mass = -5000\n\nflipper_length_range = np.linspace(data.min(), data.max(), num=300)\npredicted_body_mass = linear_model_flipper_mass(\n flipper_length_range, weight_flipper_length, intercept_body_mass)", "_____no_output_____" ] ], [ [ "We can now plot all samples and the linear model prediction.", "_____no_output_____" ] ], [ [ "label = \"{0:.2f} (g / mm) * flipper length + {1:.2f} (g)\"\n\nax = sns.scatterplot(data=penguins, x=feature_names, y=target_name)\nax.plot(flipper_length_range, predicted_body_mass, color=\"tab:orange\")\n_ = ax.set_title(label.format(weight_flipper_length, intercept_body_mass))", "_____no_output_____" ] ], [ [ "The variable `weight_flipper_length` is a weight applied to the feature\n`flipper_length` in order to make the inference. When this coefficient is\npositive, it means that penguins with longer flipper lengths will have larger\nbody masses. If the coefficient is negative, it means that penguins with\nshorter flipper lengths have larger body masses. Graphically, this\ncoefficient is represented by the slope of the curve in the plot. Below we\nshow what the curve would look like when the `weight_flipper_length`\ncoefficient is negative.", "_____no_output_____" ] ], [ [ "weight_flipper_length = -40\nintercept_body_mass = 13000\n\npredicted_body_mass = linear_model_flipper_mass(\n flipper_length_range, weight_flipper_length, intercept_body_mass)", "_____no_output_____" ] ], [ [ "We can now plot all samples and the linear model prediction.", "_____no_output_____" ] ], [ [ "ax = sns.scatterplot(data=penguins, x=feature_names, y=target_name)\nax.plot(flipper_length_range, predicted_body_mass, color=\"tab:orange\")\n_ = ax.set_title(label.format(weight_flipper_length, intercept_body_mass))", "_____no_output_____" ] ], [ [ "In our case, this coefficient has a meaningful unit: g/mm.\nFor instance, a coefficient of 40 g/mm, means that for each\nadditional millimeter in flipper length, the body weight predicted will\nincrease by 40 g.", "_____no_output_____" ] ], [ [ "body_mass_180 = linear_model_flipper_mass(\n flipper_length=180, weight_flipper_length=40, intercept_body_mass=0)\nbody_mass_181 = linear_model_flipper_mass(\n flipper_length=181, weight_flipper_length=40, intercept_body_mass=0)\n\nprint(f\"The body mass for a flipper length of 180 mm \"\n f\"is {body_mass_180} g and {body_mass_181} g \"\n f\"for a flipper length of 181 mm\")", "_____no_output_____" ] ], [ [ "We can also see that we have a parameter `intercept_body_mass` in our model.\nThis parameter corresponds to the value on the y-axis if `flipper_length=0`\n(which in our case is only a mathematical consideration, as in our data,\n the value of `flipper_length` only goes from 170mm to 230mm). This y-value\nwhen x=0 is called the y-intercept. If `intercept_body_mass` is 0, the curve\nwill pass through the origin:", "_____no_output_____" ] ], [ [ "weight_flipper_length = 25\nintercept_body_mass = 0\n\n# redefined the flipper length to start at 0 to plot the intercept value\nflipper_length_range = np.linspace(0, data.max(), num=300)\npredicted_body_mass = linear_model_flipper_mass(\n flipper_length_range, weight_flipper_length, intercept_body_mass)", "_____no_output_____" ], [ "ax = sns.scatterplot(data=penguins, x=feature_names, y=target_name)\nax.plot(flipper_length_range, predicted_body_mass, color=\"tab:orange\")\n_ = ax.set_title(label.format(weight_flipper_length, intercept_body_mass))", "_____no_output_____" ] ], [ [ "Otherwise, it will pass through the `intercept_body_mass` value:", "_____no_output_____" ] ], [ [ "weight_flipper_length = 45\nintercept_body_mass = -5000\n\npredicted_body_mass = linear_model_flipper_mass(\n flipper_length_range, weight_flipper_length, intercept_body_mass)", "_____no_output_____" ], [ "ax = sns.scatterplot(data=penguins, x=feature_names, y=target_name)\nax.plot(flipper_length_range, predicted_body_mass, color=\"tab:orange\")\n_ = ax.set_title(label.format(weight_flipper_length, intercept_body_mass))", "_____no_output_____" ] ], [ [ " In this notebook, we have seen the parametrization of a linear regression\n model and more precisely meaning of the terms weights and intercepts.", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/pbcquoc/vietocr/blob/master/vietocr_gettingstart.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "\n# Introduction\n<p align=\"center\">\n<img src=\"https://raw.githubusercontent.com/pbcquoc/vietocr/master/image/vietocr.jpg\" width=\"512\" height=\"512\">\n</p>\nThis notebook describe how you can use VietOcr to train OCR model\n\n\n", "_____no_output_____" ] ], [ [ "# pip install --quiet vietocr==0.3.2", "\u001b[?25l\r\u001b[K |█████▌ | 10kB 26.4MB/s eta 0:00:01\r\u001b[K |███████████ | 20kB 1.7MB/s eta 0:00:01\r\u001b[K |████████████████▋ | 30kB 2.3MB/s eta 0:00:01\r\u001b[K |██████████████████████▏ | 40kB 2.5MB/s eta 0:00:01\r\u001b[K |███████████████████████████▋ | 51kB 2.0MB/s eta 0:00:01\r\u001b[K |████████████████████████████████| 61kB 1.8MB/s \n\u001b[?25h Installing build dependencies ... \u001b[?25l\u001b[?25hdone\n Getting requirements to build wheel ... \u001b[?25l\u001b[?25hdone\n Preparing wheel metadata ... \u001b[?25l\u001b[?25hdone\n\u001b[K |████████████████████████████████| 880kB 7.2MB/s \n\u001b[K |████████████████████████████████| 952kB 17.0MB/s \n\u001b[?25h Building wheel for gdown (PEP 517) ... \u001b[?25l\u001b[?25hdone\n Building wheel for lmdb (setup.py) ... \u001b[?25l\u001b[?25hdone\n\u001b[31mERROR: albumentations 0.1.12 has requirement imgaug<0.2.7,>=0.2.5, but you'll have imgaug 0.4.0 which is incompatible.\u001b[0m\n" ] ], [ [ "# Inference", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\nfrom PIL import Image\n\nfrom vietocr.tool.predictor import Predictor\nfrom vietocr.tool.config import Cfg", "_____no_output_____" ], [ "config = Cfg.load_config_from_name('vgg_transformer')", "_____no_output_____" ] ], [ [ "Change weights to your weights or using default weights from our pretrained model. Path can be url or local file", "_____no_output_____" ] ], [ [ "# config['weights'] = './weights/transformerocr.pth'\nconfig['weights'] = 'https://drive.google.com/uc?id=13327Y1tz1ohsm5YZMyXVMPIOjoOA0OaA'\nconfig['cnn']['pretrained']=False\nconfig['device'] = 'cpu'\nconfig['predictor']['beamsearch']=False", "_____no_output_____" ], [ "detector = Predictor(config)", "Cached Downloading: C:\\Users\\thinhlx\\.cache/gdown\\https-COLON--SLASH--SLASH-drive.google.com-SLASH-uc-QUESTION-id-EQUAL-13327Y1tz1ohsm5YZMyXVMPIOjoOA0OaA\nDownloading...\nFrom: https://drive.google.com/uc?id=13327Y1tz1ohsm5YZMyXVMPIOjoOA0OaA\nTo: C:\\Users\\thinhlx\\.cache\\gdown\\tmpmvu_d_09\\dl\n100%|████████████████████████████████████████████████████████████████████████████████████████████| 152M/152M [00:15<00:00, 9.89MB/s]\n" ], [ "! gdown --id 1uMVd6EBjY4Q0G2IkU5iMOQ34X0bysm0b\n! unzip -qq -o sample.zip", "Downloading...\nFrom: https://drive.google.com/uc?id=1uMVd6EBjY4Q0G2IkU5iMOQ34X0bysm0b\nTo: D:\\HIPT\\vietocr\\sample.zip\n\n 0%| | 0.00/306k [00:00<?, ?B/s]\n100%|##########| 306k/306k [00:00<00:00, 1.42MB/s]\n100%|##########| 306k/306k [00:00<00:00, 1.41MB/s]\n'unzip' is not recognized as an internal or external command,\noperable program or batch file.\n" ], [ "! ls sample | shuf |head -n 5", "'ls' is not recognized as an internal or external command,\noperable program or batch file.\n" ], [ "img = './image/3.png'\nimg = Image.open(img)\nplt.imshow(img)\ns = detector.predict(img)\ns", "_____no_output_____" ] ], [ [ "# Download sample dataset", "_____no_output_____" ] ], [ [ "! gdown https://drive.google.com/uc?id=19QU4VnKtgm3gf0Uw_N2QKSquW1SQ5JiE", "Downloading...\nFrom: https://drive.google.com/uc?id=19QU4VnKtgm3gf0Uw_N2QKSquW1SQ5JiE\nTo: /content/data_line.zip\n61.2MB [00:00, 67.2MB/s]\n" ], [ "! unzip -qq -o ./data_line.zip", "_____no_output_____" ] ], [ [ "# Train model", "_____no_output_____" ], [ "\n\n1. Load your config\n2. Train model using your dataset above\n\n", "_____no_output_____" ], [ "Load the default config, we adopt VGG for image feature extraction", "_____no_output_____" ] ], [ [ "from vietocr.tool.config import Cfg\nfrom vietocr.model.trainer import Trainer", "_____no_output_____" ] ], [ [ "# Change the config \n\n* *data_root*: the folder save your all images\n* *train_annotation*: path to train annotation\n* *valid_annotation*: path to valid annotation\n* *print_every*: show train loss at every n steps\n* *valid_every*: show validation loss at every n steps\n* *iters*: number of iteration to train your model\n* *export*: export weights to folder that you can use for inference\n* *metrics*: number of sample in validation annotation you use for computing full_sequence_accuracy, for large dataset it will take too long, then you can reuduce this number\n", "_____no_output_____" ] ], [ [ "config = Cfg.load_config_from_name('vgg_transformer')", "_____no_output_____" ], [ "#config['vocab'] = 'aAàÀảẢãÃáÁạẠăĂằẰẳẲẵẴắẮặẶâÂầẦẩẨẫẪấẤậẬbBcCdDđĐeEèÈẻẺẽẼéÉẹẸêÊềỀểỂễỄếẾệỆfFgGhHiIìÌỉỈĩĨíÍịỊjJkKlLmMnNoOòÒỏỎõÕóÓọỌôÔồỒổỔỗỖốỐộỘơƠờỜởỞỡỠớỚợỢpPqQrRsStTuUùÙủỦũŨúÚụỤưƯừỪửỬữỮứỨựỰvVwWxXyYỳỲỷỶỹỸýÝỵỴzZ0123456789!\"#$%&\\'()*+,-./:;<=>?@[\\\\]^_`{|}~ '\n\ndataset_params = {\n 'name':'hw',\n 'data_root':'./data_line/',\n 'train_annotation':'train_line_annotation.txt',\n 'valid_annotation':'test_line_annotation.txt'\n}\n\nparams = {\n 'print_every':200,\n 'valid_every':15*200,\n 'iters':20000,\n 'checkpoint':'./checkpoint/transformerocr_checkpoint.pth', \n 'export':'./weights/transformerocr.pth',\n 'metrics': 10000\n }\n\nconfig['trainer'].update(params)\nconfig['dataset'].update(dataset_params)\nconfig['device'] = 'cuda:0'", "_____no_output_____" ] ], [ [ "you can change any of these params in this full list below", "_____no_output_____" ] ], [ [ "config", "_____no_output_____" ] ], [ [ "You should train model from our pretrained ", "_____no_output_____" ] ], [ [ "trainer = Trainer(config, pretrained=True)", "Downloading: \"https://download.pytorch.org/models/vgg19_bn-c79401a0.pth\" to /root/.cache/torch/hub/checkpoints/vgg19_bn-c79401a0.pth\n" ] ], [ [ "Save model configuration for inference, load_config_from_file", "_____no_output_____" ] ], [ [ "trainer.config.save('config.yml')", "_____no_output_____" ] ], [ [ "Visualize your dataset to check data augmentation is appropriate", "_____no_output_____" ] ], [ [ "trainer.visualize_dataset()", "_____no_output_____" ] ], [ [ "Train now", "_____no_output_____" ] ], [ [ "trainer.train()", "iter: 000200 - train loss: 1.657 - lr: 1.91e-05 - load time: 1.08 - gpu time: 158.33\niter: 000400 - train loss: 1.429 - lr: 3.95e-05 - load time: 0.76 - gpu time: 158.76\niter: 000600 - train loss: 1.331 - lr: 7.14e-05 - load time: 0.73 - gpu time: 158.38\niter: 000800 - train loss: 1.252 - lr: 1.12e-04 - load time: 1.29 - gpu time: 158.43\niter: 001000 - train loss: 1.218 - lr: 1.56e-04 - load time: 0.84 - gpu time: 158.86\niter: 001200 - train loss: 1.192 - lr: 2.01e-04 - load time: 0.78 - gpu time: 160.20\niter: 001400 - train loss: 1.140 - lr: 2.41e-04 - load time: 1.54 - gpu time: 158.48\niter: 001600 - train loss: 1.129 - lr: 2.73e-04 - load time: 0.70 - gpu time: 159.42\niter: 001800 - train loss: 1.095 - lr: 2.93e-04 - load time: 0.74 - gpu time: 158.03\niter: 002000 - train loss: 1.098 - lr: 3.00e-04 - load time: 0.66 - gpu time: 159.21\niter: 002200 - train loss: 1.060 - lr: 3.00e-04 - load time: 1.52 - gpu time: 157.63\niter: 002400 - train loss: 1.055 - lr: 3.00e-04 - load time: 0.80 - gpu time: 159.34\niter: 002600 - train loss: 1.032 - lr: 2.99e-04 - load time: 0.74 - gpu time: 159.13\niter: 002800 - train loss: 1.019 - lr: 2.99e-04 - load time: 1.42 - gpu time: 158.27\n" ] ], [ [ "Visualize prediction from our trained model\n", "_____no_output_____" ] ], [ [ "trainer.visualize_prediction()", "_____no_output_____" ] ], [ [ "Compute full seq accuracy for full valid dataset", "_____no_output_____" ] ], [ [ "trainer.precision()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# RNN\nIn this section, we will introduce how to use recurrent neural networks for text classification.\nThe dataset we use is the IMDB Movie Reviews.\nWe use the reviews written by the users as the input and try to predict whether the they are positive or negative.\n\n## Preparing the data\n\nYou can use the following code to load the IMDB dataset.\n", "_____no_output_____" ] ], [ [ "import tensorflow as tf\ntf.random.set_seed(42)\n\nfrom tensorflow.keras.datasets import imdb\nfrom tensorflow.keras.preprocessing import sequence\n\nmax_words = 10000\nembedding_dim = 32\n\n(train_data, train_labels), (test_data, test_labels) = imdb.load_data(\n num_words=max_words)\nprint(train_data.shape)\nprint(train_labels.shape)\nprint(train_data[:2])\nprint(train_labels[:2])", "(25000,)\n(25000,)\n[list([1, 14, 22, 16, 43, 530, 973, 1622, 1385, 65, 458, 4468, 66, 3941, 4, 173, 36, 256, 5, 25, 100, 43, 838, 112, 50, 670, 2, 9, 35, 480, 284, 5, 150, 4, 172, 112, 167, 2, 336, 385, 39, 4, 172, 4536, 1111, 17, 546, 38, 13, 447, 4, 192, 50, 16, 6, 147, 2025, 19, 14, 22, 4, 1920, 4613, 469, 4, 22, 71, 87, 12, 16, 43, 530, 38, 76, 15, 13, 1247, 4, 22, 17, 515, 17, 12, 16, 626, 18, 2, 5, 62, 386, 12, 8, 316, 8, 106, 5, 4, 2223, 5244, 16, 480, 66, 3785, 33, 4, 130, 12, 16, 38, 619, 5, 25, 124, 51, 36, 135, 48, 25, 1415, 33, 6, 22, 12, 215, 28, 77, 52, 5, 14, 407, 16, 82, 2, 8, 4, 107, 117, 5952, 15, 256, 4, 2, 7, 3766, 5, 723, 36, 71, 43, 530, 476, 26, 400, 317, 46, 7, 4, 2, 1029, 13, 104, 88, 4, 381, 15, 297, 98, 32, 2071, 56, 26, 141, 6, 194, 7486, 18, 4, 226, 22, 21, 134, 476, 26, 480, 5, 144, 30, 5535, 18, 51, 36, 28, 224, 92, 25, 104, 4, 226, 65, 16, 38, 1334, 88, 12, 16, 283, 5, 16, 4472, 113, 103, 32, 15, 16, 5345, 19, 178, 32])\n list([1, 194, 1153, 194, 8255, 78, 228, 5, 6, 1463, 4369, 5012, 134, 26, 4, 715, 8, 118, 1634, 14, 394, 20, 13, 119, 954, 189, 102, 5, 207, 110, 3103, 21, 14, 69, 188, 8, 30, 23, 7, 4, 249, 126, 93, 4, 114, 9, 2300, 1523, 5, 647, 4, 116, 9, 35, 8163, 4, 229, 9, 340, 1322, 4, 118, 9, 4, 130, 4901, 19, 4, 1002, 5, 89, 29, 952, 46, 37, 4, 455, 9, 45, 43, 38, 1543, 1905, 398, 4, 1649, 26, 6853, 5, 163, 11, 3215, 2, 4, 1153, 9, 194, 775, 7, 8255, 2, 349, 2637, 148, 605, 2, 8003, 15, 123, 125, 68, 2, 6853, 15, 349, 165, 4362, 98, 5, 4, 228, 9, 43, 2, 1157, 15, 299, 120, 5, 120, 174, 11, 220, 175, 136, 50, 9, 4373, 228, 8255, 5, 2, 656, 245, 2350, 5, 4, 9837, 131, 152, 491, 18, 2, 32, 7464, 1212, 14, 9, 6, 371, 78, 22, 625, 64, 1382, 9, 8, 168, 145, 23, 4, 1690, 15, 16, 4, 1355, 5, 28, 6, 52, 154, 462, 33, 89, 78, 285, 16, 145, 95])]\n[1 0]\n" ] ], [ [ "The code above would load the reviews into train_data and test_data, load the labels (positive or negative) into train_labels and test_labels. As you can see the reviews in train_data are lists of integers instead of texts. It is because the raw texts cannot be used as an input to a neural network. Neural networks only accepts numerical data as inputs.\n\nThe integers we see above is the raw text data after a preprocessing step named tokenization. It first split each review into a list of words and assign an integer to each of the words. For example, a scentence \"How are you? How are you doing?\" will be transformed into a list of words as [\"how\", \"are\", \"you\", \"how\", \"are\", \"you\", \"doing\"]. Then transformed to [5, 8, 9, 5, 8, 9, 7]. The integers doesn't have special meanings but a representation of the words. Same integers represents the same words, different integers represents different words.\n\nThe labels are also integers, where 1 represents positive, 0 represents negative.\n\nThen, we pad the data to the same length.", "_____no_output_____" ] ], [ [ "# Pad the sequence to length max_len.\nmaxlen = 100\nprint(len(train_data[0]))\nprint(len(train_data[1]))\ntrain_data = sequence.pad_sequences(train_data, maxlen=maxlen)\ntest_data = sequence.pad_sequences(test_data, maxlen=maxlen)\nprint(train_data.shape)\nprint(train_labels.shape)", "218\n189\n(25000, 100)\n(25000,)\n" ] ], [ [ "## Building your network\nThe next step is to build your neural network model and train it.\nWe will introduce the neural network in three steps.\nThe first step is the embedding, which transform each integer list into a list of vectors.\nThe second step is to feed the vectors to the recurrent neural network.\nThe third step is to use the output of the recurrent neural network for classification.\n### Embedding\nEmbedding means find a corresponding numerical vector for each word, which is now an integer in the list.\nThe numerical vector can be seen as the coordinate of a point in the space.\nWe embed the words into specific points in the space.\nThat is why we call the process embedding.\n\nTo implement it, we use a Keras Embedding layer.\nFirst, we need to create a Keras Sequential model.\nThen, we add the layers one by one to the model.\nThe order of the layers is from the input to the output.", "_____no_output_____" ] ], [ [ "from tensorflow.keras.layers import Embedding\nfrom tensorflow.keras import Sequential\n\nmax_words = 10000\nembedding_dim = 32\n\nmodel = Sequential()\nmodel.add(Embedding(max_words, embedding_dim))\nmodel.summary()", "Model: \"sequential\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\nembedding (Embedding) (None, None, 32) 320000 \n=================================================================\nTotal params: 320,000\nTrainable params: 320,000\nNon-trainable params: 0\n_________________________________________________________________\n" ] ], [ [ "In the code above, we initialized an Embedding layer.\nThe max_words is the vocabulary size, which is an integer meaning how many different words are there in the input data.\nThe integer 16 means the length of the vector representation fro each word is 32.\nThe output tensor of the Embedding layer is (batch_size, max_len, embedding_dim).\n### Recurrent Neural Networks\nAfter the embedding layer, we need to use a recurrent neural network for the classification.\nRecurrent neural networks can handle sequential inputs.\nFor example, we input a movie review as a sequence of word embedding vectors to it, which are the output of the embedding layer.\nEach vector has a length of 32.\nEach review contains 100 vectors.\nIf we see the RNN as a whole, it takes 100 vectors of length 16 altogether.\nHowever, in the real case, it takes one vector at a time.\n\nEach time the RNN takes in a word embedding vector,\nit not only takes the word embedding vector, but another state vector as well.\nYou can think the state vector as the memory of the RNN.\nIt memorizes the previous words it taken as input.\nIn the first step, the RNN has no previous words to remember.\nIt takes an initial state, which is usually empty,\nand the first word embedding vector as input.\nThe output of the first step is actually the state to be input to the second step.\nFor the rest of the steps, the RNN will just take the previous output and the current input as input,\nand output the state for the next step.\nFor the last step, the output state is the final output we will use for the classification.\n\nWe can use the following python code to illustrate the process.\n```python\nstate = [0] * 32\nfor i in range(100):\n state = rnn(embedding[i], state)\nreturn state\n```\nThe returned state is the final output of the RNN.\n\nSometimes, we may also need to collect the output of each step as shown in the following code.\n```python\nstate = [0] * 32\noutput = []\nfor i in range(100):\n state = rnn(embedding[i], state)\n output.append(state)\nreturn output\n```\nIn the code above, the output of an RNN can also be a sequence of vectors, which is the same format as the input to the RNN.\nTherefore, we can make the RNN deeper by stacking multiple RNN layers together.\n\n\nTo implement the RNN described, we need the SimpleRNN layer in Keras.", "_____no_output_____" ] ], [ [ "from tensorflow.keras.layers import SimpleRNN\n\nmodel.add(SimpleRNN(embedding_dim, return_sequences=True))\nmodel.add(SimpleRNN(embedding_dim, return_sequences=True))\nmodel.add(SimpleRNN(embedding_dim, return_sequences=True))\nmodel.add(SimpleRNN(embedding_dim))\nmodel.summary()", "Model: \"sequential\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\nembedding (Embedding) (None, None, 32) 320000 \n_________________________________________________________________\nsimple_rnn (SimpleRNN) (None, None, 32) 2080 \n_________________________________________________________________\nsimple_rnn_1 (SimpleRNN) (None, None, 32) 2080 \n_________________________________________________________________\nsimple_rnn_2 (SimpleRNN) (None, None, 32) 2080 \n_________________________________________________________________\nsimple_rnn_3 (SimpleRNN) (None, 32) 2080 \n=================================================================\nTotal params: 328,320\nTrainable params: 328,320\nNon-trainable params: 0\n_________________________________________________________________\n" ] ], [ [ "The return_sequences parameter controlls whether to collect all the output vectors of an RNN or only collect the last output. It is set to False by default.\n### Classification Head\nThen we will use the output of the last SimpleRNN layer, which is a vector of length 32, as the input to the classification head.\nIn the classification head, we use a fully-connected layer for the classification.\nThen we compile and train the model.\n", "_____no_output_____" ] ], [ [ "from tensorflow.keras.layers import Dense\n\nmodel.add(Dense(1, activation='sigmoid'))\nmodel.compile(optimizer='adam', metrics=['acc'], loss='binary_crossentropy')\nmodel.fit(train_data, \n train_labels,\n epochs=2,\n batch_size=128)", "Epoch 1/2\n196/196 [==============================] - 30s 119ms/step - loss: 0.6258 - acc: 0.6111\nEpoch 2/2\n196/196 [==============================] - 23s 117ms/step - loss: 0.3361 - acc: 0.8573\n" ] ], [ [ "Then we can validate our model on the testing data.", "_____no_output_____" ] ], [ [ "model.evaluate(test_data, test_labels)", "782/782 [==============================] - 28s 35ms/step - loss: 0.3684 - acc: 0.8402\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import os, shutil\nfrom google.colab import drive\ndrive.mount('/content/gdrive')\n# 표시되는 화면에서 자신이 사용 중인 구글 드라이브를 선택하고 진행\nos.chdir('gdrive/My Drive')\n%cd Test/Audio/\n# gdrive/My Drive/Test/Audio/에 본 예제에 사용되는 rain.wav 파일이 존재 (테스트에 사용될 wav 파일을 미리 업로드 해 두어야 함)", "Mounted at /content/gdrive\n/content/gdrive/My Drive/Test/Audio\n" ], [ "import librosa, librosa.display\naudio_data = 'rain.wav'\nx , sr = librosa.load(audio_data, sr=44100)", "_____no_output_____" ], [ "import sklearn\nspectral_rolloff = librosa.feature.spectral_rolloff(x+0.01, sr=sr)[0]\nspectral_rolloff.shape", "_____no_output_____" ], [ "# Computing the time variable for visualization\nimport matplotlib.pyplot as plt\nplt.figure(figsize=(12, 4))\nframes = range(len(spectral_rolloff))\nt = librosa.frames_to_time(frames)", "_____no_output_____" ], [ "# Normalising for visualisation\ndef normalize(x, axis=0):\n return sklearn.preprocessing.minmax_scale(x, axis=axis)", "_____no_output_____" ], [ "# Plotting the Spectral Rolloff along the waveform\nlibrosa.display.waveplot(x, sr=sr, alpha=0.4)\nplt.plot(t, normalize(spectral_rolloff), color='r')", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Business Results", "_____no_output_____" ], [ "## Recomendação de compra dos imóveis:\n##### Para decidir quais imóveis deverão ser comprados, iremos comparar os imóveis pelo zipcode e selecionar as casas que estão abaixo da média. Como explicado anteriormente, estaremos selecionando as casas que possuem 'condition' maior ou igual a 3.", "_____no_output_____" ] ], [ [ "import pandas as pd\n\ndata = pd.read_csv('datasets/kc_house_clean.csv')\npd.set_option( 'display.float_format', lambda x: '%.2f' % x)\npd.options.display.max_columns = None\npd.options.display.max_rows = None", "_____no_output_____" ], [ "#comparar a média do 'zipcode' e se o valor for abaixo da média e a condição >= 3 então comprar\ndf = data[['price','id','zipcode','condition']].copy()\n\nrecomendations = df[['zipcode','price']].groupby('zipcode').median().reset_index()\nrecomendations.columns = ['zipcode','median_price']\ndf = pd.merge(df, recomendations, on= 'zipcode', how = 'inner')\n\ndf['recomendations'] = 'na'\nfor i in range(len(df)):\n if (df.loc[i,'price'] < df.loc[i,'median_price'])&(df.loc[i,'condition'] >= 3):\n df.loc[i,'recomendations'] = 'comprar'\n else:\n df.loc[i,'recomendations'] = 'não comprar'\ndf.to_csv('datasets/recomendacoes_compras.csv')", "_____no_output_____" ], [ "df.head()", "_____no_output_____" ], [ "## Recomendação de venda dos imóveis\n##### Para decidir o melhor momento de venda dos imóveis estaremos comparando os imóveis pelo 'zipcode' e pela estação.\n##### Se o preço for menor que a mediana, então estaremos acrescentando 30% ao valor.\n##### Se o preço for maior que a mediana, então estaremos acrescentando 10% ao valor.\n\ndata = pd.read_csv('datasets/kc_house_clean.csv')\ndf1 = data[['price','date','zipcode','id']].copy()\n\ndf1['date'] = pd.to_datetime(df1['date']).dt.month\ndf1['season'] = df1['date'].apply(lambda x: 'spring' if ( x >= 3 )&( x <= 5 ) else\n 'summer' if ( x >= 6 )&( x <= 8 ) else\n 'fall' if ( x >= 9 )&( x <= 11 ) else \n 'winter')\n\n\nestacoes = df1[['zipcode','season','price']].groupby(['zipcode','season']).median().reset_index()\nestacoes.columns = ['zipcode','season','median_price']\nestacoes['zip_season'] = estacoes['zipcode'].astype(str) + \"_\" + estacoes['season'].astype(str)\nestacoes = estacoes.drop(['zipcode','season'], axis = 1)\n\ndf1['zip_season'] = df1['zipcode'].astype(str) + \"_\" + df1['season'].astype(str)\ndf1 = pd.merge( df1, estacoes, on='zip_season', how='inner')\n\ndf1['venda'] = 'na'\nfor i in range(len(data)):\n if (df1.loc[i,'price'] <= df1.loc[i,'median_price']):\n df1.loc[i,'venda'] = df1.loc[i,'price'] * 1.30\n else:\n df1.loc[i,'venda'] = df1.loc[i,'price'] * 1.10\ndf1.to_csv('datasets/recomendacoes_venda.csv')", "_____no_output_____" ], [ "df1.head()", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "Unlicense" ]

[ "Unlicense" ]

[ "Unlicense" ]

[ [ [ "<center>\n<img src=\"../../img/ods_stickers.jpg\" />\n \n## [mlcourse.ai](https://mlcourse.ai) – Open Machine Learning Course \n\nAuthor: [Yury Kashnitskiy](https://yorko.github.io). Translated by [Sergey Oreshkov](https://www.linkedin.com/in/sergeoreshkov/). This material is subject to the terms and conditions of the [Creative Commons CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license. Free use is permitted for any non-commercial purpose.", "_____no_output_____" ], [ "# <center> Assignment #8 (demo). Solution\n\n## <center> Implementation of online regressor\n \n**Same assignment as a [Kaggle Kernel](https://www.kaggle.com/kashnitsky/a8-demo-implementing-online-regressor) + [solution](https://www.kaggle.com/kashnitsky/a8-demo-implementing-online-regressor-solution).**", "_____no_output_____" ], [ "Here we'll implement a regressor trained with stochastic gradient descent (SGD). Fill in the missing code. If you do evething right, you'll pass a simple embedded test.", "_____no_output_____" ], [ "## <center>Linear regression and Stochastic Gradient Descent", "_____no_output_____" ] ], [ [ "import numpy as np\nimport pandas as pd\nfrom sklearn.base import BaseEstimator\nfrom sklearn.metrics import log_loss, mean_squared_error, roc_auc_score\nfrom sklearn.model_selection import train_test_split\nfrom tqdm import tqdm\n\n%matplotlib inline\nimport seaborn as sns\nfrom matplotlib import pyplot as plt\nfrom sklearn.preprocessing import StandardScaler", "_____no_output_____" ] ], [ [ "Implement class `SGDRegressor`. Specification:\n- class is inherited from `sklearn.base.BaseEstimator`\n- constructor takes parameters `eta` – gradient step ($10^{-3}$ by default) and `n_epochs` – dataset pass count (3 by default)\n- constructor also creates `mse_` and `weights_` lists in order to track mean squared error and weight vector during gradient descent iterations\n- Class has `fit` and `predict` methods\n- The `fit` method takes matrix `X` and vector `y` (`numpy.array` objects) as parameters, appends column of ones to `X` on the left side, initializes weight vector `w` with **zeros** and then makes `n_epochs` iterations of weight updates (you may refer to this [article](https://medium.com/open-machine-learning-course/open-machine-learning-course-topic-8-vowpal-wabbit-fast-learning-with-gigabytes-of-data-60f750086237) for details), and for every iteration logs mean squared error and weight vector `w` in corresponding lists we created in the constructor. \n- Additionally the `fit` method will create `w_` variable to store weights which produce minimal mean squared error\n- The `fit` method returns current instance of the `SGDRegressor` class, i.e. `self`\n- The `predict` method takes `X` matrix, adds column of ones to the left side and returns prediction vector, using weight vector `w_`, created by the `fit` method.", "_____no_output_____" ] ], [ [ "class SGDRegressor(BaseEstimator):\n def __init__(self, eta=1e-3, n_epochs=3):\n self.eta = eta\n self.n_epochs = n_epochs\n self.mse_ = []\n self.weights_ = []\n\n def fit(self, X, y):\n X = np.hstack([np.ones([X.shape[0], 1]), X])\n\n w = np.zeros(X.shape[1])\n\n for it in tqdm(range(self.n_epochs)):\n for i in range(X.shape[0]):\n\n new_w = w.copy()\n new_w[0] += self.eta * (y[i] - w.dot(X[i, :]))\n for j in range(1, X.shape[1]):\n new_w[j] += self.eta * (y[i] - w.dot(X[i, :])) * X[i, j]\n w = new_w.copy()\n\n self.weights_.append(w)\n self.mse_.append(mean_squared_error(y, X.dot(w)))\n\n self.w_ = self.weights_[np.argmin(self.mse_)]\n\n return self\n\n def predict(self, X):\n X = np.hstack([np.ones([X.shape[0], 1]), X])\n\n return X.dot(self.w_)", "_____no_output_____" ] ], [ [ "Let's test out the algorithm on height/weight data. We will predict heights (in inches) based on weights (in lbs).", "_____no_output_____" ] ], [ [ "data_demo = pd.read_csv(\"../../data/weights_heights.csv\")", "_____no_output_____" ], [ "plt.scatter(data_demo[\"Weight\"], data_demo[\"Height\"])\nplt.xlabel(\"Weight (lbs)\")\nplt.ylabel(\"Height (Inch)\")\nplt.grid();", "_____no_output_____" ], [ "X, y = data_demo[\"Weight\"].values, data_demo[\"Height\"].values", "_____no_output_____" ] ], [ [ "Perform train/test split and scale data.", "_____no_output_____" ] ], [ [ "X_train, X_valid, y_train, y_valid = train_test_split(\n X, y, test_size=0.3, random_state=17\n)", "_____no_output_____" ], [ "scaler = StandardScaler()\nX_train_scaled = scaler.fit_transform(X_train.reshape([-1, 1]))\nX_valid_scaled = scaler.transform(X_valid.reshape([-1, 1]))", "_____no_output_____" ] ], [ [ "Train created `SGDRegressor` with `(X_train_scaled, y_train)` data. Leave default parameter values for now.", "_____no_output_____" ] ], [ [ "# you code here\nsgd_reg = SGDRegressor()\nsgd_reg.fit(X_train_scaled, y_train)", "_____no_output_____" ] ], [ [ "Draw a chart with training process – dependency of mean squared error from the i-th SGD iteration number.", "_____no_output_____" ] ], [ [ "# you code here\nplt.plot(range(len(sgd_reg.mse_)), sgd_reg.mse_)\nplt.xlabel(\"#updates\")\nplt.ylabel(\"MSE\");", "_____no_output_____" ] ], [ [ "Print the minimal value of mean squared error and the best weights vector.", "_____no_output_____" ] ], [ [ "# you code here\nnp.min(sgd_reg.mse_), sgd_reg.w_", "_____no_output_____" ] ], [ [ "Draw chart of model weights ($w_0$ and $w_1$) behavior during training.", "_____no_output_____" ] ], [ [ "# you code here\nplt.subplot(121)\nplt.plot(range(len(sgd_reg.weights_)), [w[0] for w in sgd_reg.weights_])\nplt.subplot(122)\nplt.plot(range(len(sgd_reg.weights_)), [w[1] for w in sgd_reg.weights_]);", "_____no_output_____" ] ], [ [ "Make a prediction for hold-out set `(X_valid_scaled, y_valid)` and check MSE value.", "_____no_output_____" ] ], [ [ "# you code here\nsgd_holdout_mse = mean_squared_error(y_valid, sgd_reg.predict(X_valid_scaled))\nsgd_holdout_mse", "_____no_output_____" ] ], [ [ "Do the same thing for `LinearRegression` class from `sklearn.linear_model`. Evaluate MSE for hold-out set.", "_____no_output_____" ] ], [ [ "# you code here\nfrom sklearn.linear_model import LinearRegression\n\nlm = LinearRegression().fit(X_train_scaled, y_train)\nprint(lm.coef_, lm.intercept_)\nlinreg_holdout_mse = mean_squared_error(y_valid, lm.predict(X_valid_scaled))\nlinreg_holdout_mse", "_____no_output_____" ], [ "try:\n assert (sgd_holdout_mse - linreg_holdout_mse) < 1e-4\n print(\"Correct!\")\nexcept AssertionError:\n print(\n \"Something's not good.\\n Linreg's holdout MSE: {}\"\n \"\\n SGD's holdout MSE: {}\".format(linreg_holdout_mse, sgd_holdout_mse)\n )", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import importlib \nimport numpy as np\nimport toolbox\nimport annoStats\nimportlib.reload(toolbox)\nimportlib.reload(annoStats)", "_____no_output_____" ], [ "#cubes, loc, F0, nonzeros, ids = annoTightAll.testMain()\ncubes, loc, Fmax = annoStats.testMain()", "Synapsin647\nVGluT1_647\n" ], [ "loc", "_____no_output_____" ], [ "idx = np.argsort([3,2,1])\nnp.transpose(loc[:,idx])", "_____no_output_____" ], [ "toolbox.mainOUT(np.transpose(loc[:,idx]), ['x','y','z'], \"locations_test.csv\")", "_____no_output_____" ], [ "len(Fmax[0])", "_____no_output_____" ], [ "Fmax", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "#Dynamic Costs Reporting\nCalculate DV360 cost at the dynamic creative combination level.\n", "_____no_output_____" ], [ "#License\n\nCopyright 2020 Google LLC,\n\nLicensed under the Apache License, Version 2.0 (the \"License\");\nyou may not use this file except in compliance with the License.\nYou may obtain a copy of the License at\n\n https://www.apache.org/licenses/LICENSE-2.0\n\nUnless required by applicable law or agreed to in writing, software\ndistributed under the License is distributed on an \"AS IS\" BASIS,\nWITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\nSee the License for the specific language governing permissions and\nlimitations under the License.\n\n", "_____no_output_____" ], [ "#Disclaimer\nThis is not an officially supported Google product. It is a reference implementation. There is absolutely NO WARRANTY provided for using this code. The code is Apache Licensed and CAN BE fully modified, white labeled, and disassembled by your team.\n\nThis code generated (see starthinker/scripts for possible source):\n - **Command**: \"python starthinker_ui/manage.py colab\"\n - **Command**: \"python starthinker/tools/colab.py [JSON RECIPE]\"\n\n", "_____no_output_____" ], [ "#1. Install Dependencies\nFirst install the libraries needed to execute recipes, this only needs to be done once, then click play.\n", "_____no_output_____" ] ], [ [ "!pip install git+https://github.com/google/starthinker\n", "_____no_output_____" ] ], [ [ "#2. Set Configuration\n\nThis code is required to initialize the project. Fill in required fields and press play.\n\n1. If the recipe uses a Google Cloud Project:\n - Set the configuration **project** value to the project identifier from [these instructions](https://github.com/google/starthinker/blob/master/tutorials/cloud_project.md).\n\n1. If the recipe has **auth** set to **user**:\n - If you have user credentials:\n - Set the configuration **user** value to your user credentials JSON.\n - If you DO NOT have user credentials:\n - Set the configuration **client** value to [downloaded client credentials](https://github.com/google/starthinker/blob/master/tutorials/cloud_client_installed.md).\n\n1. If the recipe has **auth** set to **service**:\n - Set the configuration **service** value to [downloaded service credentials](https://github.com/google/starthinker/blob/master/tutorials/cloud_service.md).\n\n", "_____no_output_____" ] ], [ [ "from starthinker.util.configuration import Configuration\n\n\nCONFIG = Configuration(\n project=\"\",\n client={},\n service={},\n user=\"/content/user.json\",\n verbose=True\n)\n\n", "_____no_output_____" ] ], [ [ "#3. Enter Dynamic Costs Reporting Recipe Parameters\n 1. Add a sheet URL. This is where you will enter advertiser and campaign level details.\n 1. Specify the CM network ID.\n 1. Click run now once, and a tab called <strong>Dynamic Costs</strong> will be added to the sheet with instructions.\n 1. Follow the instructions on the sheet; this will be your configuration.\n 1. StarThinker will create two or three (depending on the case) reports in CM named <strong>Dynamic Costs - ...</strong>.\n 1. Wait for <b>BigQuery->->->Dynamic_Costs_Analysis</b> to be created or click Run Now.\n 1. Copy <a href='https://datastudio.google.com/open/1vBvBEiMbqCbBuJTsBGpeg8vCLtg6ztqA' target='_blank'>Dynamic Costs Sample Data ( Copy From This )</a>.\n 1. Click Edit Connection, and Change to <b>BigQuery->->->Dynamic_Costs_Analysis</b>.\n 1. Copy <a href='https://datastudio.google.com/open/1xulBAdx95SnvjnUzFP6r14lhkvvVbsP8' target='_blank'>Dynamic Costs Sample Report ( Copy From This )</a>.\n 1. When prompted, choose the new data source you just created.\n 1. Edit the table to include or exclude columns as desired.\n 1. Or, give the dashboard connection intructions to the client.\nModify the values below for your use case, can be done multiple times, then click play.\n", "_____no_output_____" ] ], [ [ "FIELDS = {\n 'dcm_account': '',\n 'auth_read': 'user', # Credentials used for reading data.\n 'configuration_sheet_url': '',\n 'auth_write': 'service', # Credentials used for writing data.\n 'bigquery_dataset': 'dynamic_costs',\n}\n\nprint(\"Parameters Set To: %s\" % FIELDS)\n", "_____no_output_____" ] ], [ [ "#4. Execute Dynamic Costs Reporting\nThis does NOT need to be modified unless you are changing the recipe, click play.\n", "_____no_output_____" ] ], [ [ "from starthinker.util.configuration import execute\nfrom starthinker.util.recipe import json_set_fields\n\nTASKS = [\n {\n 'dynamic_costs': {\n 'auth': 'user', \n 'account': {'field': {'name': 'dcm_account', 'kind': 'string', 'order': 0, 'default': ''}}, \n 'sheet': {\n 'template': {\n 'url': 'https://docs.google.com/spreadsheets/d/19J-Hjln2wd1E0aeG3JDgKQN9TVGRLWxIEUQSmmQetJc/edit?usp=sharing', \n 'tab': 'Dynamic Costs', \n 'range': 'A1'\n }, \n 'url': {'field': {'name': 'configuration_sheet_url', 'kind': 'string', 'order': 1, 'default': ''}}, \n 'tab': 'Dynamic Costs', \n 'range': 'A2:B'\n }, \n 'out': {\n 'auth': 'user', \n 'dataset': {'field': {'name': 'bigquery_dataset', 'kind': 'string', 'order': 2, 'default': 'dynamic_costs'}}\n }\n }\n }\n]\n\njson_set_fields(TASKS, FIELDS)\n\nexecute(CONFIG, TASKS, force=True)\n", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## Capítulo 2: Toma de Decisiones y Neutrosofía", "_____no_output_____" ], [ "## Distancia Euclidiana entre números SVN", "_____no_output_____" ] ], [ [ "def euclideanNeu(a1,a2):\n a=0\n c=len(a1) \n for i in range(c): \n a=a+pow(a1[i][0]-a2[i][0],2)+pow(a1[i][1]-a2[i][1],2)+pow(a1[i][2]-a2[i][2],2)\n a=pow(1.0/3.0*a,0.5)\n return(a)", "_____no_output_____" ] ], [ [ "## Ejemplo de uso de la distancia Euclidiana", "_____no_output_____" ] ], [ [ "EB=(1,0,0)\nMMB=(0.9, 0.1, 0.1)\nMB=(0.8,0.15,0.20)\nB=(0.70,0.25,0.30)\nMDB=(0.60,0.35,0.40)\nM=(0.50,0.50,0.50)\nMDM=(0.40,0.65,0.60)\nMA=(0.30,0.75,0.70)\nMM=(0.20,0.85,0.80)\nMMM=(0.10,0.90,0.90)\nEM=(0,1,1)\nr1=[MDB,B,B]\ni=[MMB, MMB, MB]\neuclideanNeu(r1,i)", "_____no_output_____" ] ], [ [ "## Operador SVNWA", "_____no_output_____" ] ], [ [ "def SVNWA(list,W):\n t=1\n i=1\n f=1\n c=0\n for j in list:\n t=t*(1-j[0])**W[c]\n i=i*j[1]**W[c]\n f=f*j[2]**W[c]\n c=c+1\n return (1-t,i,f)\n ", "_____no_output_____" ] ], [ [ "## Ejemplo de uso SVNWA", "_____no_output_____" ] ], [ [ "A=[MDB,B,MDB]\nW = [0.55, 0.26, 0.19] # W:Vector de pesos\nSVNWA(A,W)", "_____no_output_____" ] ], [ [ "## Operador SVNGA", "_____no_output_____" ] ], [ [ "def SVNGA(list,W):\n t=1\n i=1\n f=1\n c=0\n for j in list:\n t=t*j[0]**W[c]\n i=i*j[1]**W[c]\n f=f*j[2]**W[c] \n c=c+1\n return (t,i,f)", "_____no_output_____" ], [ "A=[MDB,B,MDB]\nW = [0.55, 0.26, 0.19] # W:Vector de pesos\nSVNGA(A,W)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "---\nlayout: post\ntitle: Project Euler - Problem 5\npost-order: 005\n---", "_____no_output_____" ], [ "2520 is the smallest number that can be divided by each of the numbers from 1 to 10 without any remainder.\n\nWhat is the smallest positive number that is **evenly divisible** by all of the numbers from 1 to 20?", "_____no_output_____" ], [ "### Solution #1", "_____no_output_____" ], [ "Let's brute-force it and see what happens. One thing we know: the number we are trying to find **cannot be** smaller than the products of all primes up to $20$. And the reason for that is, the number we are trying to find has to be divisible by $1$ to $20$ and the only way that can be accomplished is if the prime factors (for example $13$) is in that product.", "_____no_output_____" ] ], [ [ "import math\nimport time\n\n# let's setle \"floor\" first: the products of all primes from 1 to 20\ndef primes_up_to(n):\n product = 2\n\n for candidate_for_prime in range (3,n+1):\n for i in range (2,candidate_for_prime): \n if (candidate_for_prime % i == 0):\n break\n elif ((candidate_for_prime % i != 0) & (i + 1 == candidate_for_prime)):\n product *= candidate_for_prime\n \n return product\n\nstart = time.time()\n\ndivisible_up_to = 20\nfloor = primes_up_to(divisible_up_to)\n\n# the highest we can go is to n! (n factorial)\nceiling = math.factorial(divisible_up_to)\n\nfound = False\n\nfor ii in range (floor,ceiling):\n for i in range(2,divisible_up_to+1):\n if ii % i != 0:\n break\n elif i == divisible_up_to:\n print (ii)\n elapsed = time.time() - start\n print(elapsed)\n found = True\n break\n \n if found:\n break", "232792560\n162.78695583343506\n" ] ], [ [ "As you may notice, the running time is extremely high (and it will get higher if we choose a greater `divisible_up_to`). We need to find a more efficient way to tacke this problem.", "_____no_output_____" ], [ "### Solution #2 - Greatest power of primes", "_____no_output_____" ], [ "The solution for this problem actually requires **no computation**. And the reason for that is, if we find the prime factorization of each number up to 20 and multiply the greatest power of each, we will find the correct solution.", "_____no_output_____" ], [ "$$\n\\begin{align}\n2 &= 2^1\\\\\n3 &= 3^1\\\\\n4 &= 2^2\\\\\n5 &= 5^1\\\\\n\\vdots\\\\\n18 &= 2^1 \\cdot 3^2\\\\\n19 &= 19^1\\\\\n20 &= 2^2 \\cdot 5^1\n\\end{align}\n$$", "_____no_output_____" ], [ "We can take advantage of this and build a list with the prime factors and a list with the prime factor's greatest power that shows up from 1 to `divisible_up_to`. After that, we evaluate the first item of the list of primes raised to the first item of the list of powers, multiplied by the second item of the list of primes raised to the second item of the list of powers and so on.", "_____no_output_____" ], [ "|LISTS|<td colspan=11>|\n|----------------|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|\n| list of primes | [ | 2 | 3 | 5 | 7 | 11| 13| 17| 19| ] |\n| list of powers | [ |4 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | ] |", "_____no_output_____" ], [ "Answer = $2^4 \\cdot 3^2 \\cdot 5^1 \\cdot 7^1 \\cdot 11^1 \\cdot 13^1 \\cdot 17^1 \\cdot 19^1 = 232792560$", "_____no_output_____" ] ], [ [ "def primes_list(n):\n list_of_primes = [1,2]\n\n for i in range (3,n+1):\n for ii in range (2,i): \n if (i % ii == 0):\n break\n elif ((i % ii != 0) & (ii + 1 == i)):\n list_of_primes.append(i)\n \n return list_of_primes\n\nlist_of_primes = primes_list(20)\nprint (list_of_primes)", "[1, 2, 3, 5, 7, 11, 13, 17, 19]\n" ] ], [ [ "Let's build a new list, with the same number of elements, but now filled with zeros. It will receive the greatest power for each prime:", "_____no_output_____" ] ], [ [ "list_of_powers = [1] * len(list_of_primes)\nprint(list_of_powers)", "[1, 1, 1, 1, 1, 1, 1, 1, 1]\n" ] ], [ [ "Now we need to check what is the greatest power that shows up, for each of the primes, from 1 to 20.", "_____no_output_____" ] ], [ [ "for i in list_of_primes:\n print (i)\n\n# if it's prime, there's no need to check: the power will be 1\n# do the prime factorization of all numbers up to 20", "1\n2\n3\n5\n7\n11\n13\n17\n19\n" ] ], [ [ "Placing everything on a single code,", "_____no_output_____" ], [ "The performance of this second algorithm is much better. We can even increase `divisible_up_to` without a substantial compromise on its performance:", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# tips on drawing histograms: bin alignment vs labels", "_____no_output_____" ] ], [ [ "import pandas as pd, numpy as np, seaborn as sns\nimport os\n# Remove the most annoying pandas warning\n# A value is trying to be set on a copy of a slice from a DataFrame.\npd.options.mode.chained_assignment = None\n\ndata_dir = '../../data'\nsrc_file = 'sample01.csv'\nf = os.path.join(data_dir, src_file)\n\nimport sys\nsys.path.insert(0, '../modules')\nimport handy as hd", "_____no_output_____" ], [ "df = pd.read_csv(f, sep = ';')\ndf.shape", "_____no_output_____" ], [ "# remember the original data under this variable\ndf0 = df.copy()", "_____no_output_____" ] ], [ [ "## example: 2d histogram\n\nshowing bin allignments: incorrect, and correct", "_____no_output_____" ] ], [ [ "data = df\nsns.set()\nimport matplotlib.pyplot as plt;\nimport matplotlib.colors as mcolors\n\nfig, ax = plt.subplots(1, 3, figsize=(20,4))\nax = ax.flatten()\n\nfig.suptitle(\"some histograms\", fontsize=20, y=1.1)\n\naxis = ax[0]\n\nh = axis.hist(x = data.weekday)\n\naxis.set_xlabel('days of week (0=Mon)')\naxis.set_ylabel('frequency')\naxis.set_title(\"sloppy bin allignment\", fontsize = 16)\n\naxis = ax[1]\n\naxis.hist(x = data.weekday, bins = 7)\n\naxis.set_xlabel('days of week (0=Mon)')\naxis.set_ylabel('frequency')\naxis.set_title(\"still sloppy: x labels not aligned)\", fontsize = 16)\n\naxis = ax[2]\n\naxis.hist(x = data.weekday, bins = (np.arange(7 + 1)) - 0.5)\n\naxis.set_xlabel('days of week (0=Mon)')\naxis.set_ylabel('frequency')\naxis.set_title(\"correct bin allignment\", fontsize = 16)\n\n\nplt.tight_layout()\n", "_____no_output_____" ], [ "data = df\n\nimport matplotlib.pyplot as plt;\nimport matplotlib.colors as mcolors\n\nfig, ax = plt.subplots(1, 2, figsize=(15,6))\n\nfig.suptitle(\"two 2d histograms\", fontsize=20, y=1.1)\n\nax1 = ax[0]\n# gammas = [0.8, 0.5, 0.3]\ngamma = 0.4\n\nh = ax1.hist2d(x = data.weekday, \n y = data.hour, \n bins = [7, 24], \n norm=mcolors.PowerNorm(gamma), \n cmap='Blues')\n\ncb = fig.colorbar(h[3], ax=ax1)\ncb.set_label('incidents per bin')\nax1.set_xlabel('days of week (0=Mon)')\nax1.set_ylabel('hours of day')\nax1.set_title(\"sloppy bin allignment\", fontsize = 16)\n\nplt.tight_layout()\n\nax1 =ax[1]\nxbins = np.arange(0, 7 + 1) - 0.5\nybins = np.arange(0, 24 + 1) - 0.5\n\nh = ax1.hist2d(x = data.weekday, \n y = data.hour, \n bins = [xbins, ybins], \n norm=mcolors.PowerNorm(gamma), \n cmap='Blues')\n #vmax = 100 )\ncb = fig.colorbar(h[3], ax=ax1)\ncb.set_label('incidents per bin')\nax1.set_xlabel('days of week (0=Mon)')\nax1.set_ylabel('hours of day')\nax1.set_title(\"bins aligned correctly\", fontsize = 16)\n\nplt.tight_layout()\n", "_____no_output_____" ] ], [ [ "# same problem with Seaborn (v 0.11) displot or joinplot, type hist", "_____no_output_____" ] ], [ [ "# bad \nsns.jointplot(x=\"weekday\", y=\"hour\", data=df.sample(1000), kind='hist', ax = ax[0])\n", "_____no_output_____" ], [ "# still bad\nsns.jointplot(x=\"weekday\", y=\"hour\", data=df.sample(1000), kind='hist',bins = [7, 24])", "_____no_output_____" ], [ "#good\nbins = (np.arange(7 + 1)-0.5, np.arange(24 + 1) - 0.5)\nsns.jointplot(x=\"weekday\", y=\"hour\", data=df.sample(1000), kind='hist',bins = bins)", "_____no_output_____" ], [ "# but the problem is gone, if displot type is other than kde\nsns.jointplot(x=\"weekday\", y=\"hour\", data=df.sample(1000), kind='kde', xlim=(0,6), ylim=(0,24))", "_____no_output_____" ], [ "sns.jointplot(x=\"weekday\", y=\"hour\", cmap = 'coolwarm', data=df.sample(10000), kind='kde', fill=True)", "_____no_output_____" ], [ "# same, with contour only\nsns.jointplot(x=\"weekday\", y=\"hour\", cmap = 'coolwarm', data=df.sample(10000), kind='kde')", "_____no_output_____" ] ], [ [ "# more problems with bin allignment", "_____no_output_____" ] ], [ [ "# I will now demonstrate how bad bins lead to bad conclusions", "_____no_output_____" ], [ "# store previous df under separate variable\ndf0 = df\n", "_____no_output_____" ], [ "data_dir = '../../data'\nsrc_file = 'sample02.csv'\nf = os.path.join(data_dir, src_file)\ndf = pd.read_csv(f, sep = ';')", "_____no_output_____" ], [ "df['created'] = pd.to_datetime(df['created'], format = hd.format_dash, errors = 'coerce')\ndf['resolved'] = pd.to_datetime(df['resolved'], format = hd.format_dash, errors = 'coerce')\ndf = hd.augment_columns(df)\n\n# remember this augmented data set\ndf1 = df", "_____no_output_____" ], [ "minweek, maxweek = df.week_nr.min(), df.week_nr.max()\nminweek, maxweek", "_____no_output_____" ], [ "data1 = df[(df.week_nr == maxweek -1) & (df.category == 'Alarm')].weekhour\ndata2 = df[(df.week_nr == maxweek -1)].weekhour", "_____no_output_____" ], [ "#INCORRECT\n\nslots = 24 * 7\nbins_sloppy = slots\n \nfig, ax = plt.subplots(2,1, figsize = (25,4))\n\naxis = ax[0]\nw1 = axis.hist(x= data1, bins = bins_sloppy)\naxis.set_title('category 1, sloppy bins', loc = 'left', fontsize = 16)\n\naxis = ax[1]\nw2 = axis.hist(x= data2, bins = bins_sloppy)\naxis.set_title('category 2, sloppy bins', loc = 'left', fontsize = 16)\n\nplt.tight_layout()", "_____no_output_____" ], [ "# the visualization above is wrong. Bars are mislined, due to sloppy bins definition", "_____no_output_____" ], [ "#CORRECT\n\ndef my_title_markers(axis, title, marker1, marker2):\n axis.set_title(title, loc = 'left', fontsize = 16)\n axis.axvline(x=marker1, color='r', linestyle='dashed', linewidth=2, label = str(marker1))\n axis.axvline(x=marker2, color='b', linestyle='dashed', linewidth=2, label = str(marker2))\n axis.legend(loc = 'upper left', fontsize = '12')\n\n \nslots = 24 * 7\nsloppy_bins = slots\nbins = np.arange(slots + 1) - 0.5\n\nfig, ax = plt.subplots(4,1, figsize = (25,10))\nmarker1, marker2 = 8, 40\n\naxis = ax[0]\naxis.hist(x= data1, bins = sloppy_bins)\nmy_title_markers(axis, 'category 1, sloppy bins', marker1, marker2)\n\naxis = ax[1]\naxis.hist(x= data2, bins = sloppy_bins)\nmy_title_markers(axis, 'category 2, sloppy bins', marker1, marker2)\n\naxis = ax[2]\nw1 = axis.hist(x= data1, bins = bins)\nmy_title_markers(axis, 'category 1, correct bins', marker1, marker2)\n\naxis = ax[3]\nw2 = axis.hist(x= data2, bins = bins)\nmy_title_markers(axis, 'category 2, correct bins', marker1, marker2)\n\nplt.tight_layout()", "_____no_output_____" ] ], [ [ "# Problems with binning and rounding days and weeks (pd.Timestamp)\n\nThe below is a long version. The compact version has been summarized in handy-showcase workbook.", "_____no_output_____" ] ], [ [ "data_dir = '../../data'\nsrc_file = 'sample01.csv'\nf = os.path.join(data_dir, src_file)\ndf = pd.read_csv(f, sep = ';')\ndf['created'] = pd.to_datetime(df['created'], format = hd.format_dash, errors = 'coerce')\ndf['resolved'] = pd.to_datetime(df['resolved'], format = hd.format_dash, errors = 'coerce')\ndf = hd.augment_columns(df)\n", "_____no_output_____" ], [ "import seaborn as sns, matplotlib.pyplot as plt\nsns.set()\n\nstart, end = df.created.min(), df.created.max()\ndays = (end - start).days\nweeks = days / 7\n\nfig, ax = plt.subplots(1,1, figsize = (25,3))\ndata = df[(df.created > start) & (df.created < end)].created\n\naxis = ax\nw = axis.hist(x= data, bins = int(weeks))\naxis.set_title('Naive (incorrect) weekly histogram (1 bar = 1 week)', fontsize = 20)\n\nplt.show()\nprint('Basic statistics:\\n')\nprint('Total records:\\t{}'.format(len(df)))\nprint('start:\\t{}\\t{}\\nend:\\t{}\\t{}'.format(start, start.day_name(), end, end.day_name()))\nprint('weeks: {:.1f}\\trecords per week:{:.1f},\\t weekly min:{},\\t weekly max:{}'.format( weeks, len(df) / weeks, int(min(w[0])), int(max(w[0]))))\nprint('days: {}\\trecords per day:{:.1f}'.format(days, len(df) / days))\n", "_____no_output_____" ], [ "# return Monday 00:00:00 before given moment\ndef monday_before(now):\n monday_before = now - pd.Timedelta(now.weekday(), 'days')\n # Monday 00:00:00\n return pd.Timestamp(monday_before.date())\n\n# return Monday 00:00:00 after given moment \ndef monday_after(now):\n # trick: compute Monday before 1 week from now... it's the same.\n return monday_before(now + pd.Timedelta(7, 'days'))\n\n# use this to have full week span, spanning tne entire period\n# returns: Monday before, Monday after, number of weeks between\ndef outer_week_boundaries(series):\n start, end = monday_before(series.min()), monday_after(series.max())\n return start, end, (end - start).days // 7\n\ndef inner_week_boundaries(series):\n start, end = monday_after(series.min()), monday_before(series.max())\n return start, end, (end - start).days // 7 \n\n# exact number of days, including fraction of day (float)\ndef fractional_days(data_start, data_end):\n delta = data_end - data_start\n return delta.days + delta.seconds / (60 * 60 * 24)\n\n# number of full 24-hour periods\ndef inner_days(data_start, data_end):\n return (data_end - data_start).days \n\n# number of days between midnight-before-first-record and midnight-after-last-record\ndef outer_days(data_start, data_end):\n return (data_end.date() - data_start.date()).days + 1\n\ndef weekly_bin_edges(start, howmany):\n # add 1 for we count bin edges rather than bins\n WEEK = pd.Timedelta(7, 'days')\n return [outer_start + i * WEEK for i in np.arange(howmany + 1)]\n\ndef daily_bin_edges(start, howmany):\n # add 1 for we count bin edges rather than bins\n DAY = pd.Timedelta(1, 'days')\n return [data_start.date() + i * DAY for i in np.arange(howmany + 1)]\n\n \n ", "_____no_output_____" ], [ "# weekly bins\n\n\n#s = weekly_statistics()\nouter_start, outer_end, outer_weeks = outer_week_boundaries(df.created)\ninner_start, inner_end, inner_weeks = inner_week_boundaries(df.created)\ndata_start, data_end = df.created.min(), df.created.max()\nweekly_bins = weekly_bin_edges(outer_start, outer_weeks)\n\ndays = fractional_days(data_start, data_end)\nouter_days = outer_days(data_start, data_end)\ndaily_bins = daily_bin_edges(data_start, outer_days)\nweeks = days / 7", "_____no_output_____" ], [ "def draw(axis, outer_start, outer_end, inner_start, inner_end):\n axis.axvline(x=inner_start, color='r', linestyle='dashed', linewidth=2, label = 'inner (full) weeks range')\n axis.axvline(x=outer_start, color='b', linestyle='dashed', linewidth=2, label = 'outer (incomplete) weeks range')\n axis.axvline(x=inner_end, color='r', linestyle='dashed', linewidth=2)\n axis.axvline(x=outer_end, color='b', linestyle='dashed', linewidth=2)\n axis.legend()\n \nfig, ax = plt.subplots(2,1, figsize = (25,6))\ndata = df.created\n\naxis = ax[0]\nw = axis.hist(x= data, bins = weekly_bins)\ndraw(axis, outer_start, outer_end, inner_start, inner_end)\naxis.set_title('Correct weekly histogram (1 bar = 1 week)', fontsize = 20)\n \nweek_values = w[0]\nfullweek_values = week_values[1:-1]\n\naxis = ax[1]\ndraw(axis, outer_start, outer_end, inner_start, inner_end)\nd = axis.hist(x= data, bins = daily_bins, edgecolor = 'black')\naxis.set_title('Corresponding daily histogram (1 bar = 1 day)', fontsize = 20)\n\nday_values = d[0]\nfullday_values = day_values[1:-1]\n\nplt.tight_layout()\nplt.show()\n\n\nprint('Basic statistics:\\n')\nprint('Total records:\\t{}'.format(len(df)))\nprint('Histogram range (outer weeks):{:.0f}'.format(outer_weeks))\nstart, end = outer_start, outer_end\nprint('start:\\t{}\\t{}\\nend:\\t{}\\t{}'.format(start, start.day_name(), end, end.day_name()))\nprint('Data range:')\nstart, end = data_start, data_end\nprint('start:\\t{}\\t{}\\nend:\\t{}\\t{}'.format(start, start.day_name(), end, end.day_name()))\nprint('Full weeks (inner weeks):{:.0f}'.format(inner_weeks))\nstart, end = inner_start, inner_end\nprint('start:\\t{}\\t{}\\nend:\\t{}\\t{}'.format(start, start.day_name(), end, end.day_name()))\nprint('Data stats:')\nprint('weeks: {:.1f}\\trecords per week:{:.1f},\\t weekly min:{},\\t weekly max:{}'.\\\n format( weeks, len(df) / weeks, int(min(fullweek_values)), int(max(week_values))))\nprint('days: {:.1f}\\trecords per day:{:.1f},\\t daily min:{},\\t daily max:{}'.\\\n format(days, len(df) / days, int(min(fullday_values)), int(max(day_values))))\nprint('Note: The minima do not take into account the marginal (uncomplete) weeks or days')\n", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import logging\nimport importlib\nimportlib.reload(logging) # see https://stackoverflow.com/a/21475297/1469195\nlog = logging.getLogger()\nlog.setLevel('INFO')\nimport sys\n\nlogging.basicConfig(format='%(asctime)s %(levelname)s : %(message)s',\n level=logging.INFO, stream=sys.stdout)", "_____no_output_____" ], [ "%%capture\nimport os\nimport site\nos.sys.path.insert(0, '/home/schirrmr/code/reversible/')\nos.sys.path.insert(0, '/home/schirrmr/braindecode/code/braindecode/')\nos.sys.path.insert(0, '/home/schirrmr/code/explaining/reversible//')\n\n\n%load_ext autoreload\n%autoreload 2\nimport numpy as np\nimport logging\nlog = logging.getLogger()\nlog.setLevel('INFO')\nimport sys\nlogging.basicConfig(format='%(asctime)s %(levelname)s : %(message)s',\n level=logging.INFO, stream=sys.stdout)\nimport matplotlib\nfrom matplotlib import pyplot as plt\nfrom matplotlib import cm\n%matplotlib inline\n%config InlineBackend.figure_format = 'png'\nmatplotlib.rcParams['figure.figsize'] = (12.0, 1.0)\nmatplotlib.rcParams['font.size'] = 14\nimport seaborn\nseaborn.set_style('darkgrid')\n\nfrom reversible2.sliced import sliced_from_samples\nfrom numpy.random import RandomState\n\nimport torch\nimport torch.nn as nn\nimport torch.nn.functional as F\nfrom torch.autograd import Variable\nimport numpy as np\nimport copy\nimport math\n\nimport itertools\nimport torch as th\nfrom braindecode.torch_ext.util import np_to_var, var_to_np\nfrom reversible2.splitter import SubsampleSplitter\n\nfrom reversible2.view_as import ViewAs\n\nfrom reversible2.affine import AdditiveBlock\nfrom reversible2.plot import display_text, display_close\nfrom reversible2.bhno import load_file, create_inputs\nth.backends.cudnn.benchmark = True", "_____no_output_____" ], [ "sensor_names = ['Fz', \n 'FC3','FC1','FCz','FC2','FC4',\n 'C5','C3','C1','Cz','C2','C4','C6',\n 'CP3','CP1','CPz','CP2','CP4',\n 'P1','Pz','P2',\n 'POz']\norig_train_cnt = load_file('/data/schirrmr/schirrmr/HGD-public/reduced/train/4.mat')\ntrain_cnt = orig_train_cnt.reorder_channels(sensor_names)\n\ntrain_inputs = create_inputs(train_cnt, final_hz=256, half_before=True)\nn_split = len(train_inputs[0]) - 40\ntest_inputs = [t[-40:] for t in train_inputs]\ntrain_inputs = [t[:-40] for t in train_inputs]", "_____no_output_____" ], [ "cuda = True\nif cuda:\n train_inputs = [i.cuda() for i in train_inputs]\n test_inputs = [i.cuda() for i in test_inputs]", "_____no_output_____" ], [ "from reversible2.graph import Node\nfrom reversible2.branching import CatChans, ChunkChans, Select\nfrom reversible2.constantmemory import sequential_to_constant_memory\nfrom reversible2.constantmemory import graph_to_constant_memory\ndef invert(feature_model, out):\n return feature_model.invert(out)\n\nfrom copy import deepcopy\nfrom reversible2.graph import Node\nfrom reversible2.distribution import TwoClassDist\nfrom reversible2.wrap_invertible import WrapInvertible\nfrom reversible2.blocks import dense_add_no_switch, conv_add_3x3_no_switch\nfrom reversible2.rfft import RFFT, Interleave\nfrom reversible2.util import set_random_seeds\nfrom torch.nn import ConstantPad2d\nimport torch as th\nfrom reversible2.splitter import SubsampleSplitter\n\nset_random_seeds(2019011641, cuda)\nn_chans = train_inputs[0].shape[1]\nn_time = train_inputs[0].shape[2]\nbase_model = nn.Sequential(\n SubsampleSplitter(stride=[2,1],chunk_chans_first=False),\n conv_add_3x3_no_switch(2*n_chans,32),\n conv_add_3x3_no_switch(2*n_chans,32),\n SubsampleSplitter(stride=[2,1],chunk_chans_first=True), # 4 x 128\n conv_add_3x3_no_switch(4*n_chans,32),\n conv_add_3x3_no_switch(4*n_chans,32),\n SubsampleSplitter(stride=[2,1],chunk_chans_first=True), # 8 x 64\n conv_add_3x3_no_switch(8*n_chans,32),\n conv_add_3x3_no_switch(8*n_chans,32))\nbase_model.cuda();\n\nbranch_1_a = nn.Sequential(\n SubsampleSplitter(stride=[2,1],chunk_chans_first=False), # 8 x 32\n conv_add_3x3_no_switch(8*n_chans,32),\n conv_add_3x3_no_switch(8*n_chans,32),\n SubsampleSplitter(stride=[2,1],chunk_chans_first=True),# 16 x 16\n conv_add_3x3_no_switch(16*n_chans,32),\n conv_add_3x3_no_switch(16*n_chans,32),\n SubsampleSplitter(stride=[2,1],chunk_chans_first=True), # 32 x 8\n conv_add_3x3_no_switch(32*n_chans,32),\n conv_add_3x3_no_switch(32*n_chans,32),\n)\nbranch_1_b = nn.Sequential(\n *(list(deepcopy(branch_1_a).children()) + [\n ViewAs((-1, 32*n_chans,n_time//64,1), (-1,(n_time // 2)*n_chans)),\n dense_add_no_switch((n_time // 2)*n_chans,32),\n dense_add_no_switch((n_time // 2)*n_chans,32),\n dense_add_no_switch((n_time // 2)*n_chans,32),\n dense_add_no_switch((n_time // 2)*n_chans,32),\n]))\nbranch_1_a.cuda();\nbranch_1_b.cuda();\n\nbranch_2_a = nn.Sequential(\n SubsampleSplitter(stride=[2,1], chunk_chans_first=False),# 32 x 4\n conv_add_3x3_no_switch(32*n_chans,32),\n conv_add_3x3_no_switch(32*n_chans,32),\n SubsampleSplitter(stride=[2,1],chunk_chans_first=True),# 64 x 2\n conv_add_3x3_no_switch(64*n_chans,32),\n conv_add_3x3_no_switch(64*n_chans,32),\n ViewAs((-1, (n_time // 4)*n_chans,1,1), (-1,(n_time // 4)*n_chans)),\n dense_add_no_switch((n_time // 4)*n_chans,64),\n dense_add_no_switch((n_time // 4)*n_chans,64),\n dense_add_no_switch((n_time // 4)*n_chans,64),\n dense_add_no_switch((n_time // 4)*n_chans,64),\n)\n\n\nbranch_2_b = deepcopy(branch_2_a).cuda()\nbranch_2_a.cuda();\nbranch_2_b.cuda();\n\nfinal_model = nn.Sequential(\n dense_add_no_switch(n_time*n_chans,256),\n dense_add_no_switch(n_time*n_chans,256),\n dense_add_no_switch(n_time*n_chans,256),\n dense_add_no_switch(n_time*n_chans,256),\n)\nfinal_model.cuda();\no = Node(None, base_model)\no = Node(o, ChunkChans(2))\no1a = Node(o, Select(0))\no1b = Node(o, Select(1))\no1a = Node(o1a, branch_1_a)\no1b = Node(o1b, branch_1_b)\no2 = Node(o1a, ChunkChans(2))\no2a = Node(o2, Select(0))\no2b = Node(o2, Select(1))\no2a = Node(o2a, branch_2_a)\no2b = Node(o2b, branch_2_b)\no = Node([o1b,o2a,o2b], CatChans())\no = Node(o, final_model)\no = graph_to_constant_memory(o)\nfeature_model = o\nif cuda:\n feature_model.cuda()\nfeature_model.eval();", "_____no_output_____" ] ], [ [ "### set feature model weights and distribution to good start parametersm", "_____no_output_____" ] ], [ [ "n_dims = np.prod(train_inputs[0].shape[1:])\ni_class_dims = [int(n_dims*0.25), int(n_dims * 0.75)]\n\nfrom reversible2.constantmemory import clear_ctx_dicts\nfrom reversible2.distribution import TwoClassDist\n\nfeature_model.data_init(th.cat((train_inputs[0], train_inputs[1]), dim=0))\n\n# Check that forward + inverse is really identical\nt_out = feature_model(train_inputs[0][:2])\ninverted = invert(feature_model, t_out)\nclear_ctx_dicts(feature_model)\nassert th.allclose(train_inputs[0][:2], inverted, rtol=1e-3,atol=1e-4)\ndevice = list(feature_model.parameters())[0].device\nfrom reversible2.ot_exact import ot_euclidean_loss_for_samples\nclass_dist = TwoClassDist(2, np.prod(train_inputs[0].size()[1:]) - 2,\n i_class_inds=i_class_dims)\nclass_dist.cuda()\n\nfor i_class in range(2):\n with th.no_grad():\n this_outs = feature_model(train_inputs[i_class])\n mean = th.mean(this_outs, dim=0)\n std = th.std(this_outs, dim=0)\n class_dist.set_mean_std(i_class, mean, std)\n # Just check\n setted_mean, setted_std = class_dist.get_mean_std(i_class)\n assert th.allclose(mean, setted_mean)\n assert th.allclose(std, setted_std)\nclear_ctx_dicts(feature_model)\n\noptim_model = th.optim.Adam(feature_model.parameters(), lr=1e-3, betas=(0.9,0.999))\noptim_dist = th.optim.Adam(class_dist.parameters(), lr=1e-2, betas=(0.9,0.999))", "_____no_output_____" ], [ "%%writefile plot.py\nimport torch as th\nimport matplotlib.pyplot as plt\nimport numpy as np\nfrom reversible2.util import var_to_np\nfrom reversible2.plot import display_close\nfrom matplotlib.patches import Ellipse\nimport seaborn\n\ndef plot_outs(feature_model, train_inputs, test_inputs, class_dist):\n with th.no_grad():\n # Compute dist for mean/std of encodings\n data_cls_dists = []\n for i_class in range(len(train_inputs)):\n this_class_outs = feature_model(train_inputs[i_class])[:,class_dist.i_class_inds]\n data_cls_dists.append(\n th.distributions.MultivariateNormal(th.mean(this_class_outs, dim=0),\n covariance_matrix=th.diag(th.std(this_class_outs, dim=0) ** 2)))\n for setname, set_inputs in ((\"Train\", train_inputs), (\"Test\", test_inputs)):\n\n outs = [feature_model(ins) for ins in set_inputs]\n c_outs = [o[:,class_dist.i_class_inds] for o in outs]\n\n c_outs_all = th.cat(c_outs)\n\n cls_dists = []\n for i_class in range(len(c_outs)):\n mean, std = class_dist.get_mean_std(i_class)\n cls_dists.append(\n th.distributions.MultivariateNormal(mean[class_dist.i_class_inds],\n covariance_matrix=th.diag(std[class_dist.i_class_inds] ** 2)))\n\n preds_per_class = [th.stack([cls_dists[i_cls].log_prob(c_out)\n for i_cls in range(len(cls_dists))],\n dim=-1) for c_out in c_outs]\n\n pred_labels_per_class = [np.argmax(var_to_np(preds), axis=1)\n for preds in preds_per_class]\n\n labels = np.concatenate([np.ones(len(set_inputs[i_cls])) * i_cls \n for i_cls in range(len(train_inputs))])\n\n acc = np.mean(labels == np.concatenate(pred_labels_per_class))\n\n data_preds_per_class = [th.stack([data_cls_dists[i_cls].log_prob(c_out)\n for i_cls in range(len(cls_dists))],\n dim=-1) for c_out in c_outs]\n data_pred_labels_per_class = [np.argmax(var_to_np(data_preds), axis=1)\n for data_preds in data_preds_per_class]\n data_acc = np.mean(labels == np.concatenate(data_pred_labels_per_class))\n\n print(\"{:s} Accuracy: {:.1f}%\".format(setname, acc * 100))\n fig = plt.figure(figsize=(5,5))\n ax = plt.gca()\n for i_class in range(len(c_outs)):\n #if i_class == 0:\n # continue\n o = var_to_np(c_outs[i_class]).squeeze()\n incorrect_pred_mask = pred_labels_per_class[i_class] != i_class\n plt.scatter(o[:,0], o[:,1], s=20, alpha=0.75, label=[\"Right\", \"Rest\"][i_class])\n assert len(incorrect_pred_mask) == len(o)\n plt.scatter(o[incorrect_pred_mask,0], o[incorrect_pred_mask,1], marker='x', color='black',\n alpha=1, s=5)\n means, stds = class_dist.get_mean_std(i_class)\n means = var_to_np(means)[class_dist.i_class_inds]\n stds = var_to_np(stds)[class_dist.i_class_inds]\n for sigma in [0.5,1,2,3]:\n ellipse = Ellipse(means, stds[0]*sigma, stds[1]*sigma)\n ax.add_artist(ellipse)\n ellipse.set_edgecolor(seaborn.color_palette()[i_class])\n ellipse.set_facecolor(\"None\")\n for i_class in range(len(c_outs)):\n o = var_to_np(c_outs[i_class]).squeeze()\n plt.scatter(np.mean(o[:,0]), np.mean(o[:,1]),\n color=seaborn.color_palette()[i_class+2], s=80, marker=\"^\",\n label=[\"Right Mean\", \"Rest Mean\"][i_class])\n\n plt.title(\"{:6s} Accuracy: {:.1f}%\\n\"\n \"From data mean/std: {:.1f}%\".format(setname, acc * 100, data_acc * 100))\n plt.legend(bbox_to_anchor=(1,1,0,0))\n display_close(fig)\n return", "_____no_output_____" ], [ "import pandas as pd\ndf = pd.DataFrame()\n\nfrom reversible2.training import OTTrainer\ntrainer = OTTrainer(feature_model, class_dist,\n optim_model, optim_dist)", "_____no_output_____" ], [ "from reversible2.constantmemory import clear_ctx_dicts\nfrom reversible2.timer import Timer\nfrom plot import plot_outs\nfrom reversible2.gradient_penalty import gradient_penalty\n\n\ni_start_epoch_out = 4001\nn_epochs = 10001\nfor i_epoch in range(n_epochs):\n epoch_row = {}\n with Timer(name='EpochLoop', verbose=False) as loop_time:\n loss_on_outs = i_epoch >= i_start_epoch_out\n result = trainer.train(train_inputs, loss_on_outs=loss_on_outs)\n \n epoch_row.update(result)\n epoch_row['runtime'] = loop_time.elapsed_secs * 1000\n if i_epoch % (n_epochs // 20) != 0:\n df = df.append(epoch_row, ignore_index=True)\n # otherwise add ot loss in\n else:\n for i_class in range(len(train_inputs)):\n with th.no_grad():\n class_ins = train_inputs[i_class]\n samples = class_dist.get_samples(i_class, len(train_inputs[i_class]) * 4)\n inverted = feature_model.invert(samples)\n clear_ctx_dicts(feature_model)\n ot_loss_in = ot_euclidean_loss_for_samples(class_ins.view(class_ins.shape[0], -1),\n inverted.view(inverted.shape[0], -1)[:(len(class_ins))])\n epoch_row['ot_loss_in_{:d}'.format(i_class)] = ot_loss_in.item()\n df = df.append(epoch_row, ignore_index=True)\n print(\"Epoch {:d} of {:d}\".format(i_epoch, n_epochs))\n print(\"Loop Time: {:.0f} ms\".format(loop_time.elapsed_secs * 1000))\n display(df.iloc[-3:])\n plot_outs(feature_model, train_inputs, test_inputs,\n class_dist)\n fig = plt.figure(figsize=(8,2))\n plt.plot(var_to_np(th.cat((th.exp(class_dist.class_log_stds),\n th.exp(class_dist.non_class_log_stds)))),\n marker='o')\n display_close(fig)\n \n", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "for i_class in range(len(train_inputs)):\n with th.no_grad():\n class_ins = train_inputs[i_class]\n samples = class_dist.get_samples(i_class, len(train_inputs[i_class]) * 4)\n inverted = feature_model.invert(samples)\n clear_ctx_dicts(feature_model)\n ot_loss_in = ot_euclidean_loss_for_samples(class_ins.view(class_ins.shape[0], -1),\n inverted.view(inverted.shape[0], -1)[:(len(class_ins))])\n epoch_row['ot_loss_in_{:d}'.format(i_class)] = ot_loss_in.item()\nprint(\"Epoch {:d} of {:d}\".format(i_epoch, n_epochs))\nprint(\"Loop Time: {:.0f} ms\".format(loop_time.elapsed_secs * 1000))\ndisplay(df.iloc[-3:])\nplot_outs(feature_model, train_inputs, test_inputs,\n class_dist)\nfig = plt.figure(figsize=(8,2))\nplt.plot(var_to_np(th.cat((th.exp(class_dist.class_log_stds),\n th.exp(class_dist.non_class_log_stds)))),\n marker='o')\ndisplay_close(fig)", "_____no_output_____" ], [ "def get_non_class_outs(feature_model, inputs, class_dist):\n with th.no_grad():\n outs_per_class = [feature_model(t) for t in inputs]\n clear_ctx_dicts(feature_model)\n non_class_inds = np.setdiff1d(list(range(outs_per_class[0].shape[1])),\n class_dist.i_class_inds)\n non_class_outs = [o[:,non_class_inds].detach() for o in outs_per_class]\n return non_class_outs", "_____no_output_____" ], [ "train_non_class_outs = get_non_class_outs(feature_model, train_inputs, class_dist)\n\ntest_non_class_outs = get_non_class_outs(feature_model, test_inputs, class_dist)", "_____no_output_____" ], [ "class DistClassifier(nn.Module):\n def __init__(self, n_dims):\n super(DistClassifier, self).__init__()\n self.mean0 = th.nn.Parameter(th.zeros(n_dims))\n self.mean1 = th.nn.Parameter(th.zeros(n_dims))\n self.logstd0 = th.nn.Parameter(th.zeros(n_dims))\n self.logstd1 = th.nn.Parameter(th.zeros(n_dims))\n \n def predict(self, outs):\n dist0 = th.distributions.MultivariateNormal(self.mean0,\n th.diag(th.exp(self.logstd0) ** 2))\n dist1 = th.distributions.MultivariateNormal(self.mean1,\n th.diag(th.exp(self.logstd1) ** 2))\n return th.stack((dist0.log_prob(outs), dist1.log_prob(outs)), dim=-1)\n \n def predict_log_softmax(self, outs):\n probs = self.predict(outs)\n return F.log_softmax(probs, dim=1)\n \n ", "_____no_output_____" ], [ "clf = DistClassifier(train_non_class_outs[0].shape[1])\nclf.cuda()\noptim_clf = th.optim.Adam(clf.parameters(), lr=1e-2)", "_____no_output_____" ], [ "n_epochs = 200\nfor i_epoch in range(n_epochs):\n accs = []\n for i_class in range(2):\n preds = clf.predict_log_softmax(train_non_class_outs[i_class])\n labels = np_to_var((np.ones(len(preds)) * i_class).astype(np.int64), device='cuda')\n loss = F.nll_loss(preds, labels)\n optim_clf.zero_grad()\n loss.backward()\n optim_clf.step()\n \n with th.no_grad():\n for set_non_class_outs in [train_non_class_outs, test_non_class_outs]:\n accs = []\n for i_class in range(2):\n preds = clf.predict_log_softmax(set_non_class_outs[i_class])\n labels = np_to_var((np.ones(len(preds)) * i_class).astype(np.int64), device='cuda')\n acc = np.mean(np.argmax(var_to_np(preds), axis=1) == var_to_np(labels))\n accs.append(acc)\n print(\"Acc: {:.1f}%\".format(100*np.mean(accs)))\n print(\"\")", "_____no_output_____" ], [ "with th.no_grad():\n for set_non_class_outs in [train_non_class_outs, test_non_class_outs]:\n accs = []\n for i_class in range(2):\n preds = clf.predict_log_softmax(set_non_class_outs[i_class])\n labels = np_to_var((np.ones(len(preds)) * i_class).astype(np.int64), device='cuda')\n acc = np.mean(np.argmax(var_to_np(preds), axis=1) == var_to_np(labels))\n accs.append(acc)\n print(len(preds))\n print(\"Acc: {:.1f}%\".format(100*np.mean(accs)))\nprint(\"\")", "_____no_output_____" ], [ "%%javascript\nvar kernel = IPython.notebook.kernel;\nvar thename = window.document.getElementById(\"notebook_name\").innerHTML;\nvar command = \"nbname = \" + \"'\"+thename+\"'\";\nkernel.execute(command);", "_____no_output_____" ], [ "nbname", "_____no_output_____" ], [ "folder_path = '/data/schirrmr/schirrmr/reversible/models/notebooks/{:s}/'.format(nbname)\nos.makedirs(folder_path,exist_ok=True)", "_____no_output_____" ], [ "name_and_variable = [('feature_model', feature_model),\n ('class_dist', class_dist),\n ('non_class_log_stds', class_dist.non_class_log_stds),\n ('class_log_stds', class_dist.class_log_stds,),\n ('class_means', class_dist.class_means),\n ('non_class_means', class_dist.non_class_means),\n ('feature_model_params', feature_model.state_dict()),\n ('optim_model', optim_model.state_dict()),\n ('optim_dist', optim_dist.state_dict())]\n\nfor name, variable in name_and_variable:\n th.save(variable, os.path.join(folder_path, name + '.pkl'))", "_____no_output_____" ], [ "print(\"\\n\".join([\"{:30s}\\t{:.1f}\".format(f, os.path.getsize(os.path.join(folder_path, f)) / (1024.0 *1024.0))\n for f in os.listdir(folder_path)]))", "_____no_output_____" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np", "_____no_output_____" ], [ "def softmax(x):\n \n c = np.max(x)\n exp_x = np.exp(x-c)\n exp_all = np.sum(exp_x)\n \n return exp_x / exp_all\n\ndef cross_entropy_loss(y, t):\n delta = 1e-7\n return -np.sum(t*np.log(y+delta))\n\nclass SoftmaxWithLoss:\n def __init__(self):\n self.loss = None\n self.y = None\n self.t = None\n \n def forward(self, x, t):\n self.t = t\n self.y = softmax(x)\n self.loss = cross_entropy_loss(self.y, self.t)\n \n return self.loss\n \n def backward(self, dout=1):\n batch_size = self.t.shape[0]\n dx = self.y - self.t / batch_size\n \n return dx\n ", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Installation cell\n%%shell\nif ! command -v julia 3>&1 > /dev/null\nthen\n wget -q 'https://julialang-s3.julialang.org/bin/linux/x64/1.5/julia-1.5.3-linux-x86_64.tar.gz' \\\n -O /tmp/julia.tar.gz\n tar -x -f /tmp/julia.tar.gz -C /usr/local --strip-components 1\n rm /tmp/julia.tar.gz\nfi\njulia -e 'using Pkg; pkg\"add IJulia; precompile;\"'\necho 'Done'", "_____no_output_____" ], [ "using Pkg\r\nPkg.add(PackageSpec(url=\"https://github.com/AxelvL/AHPQ.jl\", rev=\"master\"))\r\nusing AHPQ", "_____no_output_____" ], [ "Pkg.add(\"HTTP\")\r\nusing HTTP\r\nPkg.add(\"HDF5\")\r\nusing HDF5\r\nPkg.add(\"Statistics\")\r\nusing Statistics: norm\r\nPkg.add(\"Plots\")\r\nusing Plots", "_____no_output_____" ] ], [ [ "## Downloading GloVe", "_____no_output_____" ] ], [ [ "dir = HTTP.download(\"http://ann-benchmarks.com/glove-100-angular.hdf5\", update_period=60)\r\n\r\ndata = h5open(dir, \"r\") do file\r\n read(file);\r\nend", "_____no_output_____" ], [ "train = data[\"train\"]\r\nqueries = data[\"test\"]\r\ngroundtruth = data[\"neighbors\"].+1; #zero-indexed\r\ntrain = train ./ mapslices(norm, train, dims=1);\r\ntrain_backup = deepcopy(train);", "_____no_output_____" ] ], [ [ "### 100 Bits Graph", "_____no_output_____" ] ], [ [ "n_codebooks = 25\r\nn_centers = 16\r\nn_neighbors = 100\r\nstopcond=1e-1;", "_____no_output_____" ] ], [ [ "#### L2 loss", "_____no_output_____" ] ], [ [ "ahpq = builder(train; \r\n T=0,\r\n n_codebooks=n_codebooks, \r\n n_centers=n_centers,\r\n verbose=true,\r\n stopcond=stopcond,\r\n a=0,\r\n inverted_index=true,\r\n multithreading=false,\r\n training_points=25_000,\r\n increment_steps=3);", "_____no_output_____" ], [ "yhat_L2_100bits = MIPS(ahpq, queries, n_neighbors)\r\nL2_scores_1 = get1atNscores(yhat_L2_100bits, groundtruth, n_neighbors)", "_____no_output_____" ] ], [ [ "#### Anisotropic Loss", "_____no_output_____" ] ], [ [ "train = deepcopy(train_backup)\r\nahpq = builder(train; \r\n T=0.2,\r\n n_codebooks=n_codebooks, \r\n n_centers=n_centers,\r\n verbose=true,\r\n stopcond=stopcond,\r\n a=0,\r\n inverted_index=false,\r\n multithreading=true,\r\n training_points=250_000,\r\n increment_steps=3);", "_____no_output_____" ], [ "yhat_anisotropic_100bits = MIPS(ahpq, queries, n_neighbors)\r\nanisotropic_scores_1 = get1atNscores(yhat_anisotropic_100bits, groundtruth, n_neighbors);", "_____no_output_____" ] ], [ [ "#### Comparison", "_____no_output_____" ] ], [ [ "plot(1:100, anisotropic_scores_1, label=\"Anisotropic Loss\")\r\nplot!(1:100, L2_scores_1, label=\"Reconstruction Loss\")\r\nplot!(title=\"Recall of Glove-1.2M - 100 bits\", \r\n xlabel=\"N\",\r\n ylabel=\"Recall 1@N\",\r\n legend=:bottomright,\r\n xticks=0:20:100,\r\n yticks=0.1:0.1:0.9)", "_____no_output_____" ] ], [ [ "### 200 Bits Graph", "_____no_output_____" ] ], [ [ "n_codebooks = 50\r\nn_centers = 16\r\nn_neighbors = 100\r\nstopcond=1e-1;", "_____no_output_____" ] ], [ [ "#### L2 loss", "_____no_output_____" ] ], [ [ "train = deepcopy(train_backup)\r\nahpq = builder(train; \r\n T=0,\r\n n_codebooks=n_codebooks, \r\n n_centers=n_centers,\r\n verbose=true,\r\n stopcond=stopcond,\r\n a=0,\r\n inverted_index=true,\r\n multithreading=false,\r\n training_points=250_000,\r\n increment_steps=3);", "_____no_output_____" ], [ "yhat_L2_200bits = MIPS(ahpq, queries, n_neighbors)\r\nL2_scores_2 = get1atNscores(yhat_L2_200bits, groundtruth, n_neighbors);", "_____no_output_____" ] ], [ [ "#### Anisotropic Loss", "_____no_output_____" ] ], [ [ "train = deepcopy(train_backup)\r\nahpq = builder(train; \r\n T=0.2,\r\n n_codebooks=n_codebooks, \r\n n_centers=n_centers,\r\n verbose=true,\r\n stopcond=stopcond,\r\n a=0,\r\n inverted_index=true,\r\n multithreading=false,\r\n training_points=250_000,\r\n increment_steps=3);", "_____no_output_____" ], [ "yhat_anisotropic_200bits = MIPS(ahpq, queries, n_neighbors)\r\nanisotropic_scores_2 = get1atNscores(yhat_anisotropic_200bits, groundtruth, n_neighbors)", "_____no_output_____" ] ], [ [ "#### Comparison", "_____no_output_____" ] ], [ [ "plot(1:100, anisotropic_scores_2, label=\"Anisotropic Loss\")\r\nplot!(1:100, L2_scores_2, label=\"Reconstruction Loss\")\r\nplot!(title=\"Recall of Glove-1.2M - 200 bits\", \r\n xlabel=\"N\",\r\n ylabel=\"Recall 1@N\",\r\n legend=:bottomright,\r\n xticks=0:20:100,\r\n yticks=0.1:0.1:0.9)", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# <span style=\"color:orange\"> Exercise 11.2 </span>\n## <span style=\"color:green\"> Task </span>\n\nTry to extend the model to obtain a reasonable fit of the following polynomial of order 3:\n\n$$\nf(x)=4-3x-2x^2+3x^3\n$$\nfor $x \\in [-1,1]$.\n\nIn order to make practice with NN, explore reasonable different choices for:\n\n- the number of layers\n- the number of neurons in each layer\n- the activation function\n- the optimizer\n- the loss function\n \nMake graphs comparing fits for different NNs.\nCheck your NN models by seeing how well your fits predict newly generated test data (including on data outside the range you fit. How well do your NN do on points in the range of $x$ where you trained the model? How about points outside the original training data set? \nSummarize what you have learned about the relationship between model complexity (number of parameters), goodness of fit on training data, and the ability to predict well.", "_____no_output_____" ] ], [ [ "import numpy as np\nimport math\nfrom tensorflow import keras\nfrom matplotlib import pyplot as plt\n\n#function = 3x^3 - 2x^2 - 3x + 4\ndef polynomial(x_array,a,b,c,d):\n x_array = np.asfarray(x_array)\n return a*x_array**3 + b*x_array**2 + c*x_array + d\n\n#\nnp.random.seed(0)\nx_train = np.random.uniform(-1, 1, 1000) # dataset for training\nx_valid = np.random.uniform(-1, 1, 100) #dataset for testing/validation\nx_valid.sort()\n\na=3\nb=-2\nc=-3\nd=4\ny_target = polynomial(x_valid,a,b,c,d)\n\nsigma = 0.0 # noise standard deviation\ny_train = np.random.normal(polynomial(x_train,a,b,c,d), sigma) # actual measures from which we want to guess regression parameters\ny_valid = np.random.normal(polynomial(x_valid,a,b,c,d), sigma)\n\nplt.plot(x_valid, y_target)\nplt.scatter(x_valid, y_valid, color='r')\nplt.grid(True)\nplt.show()", "_____no_output_____" ] ], [ [ "### Using model from Ex11.1\nThis section utilizes the linear model used in Exercise 11.1, just to prove that it should not function and another model must be defined for polynomials.", "_____no_output_____" ] ], [ [ "# Using model from ex11.1\n\n# Load previous model for extension\noldmodel = keras.models.load_model('models/model_ex1')\noldmodel.summary()\nprint()\n\nhistory = oldmodel.fit(x=x_train, y=y_train, \n batch_size=32, epochs=100,\n shuffle=True, #\n validation_data=(x_valid, y_valid),\n verbose=0\n)\n\nscore = oldmodel.evaluate(x_valid, y_valid, batch_size=32, verbose=0)\n\n# print performance\nprint()\nprint('Test loss:', score[0])\nprint('Test accuracy:', score[1])\nprint()\nplt.plot(history.history['loss'])\nplt.plot(history.history['val_loss'])\nplt.title('Model loss')\nplt.ylabel('Loss')\nplt.xlabel('Epoch')\nplt.legend(['Train', 'Test'], loc='best')\nplt.show()\n\nx_predicted = np.random.uniform(-1, 1, 100)\ny_predicted = oldmodel.predict(x_predicted)\nplt.scatter(x_predicted, y_predicted,color='r')\nplt.plot(x_valid, y_target)\nplt.grid(True)\nplt.show()", "Model: \"sequential_184\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_184 (Dense) (None, 1) 2 \n=================================================================\nTotal params: 2\nTrainable params: 2\nNon-trainable params: 0\n_________________________________________________________________\n\n\nTest loss: 0.8131828904151917\nTest accuracy: 0.8131828904151917\n\n" ] ], [ [ "### New model\n\n", "_____no_output_____" ] ], [ [ "from tensorflow.keras import models\nfrom tensorflow.keras import layers\nfrom tensorflow.keras import optimizers\nfrom tensorflow.keras import backend as K\nfrom tensorflow.keras import callbacks\nfrom tensorflow.keras import losses\nfrom tensorflow.keras import activations\nfrom tensorflow.keras.utils import get_custom_objects, plot_model \n\ndef run(layers:list,optimizer='sgd',loss='mse',batch_size=32,epochs=60,show_summary=True,outputs=False,testing=True,graph=True,logger=True):\n global x_train, y_train, x_valid, y_valid, y_target\n model = models.Sequential(layers)\n model.compile(optimizer=optimizer, loss=loss, metrics=['mse'])\n optname = optimizer if isinstance(optimizer,str) else optimizer.__class__.__name__\n lossname = loss if isinstance(loss,str) else loss.__class__.__name__\n if logger:\n print(f\"************** {len(layers)} Layers\\t{layers[0].output_shape[1]} input neurons\\t optimizer={optname}\\tloss={lossname}\")\n if show_summary:\n model.summary()\n \n history = model.fit(x=x_train, y=y_train, \n batch_size=batch_size, epochs=epochs,\n shuffle=True, # a good idea is to shuffle input before at each epoch\n validation_data=(x_valid, y_valid),\n verbose=0\n )\n\n score = model.evaluate(x_valid, y_valid, batch_size=batch_size, verbose=0)\n if outputs:\n print()\n print(\"Validation performance\")\n print('Test loss:', score[0])\n print('Test accuracy:', score[1])\n\n score = model.evaluate(x_valid, y_target, batch_size=batch_size, verbose=0)\n if outputs:\n print()\n print(\"Testing performance\")\n print('Test loss:', score[0])\n print('Test accuracy:', score[1])\n \n if testing:\n plt.plot(history.history['loss'])\n plt.plot(history.history['val_loss'])\n plt.title('Model loss')\n plt.ylabel('Loss')\n plt.xlabel('Epoch')\n plt.legend(['Train', 'Test'], loc='best')\n plt.show()\n\n if graph:\n fig=plt.figure(figsize=(10, 5))\n x_predicted = np.random.uniform(-1.5, 1.5, 100)\n x_predicted.sort()\n y_predicted = model.predict(x_predicted)\n plt.scatter(x_predicted, y_predicted,color='r')\n plt.plot(x_predicted, polynomial(x_predicted,a,b,c,d))\n plt.title(f\"{len(layers)} layers with {layers[0].output_shape[1]} input neurons - opt.= {optname}, loss = {lossname} \")\n plt.grid(True)\n plt.tight_layout()\n plt.show()\n \n if logger:\n print(\"\\n\\n\")", "_____no_output_____" ] ], [ [ "## <span style=\"color:green\">Dependance on layers & neurons </span>\nUsing the code above, various NNs have been trained with different layers and neurons for each layer, to study the accuracy in approximating the polynomial in $[\\frac{-3}{2},\\frac{3}{2}]$\n\n### Different Neural Networks\n", "_____no_output_____" ] ], [ [ "opt = optimizers.SGD(learning_rate=0.1)\nrun([layers.Dense(500,input_shape=(1,)),layers.Dense(1,activation=\"relu\")],optimizer=opt,testing=False)\nrun([layers.Dense(1,input_shape=(1,)),layers.Dense(30,activation=\"relu\"),layers.Dense(1,activation=\"relu\")],optimizer=opt,testing=False)\nrun([layers.Dense(500,input_shape=(1,)),layers.Dense(100,activation=\"relu\"),layers.Dense(1,activation=\"relu\")],optimizer=opt,testing=False,outputs=True)\nrun([layers.Dense(500,input_shape=(1,)),layers.Dense(250,activation=\"relu\"),layers.Dense(100,activation=\"relu\"),layers.Dense(10,activation=\"relu\"),layers.Dense(1,activation=\"relu\")],optimizer=opt,testing=False)\nrun([layers.Dense(1000,input_shape=(1,)),layers.Dense(500,activation=\"relu\"),layers.Dense(250,activation=\"relu\"),layers.Dense(175,activation=\"relu\"),layers.Dense(100,activation=\"relu\"),layers.Dense(50,activation=\"relu\"),layers.Dense(1,activation=\"relu\")],testing=False)\n", "************** 2 Layers\t500 input neurons\t optimizer=SGD\tloss=mse\nModel: \"sequential_1\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_3 (Dense) (None, 500) 1000 \n_________________________________________________________________\ndense_4 (Dense) (None, 1) 501 \n=================================================================\nTotal params: 1,501\nTrainable params: 1,501\nNon-trainable params: 0\n_________________________________________________________________\n" ] ], [ [ "### Results\nDiscuss results on layers/neurons\n\n<br>", "_____no_output_____" ], [ "## <span style=\"color:green\"> <strong> <u> Dependance on optimizers </u> </strong> </span>\nThe following section defines different optimizer functions for a Neural Network with 3 layers and 500 input neurons.\n\nWhich optimizers to chose<br>\nExpectations<br>\n\n### <u> <strong> Adam Optimizer </strong> </u>\nAdam, short for adaptive moment estimation, is an optimization algorithm that can be used instead of the classical stochastic gradient descent procedure to update network weights iteratively based in training data.\nBetween the numerous advantages brought by this algorithm, the most important for this case are:\n- Computationally efficient\n- Straightforward to implement\n- Appropriate for problems with very noisy/or sparse gradients\n\n\n#### Different optimizers", "_____no_output_____" ] ], [ [ "opt_layers = [\n layers.Dense(1,input_shape=(1,)),\n layers.Dense(50,activation=\"relu\"),\n layers.Dense(1,activation=\"selu\")\n]\nopts = [\n optimizers.SGD(learning_rate=1e-1),\n optimizers.Adam(learning_rate=1e-1),\n optimizers.RMSprop(learning_rate=1e-1),\n optimizers.Adagrad(learning_rate=1e-1)\n]\nfor oo in opts:\n run(opt_layers,optimizer=oo,testing=False,batch_size=64,epochs=150,outputs=True)\nprint()", "************** 3 Layers\t1 input neurons\t optimizer=SGD\tloss=mse\nModel: \"sequential_6\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_23 (Dense) (None, 1) 2 \n_________________________________________________________________\ndense_24 (Dense) (None, 50) 100 \n_________________________________________________________________\ndense_25 (Dense) (None, 1) 51 \n=================================================================\nTotal params: 153\nTrainable params: 153\nNon-trainable params: 0\n_________________________________________________________________\n\nValidation performance\nTest loss: 0.004281164612621069\nTest accuracy: 0.004281164612621069\n\nTesting performance\nTest loss: 0.004281164612621069\nTest accuracy: 0.004281164612621069\n" ] ], [ [ "#### Learning rate\nAs with all optimization algorithms, it is defined by numerous parameters, including the learning rate, which determines the step size at each iteration while moving toward a minimum of a loss function.\nIn the following section, I will study the impact of the learning rate parameter on the learning efficiency of the model.\nAt first, I chose to explore three different values for the learning rate:<br>\n<ul>\n <li> lr = 0.001 (default value for Adam optimizer) </li>\n <li> lr = 0.1</li>\n <li> lr = 0.00001</li>\n</ul>\nin order to choose values that could be too high, too low or decent.\nThen I will use a learning rate schedule called \"Step decay\" which systematically drops the learning rate at specific times during training, formally defined by $$LR = LR_0 * \\text{droprate}^{\\text{floor}(\\text{epoch} / \\text{epochs_drop})}$$", "_____no_output_____" ] ], [ [ "#--- LR = 0.001\n\nlr = 0.001\nadam1 = optimizers.Adam(\n learning_rate=lr,\n beta_1=0.9,\n beta_2=0.999,\n epsilon=1e-07,\n amsgrad=False,\n name=\"Adam2\"\n)\nprint(\"---> Learning rate: \",lr)\n\nrun(\n [\n layers.Dense(500,input_shape=(1,)),\n layers.Dense(100, activation=\"relu\"),\n layers.Dense(1)\n ],\n optimizer=adam1\n)", "---> Learning rate: 0.001\n************** 3 Layers\t500 input neurons\t optimizer=Adam\tloss=mse\nModel: \"sequential_10\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_26 (Dense) (None, 500) 1000 \n_________________________________________________________________\ndense_27 (Dense) (None, 100) 50100 \n_________________________________________________________________\ndense_28 (Dense) (None, 1) 101 \n=================================================================\nTotal params: 51,201\nTrainable params: 51,201\nNon-trainable params: 0\n_________________________________________________________________\n" ], [ "#--- LR = 0.1\n\nlr = 0.1\nadam1 = optimizers.Adam(\n learning_rate=lr,\n beta_1=0.9,\n beta_2=0.999,\n epsilon=1e-07,\n amsgrad=False,\n name=\"Adam2\"\n)\nprint(\"---> Learning rate: \",lr)\nrun(\n [\n layers.Dense(500,input_shape=(1,)),\n layers.Dense(100, activation=\"relu\"),\n layers.Dense(1)\n ],\n optimizer=adam1\n)", "---> Learning rate: 0.1\n************** 3 Layers\t500 input neurons\t optimizer=Adam\tloss=mse\nModel: \"sequential_11\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_29 (Dense) (None, 500) 1000 \n_________________________________________________________________\ndense_30 (Dense) (None, 100) 50100 \n_________________________________________________________________\ndense_31 (Dense) (None, 1) 101 \n=================================================================\nTotal params: 51,201\nTrainable params: 51,201\nNon-trainable params: 0\n_________________________________________________________________\n" ], [ "lr = 0.00001\nadam1 = optimizers.Adam(\n learning_rate=lr,\n beta_1=0.9,\n beta_2=0.999,\n epsilon=1e-07,\n amsgrad=False,\n name=\"Adam2\"\n)\nprint(\"---> Learning rate: \",lr)\nrun(\n [\n layers.Dense(500,input_shape=(1,)),\n layers.Dense(100, activation=\"relu\"),\n layers.Dense(1)\n ],\n optimizer=adam1\n)", "---> Learning rate: 1e-05\n************** 3 Layers\t500 input neurons\t optimizer=Adam\tloss=mse\nModel: \"sequential_12\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_32 (Dense) (None, 500) 1000 \n_________________________________________________________________\ndense_33 (Dense) (None, 100) 50100 \n_________________________________________________________________\ndense_34 (Dense) (None, 1) 101 \n=================================================================\nTotal params: 51,201\nTrainable params: 51,201\nNon-trainable params: 0\n_________________________________________________________________\n" ], [ "### Step decay\nimport math\ninitial_learning_rate = 0.01\ndef lr_step_decay(epoch, lr):\n drop_rate = 0.55\n epochs_drop = 10.5\n return initial_learning_rate * math.pow(drop_rate, math.floor(epoch/epochs_drop))\n\n\nmodellayers = [\n layers.Dense(500, input_shape=(1,)),\n layers.Dense(100, activation=\"relu\"),\n layers.Dense(1)\n]\nstepmodel = models.Sequential(modellayers)\nstepmodel.compile(optimizer=\"adam\", loss=\"mse\", metrics=['mse'])\nstepmodel.summary()\n\n# Fit the model to the training data\nhistory_step_decay = stepmodel.fit(\n x_train, \n y_train, \n epochs=100, \n validation_split=0.3,\n batch_size=64,\n callbacks=[callbacks.LearningRateScheduler(lr_step_decay, verbose=0)],\n verbose=0\n)\n\nscore = stepmodel.evaluate(x_valid, y_valid, batch_size=64, verbose=0)\nprint()\nprint(\"Validation performance\")\nprint('Test loss:', score[0])\nprint('Test accuracy:', score[1])\n\nscore = stepmodel.evaluate(x_valid, y_target, batch_size=64, verbose=0)\nprint()\nprint(\"Testing performance\")\nprint('Test loss:', score[0])\nprint('Test accuracy:', score[1])\n\nplt.plot(history.history['loss'])\nplt.plot(history.history['val_loss'])\nplt.title('Model loss')\nplt.ylabel('Loss')\nplt.xlabel('Epoch')\nplt.legend(['Train', 'Test'], loc='best')\nplt.show()\n\nfig=plt.figure(figsize=(10, 5))\nx_predicted = np.random.uniform(-1.5, 1.5, 100)\nx_predicted.sort()\ny_predicted = stepmodel.predict(x_predicted)\nplt.scatter(x_predicted, y_predicted,color='r')\nplt.plot(x_predicted, polynomial(x_predicted,a,b,c,d))\nplt.title(f\"{len(modellayers)} layers with 500 input neurons, activ. fun.= 'relu', opt.= adam, loss = mse \")\nplt.grid(True)\nplt.tight_layout()\nplt.show()\nprint(\"\\n Final learning rate: \",K.eval(stepmodel.optimizer.lr))", "Model: \"sequential_13\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_35 (Dense) (None, 500) 1000 \n_________________________________________________________________\ndense_36 (Dense) (None, 100) 50100 \n_________________________________________________________________\ndense_37 (Dense) (None, 1) 101 \n=================================================================\nTotal params: 51,201\nTrainable params: 51,201\nNon-trainable params: 0\n_________________________________________________________________\n\nValidation performance\nTest loss: 0.012978918850421906\nTest accuracy: 0.012978918850421906\n\nTesting performance\nTest loss: 0.012978918850421906\nTest accuracy: 0.012978918850421906\n" ] ], [ [ "The choice of the value for learning rate can impact two things:\n<ol>\n <li>How fast the algorithm learns</li>\n <li>Whether the cost function is minimized or not</li>\n</ol>\nFor an optimal value of the learning rate, the cost function value is minimized in a few iterations. If your learning rate is too low, training will progress very slowly as the fine-tunings to the weights in the network are small, and the number of iterations/epochs required to minimize the cost function is high.<br>\nIf the learning rate is set too high, the cost function could saturate at a value higher than the minimum value, causing undesirable divergent behavior in the loss function.<br>\n", "_____no_output_____" ], [ "## <span style=\"color:green\">Dependance on loss function </span>\nIn this section, I will try to explore the dependance of the model on the type of loss function.\nFor this purpose, I will keep all the other parameters as fixed as possible to highlight as much as possible the quested dependancy.<br>\nThe parameters are:\n- A neural network with 3 layers, respectively with 500, 100 and 1 neuron, all following the ReLU activation function\n- 100 epochs\n- A batch size of 64\n- SGD optimizer\n\nThe chosen loss functions are:\n- Mean squared error (Regression Loss)\nThe average squared difference between the estimated values and the actual value: $\\text{MSE}=\\frac{1}{N}\\sum_{i=1}^N(Y_i - <Y_i>)^2$\n- Mean absolute error (Regression Loss)\nThe measure of errors between paired observations expressing the same phenomenon, including observed versus predicted.\n$\\text{MAE}=\\frac{1}{N}\\sum_{i=1}^N |Y_i - X_i|$\n- Cross-entropy (Probabilistic Loss)\nIt measures the performance of a classification model whose output is a probability value between 0 and 1\n- Poisson", "_____no_output_____" ] ], [ [ "loss_functions = [\n losses.MeanSquaredError(),\n losses.MeanAbsoluteError(reduction=\"auto\", name=\"mean_absolute_error\"),\n losses.CategoricalCrossentropy(reduction=\"auto\",name=\"categorical_crossentropy\"),\n losses.Poisson(reduction=\"auto\", name=\"poisson\")\n]\nopt = optimizers.SGD(learning_rate=0.01)\nll = [\n layers.Dense(500,input_shape=(1,)),\n layers.Dense(250, activation=\"relu\"),\n layers.Dense(1)\n]\n\nfor func in loss_functions:\n run(ll,optimizer=opt,loss=func,testing=False,outputs=True,batch_size=64,epochs=100)", "************** 3 Layers\t500 input neurons\t optimizer=SGD\tloss=MeanSquaredError\nModel: \"sequential_14\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_38 (Dense) (None, 500) 1000 \n_________________________________________________________________\ndense_39 (Dense) (None, 250) 125250 \n_________________________________________________________________\ndense_40 (Dense) (None, 1) 251 \n=================================================================\nTotal params: 126,501\nTrainable params: 126,501\nNon-trainable params: 0\n_________________________________________________________________\n\nValidation performance\nTest loss: 0.040039416402578354\nTest accuracy: 0.040039416402578354\n\nTesting performance\nTest loss: 0.040039416402578354\nTest accuracy: 0.040039416402578354\n" ] ], [ [ "### Results\n...", "_____no_output_____" ], [ "## <span style=\"color:green\">Dependance on activation function </span>\nLastly, I reported a study on the dependance of the model on the activation function of the neurons. The structure of the Neural Network is defined by 4 layers, respectively with 500, 250, 100 and 1 neurons.<br>\nThe chosen activation functions are:\n...\n\n", "_____no_output_____" ] ], [ [ "lossfunc = losses.MeanSquaredError()\nopt = optimizers.SGD(learning_rate=0.01)\nall_layers = [[\n layers.Dense(500, input_shape=(1,)),\n layers.Dense(250, activation=\"relu\"),\n layers.Dense(100, activation=\"relu\"),\n layers.Dense(1, activation=\"relu\")\n],[\n layers.Dense(500, input_shape=(1,)),\n layers.Dense(250, activation=\"relu\"),\n layers.Dense(100, activation=\"relu\"),\n layers.Dense(1, activation=\"sigmoid\")\n],[\n layers.Dense(500, input_shape=(1,)),\n layers.Dense(250, activation=\"selu\"),\n layers.Dense(100, activation=\"selu\"),\n layers.Dense(1, activation=\"softmax\")\n]\n]\n\nfor l in all_layers:\n run(l,optimizer=opt,loss=lossfunc,testing=False,outputs=True,batch_size=64,epochs=100)", "************** 4 Layers\t500 input neurons\t optimizer=SGD\tloss=MeanSquaredError\nModel: \"sequential_18\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_41 (Dense) (None, 500) 1000 \n_________________________________________________________________\ndense_42 (Dense) (None, 250) 125250 \n_________________________________________________________________\ndense_43 (Dense) (None, 100) 25100 \n_________________________________________________________________\ndense_44 (Dense) (None, 1) 101 \n=================================================================\nTotal params: 151,451\nTrainable params: 151,451\nNon-trainable params: 0\n_________________________________________________________________\n\nValidation performance\nTest loss: 0.009869810193777084\nTest accuracy: 0.009869810193777084\n\nTesting performance\nTest loss: 0.009869810193777084\nTest accuracy: 0.009869810193777084\n" ], [ "from IPython.display import clear_output\n\nclass PlotCurrentEstimate(callbacks.Callback):\n def __init__(self, x_valid, y_valid):\n \"\"\"Keras Callback which plot current model estimate against reference target\"\"\"\n \n # convert numpy arrays into lists for plotting purposes\n self.x_valid = list(x_valid[:])\n self.y_valid = list(y_valid[:])\n self.iter=0\n\n def on_epoch_end(self, epoch, logs={}):\n \n temp = self.model.predict(self.x_valid, batch_size=None, verbose=False, steps=None)\n self.y_curr = list(temp[:]) # convert numpy array into list\n \n self.iter+=1\n if self.iter%10 == 0:\n clear_output(wait=True) \n self.eplot = plt.subplot(1,1,1)\n self.eplot.clear() \n self.eplot.scatter(self.x_valid, self.y_curr, color=\"blue\", s=4, marker=\"o\", label=\"estimate\")\n self.eplot.scatter(self.x_valid, self.y_valid, color=\"red\", s=4, marker=\"x\", label=\"valid\")\n self.eplot.legend()\n\n plt.show()", "_____no_output_____" ], [ "np.random.seed(0)\nfinalx_train = np.random.uniform(-1, 1, 10000) # dataset for training\nfinalx_valid = np.random.uniform(-1, 1, 1000) #dataset for testing/validation\nfinalx_valid.sort()\n\nfinaly_target = polynomial(finalx_valid,a,b,c,d)\n\nsigma = 0.0 # noise standard deviation\nfinaly_train = np.random.normal(polynomial(finalx_train,a,b,c,d), sigma) # actual measures from which we want to guess regression parameters\nfinaly_valid = np.random.normal(polynomial(finalx_valid,a,b,c,d), sigma)\n\nfinalmodel = models.Sequential()\nfinalmodel.add(layers.Dense(units=1, input_dim=1))\nfinalmodel.add(layers.Activation('relu'))\nfinalmodel.add(layers.Dense(units=40))\nfinalmodel.add(layers.Activation('relu'))\nfinalmodel.add(layers.Dense(units=1))\nfinalmodel.compile(loss='mean_squared_error',optimizer='adam', metrics=['mse'])\nfinalmodel.summary()\n\nhistory = finalmodel.fit(x=finalx_train, y=finaly_train, \n batch_size=64, epochs=150,\n shuffle=True, # a good idea is to shuffle input before at each epoch\n validation_data=(finalx_valid, finaly_valid),\n verbose=0\n)\n\nscore = finalmodel.evaluate(finalx_valid, finaly_valid, batch_size=64, verbose=0)\nprint()\nprint(\"Validation performance\")\nprint('Test loss:', score[0])\nprint('Test accuracy:', score[1])\n\nscore = finalmodel.evaluate(finalx_valid, finaly_target, batch_size=64, verbose=0)\nprint()\nprint(\"Testing performance\")\nprint('Test loss:', score[0])\nprint('Test accuracy:', score[1])\n\nplt.plot(history.history['loss'][20:])\nplt.plot(history.history['val_loss'][20:])\nplt.title('Model loss')\nplt.ylabel('Loss')\nplt.xlabel('Epoch')\nplt.legend(['Train', 'Test'], loc='best')\nplt.show()\n\nfig=plt.figure(figsize=(10, 5))\nfinalx_predicted = np.random.uniform(-1.5, 1.5, 1000)\nfinalx_predicted.sort()\nfinaly_predicted = finalmodel.predict(finalx_predicted)\nplt.scatter(finalx_predicted, finaly_predicted,color='r')\nplt.plot(finalx_predicted, polynomial(finalx_predicted,a,b,c,d))\nplt.title(f\"{len(finalmodel.layers)} layers with {finalmodel.layers[0].output_shape[1]} input neurons - opt.= {opt.__class__.__name__}, loss = {lossfunc.__class__.__name__} \")\nplt.grid(True)\nplt.tight_layout()\nplt.show()\n\nfinalmodel.save(\"models/model_ex2\")", "Model: \"sequential_22\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense_56 (Dense) (None, 1) 2 \n_________________________________________________________________\nactivation_4 (Activation) (None, 1) 0 \n_________________________________________________________________\ndense_57 (Dense) (None, 40) 80 \n_________________________________________________________________\nactivation_5 (Activation) (None, 40) 0 \n_________________________________________________________________\ndense_58 (Dense) (None, 1) 41 \n=================================================================\nTotal params: 123\nTrainable params: 123\nNon-trainable params: 0\n_________________________________________________________________\n\nValidation performance\nTest loss: 0.022854315117001534\nTest accuracy: 0.022854315117001534\n\nTesting performance\nTest loss: 0.022854315117001534\nTest accuracy: 0.022854315117001534\n" ], [ "plot_estimate = PlotCurrentEstimate(finalx_valid, finaly_valid)\n\nearlystop = tf.keras.callbacks.EarlyStopping(monitor='val_loss',\n min_delta=0, patience=100, mode='auto')\n\nmodel.fit(finalx_valid, finaly_valid, batch_size=32, epochs=150,\n validation_data=(finalx_valid, finalx_valid),\n callbacks=[ plot_estimate, earlystop]\n )\n\nmodel.get_weights()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/khloemaritonigloria08/CPEN-21A-ECE-2-1/blob/main/Expressions_and_Operations.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "###Boolean Operator", "_____no_output_____" ] ], [ [ "x=1\ny=2\n\nprint(x>y)\nprint(10>11)\nprint(10==10)\nprint(10!=11)", "False\nFalse\nTrue\nTrue\n" ], [ "#using bool()function\n\nprint(bool(\"Hello\"))\nprint(bool(15))\nprint(bool(1))\n\nprint(bool(True))\nprint(bool(False))\nprint(bool(None))\nprint(bool(0))\nprint(bool([]))", "True\nTrue\nTrue\nTrue\nFalse\nFalse\nFalse\nFalse\n" ] ], [ [ "##Functions can return Boolean", "_____no_output_____" ] ], [ [ "def myFunction():return False\nprint(myFunction())", "False\n" ], [ "def yourFunction():return False\nif yourFunction():\n print(\"Yes!\")\nelse:\n print(\"No\")", "No\n" ] ], [ [ "### You Try!", "_____no_output_____" ] ], [ [ "a=6\nb=7\nprint(a==b)\nprint(a!=a)", "False\nFalse\n" ] ], [ [ "##Arithmetic Operators", "_____no_output_____" ] ], [ [ "print(10+5)\nprint(10-5)\nprint(10*5)\nprint(10/5)\nprint(10%5) #modulo division, remainder\nprint(10//5) #floor division\nprint(10//3) #floor division\nprint(10%3) #3x3=9+1", "15\n5\n50\n2.0\n0\n2\n3\n1\n" ] ], [ [ "##Bitwise Operators", "_____no_output_____" ] ], [ [ "a=60 #0011 1100\nb=13 #0000 1101\n\nprint(a&b)\nprint(a|b)\nprint(a^b)\nprint(~a)\nprint(a<<1) #0111 1000\nprint(a<<2) #1111 0000\nprint(b>>1) #1 0000 0110\nprint(b>>2) #0000 0110 carry flag bit = 01", "12\n61\n49\n-61\n120\n240\n6\n3\n" ] ], [ [ "##Python Assignment Operators", "_____no_output_____" ] ], [ [ "a+=3 #Same As a=a+3\n #Same As a=60+3, a=63\nprint(a)", "63\n" ] ], [ [ "###Logical Operators", "_____no_output_____" ] ], [ [ "#and Logical Operator\na=True\nb=False\n\nprint(a and b)\nprint(not(a and b))\nprint(a or b)\nprint(not(a or b))", "False\nTrue\nTrue\nFalse\n" ], [ "print(a is b)\nprint(a is not b)", "False\nTrue\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Searching For Simple Patterns\n\nBeing able to match letters and metacharacters is the simplest task that regular expressions can do. In this section we will see how we can use regular expressions to perform more complex pattern matching. We can form any pattern we want by using the metacharacters mentioned in the previous lesson.\n\nThe first metacharacter we are going to look at is the backslash (`\\`). We already saw that the backslash can be used to escape all the metacharacters, so that you can search for them directly. However, the backslash can also be followed by various characters to signal various special sequences. Here is a list of the special sequences we are going to look at in this notebook:\n\n* `\\d` - Matches any decimal digit; this is equivalent to the set [0-9]\n\n\n* `\\D` - Matches any non-digit character; this is equivalent to the set [^0-9]\n\n\n* `\\s` - Matches any whitespace character, this is equivalent to the set [ \\t\\n\\r\\f\\v]\n\n\n* `\\S` - Matches any non-whitespace character; this is equivalent to the set [^ \\t\\n\\r\\f\\v]\n\n\n* `\\w` - Matches any alphanumeric character and the underscore; this is equivalent to the set [a-zA-Z0-9_]\n\n\n* `\\W` - Matches any non-alphanumeric character; this is equivalent to the set [^a-zA-Z0-9_]\n\nWe can see that there is a difference between lowercase and uppercase sequences. For example, while `\\d` matches any digit, `\\D` matches everything that is **not** a digit. Similarly, while `\\s` matches any whitespace character, `\\S` matches everything that is **not** a whitespace character; and while `\\w` matches any alphanumeric character, `\\W` matches everything that is **not** an alphanumeric character.\n\nLet's start by learning how to use `\\d` to search for decimal digits.", "_____no_output_____" ], [ "### Matching Numbers Using `\\d`\n\nIn the code below, we will use `'\\d'` as our regular expression to find all the decimal digits in our `sample_text` string:", "_____no_output_____" ] ], [ [ "# Import re module\nimport re\n\n# Sample text\nsample_text = 'Alice lives in 1230 First St., Ocean City, MD 156789.'\n\n# Create a regular expression object with the regular expression '\\d'\nregex = re.compile(r'\\d')\n\n# Search the sample_text for the regular expression\nmatches = regex.finditer(sample_text)\n\n# Print all the matches\nfor match in matches:\n print(match)", "<_sre.SRE_Match object; span=(15, 16), match='1'>\n<_sre.SRE_Match object; span=(16, 17), match='2'>\n<_sre.SRE_Match object; span=(17, 18), match='3'>\n<_sre.SRE_Match object; span=(18, 19), match='0'>\n<_sre.SRE_Match object; span=(46, 47), match='1'>\n<_sre.SRE_Match object; span=(47, 48), match='5'>\n<_sre.SRE_Match object; span=(48, 49), match='6'>\n<_sre.SRE_Match object; span=(49, 50), match='7'>\n<_sre.SRE_Match object; span=(50, 51), match='8'>\n<_sre.SRE_Match object; span=(51, 52), match='9'>\n" ] ], [ [ "As we can see, all the matches found above correspond to only decimal digits between 0 and 9.\n\nConversely, if wanted to find all the characters that are **not** decimal digits, we will use `\\D` as our regular expression, as shown below:", "_____no_output_____" ] ], [ [ "# Import re module\nimport re\n\n# Sample text\nsample_text = 'Alice lives in 1230 First St., Ocean City, MD 156789.'\n\n# Create a regular expression object with the regular expression '\\D'\nregex = re.compile(r'\\D')\n\n# Search the sample_text for the regular expression\nmatches = regex.finditer(sample_text)\n\n# Print all the matches\nfor match in matches:\n print(match)", "<_sre.SRE_Match object; span=(0, 1), match='A'>\n<_sre.SRE_Match object; span=(1, 2), match='l'>\n<_sre.SRE_Match object; span=(2, 3), match='i'>\n<_sre.SRE_Match object; span=(3, 4), match='c'>\n<_sre.SRE_Match object; span=(4, 5), match='e'>\n<_sre.SRE_Match object; span=(5, 6), match=' '>\n<_sre.SRE_Match object; span=(6, 7), match='l'>\n<_sre.SRE_Match object; span=(7, 8), match='i'>\n<_sre.SRE_Match object; span=(8, 9), match='v'>\n<_sre.SRE_Match object; span=(9, 10), match='e'>\n<_sre.SRE_Match object; span=(10, 11), match='s'>\n<_sre.SRE_Match object; span=(11, 12), match=' '>\n<_sre.SRE_Match object; span=(12, 13), match='i'>\n<_sre.SRE_Match object; span=(13, 14), match='n'>\n<_sre.SRE_Match object; span=(14, 15), match=' '>\n<_sre.SRE_Match object; span=(19, 20), match=' '>\n<_sre.SRE_Match object; span=(20, 21), match='F'>\n<_sre.SRE_Match object; span=(21, 22), match='i'>\n<_sre.SRE_Match object; span=(22, 23), match='r'>\n<_sre.SRE_Match object; span=(23, 24), match='s'>\n<_sre.SRE_Match object; span=(24, 25), match='t'>\n<_sre.SRE_Match object; span=(25, 26), match=' '>\n<_sre.SRE_Match object; span=(26, 27), match='S'>\n<_sre.SRE_Match object; span=(27, 28), match='t'>\n<_sre.SRE_Match object; span=(28, 29), match='.'>\n<_sre.SRE_Match object; span=(29, 30), match=','>\n<_sre.SRE_Match object; span=(30, 31), match=' '>\n<_sre.SRE_Match object; span=(31, 32), match='O'>\n<_sre.SRE_Match object; span=(32, 33), match='c'>\n<_sre.SRE_Match object; span=(33, 34), match='e'>\n<_sre.SRE_Match object; span=(34, 35), match='a'>\n<_sre.SRE_Match object; span=(35, 36), match='n'>\n<_sre.SRE_Match object; span=(36, 37), match=' '>\n<_sre.SRE_Match object; span=(37, 38), match='C'>\n<_sre.SRE_Match object; span=(38, 39), match='i'>\n<_sre.SRE_Match object; span=(39, 40), match='t'>\n<_sre.SRE_Match object; span=(40, 41), match='y'>\n<_sre.SRE_Match object; span=(41, 42), match=','>\n<_sre.SRE_Match object; span=(42, 43), match=' '>\n<_sre.SRE_Match object; span=(43, 44), match='M'>\n<_sre.SRE_Match object; span=(44, 45), match='D'>\n<_sre.SRE_Match object; span=(45, 46), match=' '>\n<_sre.SRE_Match object; span=(52, 53), match='.'>\n" ] ], [ [ "We can see that none of the matches are decimal digits. We also see, that by using `\\D` we were able to match all characters, including periods (`.`) and white spaces.", "_____no_output_____" ], [ "# TODO: Find IP Addresses\n\nIn the cell below, our `sample_text` string contains three IP addresses. Write a single regular expression that can match any IP address and save the regular expression object in a variable called `regex`. Then use the `.finditer()` method to search the `sample_text` string for the given regular expression. Finally, write a loop to print all the `matches` found by the `.finditer()` method.\n\n**HINT :** Use the special sequence `\\d` and take advantage that all IP addresses have the same pattern.", "_____no_output_____" ] ], [ [ "# Import re module\nimport re\n\n# Sample text\nsample_text = 'Here are three IP address: 123.456.789.123, 999.888.777.666, 111.222.333.444'\n\n# Create a regular expression object with the regular expression\nregex = re.compile(r'\\d\\d\\d.\\d\\d\\d.\\d\\d\\d.\\d\\d\\d')\n\n# Search the sample_text for the regular expression\nmatches = regex.finditer(sample_text)\n\n# Print all the matches\nfor match in matches:\n print(match)", "<_sre.SRE_Match object; span=(27, 42), match='123.456.789.123'>\n<_sre.SRE_Match object; span=(44, 59), match='999.888.777.666'>\n<_sre.SRE_Match object; span=(61, 76), match='111.222.333.444'>\n" ] ], [ [ "If you wrote your regex correctly you should see three matches above corresponding to the three IP addresses in our `sample_text` string.", "_____no_output_____" ], [ "### Matching Whitespace Characters Using `\\s`\n\nIn the code below, we will use `\\s` as our regular expression to find all the whitespace characters in our `sample_text` string. For this example, we will use a string literal that spans multiple lines. To create this multi-line string, we will use triple-quotes (`'''`) both at the beginning and at the end of the multi-line string.", "_____no_output_____" ] ], [ [ "# Import re module\nimport re\n\n# Sample text\nsample_text = '''\n\\tAlice lives in:\\f\n1230 First St.\\r\nOcean City, MD 156789.\\v\n'''\n\n# Create a regular expression object with the regular expression '\\s'\nregex = re.compile(r'\\s')\n\n# Search the sample_text for the regular expression\nmatches = regex.finditer(sample_text)\n\n# Print all the matches\nfor match in matches:\n print(match)", "<_sre.SRE_Match object; span=(0, 1), match='\\n'>\n<_sre.SRE_Match object; span=(1, 2), match='\\t'>\n<_sre.SRE_Match object; span=(7, 8), match=' '>\n<_sre.SRE_Match object; span=(13, 14), match=' '>\n<_sre.SRE_Match object; span=(17, 18), match='\\x0c'>\n<_sre.SRE_Match object; span=(18, 19), match='\\n'>\n<_sre.SRE_Match object; span=(23, 24), match=' '>\n<_sre.SRE_Match object; span=(29, 30), match=' '>\n<_sre.SRE_Match object; span=(33, 34), match='\\r'>\n<_sre.SRE_Match object; span=(34, 35), match='\\n'>\n<_sre.SRE_Match object; span=(40, 41), match=' '>\n<_sre.SRE_Match object; span=(46, 47), match=' '>\n<_sre.SRE_Match object; span=(49, 50), match=' '>\n<_sre.SRE_Match object; span=(57, 58), match='\\x0b'>\n<_sre.SRE_Match object; span=(58, 59), match='\\n'>\n" ] ], [ [ "As we can see, all the matches found correspond to white spaces, tabs (`\\t`), newlines (`\\n`), carriage returns (`\\r`), form feeds (`\\f`), and vertical tabs (`\\v`). Notice that form feeds appear as `\\x0c` and vertical tabs as `\\x0b`. \n\nConversely, if wanted to find all the characters that are **not** whitespace characters, we will use `\\S` as our regular expression, as shown below:", "_____no_output_____" ] ], [ [ "# Import re module\nimport re\n\n# Sample text\nsample_text = '''\n\\tAlice lives in:\\f\n1230 First St.\\r\nOcean City, MD 156789.\\v\n'''\n\n# Create a regular expression object with the regular expression '\\S'\nregex = re.compile(r'\\S')\n\n# Search the sample_text for the regular expression\nmatches = regex.finditer(sample_text)\n\n# Print all the matches\nfor match in matches:\n print(match)", "<_sre.SRE_Match object; span=(2, 3), match='A'>\n<_sre.SRE_Match object; span=(3, 4), match='l'>\n<_sre.SRE_Match object; span=(4, 5), match='i'>\n<_sre.SRE_Match object; span=(5, 6), match='c'>\n<_sre.SRE_Match object; span=(6, 7), match='e'>\n<_sre.SRE_Match object; span=(8, 9), match='l'>\n<_sre.SRE_Match object; span=(9, 10), match='i'>\n<_sre.SRE_Match object; span=(10, 11), match='v'>\n<_sre.SRE_Match object; span=(11, 12), match='e'>\n<_sre.SRE_Match object; span=(12, 13), match='s'>\n<_sre.SRE_Match object; span=(14, 15), match='i'>\n<_sre.SRE_Match object; span=(15, 16), match='n'>\n<_sre.SRE_Match object; span=(16, 17), match=':'>\n<_sre.SRE_Match object; span=(19, 20), match='1'>\n<_sre.SRE_Match object; span=(20, 21), match='2'>\n<_sre.SRE_Match object; span=(21, 22), match='3'>\n<_sre.SRE_Match object; span=(22, 23), match='0'>\n<_sre.SRE_Match object; span=(24, 25), match='F'>\n<_sre.SRE_Match object; span=(25, 26), match='i'>\n<_sre.SRE_Match object; span=(26, 27), match='r'>\n<_sre.SRE_Match object; span=(27, 28), match='s'>\n<_sre.SRE_Match object; span=(28, 29), match='t'>\n<_sre.SRE_Match object; span=(30, 31), match='S'>\n<_sre.SRE_Match object; span=(31, 32), match='t'>\n<_sre.SRE_Match object; span=(32, 33), match='.'>\n<_sre.SRE_Match object; span=(35, 36), match='O'>\n<_sre.SRE_Match object; span=(36, 37), match='c'>\n<_sre.SRE_Match object; span=(37, 38), match='e'>\n<_sre.SRE_Match object; span=(38, 39), match='a'>\n<_sre.SRE_Match object; span=(39, 40), match='n'>\n<_sre.SRE_Match object; span=(41, 42), match='C'>\n<_sre.SRE_Match object; span=(42, 43), match='i'>\n<_sre.SRE_Match object; span=(43, 44), match='t'>\n<_sre.SRE_Match object; span=(44, 45), match='y'>\n<_sre.SRE_Match object; span=(45, 46), match=','>\n<_sre.SRE_Match object; span=(47, 48), match='M'>\n<_sre.SRE_Match object; span=(48, 49), match='D'>\n<_sre.SRE_Match object; span=(50, 51), match='1'>\n<_sre.SRE_Match object; span=(51, 52), match='5'>\n<_sre.SRE_Match object; span=(52, 53), match='6'>\n<_sre.SRE_Match object; span=(53, 54), match='7'>\n<_sre.SRE_Match object; span=(54, 55), match='8'>\n<_sre.SRE_Match object; span=(55, 56), match='9'>\n<_sre.SRE_Match object; span=(56, 57), match='.'>\n" ] ], [ [ "We can see that none of the matches above are whitespace characters. We also see, that by using `\\S` we were able to match all characters, including periods (`.`), letters, and numbers.", "_____no_output_____" ], [ "# TODO: Print The Numbers Between Whitespace Characters\n\nIn the cell below, our `sample_text` consists of a multi-line string with numbers in between whitespace characters:\n\n```python\n123\t45\t7895\n1\t222\t33\n```\n\nNotice that not all the numbers have the same number of digits. For example, the first number (`123` ) has three digits, but the second number (`45` ) only has two digits. \n\nNotice that not all the numbers have the same number of digits. For example, the first number (`123` ) has three digits, but the second number (`45` ) only has two digits. \n\nWrite a single regular expression that finds the tabs (`\\t`) and the newlines (`\\n`) in this multi-line string and save the regular expression object in a variable called `regex`. Then use the `.finditer()` method to search the `sample_text` string for the given regular expression. Then, write a loop that uses the span information from each `match` to only print the numbers found in the original multi-line string. Your code should work in the general case where the numbers can have any number of digits. For example, if the numbers in the string were to change your code should still be able to find them and print them. Finally, in this exercise you cannot use `\\d` in your regular expression. \n\n**HINT :** Notice that there are no whites paces in the multiline string. Use the `\\s` sequence to find the tabs and newlines. Then notice that you can use the span's `end` and `start` index from consecutive matches to figure out the number of digits of each number. Use these indices to print the numbers found in the original multi-line string. You can use the `match.span()` method we saw before to find the `start` and `end` indices of each `match`. Alternatively, you can also use the `.start()` and `.end()` methods to extract the `start` and `end` indices of each match. The `match.start()` is equivalent to `match.span()[0]` and `match.end()` is equivalent to `match.span()[1]`.", "_____no_output_____" ] ], [ [ "# Import re module\nimport re\n\n# Sample text\nsample_text = '''\n123\\t45\\t7895\n1\\t222\\t33\n'''\n\n# Print sample_text\nprint('Sample Text:\\n', sample_text)\n\n# Create a regular expression object with the regular expression\nregex = re.compile(r'\\s')\n\n# Search the sample_text for the regular expression\nmatches = regex.finditer(sample_text)\n\n\n# Write a loop to print all the numbers found in the original string\ncounter = 0\nfor match in matches:\n if counter != 0:\n start_idx = match.start()\n print(sample_text[end_idx:start_idx])\n end_idx = match.end() \n counter += 1", "Sample Text:\n \n123\t45\t7895\n1\t222\t33\n\n123\n45\n7895\n1\n222\n33\n" ] ], [ [ "### Matching Alphanumeric Characters Using `\\w`\n\nIn the code below, we will use `\\w` as our regular expression to find all the alphanumeric characters in our `sample_text` string. This includes the underscore ( `_` ), all the numbers from 0 through 9, and all the uppercase and lowercase letters:", "_____no_output_____" ] ], [ [ "# Import re module\nimport re\n\n# Sample text\nsample_text = '''\nYou can contact FAKE Company at:\[email protected].\n'''\n\n# Create a regular expression object with the regular expression '\\w'\nregex = re.compile(r'\\w')\n\n# Search the sample_text for the regular expression\nmatches = regex.finditer(sample_text)\n\n# Print all the matches\nfor match in matches:\n print(match)", "<_sre.SRE_Match object; span=(1, 2), match='Y'>\n<_sre.SRE_Match object; span=(2, 3), match='o'>\n<_sre.SRE_Match object; span=(3, 4), match='u'>\n<_sre.SRE_Match object; span=(5, 6), match='c'>\n<_sre.SRE_Match object; span=(6, 7), match='a'>\n<_sre.SRE_Match object; span=(7, 8), match='n'>\n<_sre.SRE_Match object; span=(9, 10), match='c'>\n<_sre.SRE_Match object; span=(10, 11), match='o'>\n<_sre.SRE_Match object; span=(11, 12), match='n'>\n<_sre.SRE_Match object; span=(12, 13), match='t'>\n<_sre.SRE_Match object; span=(13, 14), match='a'>\n<_sre.SRE_Match object; span=(14, 15), match='c'>\n<_sre.SRE_Match object; span=(15, 16), match='t'>\n<_sre.SRE_Match object; span=(17, 18), match='F'>\n<_sre.SRE_Match object; span=(18, 19), match='A'>\n<_sre.SRE_Match object; span=(19, 20), match='K'>\n<_sre.SRE_Match object; span=(20, 21), match='E'>\n<_sre.SRE_Match object; span=(22, 23), match='C'>\n<_sre.SRE_Match object; span=(23, 24), match='o'>\n<_sre.SRE_Match object; span=(24, 25), match='m'>\n<_sre.SRE_Match object; span=(25, 26), match='p'>\n<_sre.SRE_Match object; span=(26, 27), match='a'>\n<_sre.SRE_Match object; span=(27, 28), match='n'>\n<_sre.SRE_Match object; span=(28, 29), match='y'>\n<_sre.SRE_Match object; span=(30, 31), match='a'>\n<_sre.SRE_Match object; span=(31, 32), match='t'>\n<_sre.SRE_Match object; span=(34, 35), match='f'>\n<_sre.SRE_Match object; span=(35, 36), match='a'>\n<_sre.SRE_Match object; span=(36, 37), match='k'>\n<_sre.SRE_Match object; span=(37, 38), match='e'>\n<_sre.SRE_Match object; span=(38, 39), match='_'>\n<_sre.SRE_Match object; span=(39, 40), match='c'>\n<_sre.SRE_Match object; span=(40, 41), match='o'>\n<_sre.SRE_Match object; span=(41, 42), match='m'>\n<_sre.SRE_Match object; span=(42, 43), match='p'>\n<_sre.SRE_Match object; span=(43, 44), match='a'>\n<_sre.SRE_Match object; span=(44, 45), match='n'>\n<_sre.SRE_Match object; span=(45, 46), match='y'>\n<_sre.SRE_Match object; span=(46, 47), match='1'>\n<_sre.SRE_Match object; span=(47, 48), match='2'>\n<_sre.SRE_Match object; span=(49, 50), match='e'>\n<_sre.SRE_Match object; span=(50, 51), match='m'>\n<_sre.SRE_Match object; span=(51, 52), match='a'>\n<_sre.SRE_Match object; span=(52, 53), match='i'>\n<_sre.SRE_Match object; span=(53, 54), match='l'>\n<_sre.SRE_Match object; span=(55, 56), match='c'>\n<_sre.SRE_Match object; span=(56, 57), match='o'>\n<_sre.SRE_Match object; span=(57, 58), match='m'>\n" ] ], [ [ "As we can see, all the matches found correspond to alphanumeric characters only, including the underscore in the email address.\n\nConversely, if wanted to find all the characters that are **not** alphanumeric characters, we will use `\\W` as our regular expression, as shown below:", "_____no_output_____" ] ], [ [ "# Import re module\nimport re\n\n# Sample text\nsample_text = '''\nYou can contact FAKE Company at:\[email protected].\n'''\n\n# Create a regular expression object with the regular expression '\\W'\nregex = re.compile(r'\\W')\n\n# Search the sample_text for the regular expression\nmatches = regex.finditer(sample_text)\n\n# Print all the matches\nfor match in matches:\n print(match)", "<_sre.SRE_Match object; span=(0, 1), match='\\n'>\n<_sre.SRE_Match object; span=(4, 5), match=' '>\n<_sre.SRE_Match object; span=(8, 9), match=' '>\n<_sre.SRE_Match object; span=(16, 17), match=' '>\n<_sre.SRE_Match object; span=(21, 22), match=' '>\n<_sre.SRE_Match object; span=(29, 30), match=' '>\n<_sre.SRE_Match object; span=(32, 33), match=':'>\n<_sre.SRE_Match object; span=(33, 34), match='\\n'>\n<_sre.SRE_Match object; span=(48, 49), match='@'>\n<_sre.SRE_Match object; span=(54, 55), match='.'>\n<_sre.SRE_Match object; span=(58, 59), match='.'>\n<_sre.SRE_Match object; span=(59, 60), match='\\n'>\n" ] ], [ [ "We can see that none of the matches are alphanumeric characters. We also see, that by using `\\W` we were able to match all whitespace characters, and the `@` symbol in the email address.", "_____no_output_____" ], [ "# TODO: Find emails\n\nIn the cell below, our `sample_text` consists of a multi-line string that contains three email addresses:\n\n```\[email protected]\[email protected]\[email protected]\n```\n\nNotice, that all three email address have the same pattern, namely, the first name initial, followed by a dot (`.`), followed by the last name initial, and ending in ``` @email.com```. \n\nTake advantage of the fact that all three email addresses have the same pattern to write a single regular expression that can find all three email addresses in our `sample_text` string. As usual, save the regular expression object in a variable called `regex`. Then use the `.finditer()` method to search the `sample_text` string for the given regular expression. Finally, write a loop to print all the `matches` found by the `.finditer()` method.", "_____no_output_____" ] ], [ [ "# Import re module\nimport re\n\n# Sample text\nsample_text = '''\nJohn Sanders: [email protected]\nAlice Walters: [email protected]\nMary Jones: [email protected]\n'''\n\n# Print sample_text\nprint('Sample Text:\\n', sample_text)\n\n# Create a regular expression object with the regular expression\nregex = re.compile(r'[0-9a-zA-Z].[0-9a-zA-Z]@email.com')\n\n# Search the sample_text for the regular expression\nmatches = regex.finditer(sample_text)\n\n# Print all the matches\nfor match in matches:\n print(match)", "Sample Text:\n \nJohn Sanders: [email protected]\nAlice Walters: [email protected]\nMary Jones: [email protected]\n\n<_sre.SRE_Match object; span=(15, 28), match='[email protected]'>\n<_sre.SRE_Match object; span=(44, 57), match='[email protected]'>\n<_sre.SRE_Match object; span=(70, 83), match='[email protected]'>\n" ] ], [ [ "If you wrote your regex correctly you should see three matches above corresponding to the three email addresses found in our `sample_text` string.", "_____no_output_____" ], [ "# Solution\n\n[Solution notebook](simple_patterns_solution.ipynb)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Creating variables", "_____no_output_____" ], [ "In this notebook we will look into the concept of variables.\n\nPython, like R, is a dynamically-typed language, meaning you can change the class/type of a variable on the go. This is convenient in many places, but dangerous in many other ways. It is impossible to rely on the type of the variable, and you should always retrace your steps throughout the code to see what the variable is currently representing. This is sometimes hard, especially in places like this notebook where we can execute different bits of code in any order.\n\n**You're suggested to run this script on Colab**\n[](https://colab.research.google.com/github/Magica-Chen/WebSNA-notes/blob/main/Week0/Week0-notes-python-fundamentals.ipynb)", "_____no_output_____" ], [ "## Intro and strings", "_____no_output_____" ], [ "Let's create a variable:", "_____no_output_____" ] ], [ [ "name = \"Edinburgh\"\nname", "_____no_output_____" ] ], [ [ "This generates a string variable. They can be easily printed, although it is safer to use the print function:", "_____no_output_____" ] ], [ [ "print(name)", "Edinburgh\n" ] ], [ [ "It is also wise to check the type of the variable, in case you are lost:", "_____no_output_____" ] ], [ [ "type(name)", "_____no_output_____" ] ], [ [ "This confirms that we are dealing with a string. There are a few things we can do with strings (which we can denote by using one or two apostrophes):", "_____no_output_____" ] ], [ [ "name = 'university of edinburgh'\nprint(name.lower())\nprint(name.upper())\nprint(name.title())", "university of edinburgh\nUNIVERSITY OF EDINBURGH\nUniversity Of Edinburgh\n" ] ], [ [ "We can concatenate strings easily using +, or using a comma in a print statement:", "_____no_output_____" ] ], [ [ "print('University', 'of Edinburgh')\nprint('University' + ' ' + 'of Edinburgh')", "University of Edinburgh\nUniversity of Edinburgh\n" ] ], [ [ "Writing print('The University of Edinburgh is '+ 439) will not work, as the + operator only works for strings, we can convert any object into a string however:", "_____no_output_____" ] ], [ [ "print('The University of Edinburgh is '+ str(439))", "The University of Edinburgh is 439\n" ] ], [ [ "A few other useful tricks:", "_____no_output_____" ] ], [ [ "name = \" edinburgh \"\nprint(\"|\"+name.lstrip()+\"|\")\nprint(\"|\"+name.rstrip()+\"|\")\nprint(\"|\"+name.strip()+\"|\")", "|edinburgh |\n| edinburgh|\n|edinburgh|\n" ] ], [ [ "You can use control characters as well:", "_____no_output_____" ] ], [ [ "print('Edinburgh\\thas a university\\nrunning web & social network analytics course')", "Edinburgh\thas a university\nrunning web & social network analytics course\n" ] ], [ [ "## Numbers", "_____no_output_____" ] ], [ [ "a = 10\nb = -10.1023\n\n#Some operations illustrated (\\t stands for a tab)\nprint(\"a: \\t\\t\\t\" + str(a))\nprint(\"b: \\t\\t\\t\" + str(b))\nprint(\"absolute of b: \\t\\t\" + str(abs(b)))\nprint(\"rounded b: \\t\\t\" + str(round(b,3)))\nprint(\"square of a: \\t\\t\" + str(pow(a,2)))\nprint(\"cube of a: \\t\\t\" + str(a**3))\nprint(\"integer part of b: \\t\" + str(int(b)))", "a: \t\t\t10\nb: \t\t\t-10.1023\nabsolute of b: \t\t10.1023\nrounded b: \t\t-10.102\nsquare of a: \t\t100\ncube of a: \t\t1000\ninteger part of b: \t-10\n" ] ], [ [ "# Flow Control", "_____no_output_____" ], [ "Control flow statements help you to structure the code and direct it towards your convenience and introduce loops and so on.", "_____no_output_____" ], [ "## If statements", "_____no_output_____" ] ], [ [ "price = -5;\n\nif price <0:\n print(\"Price is negative!\")\nelif price <1:\n print(\"Price is too small!\")\nelse:\n print(\"Price is suitable.\")", "Price is negative!\n" ] ], [ [ "Especially in text mining, comparing strings is very important:", "_____no_output_____" ] ], [ [ "#Comparing strings\nname1 = \"edinburgh\"\nname2 = \"Edinburgh\"\n\nif name1 == name2:\n print(\"Equal\")\nelse:\n print(\"Not equal\")\n\nif name1.lower() == name2.lower():\n print(\"Equal\")\nelse:\n print(\"Not equal\")", "Not equal\nEqual\n" ] ], [ [ "Using multiple conditions:", "_____no_output_____" ] ], [ [ "number = 9\nif number > 1 and not number > 9:\n print(\"Number is between 1 and 10\")\n \nnumber = 9\nname = 'johannes'\nif number < 5 or 'j' in name:\n print(\"Number is lower than 5 or the name contains a 'j'\")", "Number is between 1 and 10\nNumber is lower than 5 or the name contains a 'j'\n" ] ], [ [ "## While loops", "_____no_output_____" ] ], [ [ "number = 4\nwhile number > 1:\n print(number)\n number = number -1", "4\n3\n2\n" ] ], [ [ "## For loops", "_____no_output_____" ], [ "For loops allow you to iteratre over elements in a certain collection, for example a list:", "_____no_output_____" ] ], [ [ "# We'll look into lists in a minute\nnumber_list = [1, 2, 3, 4]\nfor item in number_list:\n print(item)", "1\n2\n3\n4\n" ], [ "list = ['a', 'b', 'c']\nfor item in list:\n print(item)", "a\nb\nc\n" ] ], [ [ "Ranges are also useful. Note that the upper element is not included and we can adjust the step size:", "_____no_output_____" ] ], [ [ "for i in range(1,4):\n print(i)", "1\n2\n3\n" ], [ "for i in range(30,100, 10):\n print(i)", "30\n40\n50\n60\n70\n80\n90\n" ] ], [ [ "## Indentation", "_____no_output_____" ], [ "Please be very careful with indentation", "_____no_output_____" ] ], [ [ "number_1 = 3\nnumber_2 = 5\n\nprint('No indent (no tabs used)')\nif number_1 > 1:\n print('\\tNumber 1 higher than 1.')\n if number_2 > 5:\n print('\\t\\tnumber 2 higher than 5')\n print('\\tnumber 2 higher than 5')\n\nnumber_1 = 3\nnumber_2 = 6\n\nprint('No indent (no tabs used)')\nif number_1 > 1:\n print('\\tNumber 1 higher than 1.')\n if number_2 > 5:\n print('\\t\\tnumber 2 higher than 5')\n print('\\tnumber 2 higher than 5')", "No indent (no tabs used)\n\tNumber 1 higher than 1.\n\tnumber 2 higher than 5\nNo indent (no tabs used)\n\tNumber 1 higher than 1.\n\t\tnumber 2 higher than 5\n\tnumber 2 higher than 5\n" ] ], [ [ "# List & Tuple", "_____no_output_____" ], [ "## Lists", "_____no_output_____" ], [ "Lists are great for collecting anything. They can contain objects of different types. For example:", "_____no_output_____" ] ], [ [ "names = [5, \"Giovanni\", \"Rose\", \"Yongzhe\", \"Luciana\", \"Imani\"]", "_____no_output_____" ] ], [ [ "Although that is not best practice. Let's start with a list of names:", "_____no_output_____" ] ], [ [ "names = [\"Johannes\", \"Giovanni\", \"Rose\", \"Yongzhe\", \"Luciana\", \"Imani\"]", "_____no_output_____" ], [ "# Loop names\nfor name in names:\n print('Name: '+name)\n\n# Get 'Giovanni' from list\n# Lists start counting at 0\ngiovanni = names[1]\nprint(giovanni.upper())\n\n# Get last item\nname = names[-1]\nprint(name.upper())\n\n# Get second to last item\nname = names[-2]\nprint(name.upper())\n\nprint(\"First three: \"+str(names[0:3]))\nprint(\"First four: \"+str(names[:4]))\nprint(\"Up until the second to last one: \"+str(names[:-2]))\nprint(\"Last two: \"+str(names[-2:]))", "Name: Johannes\nName: Giovanni\nName: Rose\nName: Yongzhe\nName: Luciana\nName: Imani\nGIOVANNI\nIMANI\nLUCIANA\nFirst three: ['Johannes', 'Giovanni', 'Rose']\nFirst four: ['Johannes', 'Giovanni', 'Rose', 'Yongzhe']\nUp until the second to last one: ['Johannes', 'Giovanni', 'Rose', 'Yongzhe']\nLast two: ['Luciana', 'Imani']\n" ] ], [ [ "## Enumeration", "_____no_output_____" ], [ "We can enumerate collections/lists that adds an index to every element:", "_____no_output_____" ] ], [ [ "for index, name in enumerate(names):\n print(str(index) , \" \" , name, \" is in the list.\")", "0 Johannes is in the list.\n1 Giovanni is in the list.\n2 Rose is in the list.\n3 Yongzhe is in the list.\n4 Luciana is in the list.\n5 Imani is in the list.\n" ] ], [ [ "## Searching and editing", "_____no_output_____" ] ], [ [ "names = [\"Johannes\", \"Giovanni\", \"Rose\", \"Yongzhe\", \"Luciana\", \"Imani\"]\n\n# Finding an element\nprint(names.index(\"Johannes\"))\n\n# Adding an element\nnames.append(\"Kumiko\")\n\n# Adding an element at a specific location\nnames.insert(2, \"Roberta\")\n\nprint(names)\n\n#Removal\nfruits = [\"apple\",\"orange\",\"pear\"]\ndel fruits[0]\nfruits.remove(\"pear\")\nprint('Fruits: ', fruits)\n\n# Modifying an element\nnames[5] = \"Tom\"\nprint(names)\n\n# Test whether an item is in the list (best do this before removing to avoid raising errors)\nprint(\"Tom\" in names)\n\n# Length of a list\nprint(\"Length of the list: \" + str(len(names)))", "0\n['Johannes', 'Giovanni', 'Roberta', 'Rose', 'Yongzhe', 'Luciana', 'Imani', 'Kumiko']\nFruits: ['orange']\n['Johannes', 'Giovanni', 'Roberta', 'Rose', 'Yongzhe', 'Tom', 'Imani', 'Kumiko']\nTrue\nLength of the list: 8\n" ] ], [ [ "Python starts at 0!!!", "_____no_output_____" ], [ "## Sorting and copying", "_____no_output_____" ] ], [ [ "# Temporary sorting:\nprint(sorted(names))\nprint(names)\n\n# Make changes permanent\nnames.sort()\nprint(\"Sorted names: \" + str(names))\nnames.sort(reverse=True)\nprint(\"Reverse sorted names: \" + str(names))", "['Giovanni', 'Imani', 'Johannes', 'Kumiko', 'Roberta', 'Rose', 'Tom', 'Yongzhe']\n['Johannes', 'Giovanni', 'Roberta', 'Rose', 'Yongzhe', 'Tom', 'Imani', 'Kumiko']\nSorted names: ['Giovanni', 'Imani', 'Johannes', 'Kumiko', 'Roberta', 'Rose', 'Tom', 'Yongzhe']\nReverse sorted names: ['Yongzhe', 'Tom', 'Rose', 'Roberta', 'Kumiko', 'Johannes', 'Imani', 'Giovanni']\n" ], [ "# Copying list (a shallow copy just duplicates the pointer to the memory address)\nnamez = names\nnamez.remove(\"Johannes\")\nprint(namez)\nprint(names)\n\n# Now a 'deep' copy\nprint(\"After deep copy\")\n\nnamez = names.copy()\nnamez.remove(\"Giovanni\")\nprint(namez)\nprint(names)\n\n#Alternative\nnamez = names[:]\nprint(namez)", "['Yongzhe', 'Tom', 'Rose', 'Roberta', 'Kumiko', 'Imani', 'Giovanni']\n['Yongzhe', 'Tom', 'Rose', 'Roberta', 'Kumiko', 'Imani', 'Giovanni']\nAfter deep copy\n['Yongzhe', 'Tom', 'Rose', 'Roberta', 'Kumiko', 'Imani']\n['Yongzhe', 'Tom', 'Rose', 'Roberta', 'Kumiko', 'Imani', 'Giovanni']\n['Yongzhe', 'Tom', 'Rose', 'Roberta', 'Kumiko', 'Imani', 'Giovanni']\n" ] ], [ [ "## Strings as lists", "_____no_output_____" ], [ "Strings can be manipulated and used just like lists. This is especially handy in text mining:", "_____no_output_____" ] ], [ [ "course = \"Predictive analytics\"\nprint(\"Last nine letters: \"+course[-9:])\nprint(\"Analytics in course title? \" + str(\"analytics\" in course))\nprint(\"Start location of 'analytics': \" + str(course.find(\"analytics\")))\nprint(course.replace(\"analytics\",\"analysis\"))\nlist_of_words = course.split(\" \")\nfor index, word in enumerate(list_of_words):\n print(\"Word \", index, \": \"+word)", "Last nine letters: analytics\nAnalytics in course title? True\nStart location of 'analytics': 11\nPredictive analysis\nWord 0 : Predictive\nWord 1 : analytics\n" ] ], [ [ "## Sets", "_____no_output_____" ], [ "Sets only contain unique elements. They have to be declared upfront using set() and allow for operations such as intersection():", "_____no_output_____" ] ], [ [ "name_set = set(names)\nprint(name_set)\n\n# Add an element\nname_set.add(\"Galina\")\nprint(name_set)\n\n# Discard an element\nname_set.discard(\"Johannes\")\nprint(name_set)\n\nname_set2 = set([\"Rose\", \"Tom\"])\n# Difference and intersection\ndifference = name_set - name_set2\nprint(difference)\nintersection = name_set.intersection(name_set2)\nprint(intersection)", "{'Yongzhe', 'Kumiko', 'Roberta', 'Giovanni', 'Tom', 'Imani', 'Rose'}\n{'Yongzhe', 'Kumiko', 'Roberta', 'Giovanni', 'Tom', 'Imani', 'Galina', 'Rose'}\n{'Yongzhe', 'Kumiko', 'Roberta', 'Giovanni', 'Tom', 'Imani', 'Galina', 'Rose'}\n{'Yongzhe', 'Kumiko', 'Roberta', 'Giovanni', 'Imani', 'Galina'}\n{'Tom', 'Rose'}\n" ] ], [ [ "# Dictionary & Function", "_____no_output_____" ], [ "## Dictionaries", "_____no_output_____" ], [ "Dictionaries are a great way to store particular data as key-value pairs, which mimics the basic structure of a simple database.", "_____no_output_____" ] ], [ [ "courses = {\"Johannes\" : \"Predictive analytics\", \"Kumiko\" : \"Prescriptive analytics\", \"Luciana\" : \"Descriptive analytics\"}\n\nfor organizer in courses:\n print(organizer + \" teaches \" + courses[organizer])", "Johannes teaches Predictive analytics\nKumiko teaches Prescriptive analytics\nLuciana teaches Descriptive analytics\n" ] ], [ [ "We can also write:", "_____no_output_____" ] ], [ [ "for organizer, course in courses.items():\n print(organizer + \" teaches \" + course)", "Johannes teaches Predictive analytics\nKumiko teaches Prescriptive analytics\nLuciana teaches Descriptive analytics\n" ], [ "# Adding items\ncourses[\"Imani\"] = \"Other analytics\"\nprint(courses)\n\n# Overwrite\ncourses[\"Johannes\"] = \"Business analytics\"\nprint(courses)", "{'Johannes': 'Predictive analytics', 'Kumiko': 'Prescriptive analytics', 'Luciana': 'Descriptive analytics', 'Imani': 'Other analytics'}\n{'Johannes': 'Business analytics', 'Kumiko': 'Prescriptive analytics', 'Luciana': 'Descriptive analytics', 'Imani': 'Other analytics'}\n" ], [ "# Remove\ndel courses[\"Johannes\"]\nprint(courses)", "{'Kumiko': 'Prescriptive analytics', 'Luciana': 'Descriptive analytics', 'Imani': 'Other analytics'}\n" ], [ "# Looping values\nfor course in courses.values():\n print(course)", "Prescriptive analytics\nDescriptive analytics\nOther analytics\n" ], [ "# Sorted output (on keys)\nfor organizer, course in sorted(courses.items()):\n print(organizer +\" teaches \" + course)", "Imani teaches Other analytics\nKumiko teaches Prescriptive analytics\nLuciana teaches Descriptive analytics\n" ] ], [ [ "## Functions", "_____no_output_____" ], [ "Functions form the backbone of all code. You have already used some, like print(). They can be easily defined by yourself as well.", "_____no_output_____" ] ], [ [ "def my_function(a, b):\n a = a.title()\n b = b.upper()\n print(a+ \" \"+b)", "_____no_output_____" ], [ "def my_function2(a, b):\n a = a.title()\n b = b.upper()\n return a + \" \" + b", "_____no_output_____" ], [ "my_function(\"johannes\",\"de smedt\")\noutput = my_function2(\"johannes\",\"de smedt\")\nprint(output)", "Johannes DE SMEDT\nJohannes DE SMEDT\n" ] ], [ [ "Notice how the first function already prints, while the second returns a string we have to print ourselves. Python is weakly-typed, so a function can produce different results, like in this example:", "_____no_output_____" ] ], [ [ "# Different output type\ndef calculate_mean(a, b):\n if (a>0):\n return (a+b)/2\n else:\n return \"a is negative\"\n\noutput = calculate_mean(1,2)\nprint(output)\noutput = calculate_mean(0,1)\nprint(output)", "1.5\na is negative\n" ] ], [ [ "## Comprehensions", "_____no_output_____" ], [ "Comprehensions allow you to quickly/efficiently write lists/dictionaries:", "_____no_output_____" ] ], [ [ "# Finding even numbers\nevens = [i for i in range(1,11) if i % 2 ==0]\nprint(evens)", "[2, 4, 6, 8, 10]\n" ] ], [ [ "In Python, you can easily make tuples such as pairs, like here:", "_____no_output_____" ] ], [ [ "# Double fun\npairs = [(x,y) for x in range(1,11) for y in range(5,11) if x>y]\nprint(pairs)", "[(6, 5), (7, 5), (7, 6), (8, 5), (8, 6), (8, 7), (9, 5), (9, 6), (9, 7), (9, 8), (10, 5), (10, 6), (10, 7), (10, 8), (10, 9)]\n" ] ], [ [ "They are also useful to perform some pre-processing, e.g., on strings:", "_____no_output_____" ] ], [ [ "# Operations\nnames = [\"jamal\", \"maurizio\", \"johannes\"]\n\ntitled_names = [name.title() for name in names]\nprint(titled_names)\n\nj_s = [name.title() for name in names if name.lower()[0] == 'j']\nprint(j_s)", "['Jamal', 'Maurizio', 'Johannes']\n['Jamal', 'Johannes']\n" ] ], [ [ "# IO & Library", "_____no_output_____" ] ], [ [ "# Download some datasets\n# If you are using git, then you don't need to run the following.\n!wget -q https://raw.githubusercontent.com/Magica-Chen/WebSNA-notes/main/Week0/data/DM_1.csv\n!wget -q https://raw.githubusercontent.com/Magica-Chen/WebSNA-notes/main/Week0/data/DM_2.csv\n!wget -q https://raw.githubusercontent.com/Magica-Chen/WebSNA-notes/main/Week0/data/ordered_amounts_per_person.csv\n!mkdir data\n!mv *.csv ./data", "_____no_output_____" ] ], [ [ "## Reading files", "_____no_output_____" ], [ "In Python, we can easily open any file type. Naturally, it is most suitable for plainly-structured formats such as .txt., .csv., as so on. You can also open Excel files with appropriate packages, such as pandas (more on this later). Let's read in a .csv file:", "_____no_output_____" ] ], [ [ "# Open a file for reading ('r')\nfile = open('data/DM_1.csv','r')\n\nfor line in file:\n print(line)", "Name,Email,City,Salary\n\nBrent Hopkins,[email protected],Mount Pearl,38363\n\nColt Bender,[email protected],Castle Douglas,21506\n\nArthur Hammond,[email protected],Biloxi,27511\n\nSean Warner,[email protected],Moere,25201\n\nTate Greene,[email protected],Ipswich,35052\n\nGavin Gibson,[email protected],Oordegem,37126\n\nKelly Garza,[email protected],Kukatpalle,39420\n\nZane Preston,[email protected],Neudšrfl,28553\n\nCole Cunningham,[email protected],Catemu,27972\n\nTarik Hendricks,[email protected],Newbury,39027\n\nElvis Collier,[email protected],Paradise,22568\n\nJackson Huber,[email protected],Veere,29922\n\nMacaulay Cline,[email protected],Campobasso,24163\n\nElijah Chase,[email protected],Grantham,23881\n\nDennis Anthony,[email protected],Cedar Rapids,27969\n\nFulton Snyder,[email protected],San Pedro,21594\n\nLeo Willis,[email protected],Kester,31203\n\nMatthew Hooper,[email protected],Bellefontaine,33222\n\nTodd Jones,[email protected],Toledo,24809\n\nPalmer Byrd,[email protected],Bissegem,29045\n" ] ], [ [ "We can store this information in objects and start using it:", "_____no_output_____" ] ], [ [ "# File is looped now, hence, reread file\nfile = open('data/DM_1.csv','r')\n# ignore the header\nnext(file)\n\n# Store names with amount (i.e. columns 1 & 2)\namount_per_person = {}\nfor line in file:\n cells = line.split(\",\")\n amount_per_person[cells[0]] = int(cells[3])\n\nfor person, amount in sorted(amount_per_person.items()):\n if amount > 25000:\n print(person , \" has \" , amount)", "Arthur Hammond has 27511\nBrent Hopkins has 38363\nCole Cunningham has 27972\nDennis Anthony has 27969\nGavin Gibson has 37126\nJackson Huber has 29922\nKelly Garza has 39420\nLeo Willis has 31203\nMatthew Hooper has 33222\nPalmer Byrd has 29045\nSean Warner has 25201\nTarik Hendricks has 39027\nTate Greene has 35052\nZane Preston has 28553\n" ], [ "# Now we use 'w' for write \noutput_file = open('data/ordered_amounts_per_person.csv','w')\n\nfor person, amount in sorted(amount_per_person.items()):\n output_file.write(person.lower()+\",\"+str(amount)) \noutput_file.close()", "_____no_output_____" ] ], [ [ "## Libraries", "_____no_output_____" ], [ "Libraries are imported by using `import`:", "_____no_output_____" ] ], [ [ "import numpy\nimport pandas\nimport sklearn", "_____no_output_____" ] ], [ [ "If you haven't installed sklearn, please install it by:", "_____no_output_____" ] ], [ [ "!pip install sklearn", "Collecting sklearn\n Downloading sklearn-0.0.tar.gz (1.1 kB)\nRequirement already satisfied: scikit-learn in c:\\users\\zchen112\\anaconda3\\lib\\site-packages (from sklearn) (0.24.1)\nRequirement already satisfied: joblib>=0.11 in c:\\users\\zchen112\\anaconda3\\lib\\site-packages (from scikit-learn->sklearn) (1.0.1)\nRequirement already satisfied: scipy>=0.19.1 in c:\\users\\zchen112\\anaconda3\\lib\\site-packages (from scikit-learn->sklearn) (1.6.2)\nRequirement already satisfied: numpy>=1.13.3 in c:\\users\\zchen112\\anaconda3\\lib\\site-packages (from scikit-learn->sklearn) (1.20.1)\nRequirement already satisfied: threadpoolctl>=2.0.0 in c:\\users\\zchen112\\anaconda3\\lib\\site-packages (from scikit-learn->sklearn) (2.1.0)\nBuilding wheels for collected packages: sklearn\n Building wheel for sklearn (setup.py): started\n Building wheel for sklearn (setup.py): finished with status 'done'\n Created wheel for sklearn: filename=sklearn-0.0-py2.py3-none-any.whl size=1316 sha256=de54cc32dd40e89c8b3d1fd541e221f659546c0f556371dadf408a133d078ca0\n Stored in directory: c:\\users\\zchen112\\appdata\\local\\pip\\cache\\wheels\\22\\0b\\40\\fd3f795caaa1fb4c6cb738bc1f56100be1e57da95849bfc897\nSuccessfully built sklearn\nInstalling collected packages: sklearn\nSuccessfully installed sklearn-0.0\n" ] ], [ [ "We can import just a few bits using `from`, or create aliases using `as`:", "_____no_output_____" ] ], [ [ "import math as m\nfrom math import pi", "_____no_output_____" ], [ "print(numpy.add(1, 2))\nprint(pi)\nprint(m.sin(1))", "3\n3.141592653589793\n0.8414709848078965\n" ] ], [ [ "In the next part, some basic procedures that exist in NumPy, pandas, and scikit-learn are covered. This only scratches the surface of the possibilities, and many other functions and code will be used later on. Make sure to search around for the possiblities that exist yourself, and get a grasp of how the modules are called and used. Let's import them in this notebook to start with:", "_____no_output_____" ] ], [ [ "import numpy as np\nimport pandas as pd\nimport sklearn", "_____no_output_____" ] ], [ [ "## Numpy", "_____no_output_____" ] ], [ [ "# Create empty arrays/matrices\nempty_array = np.zeros(5)\n\nempty_matrix = np.zeros((5,2))\n\nprint('Empty array: \\n',empty_array)\nprint('Empty matrix: \\n',empty_matrix)", "Empty array: \n [0. 0. 0. 0. 0.]\nEmpty matrix: \n [[0. 0.]\n [0. 0.]\n [0. 0.]\n [0. 0.]\n [0. 0.]]\n" ], [ "# Create matrices\nmat = np.array([[1,2,3],[4,5,6]])\nprint('Matrix: \\n', mat)\nprint('Transpose: \\n', mat.T)\nprint('Item 2,2: ', mat[1,1])\nprint('Item 2,3: ', mat[1,2])\nprint('rows and columns: ', np.shape(mat))\nprint('Sum total matrix: ', np.sum(mat))\nprint('Sum row 1: ' , np.sum(mat[0]))\nprint('Sum row 2: ', np.sum(mat[1]))\nprint('Sum column 2: ', np.sum(mat,axis=0)[2])", "Matrix: \n [[1 2 3]\n [4 5 6]]\nTranspose: \n [[1 4]\n [2 5]\n [3 6]]\nItem 2,2: 5\nItem 2,3: 6\nrows and columns: (2, 3)\nSum total matrix: 21\nSum row 1: 6\nSum row 2: 15\nSum column 2: 9\n" ] ], [ [ "## pandas", "_____no_output_____" ], [ "### Creating dataframes", "_____no_output_____" ], [ "pandas is great for reading and creating datasets, as well as performing basic operations on them.", "_____no_output_____" ] ], [ [ "# Creating a matrix with three rows of data\ndata = [['johannes',10], ['giovanni',2], ['john',3]]\n\n# Creating and printing a pandas DataFrame object from the matrix\ndf = pd.DataFrame(data)\nprint(df)", " 0 1\n0 johannes 10\n1 giovanni 2\n2 john 3\n" ], [ "# Adding columns to the DataFrame object\ndf.columns = ['names', 'years']\nprint(df)", " names years\n0 johannes 10\n1 giovanni 2\n2 john 3\n" ], [ "df_2 = pd.DataFrame(data = data, columns = ['names', 'years'])\nprint(df_2)", " names years\n0 johannes 10\n1 giovanni 2\n2 john 3\n" ], [ "# Taking out a single column and calculating its sum\n# This also shows the type of the variable: a 64 bit integer (array)\nprint(df['years'])\nprint('Sum of all values in column: ', df['years'].sum())", "0 10\n1 2\n2 3\nName: years, dtype: int64\nSum of all values in column: 15\n" ], [ "# Creating a larger matrix\ndata = [['johannes',10], ['giovanni',2], ['john',3], ['giovanni',2], ['john',3], ['giovanni',2], ['john',3], ['giovanni',2], ['john',3], ['johannes',10]]\n\n# Again, creating a DataFrame object, now with columns\ndf = pd.DataFrame(data, columns = ['names','years'])\n\n# Print the 5 first (head) and 5 last (tail) observations\nprint(df.head())\nprint('\\n')\nprint(df.tail())", " names years\n0 johannes 10\n1 giovanni 2\n2 john 3\n3 giovanni 2\n4 john 3\n\n\n names years\n5 giovanni 2\n6 john 3\n7 giovanni 2\n8 john 3\n9 johannes 10\n" ] ], [ [ "### Reading files", "_____no_output_____" ], [ "You can read files:", "_____no_output_____" ] ], [ [ "dataset = pd.read_csv('data/DM_1.csv')\nprint(dataset.head())", " Name Email City \\\n0 Brent Hopkins [email protected] Mount Pearl \n1 Colt Bender [email protected] Castle Douglas \n2 Arthur Hammond [email protected] Biloxi \n3 Sean Warner [email protected] Moere \n4 Tate Greene [email protected] Ipswich \n\n Salary \n0 38363 \n1 21506 \n2 27511 \n3 25201 \n4 35052 \n" ] ], [ [ "### Using dataframes", "_____no_output_____" ] ], [ [ "# Print all unique values of the column names\nprint(df['names'].unique())", "['johannes' 'giovanni' 'john']\n" ], [ "# Print all values and their frequency:\nprint(df['names'].value_counts())\nprint(df['years'].value_counts())", "john 4\ngiovanni 4\njohannes 2\nName: names, dtype: int64\n2 4\n3 4\n10 2\nName: years, dtype: int64\n" ], [ "# Add a column names 'code' with all zeros\ndf['code'] = np.zeros(10)\nprint(df)", " names years code\n0 johannes 10 0.0\n1 giovanni 2 0.0\n2 john 3 0.0\n3 giovanni 2 0.0\n4 john 3 0.0\n5 giovanni 2 0.0\n6 john 3 0.0\n7 giovanni 2 0.0\n8 john 3 0.0\n9 johannes 10 0.0\n" ] ], [ [ "You can also easily find things in a DataFrame use `.loc`:", "_____no_output_____" ] ], [ [ "# Rows 2 to 5 and all columns:\nprint(df.loc[2:5, :])", " names years code\n2 john 3 0.0\n3 giovanni 2 0.0\n4 john 3 0.0\n5 giovanni 2 0.0\n" ], [ "# Looping columns\nfor variable in df.columns:\n print(df[variable])", "0 johannes\n1 giovanni\n2 john\n3 giovanni\n4 john\n5 giovanni\n6 john\n7 giovanni\n8 john\n9 johannes\nName: names, dtype: object\n0 10\n1 2\n2 3\n3 2\n4 3\n5 2\n6 3\n7 2\n8 3\n9 10\nName: years, dtype: int64\n0 0.0\n1 0.0\n2 0.0\n3 0.0\n4 0.0\n5 0.0\n6 0.0\n7 0.0\n8 0.0\n9 0.0\nName: code, dtype: float64\n" ], [ "# Looping columns and obtaining the values (which returns an array)\nfor variable in df.columns:\n print(df[variable].values)", "['johannes' 'giovanni' 'john' 'giovanni' 'john' 'giovanni' 'john'\n 'giovanni' 'john' 'johannes']\n[10 2 3 2 3 2 3 2 3 10]\n[0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n" ] ], [ [ "### preparing datasets", "_____no_output_____" ] ], [ [ "dataset_1 = pd.read_csv('data/DM_1.csv', encoding='latin1')\ndataset_2 = pd.read_csv('data/DM_2.csv', encoding='latin1')", "_____no_output_____" ], [ "dataset_1", "_____no_output_____" ], [ "dataset_2", "_____no_output_____" ], [ "dataset_2.columns = ['First name', 'Last name', 'Days active']\ndataset_2", "_____no_output_____" ] ], [ [ "We can convert the second dataset to only have 1 column for names:", "_____no_output_____" ] ], [ [ "# .title() can be used to only make the first letter a capital\nnames = [dataset_2.loc[i,'First name'] + \" \" + dataset_2.loc[i,'Last name'].title() for i in range(0, len(dataset_2))]\n\n# Make a new column for the name\ndataset_2['Name'] = names\n\n# Remove the old columns\ndataset_2 = dataset_2.drop(['First name', 'Last name'], axis=1)\ndataset_2", "_____no_output_____" ] ], [ [ "### Bringing together the datasets", "_____no_output_____" ], [ "Now the datasets are made compatible, we can merge them in a few different ways.", "_____no_output_____" ] ], [ [ "# A left join starts from the left dataset, in this case dataset_1, and for every row matches the value in the \n# column used for joining. As you will see, the result has 22 rows since some names appear multiple times in \n# the second dataset dataset_2.\n\nboth = pd.merge(dataset_1, dataset_2, on='Name', how='left')\nboth", "_____no_output_____" ], [ "# A right join does the opposite: now, dataset_2 is used to match all names with the corresponding \n# observations in dataset_1. There are as many observations as there are in dataset_2, as the rows \n# in dataset_1 are unique. The last row cannot be matched with any observation in dataset_1.\n\nboth = pd.merge(dataset_1, dataset_2, on='Name', how='right')\nboth", "_____no_output_____" ], [ "# Inner and outer join\n# It is also possible to only retain the values that are matched in both tables, or match any value \n# that matches. This is using an inner and outer join respectively.\n\nboth = pd.merge(dataset_1, dataset_2, on='Name', how='inner')\nboth", "_____no_output_____" ] ], [ [ "Notice how observation 12 is missing, as there is no corresponding value in `dataset_1`.", "_____no_output_____" ] ], [ [ "both = pd.merge(dataset_1, dataset_2, on='Name', how='outer')\nboth", "_____no_output_____" ] ], [ [ "In the last table, we have 23 rows, as both matching and non-matching values are returned.\n\nMerging datasets can be really helpful. This code should give you ample ideas on how to do this quickly yourself. As always, there are a number of ways of achieving the same result. Don't hold back to explore other solutions that might be quicker or easier.", "_____no_output_____" ], [ "# scikit-learn", "_____no_output_____" ], [ "scikit-learn is great for performing all major data analysis operations. It also contains datasets. In this code, we will load a dataset and fit a simple linear regression.", "_____no_output_____" ] ], [ [ "from sklearn import datasets as ds", "_____no_output_____" ], [ "# Load the Boston Housing dataset\ndataset = ds.load_boston()\n\n# It is a dictionary, see the keys for details:\nprint(dataset.keys())", "dict_keys(['data', 'target', 'feature_names', 'DESCR', 'filename'])\n" ], [ "# The 'DESCR' key holds a description text for the whole dataset\nprint(dataset['DESCR'])", ".. _boston_dataset:\n\nBoston house prices dataset\n---------------------------\n\n**Data Set Characteristics:** \n\n :Number of Instances: 506 \n\n :Number of Attributes: 13 numeric/categorical predictive. Median Value (attribute 14) is usually the target.\n\n :Attribute Information (in order):\n - CRIM per capita crime rate by town\n - ZN proportion of residential land zoned for lots over 25,000 sq.ft.\n - INDUS proportion of non-retail business acres per town\n - CHAS Charles River dummy variable (= 1 if tract bounds river; 0 otherwise)\n - NOX nitric oxides concentration (parts per 10 million)\n - RM average number of rooms per dwelling\n - AGE proportion of owner-occupied units built prior to 1940\n - DIS weighted distances to five Boston employment centres\n - RAD index of accessibility to radial highways\n - TAX full-value property-tax rate per $10,000\n - PTRATIO pupil-teacher ratio by town\n - B 1000(Bk - 0.63)^2 where Bk is the proportion of blacks by town\n - LSTAT % lower status of the population\n - MEDV Median value of owner-occupied homes in $1000's\n\n :Missing Attribute Values: None\n\n :Creator: Harrison, D. and Rubinfeld, D.L.\n\nThis is a copy of UCI ML housing dataset.\nhttps://archive.ics.uci.edu/ml/machine-learning-databases/housing/\n\n\nThis dataset was taken from the StatLib library which is maintained at Carnegie Mellon University.\n\nThe Boston house-price data of Harrison, D. and Rubinfeld, D.L. 'Hedonic\nprices and the demand for clean air', J. Environ. Economics & Management,\nvol.5, 81-102, 1978. Used in Belsley, Kuh & Welsch, 'Regression diagnostics\n...', Wiley, 1980. N.B. Various transformations are used in the table on\npages 244-261 of the latter.\n\nThe Boston house-price data has been used in many machine learning papers that address regression\nproblems. \n \n.. topic:: References\n\n - Belsley, Kuh & Welsch, 'Regression diagnostics: Identifying Influential Data and Sources of Collinearity', Wiley, 1980. 244-261.\n - Quinlan,R. (1993). Combining Instance-Based and Model-Based Learning. In Proceedings on the Tenth International Conference of Machine Learning, 236-243, University of Massachusetts, Amherst. Morgan Kaufmann.\n\n" ], [ "# The data (independent variables) are stored under the 'data' key\n# The names of the independent variables are stored in the 'feature_names' key\n# Let's use them to create a DataFrame object:\ndf = pd.DataFrame(data=dataset['data'], columns=dataset['feature_names'])\nprint(df.head())", " CRIM ZN INDUS CHAS NOX RM AGE DIS RAD TAX \\\n0 0.00632 18.0 2.31 0.0 0.538 6.575 65.2 4.0900 1.0 296.0 \n1 0.02731 0.0 7.07 0.0 0.469 6.421 78.9 4.9671 2.0 242.0 \n2 0.02729 0.0 7.07 0.0 0.469 7.185 61.1 4.9671 2.0 242.0 \n3 0.03237 0.0 2.18 0.0 0.458 6.998 45.8 6.0622 3.0 222.0 \n4 0.06905 0.0 2.18 0.0 0.458 7.147 54.2 6.0622 3.0 222.0 \n\n PTRATIO B LSTAT \n0 15.3 396.90 4.98 \n1 17.8 396.90 9.14 \n2 17.8 392.83 4.03 \n3 18.7 394.63 2.94 \n4 18.7 396.90 5.33 \n" ], [ "# The dependent variable is stored separately\ndf_y = pd.DataFrame(data=dataset['target'], columns=['target'])\nprint(df_y.head())", " target\n0 24.0\n1 21.6\n2 34.7\n3 33.4\n4 36.2\n" ], [ "# Now, let's build a linear regression model\nfrom sklearn.linear_model import LinearRegression as LR\n\n# First we create a linear regression object\nregression = LR()\n\n# Then, we fit the independent and dependent data\nregression.fit(df, df_y)\n\n# We can obtain the R^2 score (more on this later)\nprint(regression.score(df, df_y))", "0.7406426641094095\n" ] ], [ [ "Very often, we need to perform an operation on a single observation. In that case, we have to reshape the data using numpy:", "_____no_output_____" ] ], [ [ "# Consider a single observation \nso = df.loc[2, :]\nprint(so)\n\n# Just the values of the observation without meta data\nprint(so.values)\n\n# Reshaping yields a new matrix with one row with as many columns as the original observation (indicated by the -1)\nprint(np.reshape(so.values, (1, -1)))", "CRIM 0.02729\nZN 0.00000\nINDUS 7.07000\nCHAS 0.00000\nNOX 0.46900\nRM 7.18500\nAGE 61.10000\nDIS 4.96710\nRAD 2.00000\nTAX 242.00000\nPTRATIO 17.80000\nB 392.83000\nLSTAT 4.03000\nName: 2, dtype: float64\n[2.7290e-02 0.0000e+00 7.0700e+00 0.0000e+00 4.6900e-01 7.1850e+00\n 6.1100e+01 4.9671e+00 2.0000e+00 2.4200e+02 1.7800e+01 3.9283e+02\n 4.0300e+00]\n[[2.7290e-02 0.0000e+00 7.0700e+00 0.0000e+00 4.6900e-01 7.1850e+00\n 6.1100e+01 4.9671e+00 2.0000e+00 2.4200e+02 1.7800e+01 3.9283e+02\n 4.0300e+00]]\n" ], [ "# For two observations:\nso_2 = df.loc[2:3, :]\nprint(np.reshape(so_2.values, (2, -1)))", "[[2.7290e-02 0.0000e+00 7.0700e+00 0.0000e+00 4.6900e-01 7.1850e+00\n 6.1100e+01 4.9671e+00 2.0000e+00 2.4200e+02 1.7800e+01 3.9283e+02\n 4.0300e+00]\n [3.2370e-02 0.0000e+00 2.1800e+00 0.0000e+00 4.5800e-01 6.9980e+00\n 4.5800e+01 6.0622e+00 3.0000e+00 2.2200e+02 1.8700e+01 3.9463e+02\n 2.9400e+00]]\n" ] ], [ [ "This concludes our quick run-through of some basic functionality of the modules. Later on, we will use more and more specialized functions and objects, but for now this allows you to play around with data already.", "_____no_output_____" ], [ "# Visualisation", "_____no_output_____" ], [ "The visualisations often require a bit of tricks and extra lines of code to make things look better. This is often confusing at first, but it will become more and more intuitive once you get the hang of how the general ideas work. We will be working mostly with Matplotlib (often imported as plt), Numpy (np), and pandas (pd). Often, both Matplotlib and pandas offer similar solutions, but one is often slightly more convenient than the other in various situations. Make sure to look up some of the alternatives, as they might also make more sense to you.", "_____no_output_____" ] ], [ [ "# First, we need to import our packages\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport pandas as pd", "_____no_output_____" ] ], [ [ "## Pie and bar chart", "_____no_output_____" ] ], [ [ "# Data to plot\nlabels = 'classification', 'regression', 'time series'\nsizes = [10, 22, 2]\n\ncolors = ['lightblue', 'lightgreen', 'pink']\n\n# Allows us to highlight a certain piece of the pie chart\nexplode = (0.1, 0, 0) \n \n# Plot a pie chart with the pie() function. Notice how various parameters are given for coloring, labels, etc.\n# They should be relatively self-explanatory\nplt.pie(sizes, explode=explode, labels=labels, colors=colors,\n autopct='%1.1f%%', shadow=True, startangle=90)\n \n# This function makes the axes equal, so the circle is round\nplt.axis('equal')\n\n# Add a title to the plot\nplt.title(\"Pie chart of modelling techniques\")\n\n# Finally, show the plot\nplt.show()", "_____no_output_____" ] ], [ [ "Adding a legend:", "_____no_output_____" ] ], [ [ "patches, texts = plt.pie(sizes, colors=colors, shadow=True, startangle=90)\nplt.legend(patches, labels, loc=\"best\")\nplt.axis('equal')\nplt.title(\"Pie chart of modelling techniques\")\nplt.show()", "_____no_output_____" ], [ "# Bar charts are relatively similar. Here we use the bar() function\nplt.bar(labels, sizes, align='center')\nplt.xticks(labels)\nplt.ylabel('#use cases')\nplt.title('Bar chart of modelling technique')\nplt.show()", "_____no_output_____" ] ], [ [ "## Histogram", "_____no_output_____" ] ], [ [ "# This function plots a diagram with the 'data' object providing the data\n# bins are calculated automatically, as indicated by the 'auto' option, which makes them relatively balanced and\n# sets appropriate boundaries\n# color sets the color of the bars\n# the rwidth sets the bars to somewhat slightly less wide than the bins are wide to leave space between the bars\ndata = np.random.normal(10, 2, 1000)\nplt.hist(x= data, bins='auto', color='#008000', rwidth=0.85)\n\n# For more information on colour codes, please visit: https://htmlcolorcodes.com/\n\n# Additionally, some options are added:\n\n# This option sets the grid of the plot to follow the values on the y-axis\nplt.grid(axis='y')\n\n# Adds a label to the x-axis\nplt.xlabel('Value')\n\n# Adds a label to the y-axis\nplt.ylabel('Frequency')\n\n# Adds a title to the plot\nplt.title('Histogram of x')\n\n# Makes the plot visible in the program\nplt.show()", "_____no_output_____" ], [ "# Here, a different color and manually-specified bins are used\nplt.hist(x= data, bins=[0,1,2,3,4,5,6,7,8,9,10], color='olive', rwidth=0.85)\nplt.grid(axis='y')\nplt.xlabel('Value')\nplt.ylabel('Frequency')\nplt.title('Histogram of x and y')\nplt.show()", "_____no_output_____" ] ], [ [ "See how we cut the tail off the distribution.", "_____no_output_____" ] ], [ [ "# Now, let's build a histogram with radomly generated data that follows a normal distribution\n# Mean = 10, stddev = 15, sample size = 1,000\n# More on random numbers will follow in module 2\ns = np.random.normal(10, 15, 1000)\n\nplt.hist(x=s, bins='auto', color='#008000', rwidth=0.85)\nplt.grid(axis='y')\nplt.xlabel('Value')\nplt.ylabel('Frequency')\nplt.title('Histogram of x')\nplt.show()", "_____no_output_____" ] ], [ [ "## Boxplot", "_____no_output_____" ] ], [ [ "# Boxplots are even easier. We can just use the boxplot() function without many parameters\n# We use the implementation of Pandas, which relies on Matplotlib in the background\n# We now use subplots.\ndata = [3,8,3,4,1,7,5,3,8,2,7,3,1,6,10,10,3,6,5,10]\n# Subplot with 1 row, 2 columns, here we add figure 1 of 2 (first row, first column)\nplt.subplot(1,2,1) \nplt.boxplot(data)\n\ndata_2 = [3,8,3,4,1,7,5,3,8,2,7,3,1,6,10,10,3,6,5,10, 99,87,45,-20]\n# Here we add figure 2 of 2, hence it will be positioned in the second column of the first row\nplt.subplot(1,2,2) \nplt.boxplot(data_2)\nplt.show()", "_____no_output_____" ] ], [ [ "Boxplot for multiple variables:", "_____no_output_____" ] ], [ [ "# Generate 4 columns with 10 observations\ndf = pd.DataFrame(data = np.random.random(size=(10,3)), columns = ['class.','reg.','time series'])\nprint(df)\n\nboxplot = df.boxplot()\nplt.title('Triple boxplot')\nplt.show()\n\ndf = pd.DataFrame(data = np.random.random(size=(10,3)), columns = ['class.','reg.','time series'])\ndf['number_of_runs'] = [0,0,0,1,1,2,0,1,2,0]\n\nboxplot = df.boxplot(by='number_of_runs')\nplt.show()", " class. reg. time series\n0 0.402362 0.348025 0.893360\n1 0.496534 0.454527 0.631422\n2 0.268591 0.815153 0.371747\n3 0.596372 0.121358 0.591864\n4 0.575830 0.964928 0.908575\n5 0.380839 0.435604 0.488436\n6 0.788519 0.562830 0.303210\n7 0.424057 0.888664 0.476388\n8 0.699300 0.380225 0.776302\n9 0.463731 0.239730 0.686004\n" ] ], [ [ "## Scatterplot", "_____no_output_____" ] ], [ [ "# We load the data gain\nx = [3,8,3,4,1,7,5,3,8,2,7,3,1,6,10,10,3,6,5,10]\ny = [10,7,2,7,5,4,2,3,4,1,5,7,8,4,10,2,3,4,5,6]\n\n# Here, we build a simple scatterplot of the two variables\nplt.scatter(x,y)\nplt.xlabel('x')\nplt.ylabel('y')\nplt.title('Simple scatterplot')\nplt.show()", "_____no_output_____" ] ], [ [ "Hard to tell which variable is what, but it gives an overall impression of the data.", "_____no_output_____" ] ], [ [ "# A simple line plot\n\n# We use the plot function for this. 'o-' indicates we want to use circles for markers and connect them with lines\nplt.plot(x,'o-',color='blue',)\n\n# Here we use 'x--' for cross-shaped markers connected with intermittent lines\nplt.plot(y,'x--',color='red')\nplt.xlabel('Time')\nplt.ylabel('Value')\nplt.title(\"x and y over time\")\n\n# This function sets the range limits for the x axis at 0 and 20\nplt.xlim(0,20)\n\n# Adding a grid\nplt.grid(True)\n\n# Adding markets on the x and y axis. We start at zero, make our way to 10 (the last integer is not included,\n# hence we use 21 and 11)\n# We add steps of 4 for the x axis, and 4 for the y axis\nplt.xticks(range(0,21,4))\nplt.yticks(range(0,11,2))\n\nplt.show()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "---\n\n_You are currently looking at **version 1.0** of this notebook. To download notebooks and datafiles, as well as get help on Jupyter notebooks in the Coursera platform, visit the [Jupyter Notebook FAQ](https://www.coursera.org/learn/python-text-mining/resources/d9pwm) course resource._\n\n---", "_____no_output_____" ], [ "# Assignment 4 - Document Similarity & Topic Modelling", "_____no_output_____" ], [ "## Part 1 - Document Similarity\n\nFor the first part of this assignment, you will complete the functions `doc_to_synsets` and `similarity_score` which will be used by `document_path_similarity` to find the path similarity between two documents.\n\nThe following functions are provided:\n* **`convert_tag:`** converts the tag given by `nltk.pos_tag` to a tag used by `wordnet.synsets`. You will need to use this function in `doc_to_synsets`.\n* **`document_path_similarity:`** computes the symmetrical path similarity between two documents by finding the synsets in each document using `doc_to_synsets`, then computing similarities using `similarity_score`.\n\nYou will need to finish writing the following functions:\n* **`doc_to_synsets:`** returns a list of synsets in document. This function should first tokenize and part of speech tag the document using `nltk.word_tokenize` and `nltk.pos_tag`. Then it should find each tokens corresponding synset using `wn.synsets(token, wordnet_tag)`. The first synset match should be used. If there is no match, that token is skipped.\n* **`similarity_score:`** returns the normalized similarity score of a list of synsets (s1) onto a second list of synsets (s2). For each synset in s1, find the synset in s2 with the largest similarity value. Sum all of the largest similarity values together and normalize this value by dividing it by the number of largest similarity values found. Be careful with data types, which should be floats. Missing values should be ignored.\n\nOnce `doc_to_synsets` and `similarity_score` have been completed, submit to the autograder which will run `test_document_path_similarity` to test that these functions are running correctly. \n\n*Do not modify the functions `convert_tag`, `document_path_similarity`, and `test_document_path_similarity`.*", "_____no_output_____" ] ], [ [ "import numpy as np\nimport nltk\n\nnltk.download('punkt')\nnltk.download('averaged_perceptron_tagger')\nnltk.download('wordnet')\n\nfrom nltk.corpus import wordnet as wn\nimport pandas as pd\n\n\ndef convert_tag(tag):\n \"\"\"Convert the tag given by nltk.pos_tag to the tag used by wordnet.synsets\"\"\"\n \n tag_dict = {'N': 'n', 'J': 'a', 'R': 'r', 'V': 'v'}\n try:\n return tag_dict[tag[0]]\n except KeyError:\n return None\n\n\ndef doc_to_synsets(doc):\n \"\"\"\n Returns a list of synsets in document.\n\n Tokenizes and tags the words in the document doc.\n Then finds the first synset for each word/tag combination.\n If a synset is not found for that combination it is skipped.\n\n Args:\n doc: string to be converted\n\n Returns:\n list of synsets\n\n Example:\n doc_to_synsets('Fish are nvqjp friends.')\n Out: [Synset('fish.n.01'), Synset('be.v.01'), Synset('friend.n.01')]\n \"\"\"\n \n # Your Code Here\n tokens = nltk.word_tokenize(doc)\n tags = [tag[1] for tag in nltk.pos_tag(tokens)]\n wordnet_tags = [convert_tag(tag) for tag in tags]\n synsets = [wn.synsets(token, wordnet_tag) for token, wordnet_tag in list(zip(tokens, wordnet_tags))]\n answer = [i[0] for i in synsets if len(i) > 0]\n \n return answer # Your Answer Here\n\n\ndef similarity_score(s1, s2):\n \"\"\"\n Calculate the normalized similarity score of s1 onto s2\n\n For each synset in s1, finds the synset in s2 with the largest similarity value.\n Sum of all of the largest similarity values and normalize this value by dividing it by the\n number of largest similarity values found.\n\n Args:\n s1, s2: list of synsets from doc_to_synsets\n\n Returns:\n normalized similarity score of s1 onto s2\n\n Example:\n synsets1 = doc_to_synsets('I like cats')\n synsets2 = doc_to_synsets('I like dogs')\n similarity_score(synsets1, synsets2)\n Out: 0.73333333333333339\n \"\"\"\n \n \n # Your Code Here\n \n lvs = [] # largest similarity values\n for i1 in s1:\n scores=[x for x in [i1.path_similarity(i2) for i2 in s2] if x is not None]\n if scores:\n lvs.append(max(scores))\n \n return sum(lvs) / len(lvs)# Your Answer Here\n\n\ndef document_path_similarity(doc1, doc2):\n \"\"\"Finds the symmetrical similarity between doc1 and doc2\"\"\"\n\n synsets1 = doc_to_synsets(doc1)\n synsets2 = doc_to_synsets(doc2)\n\n return (similarity_score(synsets1, synsets2) + similarity_score(synsets2, synsets1)) / 2", "[nltk_data] Downloading package punkt to /home/jovyan/nltk_data...\n[nltk_data] Package punkt is already up-to-date!\n[nltk_data] Downloading package averaged_perceptron_tagger to\n[nltk_data] /home/jovyan/nltk_data...\n[nltk_data] Package averaged_perceptron_tagger is already up-to-\n[nltk_data] date!\n[nltk_data] Downloading package wordnet to /home/jovyan/nltk_data...\n[nltk_data] Package wordnet is already up-to-date!\n" ] ], [ [ "### test_document_path_similarity\n\nUse this function to check if doc_to_synsets and similarity_score are correct.\n\n*This function should return the similarity score as a float.*", "_____no_output_____" ] ], [ [ "def test_document_path_similarity():\n doc1 = 'This is a function to test document_path_similarity.'\n doc2 = 'Use this function to see if your code in doc_to_synsets \\\n and similarity_score is correct!'\n return document_path_similarity(doc1, doc2)", "_____no_output_____" ], [ "test_document_path_similarity()", "_____no_output_____" ] ], [ [ "<br>\n___\n`paraphrases` is a DataFrame which contains the following columns: `Quality`, `D1`, and `D2`.\n\n`Quality` is an indicator variable which indicates if the two documents `D1` and `D2` are paraphrases of one another (1 for paraphrase, 0 for not paraphrase).", "_____no_output_____" ] ], [ [ "# Use this dataframe for questions most_similar_docs and label_accuracy\nparaphrases = pd.read_csv('paraphrases.csv')\nparaphrases.head()", "_____no_output_____" ] ], [ [ "___\n\n### most_similar_docs\n\nUsing `document_path_similarity`, find the pair of documents in paraphrases which has the maximum similarity score.\n\n*This function should return a tuple `(D1, D2, similarity_score)`*", "_____no_output_____" ] ], [ [ "def most_similar_docs():\n \n # Your Code Here\n \n return max(map(document_path_similarity, paraphrases['D1'], paraphrases['D2'])) # Your Answer Here", "_____no_output_____" ], [ "most_similar_docs()", "_____no_output_____" ] ], [ [ "### label_accuracy\n\nProvide labels for the twenty pairs of documents by computing the similarity for each pair using `document_path_similarity`. Let the classifier rule be that if the score is greater than 0.75, label is paraphrase (1), else label is not paraphrase (0). Report accuracy of the classifier using scikit-learn's accuracy_score.\n\n*This function should return a float.*", "_____no_output_____" ] ], [ [ "def label_accuracy():\n from sklearn.metrics import accuracy_score\n\n paraphrases['labels'] = [1 if i > 0.75 else 0 for i in map(document_path_similarity, paraphrases['D1'], paraphrases['D2'])] # Your Code Here\n \n return accuracy_score(paraphrases['Quality'], paraphrases['labels']) # Your Answer Here", "_____no_output_____" ], [ "label_accuracy()", "_____no_output_____" ] ], [ [ "## Part 2 - Topic Modelling\n\nFor the second part of this assignment, you will use Gensim's LDA (Latent Dirichlet Allocation) model to model topics in `newsgroup_data`. You will first need to finish the code in the cell below by using gensim.models.ldamodel.LdaModel constructor to estimate LDA model parameters on the corpus, and save to the variable `ldamodel`. Extract 10 topics using `corpus` and `id_map`, and with `passes=25` and `random_state=34`.", "_____no_output_____" ] ], [ [ "import pickle\nimport gensim\nfrom sklearn.feature_extraction.text import CountVectorizer\n\n# Load the list of documents\nwith open('newsgroups', 'rb') as f:\n newsgroup_data = pickle.load(f)\n\n# Use CountVectorizor to find three letter tokens, remove stop_words, \n# remove tokens that don't appear in at least 20 documents,\n# remove tokens that appear in more than 20% of the documents\nvect = CountVectorizer(min_df=20, max_df=0.2, stop_words='english', \n token_pattern='(?u)\\\\b\\\\w\\\\w\\\\w+\\\\b')\n# Fit and transform\nX = vect.fit_transform(newsgroup_data)\n\n# Convert sparse matrix to gensim corpus.\ncorpus = gensim.matutils.Sparse2Corpus(X, documents_columns=False)\n\n# Mapping from word IDs to words (To be used in LdaModel's id2word parameter)\nid_map = dict((v, k) for k, v in vect.vocabulary_.items())\n", "_____no_output_____" ], [ "# Use the gensim.models.ldamodel.LdaModel constructor to estimate \n# LDA model parameters on the corpus, and save to the variable `ldamodel`\n\n# Your code here:\nldamodel = gensim.models.ldamodel.LdaModel(corpus=corpus, num_topics=10, id2word=id_map, passes=25, random_state=34)", "_____no_output_____" ] ], [ [ "### lda_topics\n\nUsing `ldamodel`, find a list of the 10 topics and the most significant 10 words in each topic. This should be structured as a list of 10 tuples where each tuple takes on the form:\n\n`(9, '0.068*\"space\" + 0.036*\"nasa\" + 0.021*\"science\" + 0.020*\"edu\" + 0.019*\"data\" + 0.017*\"shuttle\" + 0.015*\"launch\" + 0.015*\"available\" + 0.014*\"center\" + 0.014*\"sci\"')`\n\nfor example.\n\n*This function should return a list of tuples.*", "_____no_output_____" ] ], [ [ "def lda_topics():\n \n # Your Code Here\n \n return ldamodel.print_topics(num_topics=10, num_words=10) # Your Answer Here", "_____no_output_____" ], [ "lda_topics()", "_____no_output_____" ] ], [ [ "### topic_distribution\n\nFor the new document `new_doc`, find the topic distribution. Remember to use vect.transform on the the new doc, and Sparse2Corpus to convert the sparse matrix to gensim corpus.\n\n*This function should return a list of tuples, where each tuple is `(#topic, probability)`*", "_____no_output_____" ] ], [ [ "new_doc = [\"\\n\\nIt's my understanding that the freezing will start to occur because \\\nof the\\ngrowing distance of Pluto and Charon from the Sun, due to it's\\nelliptical orbit. \\\nIt is not due to shadowing effects. \\n\\n\\nPluto can shadow Charon, and vice-versa.\\n\\nGeorge \\\nKrumins\\n-- \"]", "_____no_output_____" ], [ "def topic_distribution():\n \n # Your Code Here\n \n # Transform\n X = vect.transform(new_doc)\n\n # Convert sparse matrix to gensim corpus.\n corpus = gensim.matutils.Sparse2Corpus(X, documents_columns=False)\n \n return list(ldamodel[corpus])[0] # Your Answer Here", "_____no_output_____" ], [ "topic_distribution()", "_____no_output_____" ] ], [ [ "### topic_names\n\nFrom the list of the following given topics, assign topic names to the topics you found. If none of these names best matches the topics you found, create a new 1-3 word \"title\" for the topic.\n\nTopics: Health, Science, Automobiles, Politics, Government, Travel, Computers & IT, Sports, Business, Society & Lifestyle, Religion, Education.\n\n*This function should return a list of 10 strings.*", "_____no_output_____" ] ], [ [ "def topic_names():\n \n # Your Code Here\n \n return ['Automobiles', 'Health', 'Science',\n 'Politics',\n 'Sports',\n 'Business', 'Society & Lifestyle',\n 'Religion', 'Education', 'Computers & IT'] # Your Answer Here", "_____no_output_____" ], [ "topic_names()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# NOAA extreme weather events\nThe [National Oceanic and Atmospheric Administration](https://en.wikipedia.org/wiki/National_Oceanic_and_Atmospheric_Administration) has a database of extreme weather events that contains lots of detail for every year ([Link](https://www.climate.gov/maps-data/dataset/severe-storms-and-extreme-events-data-table)). In this notebook I will create map files for individual weather events, mapped to their coordinates.", "_____no_output_____" ] ], [ [ "import pandas as pd\nimport numpy as np\nimport random\nimport geopandas\nimport matplotlib.pyplot as plt\npd.set_option('display.max_columns', None) # Unlimited columns\n\n# Custom function for displaying the shape and head of a dataframe\ndef display(df, n=5):\n print(df.shape)\n return df.head(n) ", "_____no_output_____" ] ], [ [ "# Get map of US counties", "_____no_output_____" ] ], [ [ "# Import a shape file with all the counties in the US.\n# Note how it doesn't include all the same territories as the \n# quake contour map.\ncounties = geopandas.read_file('../data_input/1_USCounties/')\n\n# Turn state codes from strings to integers\nfor col in ['STATE_FIPS', 'CNTY_FIPS', 'FIPS']:\n counties[col] = counties[col].astype(int)", "_____no_output_____" ] ], [ [ "# Process NOAA data for one year only\nAs a starting point that I'll generalize later.", "_____no_output_____" ] ], [ [ "# Get NOAA extreme weather event data for one year\ndf1 = pd.read_csv('../data_local/NOAA/StormEvents_details-ftp_v1.0_d2018_c20190422.csv')\nprint(df1.shape)\nprint(df1.columns)\ndf1.head(2)", "(62169, 51)\nIndex(['BEGIN_YEARMONTH', 'BEGIN_DAY', 'BEGIN_TIME', 'END_YEARMONTH',\n 'END_DAY', 'END_TIME', 'EPISODE_ID', 'EVENT_ID', 'STATE', 'STATE_FIPS',\n 'YEAR', 'MONTH_NAME', 'EVENT_TYPE', 'CZ_TYPE', 'CZ_FIPS', 'CZ_NAME',\n 'WFO', 'BEGIN_DATE_TIME', 'CZ_TIMEZONE', 'END_DATE_TIME',\n 'INJURIES_DIRECT', 'INJURIES_INDIRECT', 'DEATHS_DIRECT',\n 'DEATHS_INDIRECT', 'DAMAGE_PROPERTY', 'DAMAGE_CROPS', 'SOURCE',\n 'MAGNITUDE', 'MAGNITUDE_TYPE', 'FLOOD_CAUSE', 'CATEGORY', 'TOR_F_SCALE',\n 'TOR_LENGTH', 'TOR_WIDTH', 'TOR_OTHER_WFO', 'TOR_OTHER_CZ_STATE',\n 'TOR_OTHER_CZ_FIPS', 'TOR_OTHER_CZ_NAME', 'BEGIN_RANGE',\n 'BEGIN_AZIMUTH', 'BEGIN_LOCATION', 'END_RANGE', 'END_AZIMUTH',\n 'END_LOCATION', 'BEGIN_LAT', 'BEGIN_LON', 'END_LAT', 'END_LON',\n 'EPISODE_NARRATIVE', 'EVENT_NARRATIVE', 'DATA_SOURCE'],\n dtype='object')\n" ], [ "# Extract only a few useful columns\ndf2 = df1[['TOR_F_SCALE','EVENT_TYPE','BEGIN_LAT','BEGIN_LON']].copy()\n\n# Remove any rows with null coordinates\ndf2 = df2.dropna(subset=['BEGIN_LAT','BEGIN_LON'])\n\n# Create geoDF of all the points \ndf3 = geopandas.GeoDataFrame(\n df2, geometry=geopandas.points_from_xy(df2.BEGIN_LON, df2.BEGIN_LAT))\n\n# Trim the list of events to only include those that happened within one of our official counties.\ndf4 = geopandas.sjoin(df3, counties, how='left', op='within').dropna(subset=['FIPS'])\n\n# Drop useless columns\ndf4 = df4[['TOR_F_SCALE','EVENT_TYPE','geometry']]\n\n# Add new columns for event categories\n\n\nflood_types =['Flood','Flash Flood','Coastal Flood',\n 'Storm Surge/Tide','Lakeshore Flood','Debris Flow'] \ndf4['Flood'] = df4['EVENT_TYPE'].isin(flood_types)\n\n\n\nstorm_types = ['Thunderstorm Wind','Marine Thunderstorm Wind','Marine High Wind',\n 'High Wind','Funnel Cloud','Dust Storm',\n 'Strong Wind','Dust Devil','Tropical Depression','Lightning',\n 'Tropical Storm','High Surf','Heavy Rain','Hail','Marine Hail',\n 'Marine Strong Wind','Waterspout']\ndf4['Storm'] = df4['EVENT_TYPE'].isin(storm_types)\n\n\ndf4['Tornado'] = df4['EVENT_TYPE'].isin(['Tornado'])\n\n\n# Reorganize columns\ntype_columns = ['Storm','Flood','Tornado']\ndf4 = df4[['TOR_F_SCALE','EVENT_TYPE','geometry'] + type_columns]\n\ndisplay(df4)", "(35940, 6)\n" ], [ "# Plot over a map of US counties\nfig, ax = plt.subplots(figsize=(20,20))\ncounties.plot(ax=ax, color='white', edgecolor='black');\ndf4.plot(ax=ax, marker='o')\n# ax.set_xlim(-125,-114)\nax.set_ylim(15,75)\nplt.show()", "_____no_output_____" ] ], [ [ "## NOAA file processing function\nGeneralize the previous operations so they can apply to the data for any year", "_____no_output_____" ] ], [ [ "def process_noaa(filepath):\n \"\"\"\n Process one year of NOAA Extreme weather events. Requires\n the list of official counties and the list of official weather\n event types.\n \n Inputs\n ------\n filepath (string) : file path for the list of events from one year.\n \n \n Outputs\n -------\n result (pandas.DataFrame) : Dataframe each event for that year, with\n boolean columns for each event category.\n \n \"\"\"\n \n df1 = pd.read_csv(filepath)\n \n # Extract only a few useful columns\n df2 = df1[['TOR_F_SCALE','EVENT_TYPE','BEGIN_LAT','BEGIN_LON']].copy()\n\n # Remove any rows with null coordinates\n df2 = df2.dropna(subset=['BEGIN_LAT','BEGIN_LON'])\n\n # Create geoDF of all the points \n df3 = geopandas.GeoDataFrame(\n df2, geometry=geopandas.points_from_xy(df2.BEGIN_LON, df2.BEGIN_LAT))\n\n # Trim the list of events to only include those that happened within one of our official counties.\n df4 = geopandas.sjoin(df3, counties, how='left', op='within').dropna(subset=['FIPS'])\n\n # Drop useless columns\n df4 = df4[['TOR_F_SCALE','EVENT_TYPE','geometry']]\n\n # Add new columns for event categories\n\n\n flood_types =['Flood','Flash Flood','Coastal Flood',\n 'Storm Surge/Tide','Lakeshore Flood','Debris Flow'] \n df4['Flood'] = df4['EVENT_TYPE'].isin(flood_types)\n\n\n\n storm_types = ['Thunderstorm Wind','Marine Thunderstorm Wind','Marine High Wind',\n 'High Wind','Funnel Cloud','Dust Storm',\n 'Strong Wind','Dust Devil','Tropical Depression','Lightning',\n 'Tropical Storm','High Surf','Heavy Rain','Hail','Marine Hail',\n 'Marine Strong Wind','Waterspout']\n df4['Storm'] = df4['EVENT_TYPE'].isin(storm_types)\n\n\n df4['Tornado'] = df4['EVENT_TYPE'].isin(['Tornado'])\n\n\n # Reorganize columns\n type_columns = ['Storm','Flood','Tornado']\n df4 = df4[['TOR_F_SCALE','EVENT_TYPE','geometry'] + type_columns]\n \n # Add a column for the year of this file\n year = int(filepath[49:53])\n df4['year'] = year\n\n return df4", "_____no_output_____" ], [ "# Example\ntest_2018 = process_noaa('../data_local/NOAA/StormEvents_details-ftp_v1.0_d2018_c20190422.csv')\ndisplay(test_2018)", "(35940, 7)\n" ], [ "# These are the extreme weather events recorded in 2018\ntest_2018[type_columns].sum().sort_values(ascending=False)", "_____no_output_____" ] ], [ [ "# Process all the available data", "_____no_output_____" ] ], [ [ "import glob\nimport os\n\n# Read the CSV files for each year going back to 1996 (the first year \n# when many of these event types started being recorded)\npath = '../data_local/NOAA/'\nfilenames = sorted(glob.glob(os.path.join(path, '*.csv')))\nlayers = []\n\n# Aggregate the dataframes in a list\nfor name in filenames:\n year = int(name[49:53])\n print(f'Processing {year}')\n layers.append(process_noaa(name))\n\n# Concatenate all these dataframes into a single dataframe\nnoaa = pd.concat(layers)", "Processing 1996\nProcessing 1997\nProcessing 1998\nProcessing 1999\nProcessing 2000\nProcessing 2001\nProcessing 2002\nProcessing 2003\nProcessing 2004\nProcessing 2005\nProcessing 2006\nProcessing 2007\nProcessing 2008\nProcessing 2009\nProcessing 2010\nProcessing 2011\nProcessing 2012\nProcessing 2013\nProcessing 2014\nProcessing 2015\nProcessing 2016\nProcessing 2017\nProcessing 2018\n" ], [ "display(noaa)", "(731944, 7)\n" ], [ "# total events per type\nnoaa[type_columns].sum()", "_____no_output_____" ], [ "# Aggregate event types into different geopandas dataframes.\nstorms = noaa[noaa['Storm']][['EVENT_TYPE','year','geometry']].reset_index(drop=True)\nfloods = noaa[noaa['Flood']][['EVENT_TYPE','year','geometry']].reset_index(drop=True)\ntornadoes = noaa[noaa['Tornado']][['TOR_F_SCALE','year','geometry']].reset_index(drop=True)\n\nstorms.shape, floods.shape, tornadoes.shape", "_____no_output_____" ] ], [ [ "## Process tornado data\nIn 2007, the National Weather Service (NWS) switched their scale for measuring tornado intensity, from the Fujita (F) scale to the Enhanced Fujita (EF) scale. I will lump them together here and just make a note for the user that the scale means something slightly different before and after 2007. Also, I'll cast unknown magnitudes (EFU) as if they were EF0.", "_____no_output_____" ] ], [ [ "# Tornadoes by magnitude, using the NWS's original labels.\n# Notice the two different scales and also a label for 'unknown'\ntornadoes.TOR_F_SCALE.value_counts()", "_____no_output_____" ], [ "# Function that extracts the scale level and sets unkwnown to zero.\ndef process_fujita(x):\n if x[-1] == 'U':\n return 0\n else:\n return int(x[-1])\n\ntornadoes['intensity'] = tornadoes['TOR_F_SCALE'].apply(process_fujita)\ntornadoes = tornadoes.drop(columns='TOR_F_SCALE')\n\ndisplay(tornadoes)", "(30898, 3)\n" ], [ "# Distribution of tornado intensities.\ntornadoes.intensity.hist();", "_____no_output_____" ] ], [ [ "# Visualizing the data", "_____no_output_____" ] ], [ [ "# Sample of 2000 storms in the Lower48\nfig, ax = plt.subplots(figsize=(20,20))\ncounties.plot(ax=ax, color='white', edgecolor='black');\nstorms.sample(2000).plot(ax=ax, marker='o')\nax.set_xlim(-125.0011,-66.9326)\nax.set_ylim(24.9493, 49.5904)\nplt.show()", "_____no_output_____" ] ], [ [ "Floods and tornadoes show basically the same distribution, so I won't plot them separately. For reference, this is what the dataframes that we're about to export look like.", "_____no_output_____" ] ], [ [ "display(storms)", "(622110, 3)\n" ], [ "display(floods)", "(78936, 3)\n" ], [ "display(tornadoes)", "(30898, 3)\n" ] ], [ [ "# Export!", "_____no_output_____" ] ], [ [ "storms.to_file(\"../data_output/5__NOAA/storms.geojson\",\n driver='GeoJSON')\nfloods.to_file(\"../data_output/5__NOAA/floods.geojson\",\n driver='GeoJSON')\ntornadoes.to_file(\"../data_output/5__NOAA/tornadoes.geojson\",\n driver='GeoJSON')", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import local_models.local_models as local_models\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport sklearn.linear_model\nimport sklearn.cluster\nfrom importlib import reload\nfrom ml_battery.utils import cmap\nimport matplotlib as mpl\nreload(local_models)\nmpl.rcParams['figure.figsize'] = [8.0, 8.0]\nmpl.rcParams['font.size'] = 20\nmpl.rcParams['figure.autolayout'] = True\nnp.random.seed(1)", "_____no_output_____" ], [ "n = 100\nsigma = 0.05\nk = 15", "_____no_output_____" ], [ "X1 = np.linspace(1.,3.,n)\nX2 = np.linspace(0.,2.,n)", "_____no_output_____" ], [ "y1 = -3*X1 + 8 + np.random.normal(0,sigma,n)", "_____no_output_____" ], [ "y2 = -3*X2 + 0 + np.random.normal(0,sigma,n)", "_____no_output_____" ], [ "colors = np.concatenate((np.ones(n), np.zeros(n)))", "_____no_output_____" ], [ "y = np.concatenate((y1,y2))\nx = np.concatenate((X1,X2))", "_____no_output_____" ], [ "print(x.shape, y.shape)", "(200,) (200,)\n" ], [ "plt.scatter(x,y,c=cmap(colors))\nplt.show()", "/usr/local/lib/python3.5/dist-packages/matplotlib/figure.py:1999: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.\n warnings.warn(\"This figure includes Axes that are not compatible \"\n" ], [ "lr = sklearn.linear_model.LinearRegression()\nlr.fit(x.reshape((-1,1)),y)\nxx = np.linspace(0,4,100)\npred = lr.predict(xx.reshape((-1,1)))\nplt.plot(xx, pred, c='r')\nplt.scatter(x,y)#,c=cmap(colors))\nplt.xlabel('x')\nplt.ylabel('y')\nplt.savefig(\"parallel_lines_w_regression.png\")\nplt.show()", "/usr/local/lib/python3.5/dist-packages/matplotlib/figure.py:1999: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.\n warnings.warn(\"This figure includes Axes that are not compatible \"\n" ], [ "km = sklearn.cluster.KMeans(2)\nkm.fit(x.reshape((-1,1)))\npred = km.predict(x.reshape((-1,1)))\nplt.scatter(x,y,c=cmap(pred))\nplt.show()", "/usr/local/lib/python3.5/dist-packages/matplotlib/figure.py:1999: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.\n warnings.warn(\"This figure includes Axes that are not compatible \"\n" ], [ "models = local_models.LocalModels(sklearn.linear_model.LinearRegression())", "_____no_output_____" ], [ "models.fit(x.reshape((-1,1)), y, np.stack((x,y)).T)", "_____no_output_____" ], [ "lm_params = models.transform(np.stack((x,y)).T,k=k)", "_____no_output_____" ], [ "lr.__dict__", "_____no_output_____" ], [ "plt.scatter(lm_params[:,0], lm_params[:,1])\nplt.scatter([lr.coef_], [lr.intercept_], c='r')\nplt.xlabel('m')\nplt.ylabel('b')\nplt.savefig('parallel_lines_local_linear_feature_transform.png')\nplt.show()", "/usr/local/lib/python3.5/dist-packages/matplotlib/figure.py:1999: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.\n warnings.warn(\"This figure includes Axes that are not compatible \"\n" ], [ "clf = sklearn.cluster.KMeans(2)", "_____no_output_____" ], [ "clf.fit(lm_params)", "_____no_output_____" ], [ "plt.scatter(x,y,c=cmap(clf.predict(lm_params)))\nplt.show()", "/usr/local/lib/python3.5/dist-packages/matplotlib/figure.py:1999: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.\n warnings.warn(\"This figure includes Axes that are not compatible \"\n" ], [ "g = np.mgrid[0:4:0.02, -6:6:0.06]\nxx = np.vstack(map(np.ravel, g)).T", "_____no_output_____" ], [ "pred = clf.predict(models.transform(xx,k=k))", "_____no_output_____" ], [ "pred.shape", "_____no_output_____" ], [ "cc = np.array(list(map(mpl.colors.to_rgb, np.unique(cmap(pred)))))\nplt.contourf(g[0], g[1], pred.reshape(g[0].shape), 1, colors=cc)\nplt.scatter(x,y, c=cmap(clf.predict(lm_params)), linewidths=1, edgecolors='k')\nplt.xlabel('x')\nplt.ylabel('y')\nplt.savefig('parallel_lines_kmeans_w_local_linear_features.png')\nplt.show()", "/usr/local/lib/python3.5/dist-packages/matplotlib/figure.py:1999: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.\n warnings.warn(\"This figure includes Axes that are not compatible \"\n" ], [ "clf2 = sklearn.cluster.KMeans(2)", "_____no_output_____" ], [ "clf2.fit(np.stack((x,y)).T)", "_____no_output_____" ], [ "pred2 = clf2.predict(xx)", "_____no_output_____" ], [ "plt.scatter(xx[:,0], xx[:,1], c=cmap(pred2), s=6)\nplt.scatter(x,y, c=cmap(clf2.predict(np.stack((x,y)).T)), linewidths=1, edgecolors=\"k\")\nplt.xlabel('x')\nplt.ylabel('y')\nplt.savefig('parallel_lines_kmeans.png')\nplt.show()", "/usr/local/lib/python3.5/dist-packages/matplotlib/figure.py:1999: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.\n warnings.warn(\"This figure includes Axes that are not compatible \"\n" ], [ "import numpy as np\nimport matplotlib.pyplot as plt\nfrom matplotlib.animation import FuncAnimation\nimport matplotlib.animation as animation\nfrom IPython.display import HTML\n%matplotlib inline", "_____no_output_____" ], [ "print(animation.writers.list())", "['imagemagick_file', 'ffmpeg_file', 'imagemagick', 'ffmpeg', 'html']\n" ], [ "xdata, ydata = x, y\noffsets = np.stack((xdata, ydata)).T\nmultioffsets = []\nfor i in range(100):\n offsets[:,1] += np.random.normal(0,0.05,offsets.shape[0])\n multioffsets.append(np.copy(offsets))\n\nfig = plt.figure()\ndecision_surface = plt.scatter(xx[:,0], xx[:,1], c=cmap(pred), s=6, animated=True)\nscat = plt.scatter(x,y, c=cmap(clf.predict(lm_params)), linewidths=1, edgecolors='k', animated=True)\n\ndef init():\n return scat, decision_surface\n\ndef update(frame):\n offsets = multioffsets[frame]\n models.fit(offsets[:,:1], offsets[:,1:], offsets)\n pred = clf.predict(models.transform(xx,k=k))\n decision_surface.set_color(cmap(pred))\n scat.set_offsets(np.copy(multioffsets[frame]))\n return scat, decision_surface\n\n\nani = FuncAnimation(fig, update, frames=range(100),\n init_func=init, blit=True)\n\n# Set up formatting for the movie files\nWriter = animation.writers['ffmpeg']\nwriter = Writer(fps=15, metadata=dict(artist='CScott!'), bitrate=1800)\nani.save('ani_k10_double.mp4', writer=writer)\n\nHTML(ani.to_html5_video())\n#offsets[:,1] += np.random.normal(0,0.05,offsets.shape[0])\n#scat.set_offsets(offsets)\n#scat", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/iotanalytics/IoTTutorial/blob/main/code/clustering_and_classification/1D_CNN.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "## **1D Convolutional Neural Networks**\n\n\"A Convolutional Neural Network (ConvNet/CNN) is a Deep Learning algorithm which can take in an input image, assign importance (learnable weights and biases) to various aspects/objects in the image and be able to differentiate one from the other. The pre-processing required in a ConvNet is much lower as compared to other classification algorithms. While in primitive methods filters are hand-engineered, with enough training, ConvNets have the ability to learn these filters/characteristics.\" [4]\n\n\"The architecture of a ConvNet is analogous to that of the connectivity pattern of Neurons in the Human Brain and was inspired by the organization of the Visual Cortex. Individual neurons respond to stimuli only in a restricted region of the visual field known as the Receptive Field. A collection of such fields overlap to cover the entire visual area.\" [4]\n\n\"Convolutional neural network models were developed for image classification problems, where the model learns an internal representation of a two-dimensional input, in a process referred to as feature learning.\" [1]\n\n\"This same process can be harnessed on one-dimensional sequences of data, such as in the case of acceleration and gyroscopic data for human activity recognition. The model learns to extract features from sequences of observations and how to map the internal features to different activity types.\" [1]\n\n\"The benefit of using CNNs for sequence classification is that they can learn from the raw time series data directly, and in turn do not require domain expertise to manually engineer input features. The model can learn an internal representation of the time series data and ideally achieve comparable performance to models fit on a version of the dataset with engineered features.\" [1]\n\n**Convolutional Neural Network Architecture**\n\"A CNN typically has three layers: a convolutional layer, a pooling layer, and a fully connected layer.\" [5]\n\n\n\n**Convolution Layer**\n\"The convolution layer is the core building block of the CNN. It carries the main portion of the network’s computational load.\" [5]\n\n\"This layer performs a dot product between two matrices, where one matrix is the set of learnable parameters otherwise known as a kernel, and the other matrix is the restricted portion of the receptive field. The kernel is spatially smaller than an image but is more in-depth. This means that, if the image is composed of three (RGB) channels, the kernel height and width will be spatially small, but the depth extends up to all three channels.\" [5]\n\n\"During the forward pass, the kernel slides across the height and width of the image-producing the image representation of that receptive region. This produces a two-dimensional representation of the image known as an activation map that gives the response of the kernel at each spatial position of the image. The sliding size of the kernel is called a stride.\nIf we have an input of size W x W x D and Dout number of kernels with a spatial size of F with stride S and amount of padding P, then the size of output volume can be determined by the following formula:\" [5]\n\n\n\n**Pooling Layer**\n\"The pooling layer replaces the output of the network at certain locations by deriving a summary statistic of the nearby outputs. This helps in reducing the spatial size of the representation, which decreases the required amount of computation and weights. The pooling operation is processed on every slice of the representation individually.\" [5]\n\n\"There are several pooling functions such as the average of the rectangular neighborhood, L2 norm of the rectangular neighborhood, and a weighted average based on the distance from the central pixel. However, the most popular process is max pooling, which reports the maximum output from the neighborhood.\" [5]\n\n\"If we have an activation map of size W x W x D, a pooling kernel of spatial size F, and stride S, then the size of output volume can be determined by the following formula:\" [5]\n\n\n\n\"This will yield an output volume of size Wout x Wout x D.\nIn all cases, pooling provides some translation invariance which means that an object would be recognizable regardless of where it appears on the frame.\" [5]\n\n**Fully Connected Layer**\n\"Neurons in this layer have full connectivity with all neurons in the preceding and succeeding layer as seen in regular FCNN. This is why it can be computed as usual by a matrix multiplication followed by a bias effect.\" [5]\n\n\"The FC layer helps to map the representation between the input and the output.\" [5]\n\n**Non-Linearity Layers**\n\"Since convolution is a linear operation and images are far from linear, non-linearity layers are often placed directly after the convolutional layer to introduce non-linearity to the activation map.\" [5]\n\n\"There are several types of non-linear operations, the popular ones being:\" [5]\n1. Sigmoid\n2. Tanh\n3. ReLU\n\n</br>\n\n**Advantages:**\n1. Speed vs. other type of neural networks [2]\n2. Capacity to extract the most important features automatically [2]\n\n**Disadvantages:**\n1. Classification of similar objects with different Positions [3]\n2. Vulnerable to adversarial examples [3]\n3. Coordinate Frame [3]\n4. Other minor disadvantages like performance [3]\n\n**References:**\n1. https://machinelearningmastery.com/cnn-models-for-human-activity-recognition-time-series-classification/\n2. https://cai.tools.sap/blog/ml-spotlight-cnn/\n3. https://iq.opengenus.org/disadvantages-of-cnn/\n4. https://towardsdatascience.com/a-comprehensive-guide-to-convolutional-neural-networks-the-eli5-way-3bd2b1164a53\n5. https://towardsdatascience.com/convolutional-neural-networks-explained-9cc5188c4939\n\n", "_____no_output_____" ] ], [ [ "pip install tensorflow", "Requirement already satisfied: tensorflow in /usr/local/lib/python3.7/dist-packages (2.5.0)\nRequirement already satisfied: tensorflow-estimator<2.6.0,>=2.5.0rc0 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (2.5.0)\nRequirement already satisfied: flatbuffers~=1.12.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (1.12)\nRequirement already satisfied: protobuf>=3.9.2 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (3.17.3)\nRequirement already satisfied: astunparse~=1.6.3 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (1.6.3)\nRequirement already satisfied: keras-nightly~=2.5.0.dev in /usr/local/lib/python3.7/dist-packages (from tensorflow) (2.5.0.dev2021032900)\nRequirement already satisfied: wrapt~=1.12.1 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (1.12.1)\nRequirement already satisfied: absl-py~=0.10 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (0.12.0)\nRequirement already satisfied: wheel~=0.35 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (0.37.0)\nRequirement already satisfied: numpy~=1.19.2 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (1.19.5)\nRequirement already satisfied: google-pasta~=0.2 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (0.2.0)\nRequirement already satisfied: tensorboard~=2.5 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (2.5.0)\nRequirement already satisfied: h5py~=3.1.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (3.1.0)\nRequirement already satisfied: six~=1.15.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (1.15.0)\nRequirement already satisfied: opt-einsum~=3.3.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (3.3.0)\nRequirement already satisfied: typing-extensions~=3.7.4 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (3.7.4.3)\nRequirement already satisfied: grpcio~=1.34.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (1.34.1)\nRequirement already satisfied: keras-preprocessing~=1.1.2 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (1.1.2)\nRequirement already satisfied: gast==0.4.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (0.4.0)\nRequirement already satisfied: termcolor~=1.1.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow) (1.1.0)\nRequirement already satisfied: cached-property in /usr/local/lib/python3.7/dist-packages (from h5py~=3.1.0->tensorflow) (1.5.2)\nRequirement already satisfied: werkzeug>=0.11.15 in /usr/local/lib/python3.7/dist-packages (from tensorboard~=2.5->tensorflow) (1.0.1)\nRequirement already satisfied: google-auth<2,>=1.6.3 in /usr/local/lib/python3.7/dist-packages (from tensorboard~=2.5->tensorflow) (1.34.0)\nRequirement already satisfied: requests<3,>=2.21.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard~=2.5->tensorflow) (2.23.0)\nRequirement already satisfied: google-auth-oauthlib<0.5,>=0.4.1 in /usr/local/lib/python3.7/dist-packages (from tensorboard~=2.5->tensorflow) (0.4.5)\nRequirement already satisfied: tensorboard-data-server<0.7.0,>=0.6.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard~=2.5->tensorflow) (0.6.1)\nRequirement already satisfied: markdown>=2.6.8 in /usr/local/lib/python3.7/dist-packages (from tensorboard~=2.5->tensorflow) (3.3.4)\nRequirement already satisfied: tensorboard-plugin-wit>=1.6.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard~=2.5->tensorflow) (1.8.0)\nRequirement already satisfied: setuptools>=41.0.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard~=2.5->tensorflow) (57.2.0)\nRequirement already satisfied: rsa<5,>=3.1.4 in /usr/local/lib/python3.7/dist-packages (from google-auth<2,>=1.6.3->tensorboard~=2.5->tensorflow) (4.7.2)\nRequirement already satisfied: cachetools<5.0,>=2.0.0 in /usr/local/lib/python3.7/dist-packages (from google-auth<2,>=1.6.3->tensorboard~=2.5->tensorflow) (4.2.2)\nRequirement already satisfied: pyasn1-modules>=0.2.1 in /usr/local/lib/python3.7/dist-packages (from google-auth<2,>=1.6.3->tensorboard~=2.5->tensorflow) (0.2.8)\nRequirement already satisfied: requests-oauthlib>=0.7.0 in /usr/local/lib/python3.7/dist-packages (from google-auth-oauthlib<0.5,>=0.4.1->tensorboard~=2.5->tensorflow) (1.3.0)\nRequirement already satisfied: importlib-metadata in /usr/local/lib/python3.7/dist-packages (from markdown>=2.6.8->tensorboard~=2.5->tensorflow) (4.6.3)\nRequirement already satisfied: pyasn1<0.5.0,>=0.4.6 in /usr/local/lib/python3.7/dist-packages (from pyasn1-modules>=0.2.1->google-auth<2,>=1.6.3->tensorboard~=2.5->tensorflow) (0.4.8)\nRequirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard~=2.5->tensorflow) (2.10)\nRequirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard~=2.5->tensorflow) (1.24.3)\nRequirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard~=2.5->tensorflow) (2021.5.30)\nRequirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard~=2.5->tensorflow) (3.0.4)\nRequirement already satisfied: oauthlib>=3.0.0 in /usr/local/lib/python3.7/dist-packages (from requests-oauthlib>=0.7.0->google-auth-oauthlib<0.5,>=0.4.1->tensorboard~=2.5->tensorflow) (3.1.1)\nRequirement already satisfied: zipp>=0.5 in /usr/local/lib/python3.7/dist-packages (from importlib-metadata->markdown>=2.6.8->tensorboard~=2.5->tensorflow) (3.5.0)\n" ], [ "!pip install fsspec", "Collecting fsspec\n Downloading fsspec-2021.7.0-py3-none-any.whl (118 kB)\n\u001b[?25l\r\u001b[K |██▊ | 10 kB 21.1 MB/s eta 0:00:01\r\u001b[K |█████▌ | 20 kB 27.4 MB/s eta 0:00:01\r\u001b[K |████████▎ | 30 kB 12.9 MB/s eta 0:00:01\r\u001b[K |███████████ | 40 kB 9.5 MB/s eta 0:00:01\r\u001b[K |█████████████▉ | 51 kB 5.2 MB/s eta 0:00:01\r\u001b[K |████████████████▋ | 61 kB 5.3 MB/s eta 0:00:01\r\u001b[K |███████████████████▍ | 71 kB 6.0 MB/s eta 0:00:01\r\u001b[K |██████████████████████▏ | 81 kB 6.7 MB/s eta 0:00:01\r\u001b[K |█████████████████████████ | 92 kB 6.3 MB/s eta 0:00:01\r\u001b[K |███████████████████████████▊ | 102 kB 5.4 MB/s eta 0:00:01\r\u001b[K |██████████████████████████████▌ | 112 kB 5.4 MB/s eta 0:00:01\r\u001b[K |████████████████████████████████| 118 kB 5.4 MB/s \n\u001b[?25hInstalling collected packages: fsspec\nSuccessfully installed fsspec-2021.7.0\n" ], [ "# cnn model\nfrom numpy import mean\nfrom numpy import std\nfrom numpy import dstack\nfrom pandas import read_csv\nfrom matplotlib import pyplot\nfrom keras.models import Sequential\nfrom keras.layers import Dense\nfrom keras.layers import Flatten\nfrom keras.layers import Dropout\nfrom keras.layers.convolutional import Conv1D\nfrom keras.layers.convolutional import MaxPooling1D\n#from keras.utils import to_categorical\nfrom tensorflow.keras.utils import to_categorical # previous commented out line does not work\n \n# load a single file as a numpy array\ndef load_file(filepath):\n\tdataframe = read_csv(filepath, header=None, delim_whitespace=True)\n\treturn dataframe.values\n \n# load a list of files and return as a 3d numpy array\ndef load_group(filenames, prefix=''):\n\tloaded = list()\n\tfor name in filenames:\n\t\tdata = load_file(prefix + name)\n\t\tloaded.append(data)\n\t# stack group so that features are the 3rd dimension\n\tloaded = dstack(loaded)\n\treturn loaded\n \n# load a dataset group, such as train or test\ndef load_dataset_group(group, prefix=''):\n #filepath = prefix + group + '/Inertial Signals/' \n filepath = 'https://raw.githubusercontent.com/iotanalytics/IoTTutorial/main/data/UCI%20HAR%20Dataset/' + group + '/Inertial%20Signals/'\n\t# load all 9 files as a single array\n filenames = list()\n\t# total acceleration\n filenames += ['total_acc_x_'+group+'.txt', 'total_acc_y_'+group+'.txt', 'total_acc_z_'+group+'.txt']\n\t# body acceleration\n filenames += ['body_acc_x_'+group+'.txt', 'body_acc_y_'+group+'.txt', 'body_acc_z_'+group+'.txt']\n\t# body gyroscope\n filenames += ['body_gyro_x_'+group+'.txt', 'body_gyro_y_'+group+'.txt', 'body_gyro_z_'+group+'.txt']\n\t# load input data\n X = load_group(filenames, filepath)\n\t# load class output\n #y = load_file(prefix + group + '/y_'+group+'.txt')\n y = load_file('https://raw.githubusercontent.com/iotanalytics/IoTTutorial/main/data/UCI%20HAR%20Dataset/'+group+'/y_'+group+'.txt')\n return X, y\n\n# load the dataset, returns train and test X and y elements\ndef load_dataset(prefix=''):\n\t# load all train\n #trainX, trainy = load_dataset_group('train', prefix + 'HARDataset/')\n trainX, trainy = load_dataset_group('train', prefix) \n print(trainX.shape, trainy.shape)\n\t# load all test\n #testX, testy = load_dataset_group('test', prefix + 'HARDataset/')\n testX, testy = load_dataset_group('test', prefix)\n print(testX.shape, testy.shape)\n\t# zero-offset class values\n trainy = trainy - 1\n testy = testy - 1\n\t# one hot encode y\n trainy = to_categorical(trainy)\n testy = to_categorical(testy)\n print(trainX.shape, trainy.shape, testX.shape, testy.shape)\n return trainX, trainy, testX, testy\n \n# fit and evaluate a model\ndef evaluate_model(trainX, trainy, testX, testy):\n\tverbose, epochs, batch_size = 0, 10, 32\n\tn_timesteps, n_features, n_outputs = trainX.shape[1], trainX.shape[2], trainy.shape[1]\n\tmodel = Sequential()\n\tmodel.add(Conv1D(filters=64, kernel_size=3, activation='relu', input_shape=(n_timesteps,n_features)))\n\tmodel.add(Conv1D(filters=64, kernel_size=3, activation='relu'))\n\tmodel.add(Dropout(0.5))\n\tmodel.add(MaxPooling1D(pool_size=2))\n\tmodel.add(Flatten())\n\tmodel.add(Dense(100, activation='relu'))\n\tmodel.add(Dense(n_outputs, activation='softmax'))\n\tmodel.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n\t# fit network\n\tmodel.fit(trainX, trainy, epochs=epochs, batch_size=batch_size, verbose=verbose)\n\t# evaluate model\n\t_, accuracy = model.evaluate(testX, testy, batch_size=batch_size, verbose=0)\n\treturn accuracy\n \n# summarize scores\ndef summarize_results(scores):\n\tprint(scores)\n\tm, s = mean(scores), std(scores)\n\tprint('Accuracy: %.3f%% (+/-%.3f)' % (m, s))\n \n# run an experiment\ndef run_experiment(repeats=10):\n\t# load data\n\ttrainX, trainy, testX, testy = load_dataset()\n\t# repeat experiment\n\tscores = list()\n\tfor r in range(repeats):\n\t\tscore = evaluate_model(trainX, trainy, testX, testy)\n\t\tscore = score * 100.0\n\t\tprint('>#%d: %.3f' % (r+1, score))\n\t\tscores.append(score)\n\t# summarize results\n\tsummarize_results(scores)\n \n# run the experiment\nrun_experiment()", "(7352, 128, 9) (7352, 1)\n(2947, 128, 9) (2947, 1)\n(7352, 128, 9) (7352, 6) (2947, 128, 9) (2947, 6)\n>#1: 90.363\n>#2: 88.157\n>#3: 92.467\n>#4: 90.601\n>#5: 90.227\n>#6: 90.058\n>#7: 91.992\n>#8: 90.363\n>#9: 89.786\n>#10: 91.211\n[90.3630793094635, 88.15745115280151, 92.46691465377808, 90.60060977935791, 90.22734761238098, 90.05768299102783, 91.99185371398926, 90.3630793094635, 89.78622555732727, 91.21140241622925]\nAccuracy: 90.523% (+/-1.136)\n" ] ] ]

[ "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code", "code" ] ]

[ "ADSL" ]

[ "ADSL" ]

[ "ADSL" ]

[ [ [ "import this", "The Zen of Python, by Tim Peters\n\nBeautiful is better than ugly.\nExplicit is better than implicit.\nSimple is better than complex.\nComplex is better than complicated.\nFlat is better than nested.\nSparse is better than dense.\nReadability counts.\nSpecial cases aren't special enough to break the rules.\nAlthough practicality beats purity.\nErrors should never pass silently.\nUnless explicitly silenced.\nIn the face of ambiguity, refuse the temptation to guess.\nThere should be one-- and preferably only one --obvious way to do it.\nAlthough that way may not be obvious at first unless you're Dutch.\nNow is better than never.\nAlthough never is often better than *right* now.\nIf the implementation is hard to explain, it's a bad idea.\nIf the implementation is easy to explain, it may be a good idea.\nNamespaces are one honking great idea -- let's do more of those!\n" ] ], [ [ "# Pymaceuticals Inc.\n---\n\n### Analysis\n* Overall, it is clear that Capomulin is a viable drug regimen to reduce tumor growth.\n* Capomulin had the most number of mice complete the study, with the exception of Remicane, all other regimens observed a number of mice deaths across the duration of the study. \n* There is a strong correlation between mouse weight and tumor volume, indicating that mouse weight may be contributing to the effectiveness of any drug regimen.\n* There was one potential outlier within the Infubinol regimen. While most mice showed tumor volume increase, there was one mouse that had a reduction in tumor growth in the study. ", "_____no_output_____" ] ], [ [ "# Dependencies and Setup\nimport matplotlib.pyplot as plt\nimport pandas as pd\nimport scipy.stats as st\n\n# Study data files\nmouse_metadata_path = \"data/Mouse_metadata.csv\"\nstudy_results_path = \"data/Study_results.csv\"\n\n# Read the mouse data and the study results\nmouse_metadata = pd.read_csv(mouse_metadata_path)\nstudy_results = pd.read_csv(study_results_path)\n\n# Combine the data into a single dataset\n\n\n# Display the data table for preview\n", "_____no_output_____" ], [ "# Checking the number of mice.\n", "_____no_output_____" ], [ "# Getting the duplicate mice by ID number that shows up for Mouse ID and Timepoint. \n", "_____no_output_____" ], [ "# Optional: Get all the data for the duplicate mouse ID. \n", "_____no_output_____" ], [ "# Create a clean DataFrame by dropping the duplicate mouse by its ID.\n", "_____no_output_____" ], [ "# Checking the number of mice in the clean DataFrame.\n", "_____no_output_____" ] ], [ [ "## Summary Statistics", "_____no_output_____" ] ], [ [ "# Generate a summary statistics table of mean, median, variance, standard deviation, and SEM of the tumor volume for each regimen\n\n# Use groupby and summary statistical methods to calculate the following properties of each drug regimen: \n# mean, median, variance, standard deviation, and SEM of the tumor volume. \n# Assemble the resulting series into a single summary dataframe.\n\n", "_____no_output_____" ], [ "# Generate a summary statistics table of mean, median, variance, standard deviation, and SEM of the tumor volume for each regimen\n\n# Using the aggregation method, produce the same summary statistics in a single line\n", "_____no_output_____" ] ], [ [ "## Bar and Pie Charts", "_____no_output_____" ] ], [ [ "# Generate a bar plot showing the total number of measurements taken on each drug regimen using pandas.\n", "_____no_output_____" ], [ "# Generate a bar plot showing the total number of measurements taken on each drug regimen using using pyplot.\n", "_____no_output_____" ], [ "# Generate a pie plot showing the distribution of female versus male mice using pandas\n", "_____no_output_____" ], [ "# Generate a pie plot showing the distribution of female versus male mice using pyplot\n", "_____no_output_____" ] ], [ [ "## Quartiles, Outliers and Boxplots", "_____no_output_____" ] ], [ [ "# Calculate the final tumor volume of each mouse across four of the treatment regimens: \n# Capomulin, Ramicane, Infubinol, and Ceftamin\n\n# Start by getting the last (greatest) timepoint for each mouse\n\n\n# Merge this group df with the original dataframe to get the tumor volume at the last timepoint\n", "_____no_output_____" ], [ "# Put treatments into a list for for loop (and later for plot labels)\n\n\n# Create empty list to fill with tumor vol data (for plotting)\n\n\n# Calculate the IQR and quantitatively determine if there are any potential outliers. \n\n \n # Locate the rows which contain mice on each drug and get the tumor volumes\n \n \n # add subset \n \n \n # Determine outliers using upper and lower bounds\n ", "Capomulin's potential outliers: Series([], Name: Tumor Volume (mm3), dtype: float64)\nRamicane's potential outliers: Series([], Name: Tumor Volume (mm3), dtype: float64)\nInfubinol's potential outliers: 31 36.321346\nName: Tumor Volume (mm3), dtype: float64\nCeftamin's potential outliers: Series([], Name: Tumor Volume (mm3), dtype: float64)\n" ], [ "# Generate a box plot of the final tumor volume of each mouse across four regimens of interest\n", "_____no_output_____" ] ], [ [ "## Line and Scatter Plots", "_____no_output_____" ] ], [ [ "# Generate a line plot of tumor volume vs. time point for a mouse treated with Capomulin\n", "_____no_output_____" ], [ "# Generate a scatter plot of average tumor volume vs. mouse weight for the Capomulin regimen\n", "_____no_output_____" ] ], [ [ "## Correlation and Regression", "_____no_output_____" ] ], [ [ "# Calculate the correlation coefficient and linear regression model \n# for mouse weight and average tumor volume for the Capomulin regimen\n", "The correlation between mouse weight and the average tumor volume is 0.84\n" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "CC-BY-4.0", "MIT" ]

[ "CC-BY-4.0", "MIT" ]

[ "CC-BY-4.0", "MIT" ]

[ [ [ "# Microsoft Insights Module Example Notebook", "_____no_output_____" ] ], [ [ "%run /OEA_py", "_____no_output_____" ], [ "%run /NEW_Insights_py", "_____no_output_____" ], [ "# 0) Initialize the OEA framework and Insights module class notebook.\r\noea = OEA()\r\ninsights = Insights()", "_____no_output_____" ], [ "insights.ingest()", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code" ] ]

[ "ADSL" ]

[ "ADSL" ]

[ "ADSL" ]

[ [ [ "# WeatherPy\n----\n\n#### Note\n* Instructions have been included for each segment. You do not have to follow them exactly, but they are included to help you think through the steps.", "_____no_output_____" ] ], [ [ "w_api = 'f85af5acc7275a9eb032d03a3cca5913'", "_____no_output_____" ], [ "# Dependencies and Setup\nimport matplotlib.pyplot as plt\nimport pandas as pd\nimport numpy as np\nimport requests\nimport time\nfrom scipy.stats import linregress\n\n# Import API key\n# from api_keys import weather_api_key\n\n\n# Incorporated citipy to determine city based on latitude and longitude\nfrom citipy import citipy\n\n# Output File (CSV)\noutput_data_file = \"output_data/cities.csv\"\n\n# Range of latitudes and longitudes\nlat_range = (-90, 90)\nlng_range = (-180, 180)", "_____no_output_____" ] ], [ [ "## Generate Cities List", "_____no_output_____" ] ], [ [ "# List for holding lat_lngs and cities\nlat_lngs = []\ncities = []\n\n# Create a set of random lat and lng combinations\nlats = np.random.uniform(lat_range[0], lat_range[1], size=1500)\nlngs = np.random.uniform(lng_range[0], lng_range[1], size=1500)\nlat_lngs = zip(lats, lngs)\n\n# Identify nearest city for each lat, lng combination\nfor lat_lng in lat_lngs:\n city = citipy.nearest_city(lat_lng[0], lat_lng[1]).city_name\n \n # If the city is unique, then add it to a our cities list\n if city not in cities:\n cities.append(city)\n\n# Print the city count to confirm sufficient count\nlen(cities)", "_____no_output_____" ] ], [ [ "### Perform API Calls\n* Perform a weather check on each city using a series of successive API calls.\n* Include a print log of each city as it'sbeing processed (with the city number and city name).\n", "_____no_output_____" ], [ "### Convert Raw Data to DataFrame\n* Export the city data into a .csv.\n* Display the DataFrame", "_____no_output_____" ], [ "## Inspect the data and remove the cities where the humidity > 100%.\n----\nSkip this step if there are no cities that have humidity > 100%. ", "_____no_output_____" ] ], [ [ "# Get the indices of cities that have humidity over 100%.\n", "_____no_output_____" ], [ "# Make a new DataFrame equal to the city data to drop all humidity outliers by index.\n# Passing \"inplace=False\" will make a copy of the city_data DataFrame, which we call \"clean_city_data\".\n", "_____no_output_____" ], [ "\n", "_____no_output_____" ] ], [ [ "## Plotting the Data\n* Use proper labeling of the plots using plot titles (including date of analysis) and axes labels.\n* Save the plotted figures as .pngs.", "_____no_output_____" ], [ "## Latitude vs. Temperature Plot", "_____no_output_____" ], [ "## Latitude vs. Humidity Plot", "_____no_output_____" ], [ "## Latitude vs. Cloudiness Plot", "_____no_output_____" ], [ "## Latitude vs. Wind Speed Plot", "_____no_output_____" ], [ "## Linear Regression", "_____no_output_____" ], [ "#### Northern Hemisphere - Max Temp vs. Latitude Linear Regression", "_____no_output_____" ], [ "#### Southern Hemisphere - Max Temp vs. Latitude Linear Regression", "_____no_output_____" ], [ "#### Northern Hemisphere - Humidity (%) vs. Latitude Linear Regression", "_____no_output_____" ], [ "#### Southern Hemisphere - Humidity (%) vs. Latitude Linear Regression", "_____no_output_____" ], [ "#### Northern Hemisphere - Cloudiness (%) vs. Latitude Linear Regression", "_____no_output_____" ], [ "#### Southern Hemisphere - Cloudiness (%) vs. Latitude Linear Regression", "_____no_output_____" ], [ "#### Northern Hemisphere - Wind Speed (mph) vs. Latitude Linear Regression", "_____no_output_____" ], [ "#### Southern Hemisphere - Wind Speed (mph) vs. Latitude Linear Regression", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Matplotlib Applied", "_____no_output_____" ], [ "**Aim: SWBAT create a figure with 4 subplots of varying graph types.**", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\nimport numpy as np", "_____no_output_____" ], [ "from numpy.random import seed, randint\nseed(100)\n\n# Create Figure and Subplots\nfig, axes = plt.subplots(2,2, figsize=(10,6), sharex=True, sharey=True, dpi=100)\n\n# Define the colors and markers to use\ncolors = {0:'g', 1:'b', 2:'r', 3:'y'}\nmarkers = {0:'o', 1:'x', 2:'*', 3:'p'}\n\n# Plot each axes\nfor i, ax in enumerate(axes.ravel()):\n ax.plot(sorted(randint(0,10,10)), sorted(randint(0,10,10)), marker=markers[i], color=colors[i]) \n ax.set_title('Ax: ' + str(i))\n ax.yaxis.set_ticks_position('right')\n\nplt.suptitle('Four Subplots in One Figure', verticalalignment='bottom', fontsize=16) \nplt.tight_layout()\n# plt.show()", "_____no_output_____" ] ], [ [ "### Go through and play with the code above to try answer the questions below:\n\n- What do you think `sharex` and `sharey` do?\n\n- What does the `dpi` argument control?\n\n- What does `numpy.ravel()` do, and why do they call it here?\n\n- What does `yaxis.set_ticks_position()` do?\n\n- How do they use the `colors` and `markers` dictionaries?", "_____no_output_____" ], [ "### Your turn:\n\n- Create a figure that has 4 sub plots on it.\n\n- Plot 1: a line blue graph (`.plot()`) using data `x` and `y`\n\n- Plot 2: a scatter plot (`.scatter()`) using data `x2` and `y2` with red markers that are non-filled circles.\n\n- Plot 3: a plot that has both a line graph (x and y data) and a scatterplot (x2, y2) that only use 1 y axis\n\n- plot 4: a plot that is similiar to plot3 except the scatterplot has it own axis on the right hand side. \n\n- Put titles on each subplot.\n\n- Create a title for the entire figure.\n\n- Save figure as png.", "_____no_output_____" ] ], [ [ "from numpy.random import seed, randint\nseed(100)\n\nx = sorted(randint(0,10,10))\nx2 = sorted(randint(0,20,10))\ny = sorted(randint(0,10,10))\ny2 = sorted(randint(0,20,10))", "_____no_output_____" ] ], [ [ "## Great tutorial on matplotlib\n\nhttps://www.machinelearningplus.com/plots/matplotlib-tutorial-complete-guide-python-plot-examples/", "_____no_output_____" ] ], [ [ "fig", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Now You Code 2: Paint Pricing\n\nHouse Depot, a big-box hardware retailer, has contracted you to create an app to calculate paint prices. \n\nThe price of paint is determined by the following factors:\n- Everyday quality paint is `$19.99` per gallon.\n- Select quality paint is `$24.99` per gallon.\n- Premium quality paint is `$32.99` per gallon.\n\nIn addition if the customer wants computerized color-matching that incurs an additional fee of `$4.99` per gallon. \n\nWrite a program to ask the user to select a paint quality: 'everyday', 'select' or 'premium' and then whether they need color matching and then outputs the price per gallon of the paint.\n\nExample Run 1:\n\n```\nWhich paint quality do you require ['everyday', 'select', 'premium'] ?select\nDo you require color matching [y/n] ?y\nTotal price of select paint with color matching is $29.98\n```\n\nExample Run 2:\n\n```\nWhich paint quality do you require ['everyday', 'select', 'premium'] ?premium\nDo you require color matching [y/n] ?n\nTotal price of premium paint without color matching is $32.99\n```", "_____no_output_____" ], [ "## Step 1: Problem Analysis\n\nInputs:\n\nOutputs:\n\nAlgorithm (Steps in Program):\n\n", "_____no_output_____" ] ], [ [ "# Step 2: Write code here\nchoices = [\"everyday\", \"select\", \"premium\"]\ncolorChoices = [\"y\", \"n\"]\nquality = input(\"which paint quality would you like? [\"everyday\", \"select\", \"premium\"]\")\nif quality in choices \n if quality == \"everyday\":\n quality =19.99\n elif quality == \"select\":\n quality = 24.99\n elif quality ==\"premium\":\n quality = 32.99\ncolorMatching = input(\"do you requrie color matching [yes/no]\")\n if colorMatching in colorChoices:\n if colorMatching == \"yes\":\n colorMatching = 4.99 \n elif colorMatching == \"no\": \n colorMatching = 0\n final = quality + colorMatching\n if colorMatching == \"yes\":\n print(\"total price of paint with color matching is %.2f\" %(final))\n else:\n print(\"total price of paint without color matching is %.2f\" %(final))\n else:\n print(\"you must enter yes or no\")\n else:\n print(\"That is not a paint quality\")\n ", "_____no_output_____" ] ], [ [ "## Step 3: Questions\n\n1. When you enter something other than `'everyday', 'select',` or `'premium'` what happens? Modify the program to print `that is not a paint quality` and then exit in those cases. the program errors \n2. What happens when you enter something other than `'y'` or `'n'` for color matching? Re-write the program to print `you must enter y or n` whenever you enter something other than those two values.the program does nothing \n3. Why can't we use Python's `try...except` in this example?\n4. How many times (at minimum) must we execute this program and check the results before we can be reasonably assured it is correct?\n", "_____no_output_____" ], [ "## Reminder of Evaluation Criteria\n\n1. What the problem attempted (analysis, code, and answered questions) ?\n2. What the problem analysis thought out? (does the program match the plan?)\n3. Does the code execute without syntax error?\n4. Does the code solve the intended problem?\n5. Is the code well written? (easy to understand, modular, and self-documenting, handles errors)\n", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "%matplotlib inline\nimport matplotlib.pyplot as plt\nplt.style.use(\"seaborn\")\nimport os\nfrom urllib.request import urlretrieve\nimport pandas as pd\nURL = 'https://data.seattle.gov/api/views/65db-xm6k/rows.csv?accessType=DOWNLOAD'\n\ndef get_fremont_data(filename = 'Fremont.csv', url=URL, force_download = False):\n if force_download or not os.path.exists(filename):\n urlretrieve(url,filename)\n data = pd.read_csv(filename,index_col='Date',parse_dates = True)\n data.columns = ['Total','West','East']\n return data", "_____no_output_____" ], [ "data = get_fremont_data()\ndata.head()", "_____no_output_____" ], [ "# visualisation\n\n# put all the plots in the notebook itself, rather than separate windows\ndata.plot()", "_____no_output_____" ], [ "data.resample('W').sum().plot()", "_____no_output_____" ], [ "%matplotlib inline\ndata.resample('W').sum().plot()", "_____no_output_____" ], [ "\n\n\ndata.resample('W').sum().plot()", "_____no_output_____" ], [ "data.head()", "_____no_output_____" ], [ "data.resample('W').sum().plot()", "_____no_output_____" ], [ "data.resample('D').sum().rolling(365).sum().plot()", "_____no_output_____" ], [ "ax = data.resample('D').sum().rolling(365).sum().plot()\nax.set_ylim(0,None) # setting bottom line of y to be y = 0", "_____no_output_____" ], [ "ax = data.resample('D').sum().rolling(365).sum().plot()\nax.set_ylim(0,None) # setting bottom line of y to be y = 0", "_____no_output_____" ], [ "# look at trends at individual days\ndata.groupby(data.index.time).mean().plot()", "_____no_output_____" ], [ "pivoted = data.pivot_table('Total', index = data.index.time, columns = data.index.date)\npivoted.iloc[:5,:5]", "_____no_output_____" ], [ "pivoted.plot(legend=False)", "_____no_output_____" ], [ "pivoted.plot(legend=False, alpha=0.01) # add transparency to better visualise", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "### Kaggle ML and Data Science Survey Analysis\n#### Data 512, Final Project Plan - Zicong Liang", "_____no_output_____" ], [ "### Project Motivation\nThis project an analysis for a survey about Machine Learning and Data Science. Recently, lots of people are talking about machine learning and Data Science. In addition, more and more companies hire data science talents and invest in data science in order to make their business more data-driven. According to Wikipedia, Data science is also known as data-driven science, is an interdisciplinary field combined from mathematics, statistics and computer science to extract insights from data. On the one hand, I remember Oliver commented about nobody's really done a proper survey related to Data Science. I think it is a great opportunity to do some research in this field of study because I find out Kaggle releases a survey responses about Data Science and Machine Learning. On the other hand, there are a few reasons drive me to know more about data science and machine learning field of study.\n1.\tI am one of the graduate students in the master of science data science program at University of Washington.\n2.\tI'd like to get a job in data science filed after graduating from this program.\n3.\tI'd like to know how those people related to this field think about data science and machine learning, then analyses their response to draw some conclusions that I am looking for. \n\nAs one of the beginner in Data Science, it is better to know more about what kind of skills we need to well-equipped in order to be ready to step into this field. That's why I am planning to do this analysis.\nThroughout this project, we are going to conduct some analyses for Data Science and Machine Learning in various aspects, such as gender, age, programming skills, algorithms, education degree and salary and so on. It helps us gain new insights of Data Science that different from what we have learnt from school. In addition, the analysis of this project can be a reference to those who people for pursuing a career in Data Science, not only for switching job but also looking for their first job in this field.", "_____no_output_____" ], [ "### Project Dataset\nThe data I am going to use for this project is from Kaggle. It is one of the most popular datasets in Kaggle recently. To be more specifically, the data is about a survey which conducted by Kaggle for an industry-wide to establish a comprehensive view of the state of Data Science and Machine Learning. This survey received more than 16000 responses. According to this data, we would be able to learn a ton about who is working in the Data Science Field, what’s happening at the cutting edge of machine learning across industries, and how new data scientists can best break into the field.\n\nHere is the link to the main page of [Kaggle ML and Data Science Survey, 2017](https://www.kaggle.com/kaggle/kaggle-survey-2017).\n\nThe dataset consists of 5 files:\n1. **schema.csv**: a CSV file with survey schema. This schema includes the questions that correspond to each column name in both the **multipleChoiceResponses.csv** and **freeformResponses.csv**.\n2. **multipleChoiceResponses.csv**: Respondents' answers to multiple choice and ranking questions.\n3. **freeformResponses.csv**: Respondents' freeform answers to Kaggle's survey questions.\n4. **conversionRates.csv**: Currency conversion rates to USD (accessed from the R package \"quantmod\" on September 14, 2017)\n5. **RespondentTypeREADME.txt**: This is a schema for decoding the responses in the \"Asked\" column of the **schema.csv** file.\n\nClick [here](https://www.kaggle.com/kaggle/kaggle-survey-2017/data) to the data page.\n\nBase on the description above, I am going to use two files from the data.\n1. **multipleChoiceResponses.csv**\n2. **conversionRates.csv**\nThe **conversionRates.csv** data is pretty straightforward, so let's look at the **multipleChoiceResponses.csv** before the start any analysis.", "_____no_output_____" ] ], [ [ "import pandas as pd\nmy_data = pd.read_csv(\"multipleChoiceResponses.csv\", encoding='ISO-8859-1', delimiter=',', low_memory=False)\nmy_data.head()", "_____no_output_____" ], [ "my_data.shape", "_____no_output_____" ] ], [ [ "We can see the first 5 rows from the above data (**multipleChoiceResponses.csv**). There are lots of variables that contain NaN values. This file contains 16716 observations and 228 variables. During the project, I will focus on the subset of this file, keeping those varibles that necessary for my analysis and dropping those NaN values.\n\n**Note: Since there are 228 columns in the *multipleChoiceResponses.csv*, I might need further exploration of this file in order to determine how many variables I will need throughout this project. As long as I have my decision while answering the questions and hypotheses below, I would follow what I have done for the previous assignments to make a final dataset that include all the variables I need for this project as well as the descriptions.**\n\n**Note: Since the data is from an online survey responses, I think bias would be possibly exist in the data, I will analysis the findings from data as well as combing facts to draw the my conclusions.**\n\n\n#### License\nReleased UnderDatabase: [Open Database](http://opendatacommons.org/licenses/odbl/1.0/), Contents: © Original Authors\n\nHere is some information about the Open Database License on Wikipedia. Click [here](https://en.wikipedia.org/wiki/Open_Database_License) to see more about it.\n\n**Note: It seems there is some issue with the site, hopefully it will be fixed soon. (Mark at 4:47PM November 9th 2017)**\n\nFor more information about Kaggle please see [Terms](https://www.kaggle.com/terms) and [Privacy](https://www.kaggle.com/about/privacy)", "_____no_output_____" ], [ "### Research Question & Hypothesis", "_____no_output_____" ], [ "#### Research Question\nI did some research that most people are using Python and R in the Data Science field, some use MATLAB and SAS. I'd \nAs a student in the data science program, most of our courses suggest us to use Python as our main programming language instead of R. As I mention above, I'd like to become a data scientist after I graduate from the program, it's good to know what kind of companies that data scientists would most interested in working for and the salary of course. Moreover, I have listed some of the questions that raise in mind I want to know about from this research.\n\nQ1: What's the gender diversity among this survey? (What's the difference between male and female in this field?)\n\nQ2: As a student, really want to know much time should I spend studying in Data Science every week? And how important my MS Data Science Degree would help me success in this filed?\n\nQ3: What is dominant programming language that people would suggestion to use or learn as in Data Science and Machine Learning, Python or R?\n\nQ4: What are 10 most popular algorithms that people think are important to know as in the Data Science filed?\n\nQ5: What are the challenges as working in Data Science filed that most people take for?\n\nQ6: What's the mean/median salary as a data scientist? Moreover, what are the important factors that people consider would consider affecting the salary?\n\nMore questions might be added later\n\n**Note: During the process of the project, I think there will be more interesting findings from the data (since there are lots of variables in the data file, I will explore more about the context and try to raise more related questions throughout the process in order to expand my analysis. **\n\n#### Hypothesis\nThe hypothesis that first come to my mind is about gender imbalance.\n\n*Hypothesis: There is a gender gap between male and female in Data Science field.*\nAnother hypothesis is about dominant programming language. \n\n*Hypothesis: Python is the most popular tool for data scientist compare to the other programming language.*\n\nMoreover, I'd like to raise a hypothesis about the relationship between some important factors (e.g. skills, degree, experience) and salary. However, to address this hypothesis properly, I will need to finalize the variables selection.\n\n*Hypothesis: There is relationship between some factors (e.g. skills, degree, experience) and salary.*\n\n**Note: the project will focus on raising questions and try to answer those questions from the analysis more emphasis on using visulizations.**", "_____no_output_____" ], [ "### Technical Approach & Method", "_____no_output_____" ], [ "During this project, I am going to Using python and the following packages to acheive my analysis\n1. **Pandas**\n2. **Matplotlib**\n3. **Seaborn**\n4. **Scikit Learn**\n\nThe purpose of this project is to analyze the survey response and draw some conclusion base on the analysis. Therefore, I will use the Python Package Pandas to manipulate the data such as grouping, filtering and aggregating the data. Moreover, I will also use the Matplotlib and Seaborn libraries together to generate some visualizations from the data for the questions I raise above. As in human-centered Data Science, providing appropriate visualizations corresponding to the questions should be better for interpretation as well as easier for others to understand the questions and results. In the end, as time permits, I might also consider developing a regression model to determine what factors would impact the salary the most by using the Sklearn package (e.g. education degree, programming skills, machine learning algorithms, gender, experience and location). The final approach might be change a bit while working on the project (I might need more packages later if it is necessary). However, the overall analysis should be supported by visualizations as well as explanation to address the research questions and hypotheses.\n\nTo be more specifically, I will follow what I have learnt from the course HCDE 511 (human centered design & engineering) to encode the data properly for all the visualizations. Since there are lots of comparisons within each visualization, I think using bar charts (both vertical and horizontal) and pie charts should be the best choice to meet the requirement. Moreover, hue is also helpful to identify different categories in the visualizations.", "_____no_output_____" ], [ "### Any unknowns or dependencies", "_____no_output_____" ], [ "As I mention above, the data file (*multipleChoiceResponses.csv*) I am going to use has 228 variables. I need further exploration about the content to determine how many variables are needed throughout this project. At the end, I will make a final data set that contains some of the original variable and might have some new generated variables (such as counting, mean, sum) which depend on the other variables. In addition, I will have a full description about the final data set. The analysis will base on the final dataset.", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import pandas as pd\ndf = pd.read_csv(\"dmvpn.csv\")\n", "_____no_output_____" ], [ "df.groupby(df.ip).count().sort_values(\"status\")", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Continuous training pipeline with Kubeflow Pipeline and AI Platform", "_____no_output_____" ], [ "**Learning Objectives:**\n1. Learn how to use Kubeflow Pipeline(KFP) pre-build components (BiqQuery, AI Platform training and predictions)\n1. Learn how to use KFP lightweight python components\n1. Learn how to build a KFP with these components\n1. Learn how to compile, upload, and run a KFP with the command line\n\n\nIn this lab, you will build, deploy, and run a KFP pipeline that orchestrates **BigQuery** and **AI Platform** services to train, tune, and deploy a **scikit-learn** model.", "_____no_output_____" ], [ "## Understanding the pipeline design\n", "_____no_output_____" ], [ "The workflow implemented by the pipeline is defined using a Python based Domain Specific Language (DSL). The pipeline's DSL is in the `covertype_training_pipeline.py` file that we will generate below.\n\nThe pipeline's DSL has been designed to avoid hardcoding any environment specific settings like file paths or connection strings. These settings are provided to the pipeline code through a set of environment variables.\n", "_____no_output_____" ] ], [ [ "#!grep 'BASE_IMAGE =' -A 5 pipeline/covertype_training_pipeline.py\n!pip list | grep kfp", "kfp 1.0.0\nkfp-pipeline-spec 0.1.7\nkfp-server-api 1.5.0\n" ] ], [ [ "The pipeline uses a mix of custom and pre-build components.\n\n- Pre-build components. The pipeline uses the following pre-build components that are included with the KFP distribution:\n - [BigQuery query component](https://github.com/kubeflow/pipelines/tree/0.2.5/components/gcp/bigquery/query)\n - [AI Platform Training component](https://github.com/kubeflow/pipelines/tree/0.2.5/components/gcp/ml_engine/train)\n - [AI Platform Deploy component](https://github.com/kubeflow/pipelines/tree/0.2.5/components/gcp/ml_engine/deploy)\n- Custom components. The pipeline uses two custom helper components that encapsulate functionality not available in any of the pre-build components. The components are implemented using the KFP SDK's [Lightweight Python Components](https://www.kubeflow.org/docs/pipelines/sdk/lightweight-python-components/) mechanism. The code for the components is in the `helper_components.py` file:\n - **Retrieve Best Run**. This component retrieves a tuning metric and hyperparameter values for the best run of a AI Platform Training hyperparameter tuning job.\n - **Evaluate Model**. This component evaluates a *sklearn* trained model using a provided metric and a testing dataset.\n ", "_____no_output_____" ] ], [ [ "%%writefile ./pipeline/covertype_training_pipeline.py\n# Copyright 2019 Google LLC\n#\n# Licensed under the Apache License, Version 2.0 (the \"License\");\n# you may not use this file except in compliance with the License.\n# You may obtain a copy of the License at\n#\n# http://www.apache.org/licenses/LICENSE-2.0\n#\n# Unless required by applicable law or agreed to in writing, software\n# distributed under the License is distributed on an \"AS IS\" BASIS,\n# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n# See the License for the specific language governing permissions and\n# limitations under the License.\n\"\"\"KFP orchestrating BigQuery and Cloud AI Platform services.\"\"\"\n\nimport os\n\nfrom helper_components import evaluate_model\nfrom helper_components import retrieve_best_run\nfrom jinja2 import Template\nimport kfp\nfrom kfp.components import func_to_container_op\nfrom kfp.dsl.types import Dict\nfrom kfp.dsl.types import GCPProjectID\nfrom kfp.dsl.types import GCPRegion\nfrom kfp.dsl.types import GCSPath\nfrom kfp.dsl.types import String\nfrom kfp.gcp import use_gcp_secret\n\n# Defaults and environment settings\nBASE_IMAGE = os.getenv('BASE_IMAGE')\nTRAINER_IMAGE = os.getenv('TRAINER_IMAGE')\nRUNTIME_VERSION = os.getenv('RUNTIME_VERSION')\nPYTHON_VERSION = os.getenv('PYTHON_VERSION')\nCOMPONENT_URL_SEARCH_PREFIX = os.getenv('COMPONENT_URL_SEARCH_PREFIX')\nUSE_KFP_SA = os.getenv('USE_KFP_SA')\n\nTRAINING_FILE_PATH = 'datasets/training/data.csv'\nVALIDATION_FILE_PATH = 'datasets/validation/data.csv'\nTESTING_FILE_PATH = 'datasets/testing/data.csv'\n\n# Parameter defaults\nSPLITS_DATASET_ID = 'splits'\nHYPERTUNE_SETTINGS = \"\"\"\n{\n \"hyperparameters\": {\n \"goal\": \"MAXIMIZE\",\n \"maxTrials\": 6,\n \"maxParallelTrials\": 3,\n \"hyperparameterMetricTag\": \"accuracy\",\n \"enableTrialEarlyStopping\": True,\n \"params\": [\n {\n \"parameterName\": \"max_iter\",\n \"type\": \"DISCRETE\",\n \"discreteValues\": [500, 1000]\n },\n {\n \"parameterName\": \"alpha\",\n \"type\": \"DOUBLE\",\n \"minValue\": 0.0001,\n \"maxValue\": 0.001,\n \"scaleType\": \"UNIT_LINEAR_SCALE\"\n }\n ]\n }\n}\n\"\"\"\n\n\n# Helper functions\ndef generate_sampling_query(source_table_name, num_lots, lots):\n \"\"\"Prepares the data sampling query.\"\"\"\n\n sampling_query_template = \"\"\"\n SELECT *\n FROM \n `{{ source_table }}` AS cover\n WHERE \n MOD(ABS(FARM_FINGERPRINT(TO_JSON_STRING(cover))), {{ num_lots }}) IN ({{ lots }})\n \"\"\"\n query = Template(sampling_query_template).render(\n source_table=source_table_name, num_lots=num_lots, lots=str(lots)[1:-1])\n\n return query\n\n\n# Create component factories\ncomponent_store = kfp.components.ComponentStore(\n local_search_paths=None, url_search_prefixes=[COMPONENT_URL_SEARCH_PREFIX])\n\nbigquery_query_op = component_store.load_component('bigquery/query')\nmlengine_train_op = component_store.load_component('ml_engine/train')\nmlengine_deploy_op = component_store.load_component('ml_engine/deploy')\nretrieve_best_run_op = func_to_container_op(\n retrieve_best_run, base_image=BASE_IMAGE)\nevaluate_model_op = func_to_container_op(evaluate_model, base_image=BASE_IMAGE)\n\n\[email protected](\n name='Covertype Classifier Training',\n description='The pipeline training and deploying the Covertype classifierpipeline_yaml'\n)\ndef covertype_train(project_id,\n region,\n source_table_name,\n gcs_root,\n dataset_id,\n evaluation_metric_name,\n evaluation_metric_threshold,\n model_id,\n version_id,\n replace_existing_version,\n hypertune_settings=HYPERTUNE_SETTINGS,\n dataset_location='US'):\n \"\"\"Orchestrates training and deployment of an sklearn model.\"\"\"\n\n # Create the training split\n query = generate_sampling_query(\n source_table_name=source_table_name, num_lots=10, lots=[1, 2, 3, 4])\n\n training_file_path = '{}/{}'.format(gcs_root, TRAINING_FILE_PATH)\n\n create_training_split = bigquery_query_op(\n query=query,\n project_id=project_id,\n dataset_id=dataset_id,\n table_id='',\n output_gcs_path=training_file_path,\n dataset_location=dataset_location)\n\n # Create the validation split\n query = generate_sampling_query(\n source_table_name=source_table_name, num_lots=10, lots=[8])\n\n validation_file_path = '{}/{}'.format(gcs_root, VALIDATION_FILE_PATH)\n\n create_validation_split = bigquery_query_op(\n query=query,\n project_id=project_id,\n dataset_id=dataset_id,\n table_id='',\n output_gcs_path=validation_file_path,\n dataset_location=dataset_location)\n\n # Create the testing split\n query = generate_sampling_query(\n source_table_name=source_table_name, num_lots=10, lots=[9])\n\n testing_file_path = '{}/{}'.format(gcs_root, TESTING_FILE_PATH)\n\n create_testing_split = bigquery_query_op(\n query=query,\n project_id=project_id,\n dataset_id=dataset_id,\n table_id='',\n output_gcs_path=testing_file_path,\n dataset_location=dataset_location)\n\n # Tune hyperparameters\n tune_args = [\n '--training_dataset_path',\n create_training_split.outputs['output_gcs_path'],\n '--validation_dataset_path',\n create_validation_split.outputs['output_gcs_path'], '--hptune', 'True'\n ]\n\n job_dir = '{}/{}/{}'.format(gcs_root, 'jobdir/hypertune',\n kfp.dsl.RUN_ID_PLACEHOLDER)\n\n hypertune = mlengine_train_op(\n project_id=project_id,\n region=region,\n master_image_uri=TRAINER_IMAGE,\n job_dir=job_dir,\n args=tune_args,\n training_input=hypertune_settings)\n\n # Retrieve the best trial\n get_best_trial = retrieve_best_run_op(\n project_id, hypertune.outputs['job_id'])\n\n # Train the model on a combined training and validation datasets\n job_dir = '{}/{}/{}'.format(gcs_root, 'jobdir', kfp.dsl.RUN_ID_PLACEHOLDER)\n\n train_args = [\n '--training_dataset_path',\n create_training_split.outputs['output_gcs_path'],\n '--validation_dataset_path',\n create_validation_split.outputs['output_gcs_path'], '--alpha',\n get_best_trial.outputs['alpha'], '--max_iter',\n get_best_trial.outputs['max_iter'], '--hptune', 'False'\n ]\n\n train_model = mlengine_train_op(\n project_id=project_id,\n region=region,\n master_image_uri=TRAINER_IMAGE,\n job_dir=job_dir,\n args=train_args)\n\n # Evaluate the model on the testing split\n eval_model = evaluate_model_op(\n dataset_path=str(create_testing_split.outputs['output_gcs_path']),\n model_path=str(train_model.outputs['job_dir']),\n metric_name=evaluation_metric_name)\n\n # Deploy the model if the primary metric is better than threshold\n with kfp.dsl.Condition(eval_model.outputs['metric_value'] > evaluation_metric_threshold):\n deploy_model = mlengine_deploy_op(\n model_uri=train_model.outputs['job_dir'],\n project_id=project_id,\n model_id=model_id,\n version_id=version_id,\n runtime_version=RUNTIME_VERSION,\n python_version=PYTHON_VERSION,\n replace_existing_version=replace_existing_version)\n\n # Configure the pipeline to run using the service account defined\n # in the user-gcp-sa k8s secret\n if USE_KFP_SA == 'True':\n kfp.dsl.get_pipeline_conf().add_op_transformer(\n use_gcp_secret('user-gcp-sa'))", "_____no_output_____" ] ], [ [ "The custom components execute in a container image defined in `base_image/Dockerfile`.", "_____no_output_____" ] ], [ [ "!cat base_image/Dockerfile", "_____no_output_____" ] ], [ [ "The training step in the pipeline employes the AI Platform Training component to schedule a AI Platform Training job in a custom training container. The custom training image is defined in `trainer_image/Dockerfile`.", "_____no_output_____" ] ], [ [ "!cat trainer_image/Dockerfile", "_____no_output_____" ] ], [ [ "## Building and deploying the pipeline\n\nBefore deploying to AI Platform Pipelines, the pipeline DSL has to be compiled into a pipeline runtime format, also refered to as a pipeline package. The runtime format is based on [Argo Workflow](https://github.com/argoproj/argo), which is expressed in YAML. \n", "_____no_output_____" ], [ "### Configure environment settings\n\nUpdate the below constants with the settings reflecting your lab environment. \n\n- `REGION` - the compute region for AI Platform Training and Prediction\n- `ARTIFACT_STORE` - the GCS bucket created during installation of AI Platform Pipelines. The bucket name will be similar to `qwiklabs-gcp-xx-xxxxxxx-kubeflowpipelines-default`.\n- `ENDPOINT` - set the `ENDPOINT` constant to the endpoint to your AI Platform Pipelines instance. Then endpoint to the AI Platform Pipelines instance can be found on the [AI Platform Pipelines](https://console.cloud.google.com/ai-platform/pipelines/clusters) page in the Google Cloud Console.\n\n1. Open the **SETTINGS** for your instance\n2. Use the value of the `host` variable in the **Connect to this Kubeflow Pipelines instance from a Python client via Kubeflow Pipelines SKD** section of the **SETTINGS** window.\n\nRun gsutil ls without URLs to list all of the Cloud Storage buckets under your default project ID.", "_____no_output_____" ] ], [ [ "!gsutil ls", "_____no_output_____" ] ], [ [ "**HINT:** \n\nFor **ENDPOINT**, use the value of the `host` variable in the **Connect to this Kubeflow Pipelines instance from a Python client via Kubeflow Pipelines SDK** section of the **SETTINGS** window.\n\nFor **ARTIFACT_STORE_URI**, copy the bucket name which starts with the qwiklabs-gcp-xx-xxxxxxx-kubeflowpipelines-default prefix from the previous cell output. Your copied value should look like **'gs://qwiklabs-gcp-xx-xxxxxxx-kubeflowpipelines-default'**", "_____no_output_____" ] ], [ [ "REGION = 'us-central1'\nENDPOINT = '627be4a1d4049ed3-dot-us-central1.pipelines.googleusercontent.com' # TO DO: REPLACE WITH YOUR ENDPOINT\nARTIFACT_STORE_URI = 'gs://dna-gcp-data-kubeflowpipelines-default' # TO DO: REPLACE WITH YOUR ARTIFACT_STORE NAME \nPROJECT_ID = !(gcloud config get-value core/project)\nPROJECT_ID = PROJECT_ID[0]", "_____no_output_____" ] ], [ [ "### Build the trainer image", "_____no_output_____" ] ], [ [ "IMAGE_NAME='trainer_image'\nTAG='test'\nTRAINER_IMAGE='gcr.io/{}/{}:{}'.format(PROJECT_ID, IMAGE_NAME, TAG)", "_____no_output_____" ] ], [ [ "#### **Note**: Please ignore any **incompatibility ERROR** that may appear for the packages visions as it will not affect the lab's functionality.", "_____no_output_____" ] ], [ [ "!gcloud builds submit --timeout 15m --tag $TRAINER_IMAGE trainer_image", "_____no_output_____" ] ], [ [ "### Build the base image for custom components", "_____no_output_____" ] ], [ [ "IMAGE_NAME='base_image'\nTAG='test2'\nBASE_IMAGE='gcr.io/{}/{}:{}'.format(PROJECT_ID, IMAGE_NAME, TAG)\n!pwd", "/home/jupyter/demo/mlops-on-gcp/on_demand/kfp-caip-sklearn/lab-02-kfp-pipeline\n" ], [ "!gcloud builds submit --timeout 15m --tag $BASE_IMAGE base_image", "Creating temporary tarball archive of 2 file(s) totalling 290 bytes before compression.\nUploading tarball of [base_image] to [gs://dna-gcp-data_cloudbuild/source/1621581960.433286-cef9441cb3234402ad8faeccf31ce5fe.tgz]\nCreated [https://cloudbuild.googleapis.com/v1/projects/dna-gcp-data/locations/global/builds/d2e1016b-599c-4537-b03f-3a8e0039c2fc].\nLogs are available at [https://console.cloud.google.com/cloud-build/builds/d2e1016b-599c-4537-b03f-3a8e0039c2fc?project=1011566672334].\n----------------------------- REMOTE BUILD OUTPUT ------------------------------\nstarting build \"d2e1016b-599c-4537-b03f-3a8e0039c2fc\"\n\nFETCHSOURCE\nFetching storage object: gs://dna-gcp-data_cloudbuild/source/1621581960.433286-cef9441cb3234402ad8faeccf31ce5fe.tgz#1621581960754721\nCopying gs://dna-gcp-data_cloudbuild/source/1621581960.433286-cef9441cb3234402ad8faeccf31ce5fe.tgz#1621581960754721...\n/ [1 files][ 299.0 B/ 299.0 B] \nOperation completed over 1 objects/299.0 B. \nBUILD\nAlready have image (with digest): gcr.io/cloud-builders/docker\nSending build context to Docker daemon 3.584kB\nStep 1/3 : FROM gcr.io/deeplearning-platform-release/base-cpu\nlatest: Pulling from deeplearning-platform-release/base-cpu\n01bf7da0a88c: Pulling fs layer\nf3b4a5f15c7a: Pulling fs layer\n57ffbe87baa1: Pulling fs layer\n424e7c9d5d89: Pulling fs layer\n9b397537aef0: Pulling fs layer\n2bd5028f4b85: Pulling fs layer\n4f4fb700ef54: Pulling fs layer\nb2ed56b85d3a: Pulling fs layer\n8bfb788e9874: Pulling fs layer\n0618fb353339: Pulling fs layer\n42045a665612: Pulling fs layer\n031d8d7b75f7: Pulling fs layer\n5780cc9addac: Pulling fs layer\n8fbe78107b3d: Pulling fs layer\neee173fc570a: Pulling fs layer\n424e7c9d5d89: Waiting\n9b397537aef0: Waiting\n2bd5028f4b85: Waiting\n4f4fb700ef54: Waiting\nb2ed56b85d3a: Waiting\n8bfb788e9874: Waiting\n0618fb353339: Waiting\n42045a665612: Waiting\n031d8d7b75f7: Waiting\n5780cc9addac: Waiting\n8fbe78107b3d: Waiting\neee173fc570a: Waiting\n9334ecc802d5: Pulling fs layer\nc631c38965fd: Pulling fs layer\n9334ecc802d5: Waiting\nc631c38965fd: Waiting\n1aada407354a: Pulling fs layer\n151efcb7d3c3: Pulling fs layer\n151efcb7d3c3: Waiting\n1aada407354a: Waiting\n57ffbe87baa1: Verifying Checksum\n57ffbe87baa1: Download complete\nf3b4a5f15c7a: Verifying Checksum\nf3b4a5f15c7a: Download complete\n424e7c9d5d89: Verifying Checksum\n424e7c9d5d89: Download complete\n01bf7da0a88c: Verifying Checksum\n01bf7da0a88c: Download complete\n4f4fb700ef54: Verifying Checksum\n4f4fb700ef54: Download complete\nb2ed56b85d3a: Verifying Checksum\nb2ed56b85d3a: Download complete\n2bd5028f4b85: Verifying Checksum\n2bd5028f4b85: Download complete\n0618fb353339: Download complete\n42045a665612: Verifying Checksum\n42045a665612: Download complete\n031d8d7b75f7: Verifying Checksum\n031d8d7b75f7: Download complete\n5780cc9addac: Download complete\n8fbe78107b3d: Verifying Checksum\n8fbe78107b3d: Download complete\neee173fc570a: Verifying Checksum\neee173fc570a: Download complete\n9334ecc802d5: Verifying Checksum\n9334ecc802d5: Download complete\nc631c38965fd: Verifying Checksum\nc631c38965fd: Download complete\n9b397537aef0: Verifying Checksum\n9b397537aef0: Download complete\n8bfb788e9874: Verifying Checksum\n8bfb788e9874: Download complete\n151efcb7d3c3: Verifying Checksum\n151efcb7d3c3: Download complete\n01bf7da0a88c: Pull complete\nf3b4a5f15c7a: Pull complete\n57ffbe87baa1: Pull complete\n424e7c9d5d89: Pull complete\n1aada407354a: Verifying Checksum\n1aada407354a: Download complete\n9b397537aef0: Pull complete\n2bd5028f4b85: Pull complete\n4f4fb700ef54: Pull complete\nb2ed56b85d3a: Pull complete\n8bfb788e9874: Pull complete\n0618fb353339: Pull complete\n42045a665612: Pull complete\n031d8d7b75f7: Pull complete\n5780cc9addac: Pull complete\n8fbe78107b3d: Pull complete\neee173fc570a: Pull complete\n9334ecc802d5: Pull complete\nc631c38965fd: Pull complete\n1aada407354a: Pull complete\n151efcb7d3c3: Pull complete\nDigest: sha256:76ee9c0261dbcfb75e201ce21fd666f61127fe6b9ff74e6cf78b6ef09751de95\nStatus: Downloaded newer image for gcr.io/deeplearning-platform-release/base-cpu:latest\n ---> c1e1d5999dc3\nStep 2/3 : RUN pip install -U fire scikit-learn==0.20.4 pandas==0.24.2 kfp==1.0.0\n ---> Running in c4bf640bbcee\nCollecting fire\n Downloading fire-0.4.0.tar.gz (87 kB)\nCollecting scikit-learn==0.20.4\n Downloading scikit_learn-0.20.4-cp37-cp37m-manylinux1_x86_64.whl (5.4 MB)\nCollecting pandas==0.24.2\n Downloading pandas-0.24.2-cp37-cp37m-manylinux1_x86_64.whl (10.1 MB)\nCollecting kfp==1.0.0\n Downloading kfp-1.0.0.tar.gz (116 kB)\nRequirement already satisfied: scipy>=0.13.3 in /opt/conda/lib/python3.7/site-packages (from scikit-learn==0.20.4) (1.6.3)\nRequirement already satisfied: numpy>=1.8.2 in /opt/conda/lib/python3.7/site-packages (from scikit-learn==0.20.4) (1.19.5)\nRequirement already satisfied: python-dateutil>=2.5.0 in /opt/conda/lib/python3.7/site-packages (from pandas==0.24.2) (2.8.1)\nRequirement already satisfied: pytz>=2011k in /opt/conda/lib/python3.7/site-packages (from pandas==0.24.2) (2021.1)\nRequirement already satisfied: PyYAML in /opt/conda/lib/python3.7/site-packages (from kfp==1.0.0) (5.4.1)\nRequirement already satisfied: google-cloud-storage>=1.13.0 in /opt/conda/lib/python3.7/site-packages (from kfp==1.0.0) (1.38.0)\nCollecting kubernetes<12.0.0,>=8.0.0\n Downloading kubernetes-11.0.0-py3-none-any.whl (1.5 MB)\nRequirement already satisfied: google-auth>=1.6.1 in /opt/conda/lib/python3.7/site-packages (from kfp==1.0.0) (1.30.0)\nCollecting requests_toolbelt>=0.8.0\n Downloading requests_toolbelt-0.9.1-py2.py3-none-any.whl (54 kB)\nRequirement already satisfied: cloudpickle in /opt/conda/lib/python3.7/site-packages (from kfp==1.0.0) (1.6.0)\nCollecting kfp-server-api<2.0.0,>=0.2.5\n Downloading kfp-server-api-1.5.0.tar.gz (50 kB)\nRequirement already satisfied: jsonschema>=3.0.1 in /opt/conda/lib/python3.7/site-packages (from kfp==1.0.0) (3.2.0)\nCollecting tabulate\n Downloading tabulate-0.8.9-py3-none-any.whl (25 kB)\nRequirement already satisfied: click in /opt/conda/lib/python3.7/site-packages (from kfp==1.0.0) (7.1.2)\nCollecting Deprecated\n Downloading Deprecated-1.2.12-py2.py3-none-any.whl (9.5 kB)\nCollecting strip-hints\n Downloading strip-hints-0.1.9.tar.gz (30 kB)\nRequirement already satisfied: setuptools>=40.3.0 in /opt/conda/lib/python3.7/site-packages (from google-auth>=1.6.1->kfp==1.0.0) (49.6.0.post20210108)\nRequirement already satisfied: cachetools<5.0,>=2.0.0 in /opt/conda/lib/python3.7/site-packages (from google-auth>=1.6.1->kfp==1.0.0) (4.2.2)\nRequirement already satisfied: rsa<5,>=3.1.4 in /opt/conda/lib/python3.7/site-packages (from google-auth>=1.6.1->kfp==1.0.0) (4.7.2)\nRequirement already satisfied: pyasn1-modules>=0.2.1 in /opt/conda/lib/python3.7/site-packages (from google-auth>=1.6.1->kfp==1.0.0) (0.2.7)\nRequirement already satisfied: six>=1.9.0 in /opt/conda/lib/python3.7/site-packages (from google-auth>=1.6.1->kfp==1.0.0) (1.16.0)\nRequirement already satisfied: requests<3.0.0dev,>=2.18.0 in /opt/conda/lib/python3.7/site-packages (from google-cloud-storage>=1.13.0->kfp==1.0.0) (2.25.1)\nRequirement already satisfied: google-resumable-media<2.0dev,>=1.2.0 in /opt/conda/lib/python3.7/site-packages (from google-cloud-storage>=1.13.0->kfp==1.0.0) (1.2.0)\nRequirement already satisfied: google-cloud-core<2.0dev,>=1.4.1 in /opt/conda/lib/python3.7/site-packages (from google-cloud-storage>=1.13.0->kfp==1.0.0) (1.6.0)\nRequirement already satisfied: google-api-core<2.0.0dev,>=1.21.0 in /opt/conda/lib/python3.7/site-packages (from google-cloud-core<2.0dev,>=1.4.1->google-cloud-storage>=1.13.0->kfp==1.0.0) (1.26.3)\nRequirement already satisfied: packaging>=14.3 in /opt/conda/lib/python3.7/site-packages (from google-api-core<2.0.0dev,>=1.21.0->google-cloud-core<2.0dev,>=1.4.1->google-cloud-storage>=1.13.0->kfp==1.0.0) (20.9)\nRequirement already satisfied: googleapis-common-protos<2.0dev,>=1.6.0 in /opt/conda/lib/python3.7/site-packages (from google-api-core<2.0.0dev,>=1.21.0->google-cloud-core<2.0dev,>=1.4.1->google-cloud-storage>=1.13.0->kfp==1.0.0) (1.53.0)\nRequirement already satisfied: protobuf>=3.12.0 in /opt/conda/lib/python3.7/site-packages (from google-api-core<2.0.0dev,>=1.21.0->google-cloud-core<2.0dev,>=1.4.1->google-cloud-storage>=1.13.0->kfp==1.0.0) (3.16.0)\nRequirement already satisfied: google-crc32c<2.0dev,>=1.0 in /opt/conda/lib/python3.7/site-packages (from google-resumable-media<2.0dev,>=1.2.0->google-cloud-storage>=1.13.0->kfp==1.0.0) (1.1.2)\nRequirement already satisfied: cffi>=1.0.0 in /opt/conda/lib/python3.7/site-packages (from google-crc32c<2.0dev,>=1.0->google-resumable-media<2.0dev,>=1.2.0->google-cloud-storage>=1.13.0->kfp==1.0.0) (1.14.5)\nRequirement already satisfied: pycparser in /opt/conda/lib/python3.7/site-packages (from cffi>=1.0.0->google-crc32c<2.0dev,>=1.0->google-resumable-media<2.0dev,>=1.2.0->google-cloud-storage>=1.13.0->kfp==1.0.0) (2.20)\nRequirement already satisfied: attrs>=17.4.0 in /opt/conda/lib/python3.7/site-packages (from jsonschema>=3.0.1->kfp==1.0.0) (21.2.0)\nRequirement already satisfied: importlib-metadata in /opt/conda/lib/python3.7/site-packages (from jsonschema>=3.0.1->kfp==1.0.0) (4.0.1)\nRequirement already satisfied: pyrsistent>=0.14.0 in /opt/conda/lib/python3.7/site-packages (from jsonschema>=3.0.1->kfp==1.0.0) (0.17.3)\nRequirement already satisfied: urllib3>=1.15 in /opt/conda/lib/python3.7/site-packages (from kfp-server-api<2.0.0,>=0.2.5->kfp==1.0.0) (1.26.4)\nRequirement already satisfied: certifi in /opt/conda/lib/python3.7/site-packages (from kfp-server-api<2.0.0,>=0.2.5->kfp==1.0.0) (2020.12.5)\nRequirement already satisfied: websocket-client!=0.40.0,!=0.41.*,!=0.42.*,>=0.32.0 in /opt/conda/lib/python3.7/site-packages (from kubernetes<12.0.0,>=8.0.0->kfp==1.0.0) (0.57.0)\nRequirement already satisfied: requests-oauthlib in /opt/conda/lib/python3.7/site-packages (from kubernetes<12.0.0,>=8.0.0->kfp==1.0.0) (1.3.0)\nRequirement already satisfied: pyparsing>=2.0.2 in /opt/conda/lib/python3.7/site-packages (from packaging>=14.3->google-api-core<2.0.0dev,>=1.21.0->google-cloud-core<2.0dev,>=1.4.1->google-cloud-storage>=1.13.0->kfp==1.0.0) (2.4.7)\nRequirement already satisfied: pyasn1<0.5.0,>=0.4.6 in /opt/conda/lib/python3.7/site-packages (from pyasn1-modules>=0.2.1->google-auth>=1.6.1->kfp==1.0.0) (0.4.8)\nRequirement already satisfied: idna<3,>=2.5 in /opt/conda/lib/python3.7/site-packages (from requests<3.0.0dev,>=2.18.0->google-cloud-storage>=1.13.0->kfp==1.0.0) (2.10)\nRequirement already satisfied: chardet<5,>=3.0.2 in /opt/conda/lib/python3.7/site-packages (from requests<3.0.0dev,>=2.18.0->google-cloud-storage>=1.13.0->kfp==1.0.0) (4.0.0)\nCollecting termcolor\n Downloading termcolor-1.1.0.tar.gz (3.9 kB)\nRequirement already satisfied: wrapt<2,>=1.10 in /opt/conda/lib/python3.7/site-packages (from Deprecated->kfp==1.0.0) (1.12.1)\nRequirement already satisfied: zipp>=0.5 in /opt/conda/lib/python3.7/site-packages (from importlib-metadata->jsonschema>=3.0.1->kfp==1.0.0) (3.4.1)\nRequirement already satisfied: typing-extensions>=3.6.4 in /opt/conda/lib/python3.7/site-packages (from importlib-metadata->jsonschema>=3.0.1->kfp==1.0.0) (3.7.4.3)\nRequirement already satisfied: oauthlib>=3.0.0 in /opt/conda/lib/python3.7/site-packages (from requests-oauthlib->kubernetes<12.0.0,>=8.0.0->kfp==1.0.0) (3.0.1)\nRequirement already satisfied: wheel in /opt/conda/lib/python3.7/site-packages (from strip-hints->kfp==1.0.0) (0.36.2)\nBuilding wheels for collected packages: kfp, kfp-server-api, fire, strip-hints, termcolor\n Building wheel for kfp (setup.py): started\n Building wheel for kfp (setup.py): finished with status 'done'\n Created wheel for kfp: filename=kfp-1.0.0-py3-none-any.whl size=159769 sha256=74a8b1d2b91b9957b95f2f878e91ba0a2b847ba6b479ba6090b199fee94f8de1\n Stored in directory: /root/.cache/pip/wheels/81/39/f2/ee01d785a5bd135e42e7721fedb05857badf763fc465a4e822\n Building wheel for kfp-server-api (setup.py): started\n Building wheel for kfp-server-api (setup.py): finished with status 'done'\n Created wheel for kfp-server-api: filename=kfp_server_api-1.5.0-py3-none-any.whl size=92524 sha256=b672a04ca3bcf1257f2981061c5a9b460c20f5e816ede414eb22892649d86973\n Stored in directory: /root/.cache/pip/wheels/1e/ab/eb/1608f904a1a3f2a28696129c6dbd3cac00bea2cdad226ee60e\n Building wheel for fire (setup.py): started\n Building wheel for fire (setup.py): finished with status 'done'\n Created wheel for fire: filename=fire-0.4.0-py2.py3-none-any.whl size=115928 sha256=040e97679ac4b63443c3b3127e2f2b0afb4d70d927adcc11efa1f94be0084ba3\n Stored in directory: /root/.cache/pip/wheels/8a/67/fb/2e8a12fa16661b9d5af1f654bd199366799740a85c64981226\n Building wheel for strip-hints (setup.py): started\n Building wheel for strip-hints (setup.py): finished with status 'done'\n Created wheel for strip-hints: filename=strip_hints-0.1.9-py2.py3-none-any.whl size=20993 sha256=9481cd3b4b52c0e9713db3553c23c1bda587a1075bcd59d71e99110b6f1b6533\n Stored in directory: /root/.cache/pip/wheels/2d/b8/4e/a3ec111d2db63cec88121bd7c0ab1a123bce3b55dd19dda5c1\n Building wheel for termcolor (setup.py): started\n Building wheel for termcolor (setup.py): finished with status 'done'\n Created wheel for termcolor: filename=termcolor-1.1.0-py3-none-any.whl size=4829 sha256=b67e2ecf8cbb39455760408250cf67d6ddf4f873c94b486c5e753b4e84f4c8db\n Stored in directory: /root/.cache/pip/wheels/3f/e3/ec/8a8336ff196023622fbcb36de0c5a5c218cbb24111d1d4c7f2\nSuccessfully built kfp kfp-server-api fire strip-hints termcolor\nInstalling collected packages: termcolor, tabulate, strip-hints, requests-toolbelt, kubernetes, kfp-server-api, Deprecated, scikit-learn, pandas, kfp, fire\n Attempting uninstall: kubernetes\n Found existing installation: kubernetes 12.0.1\n Uninstalling kubernetes-12.0.1:\n Successfully uninstalled kubernetes-12.0.1\n Attempting uninstall: scikit-learn\n Found existing installation: scikit-learn 0.24.2\n Uninstalling scikit-learn-0.24.2:\n Successfully uninstalled scikit-learn-0.24.2\n Attempting uninstall: pandas\n Found existing installation: pandas 1.2.4\n Uninstalling pandas-1.2.4:\n Successfully uninstalled pandas-1.2.4\n\u001b[91mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\nvisions 0.7.1 requires pandas>=0.25.3, but you have pandas 0.24.2 which is incompatible.\nphik 0.11.2 requires pandas>=0.25.1, but you have pandas 0.24.2 which is incompatible.\npandas-profiling 3.0.0 requires pandas!=1.0.0,!=1.0.1,!=1.0.2,!=1.1.0,>=0.25.3, but you have pandas 0.24.2 which is incompatible.\n\u001b[0m\u001b[91mWARNING: Running pip as root will break packages and permissions. You should install packages reliably by using venv: https://pip.pypa.io/warnings/venv\n\u001b[0mSuccessfully installed Deprecated-1.2.12 fire-0.4.0 kfp-1.0.0 kfp-server-api-1.5.0 kubernetes-11.0.0 pandas-0.24.2 requests-toolbelt-0.9.1 scikit-learn-0.20.4 strip-hints-0.1.9 tabulate-0.8.9 termcolor-1.1.0\nRemoving intermediate container c4bf640bbcee\n ---> 94e41aa9f2e0\nStep 3/3 : RUN pip list | grep kfp\n ---> Running in cf3f27276384\nkfp 1.0.0\nkfp-server-api 1.5.0\nRemoving intermediate container cf3f27276384\n ---> e31322a505ba\nSuccessfully built e31322a505ba\nSuccessfully tagged gcr.io/dna-gcp-data/base_image:test2\nPUSH\nPushing gcr.io/dna-gcp-data/base_image:test2\nThe push refers to repository [gcr.io/dna-gcp-data/base_image]\nce2e56091c30: Preparing\nfed09d378d72: Preparing\n896901ec1a67: Preparing\n06a5bf49b163: Preparing\nb34dae69fc5d: Preparing\n0ffb7465dde9: Preparing\ne2563d1ada9a: Preparing\n42b027d1e826: Preparing\n636a7c2e7d03: Preparing\n1ba1158adf89: Preparing\n96e46d1341e8: Preparing\n954f6dc3f7f5: Preparing\n8760a171b659: Preparing\n5f70bf18a086: Preparing\na0710233fd2d: Preparing\n05449afa4be9: Preparing\n5b9e34b5cf74: Preparing\n8cafc6d2db45: Preparing\na5d4bacb0351: Preparing\n5153e1acaabc: Preparing\n96e46d1341e8: Waiting\n954f6dc3f7f5: Waiting\n8760a171b659: Waiting\n5f70bf18a086: Waiting\na0710233fd2d: Waiting\n05449afa4be9: Waiting\n5b9e34b5cf74: Waiting\n8cafc6d2db45: Waiting\n0ffb7465dde9: Waiting\ne2563d1ada9a: Waiting\n42b027d1e826: Waiting\n636a7c2e7d03: Waiting\n1ba1158adf89: Waiting\n5153e1acaabc: Waiting\n896901ec1a67: Layer already exists\nfed09d378d72: Layer already exists\nb34dae69fc5d: Layer already exists\n06a5bf49b163: Layer already exists\n0ffb7465dde9: Layer already exists\ne2563d1ada9a: Layer already exists\n42b027d1e826: Layer already exists\n636a7c2e7d03: Layer already exists\n1ba1158adf89: Layer already exists\n954f6dc3f7f5: Layer already exists\n8760a171b659: Layer already exists\n5f70bf18a086: Layer already exists\n96e46d1341e8: Layer already exists\n8cafc6d2db45: Layer already exists\n5b9e34b5cf74: Layer already exists\n05449afa4be9: Layer already exists\na0710233fd2d: Layer already exists\na5d4bacb0351: Layer already exists\n5153e1acaabc: Layer already exists\nce2e56091c30: Pushed\ntest2: digest: sha256:09e11684e6de0ac905077560c50a24b94b8b3d22afa167ccb7f2b92345068511 size: 4499\nDONE\n--------------------------------------------------------------------------------\n\nID CREATE_TIME DURATION SOURCE IMAGES STATUS\nd2e1016b-599c-4537-b03f-3a8e0039c2fc 2021-05-21T07:26:00+00:00 2M25S gs://dna-gcp-data_cloudbuild/source/1621581960.433286-cef9441cb3234402ad8faeccf31ce5fe.tgz gcr.io/dna-gcp-data/base_image:test2 SUCCESS\n" ] ], [ [ "### Compile the pipeline\n\nYou can compile the DSL using an API from the **KFP SDK** or using the **KFP** compiler.\n\nTo compile the pipeline DSL using the **KFP** compiler.", "_____no_output_____" ], [ "#### Set the pipeline's compile time settings\n\nThe pipeline can run using a security context of the GKE default node pool's service account or the service account defined in the `user-gcp-sa` secret of the Kubernetes namespace hosting KFP. If you want to use the `user-gcp-sa` service account you change the value of `USE_KFP_SA` to `True`.\n\nNote that the default AI Platform Pipelines configuration does not define the `user-gcp-sa` secret.", "_____no_output_____" ] ], [ [ "USE_KFP_SA = False\n\nCOMPONENT_URL_SEARCH_PREFIX = 'https://raw.githubusercontent.com/kubeflow/pipelines/0.2.5/components/gcp/'\nRUNTIME_VERSION = '1.15'\nPYTHON_VERSION = '3.7'\nENDPOINT='https://627be4a1d4049ed3-dot-us-central1.pipelines.googleusercontent.com'\n\n%env USE_KFP_SA={USE_KFP_SA}\n%env BASE_IMAGE={BASE_IMAGE}\n%env TRAINER_IMAGE={TRAINER_IMAGE}\n%env COMPONENT_URL_SEARCH_PREFIX={COMPONENT_URL_SEARCH_PREFIX}\n%env RUNTIME_VERSION={RUNTIME_VERSION}\n%env PYTHON_VERSION={PYTHON_VERSION}\n%env ENDPOINT={ENDPOINT}", "env: USE_KFP_SA=False\nenv: BASE_IMAGE=gcr.io/dna-gcp-data/base_image:test2\nenv: TRAINER_IMAGE=gcr.io/dna-gcp-data/trainer_image:test\nenv: COMPONENT_URL_SEARCH_PREFIX=https://raw.githubusercontent.com/kubeflow/pipelines/0.2.5/components/gcp/\nenv: RUNTIME_VERSION=1.15\nenv: PYTHON_VERSION=3.7\nenv: ENDPOINT=https://627be4a1d4049ed3-dot-us-central1.pipelines.googleusercontent.com\n" ] ], [ [ "#### Use the CLI compiler to compile the pipeline", "_____no_output_____" ] ], [ [ "!dsl-compile --py pipeline/covertype_training_pipeline.py --output covertype_training_pipeline.yaml", "_____no_output_____" ] ], [ [ "The result is the `covertype_training_pipeline.yaml` file. ", "_____no_output_____" ] ], [ [ "!head covertype_training_pipeline.yaml", "_____no_output_____" ] ], [ [ "### Deploy the pipeline package", "_____no_output_____" ] ], [ [ "PIPELINE_NAME='covertype_continuous_training'\n\n!kfp --endpoint $ENDPOINT pipeline upload \\\n-p $PIPELINE_NAME \\\ncovertype_training_pipeline.yaml", "Pipeline 7eda6268-681e-41eb-8f65-a9c853030888 has been submitted\n\nPipeline Details\n------------------\nID 7eda6268-681e-41eb-8f65-a9c853030888\nName covertype_continuous_training\nDescription\nUploaded at 2021-05-21T08:50:00+00:00\n+--------------------------+--------------------------------------------------+\n| Parameter Name | Default Value |\n+==========================+==================================================+\n| project_id | |\n+--------------------------+--------------------------------------------------+\n| region | |\n+--------------------------+--------------------------------------------------+\n| source_table_name | |\n+--------------------------+--------------------------------------------------+\n| gcs_root | |\n+--------------------------+--------------------------------------------------+\n| dataset_id | |\n+--------------------------+--------------------------------------------------+\n| evaluation_metric_name | |\n+--------------------------+--------------------------------------------------+\n| model_id | |\n+--------------------------+--------------------------------------------------+\n| version_id | |\n+--------------------------+--------------------------------------------------+\n| replace_existing_version | |\n+--------------------------+--------------------------------------------------+\n| experiment_id | |\n+--------------------------+--------------------------------------------------+\n| hypertune_settings | { |\n| | \"hyperparameters\": { |\n| | \"goal\": \"MAXIMIZE\", |\n| | \"maxTrials\": 6, |\n| | \"maxParallelTrials\": 3, |\n| | \"hyperparameterMetricTag\": \"accuracy\", |\n| | \"enableTrialEarlyStopping\": True, |\n| | \"params\": [ |\n| | { |\n| | \"parameterName\": \"max_iter\", |\n| | \"type\": \"DISCRETE\", |\n| | \"discreteValues\": [500, 1000] |\n| | }, |\n| | { |\n| | \"parameterName\": \"alpha\", |\n| | \"type\": \"DOUBLE\", |\n| | \"minValue\": 0.0001, |\n| | \"maxValue\": 0.001, |\n| | \"scaleType\": \"UNIT_LINEAR_SCALE\" |\n| | } |\n| | ] |\n| | } |\n| | } |\n+--------------------------+--------------------------------------------------+\n| dataset_location | US |\n+--------------------------+--------------------------------------------------+\n" ] ], [ [ "## Submitting pipeline runs\n\nYou can trigger pipeline runs using an API from the KFP SDK or using KFP CLI. To submit the run using KFP CLI, execute the following commands. Notice how the pipeline's parameters are passed to the pipeline run.\n\n### List the pipelines in AI Platform Pipelines", "_____no_output_____" ] ], [ [ "!kfp --endpoint $ENDPOINT experiment list", "+--------------------------------------+-------------------------------+---------------------------+\n| Experiment ID | Name | Created at |\n+======================================+===============================+===========================+\n| 889c1532-fee9-4b06-bc2b-10b1cd332c9a | Covertype_Classifier_Training | 2021-05-19T12:54:04+00:00 |\n+--------------------------------------+-------------------------------+---------------------------+\n| 3794c159-24a7-41e0-89be-f23152971870 | helloworld-dev | 2021-05-06T16:07:23+00:00 |\n+--------------------------------------+-------------------------------+---------------------------+\n| 821a36b0-8db9-4604-9e65-035b8f70c77d | my_pipeline | 2021-05-06T10:35:19+00:00 |\n+--------------------------------------+-------------------------------+---------------------------+\n| 6587995a-9b11-4a8e-a2fc-d0b80534dfe8 | Default | 2021-05-04T02:29:12+00:00 |\n+--------------------------------------+-------------------------------+---------------------------+\n" ] ], [ [ "### Submit a run\n\nFind the ID of the `covertype_continuous_training` pipeline you uploaded in the previous step and update the value of `PIPELINE_ID` .\n\n", "_____no_output_____" ] ], [ [ "PIPELINE_ID='7eda6268-681e-41eb-8f65-a9c853030888' # TO DO: REPLACE WITH YOUR PIPELINE ID ", "_____no_output_____" ], [ "EXPERIMENT_NAME = 'Covertype_Classifier_Training'\nRUN_ID = 'Run_001'\nSOURCE_TABLE = 'covertype_dataset.covertype'\nDATASET_ID = 'covertype_dataset'\nEVALUATION_METRIC = 'accuracy'\nMODEL_ID = 'covertype_classifier'\nVERSION_ID = 'v01'\nREPLACE_EXISTING_VERSION = 'True'\nEXPERIMENT_ID = '889c1532-fee9-4b06-bc2b-10b1cd332c9a'\nGCS_STAGING_PATH = '{}/staging'.format(ARTIFACT_STORE_URI)", "_____no_output_____" ] ], [ [ "Run the pipeline using the `kfp` command line by retrieving the variables from the environment to pass to the pipeline where:\n\n- EXPERIMENT_NAME is set to the experiment used to run the pipeline. You can choose any name you want. If the experiment does not exist it will be created by the command\n- RUN_ID is the name of the run. You can use an arbitrary name\n- PIPELINE_ID is the id of your pipeline. Use the value retrieved by the `kfp pipeline list` command\n- GCS_STAGING_PATH is the URI to the Cloud Storage location used by the pipeline to store intermediate files. By default, it is set to the `staging` folder in your artifact store.\n- REGION is a compute region for AI Platform Training and Prediction. \n\nYou should be already familiar with these and other parameters passed to the command. If not go back and review the pipeline code.", "_____no_output_____" ] ], [ [ "!kfp --endpoint $ENDPOINT run submit \\\n-e $EXPERIMENT_NAME \\\n-r $RUN_ID \\\n-p $PIPELINE_ID \\\nproject_id=$PROJECT_ID \\\ngcs_root=$GCS_STAGING_PATH \\\nregion=$REGION \\\nsource_table_name=$SOURCE_TABLE \\\ndataset_id=$DATASET_ID \\\nevaluation_metric_name=$EVALUATION_METRIC \\\nmodel_id=$MODEL_ID \\\nversion_id=$VERSION_ID \\\nreplace_existing_version=$REPLACE_EXISTING_VERSION \\\nexperiment_id=$EXPERIMENT_ID", "Run b1b45663-292c-4d0d-ba93-66671abde854 is submitted\n+--------------------------------------+---------+----------+---------------------------+\n| run id | name | status | created at |\n+======================================+=========+==========+===========================+\n| b1b45663-292c-4d0d-ba93-66671abde854 | Run_001 | | 2021-05-21T09:13:53+00:00 |\n+--------------------------------------+---------+----------+---------------------------+\n" ], [ "#!kfp --endpoint $ENDPOINT experiment list\nfrom typing import NamedTuple\ndef get_previous_run_metric( ENDPOINT: str, experiment_id: str ) -> NamedTuple('Outputs', [('run_id', str), ('accuracy', float)]):\n\n import kfp as kfp\n runs_details= kfp.Client(host=ENDPOINT).list_runs(experiment_id=experiment_id, sort_by='created_at desc').to_dict()\n# print(runs_details)\n latest_success_run_details=''\n print(\"runs_details['runs'] type {}\".format(type(runs_details['runs'])))\n for i in runs_details['runs']:\n print(\"i['status'] type {}\".format(type(i['status'])))\n if i['status'] == 'Succeeded':\n run_id=i['id']\n accuracy=i['metrics'][0]['number_value']\n break;\n print(\"accuracy={}\".format(accuracy))\n print(type(run_id))\n return (run_id, accuracy)\n\na=get_previous_run_metric(ENDPOINT, experiment_id)\nprint(a)\n", "runs_details['runs'] type <class 'list'>\ni['status'] type <class 'str'>\naccuracy=0.7225525168450257\n<class 'str'>\n('d16368c5-24bb-4292-9060-315755f79b0b', 0.7225525168450257)\n" ], [ "import kfp as kfp\nruns_details= kfp.Client(host=ENDPOINT).list_runs(experiment_id=experiment_id, sort_by='created_at desc').to_dict()\nlatest_success_run_details=''\nprint(\"runs_details['runs'] type {}\".format(type(runs_details['runs'])))\nfor i in runs_details['runs']:\n print(\"i['status'] type {}\".format(type(i['status'])))\n if i['status'] == 'Succeeded':\n latest_success_run_details=i\n break;\nrun_id=latest_success_run_details['id']\nrun_id_details=kfp.Client(host=ENDPOINT).get_run(run_id=run_id).to_dict()\nprint(run_id_details)\naccuracy=run_id_details['run']['metrics'][0]['number_value']\nprint(accuracy)", "runs_details['runs'] type <class 'list'>\ni['status'] type <class 'str'>\n{'run': {'id': 'd16368c5-24bb-4292-9060-315755f79b0b', 'name': 'Run_001', 'storage_state': None, 'description': None, 'pipeline_spec': {'pipeline_id': None, 'pipeline_name': None, 'workflow_manifest': '{\"kind\":\"Workflow\",\"apiVersion\":\"argoproj.io/v1alpha1\",\"metadata\":{\"generateName\":\"covertype-classifier-training-\",\"creationTimestamp\":null,\"annotations\":{\"pipelines.kubeflow.org/pipeline_spec\":\"{\\\\\"description\\\\\": \\\\\"The pipeline training and deploying the Covertype classifierpipeline_yaml\\\\\", \\\\\"inputs\\\\\": [{\\\\\"name\\\\\": \\\\\"project_id\\\\\"}, {\\\\\"name\\\\\": \\\\\"region\\\\\"}, {\\\\\"name\\\\\": \\\\\"source_table_name\\\\\"}, {\\\\\"name\\\\\": \\\\\"gcs_root\\\\\"}, {\\\\\"name\\\\\": \\\\\"dataset_id\\\\\"}, {\\\\\"name\\\\\": \\\\\"evaluation_metric_name\\\\\"}, {\\\\\"name\\\\\": \\\\\"evaluation_metric_threshold\\\\\"}, {\\\\\"name\\\\\": \\\\\"model_id\\\\\"}, {\\\\\"name\\\\\": \\\\\"version_id\\\\\"}, {\\\\\"name\\\\\": \\\\\"replace_existing_version\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\\\\n{\\\\\\\\n \\\\\\\\\\\\\"hyperparameters\\\\\\\\\\\\\": {\\\\\\\\n \\\\\\\\\\\\\"goal\\\\\\\\\\\\\": \\\\\\\\\\\\\"MAXIMIZE\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"maxTrials\\\\\\\\\\\\\": 6,\\\\\\\\n \\\\\\\\\\\\\"maxParallelTrials\\\\\\\\\\\\\": 3,\\\\\\\\n \\\\\\\\\\\\\"hyperparameterMetricTag\\\\\\\\\\\\\": \\\\\\\\\\\\\"accuracy\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"enableTrialEarlyStopping\\\\\\\\\\\\\": True,\\\\\\\\n \\\\\\\\\\\\\"params\\\\\\\\\\\\\": [\\\\\\\\n {\\\\\\\\n \\\\\\\\\\\\\"parameterName\\\\\\\\\\\\\": \\\\\\\\\\\\\"max_iter\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"type\\\\\\\\\\\\\": \\\\\\\\\\\\\"DISCRETE\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"discreteValues\\\\\\\\\\\\\": [500, 1000]\\\\\\\\n },\\\\\\\\n {\\\\\\\\n \\\\\\\\\\\\\"parameterName\\\\\\\\\\\\\": \\\\\\\\\\\\\"alpha\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"type\\\\\\\\\\\\\": \\\\\\\\\\\\\"DOUBLE\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"minValue\\\\\\\\\\\\\": 0.0001,\\\\\\\\n \\\\\\\\\\\\\"maxValue\\\\\\\\\\\\\": 0.001,\\\\\\\\n \\\\\\\\\\\\\"scaleType\\\\\\\\\\\\\": \\\\\\\\\\\\\"UNIT_LINEAR_SCALE\\\\\\\\\\\\\"\\\\\\\\n }\\\\\\\\n ]\\\\\\\\n }\\\\\\\\n}\\\\\\\\n\\\\\", \\\\\"name\\\\\": \\\\\"hypertune_settings\\\\\", \\\\\"optional\\\\\": true}, {\\\\\"default\\\\\": \\\\\"US\\\\\", \\\\\"name\\\\\": \\\\\"dataset_location\\\\\", \\\\\"optional\\\\\": true}], \\\\\"name\\\\\": \\\\\"Covertype Classifier Training\\\\\"}\"}},\"spec\":{\"templates\":[{\"name\":\"bigquery-query\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\"},{\"name\":\"dataset_location\"},{\"name\":\"gcs_root\"},{\"name\":\"project_id\"},{\"name\":\"source_table_name\"}]},\"outputs\":{\"parameters\":[{\"name\":\"bigquery-query-output_gcs_path\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\"},{\"name\":\"bigquery-query-output_gcs_path\",\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a query to Google Cloud Bigquery \\\\\\\\nservice and dump outputs to a Google Cloud Storage blob. \\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The query used by Bigquery service to fetch the results.\\\\\", \\\\\"name\\\\\": \\\\\"query\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The project to execute the query job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the persistent dataset to keep the results of the query.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the table to keep the results of the query. If absent, the operation will generate a random id for the table.\\\\\", \\\\\"name\\\\\": \\\\\"table_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket to store the query output.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"US\\\\\", \\\\\"description\\\\\": \\\\\"The location to create the dataset. Defaults to `US`.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_location\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The full config spec for the query job.See [QueryJobConfig](https://googleapis.github.io/google-cloud-python/latest/bigquery/generated/google.cloud.bigquery.job.QueryJobConfig.html#google.cloud.bigquery.job.QueryJobConfig) for details.\\\\\", \\\\\"name\\\\\": \\\\\"job_config\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Bigquery - Query\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket containing the query output in CSV format.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\"},\"labels\":{\"add-pod-env\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.bigquery\",\"query\",\"--query\",\"\\\\n SELECT *\\\\n FROM \\\\n `{{inputs.parameters.source_table_name}}` AS cover\\\\n WHERE \\\\n MOD(ABS(FARM_FINGERPRINT(TO_JSON_STRING(cover))), 10) IN (1, 2, 3, 4)\\\\n \",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--dataset_id\",\"{{inputs.parameters.dataset_id}}\",\"--table_id\",\"\",\"--dataset_location\",\"{{inputs.parameters.dataset_location}}\",\"--output_gcs_path\",\"{{inputs.parameters.gcs_root}}/datasets/training/data.csv\",\"--job_config\",\"\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"bigquery-query-2\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\"},{\"name\":\"dataset_location\"},{\"name\":\"gcs_root\"},{\"name\":\"project_id\"},{\"name\":\"source_table_name\"}]},\"outputs\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\"},{\"name\":\"bigquery-query-2-output_gcs_path\",\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a query to Google Cloud Bigquery \\\\\\\\nservice and dump outputs to a Google Cloud Storage blob. \\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The query used by Bigquery service to fetch the results.\\\\\", \\\\\"name\\\\\": \\\\\"query\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The project to execute the query job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the persistent dataset to keep the results of the query.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the table to keep the results of the query. If absent, the operation will generate a random id for the table.\\\\\", \\\\\"name\\\\\": \\\\\"table_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket to store the query output.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"US\\\\\", \\\\\"description\\\\\": \\\\\"The location to create the dataset. Defaults to `US`.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_location\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The full config spec for the query job.See [QueryJobConfig](https://googleapis.github.io/google-cloud-python/latest/bigquery/generated/google.cloud.bigquery.job.QueryJobConfig.html#google.cloud.bigquery.job.QueryJobConfig) for details.\\\\\", \\\\\"name\\\\\": \\\\\"job_config\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Bigquery - Query\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket containing the query output in CSV format.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\"},\"labels\":{\"add-pod-env\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.bigquery\",\"query\",\"--query\",\"\\\\n SELECT *\\\\n FROM \\\\n `{{inputs.parameters.source_table_name}}` AS cover\\\\n WHERE \\\\n MOD(ABS(FARM_FINGERPRINT(TO_JSON_STRING(cover))), 10) IN (8)\\\\n \",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--dataset_id\",\"{{inputs.parameters.dataset_id}}\",\"--table_id\",\"\",\"--dataset_location\",\"{{inputs.parameters.dataset_location}}\",\"--output_gcs_path\",\"{{inputs.parameters.gcs_root}}/datasets/validation/data.csv\",\"--job_config\",\"\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"bigquery-query-3\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\"},{\"name\":\"dataset_location\"},{\"name\":\"gcs_root\"},{\"name\":\"project_id\"},{\"name\":\"source_table_name\"}]},\"outputs\":{\"parameters\":[{\"name\":\"bigquery-query-3-output_gcs_path\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\"},{\"name\":\"bigquery-query-3-output_gcs_path\",\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a query to Google Cloud Bigquery \\\\\\\\nservice and dump outputs to a Google Cloud Storage blob. \\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The query used by Bigquery service to fetch the results.\\\\\", \\\\\"name\\\\\": \\\\\"query\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The project to execute the query job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the persistent dataset to keep the results of the query.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the table to keep the results of the query. If absent, the operation will generate a random id for the table.\\\\\", \\\\\"name\\\\\": \\\\\"table_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket to store the query output.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"US\\\\\", \\\\\"description\\\\\": \\\\\"The location to create the dataset. Defaults to `US`.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_location\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The full config spec for the query job.See [QueryJobConfig](https://googleapis.github.io/google-cloud-python/latest/bigquery/generated/google.cloud.bigquery.job.QueryJobConfig.html#google.cloud.bigquery.job.QueryJobConfig) for details.\\\\\", \\\\\"name\\\\\": \\\\\"job_config\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Bigquery - Query\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket containing the query output in CSV format.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\"},\"labels\":{\"add-pod-env\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.bigquery\",\"query\",\"--query\",\"\\\\n SELECT *\\\\n FROM \\\\n `{{inputs.parameters.source_table_name}}` AS cover\\\\n WHERE \\\\n MOD(ABS(FARM_FINGERPRINT(TO_JSON_STRING(cover))), 10) IN (9)\\\\n \",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--dataset_id\",\"{{inputs.parameters.dataset_id}}\",\"--table_id\",\"\",\"--dataset_location\",\"{{inputs.parameters.dataset_location}}\",\"--output_gcs_path\",\"{{inputs.parameters.gcs_root}}/datasets/testing/data.csv\",\"--job_config\",\"\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"condition-1\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"model_id\"},{\"name\":\"project_id\"},{\"name\":\"replace_existing_version\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\"},{\"name\":\"version_id\"}]},\"outputs\":{},\"metadata\":{},\"dag\":{\"tasks\":[{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine\",\"template\":\"deploying-a-trained-model-to-cloud-machine-learning-engine\",\"arguments\":{\"parameters\":[{\"name\":\"model_id\",\"value\":\"{{inputs.parameters.model_id}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"replace_existing_version\",\"value\":\"{{inputs.parameters.replace_existing_version}}\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"{{inputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\"},{\"name\":\"version_id\",\"value\":\"{{inputs.parameters.version_id}}\"}]}}]}},{\"name\":\"covertype-classifier-training\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\"},{\"name\":\"dataset_location\"},{\"name\":\"evaluation_metric_name\"},{\"name\":\"evaluation_metric_threshold\"},{\"name\":\"gcs_root\"},{\"name\":\"hypertune_settings\"},{\"name\":\"model_id\"},{\"name\":\"project_id\"},{\"name\":\"region\"},{\"name\":\"replace_existing_version\"},{\"name\":\"source_table_name\"},{\"name\":\"version_id\"}]},\"outputs\":{},\"metadata\":{},\"dag\":{\"tasks\":[{\"name\":\"bigquery-query\",\"template\":\"bigquery-query\",\"arguments\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"{{inputs.parameters.dataset_id}}\"},{\"name\":\"dataset_location\",\"value\":\"{{inputs.parameters.dataset_location}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"source_table_name\",\"value\":\"{{inputs.parameters.source_table_name}}\"}]}},{\"name\":\"bigquery-query-2\",\"template\":\"bigquery-query-2\",\"arguments\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"{{inputs.parameters.dataset_id}}\"},{\"name\":\"dataset_location\",\"value\":\"{{inputs.parameters.dataset_location}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"source_table_name\",\"value\":\"{{inputs.parameters.source_table_name}}\"}]}},{\"name\":\"bigquery-query-3\",\"template\":\"bigquery-query-3\",\"arguments\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"{{inputs.parameters.dataset_id}}\"},{\"name\":\"dataset_location\",\"value\":\"{{inputs.parameters.dataset_location}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"source_table_name\",\"value\":\"{{inputs.parameters.source_table_name}}\"}]}},{\"name\":\"condition-1\",\"template\":\"condition-1\",\"arguments\":{\"parameters\":[{\"name\":\"model_id\",\"value\":\"{{inputs.parameters.model_id}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"replace_existing_version\",\"value\":\"{{inputs.parameters.replace_existing_version}}\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"{{tasks.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2.outputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\"},{\"name\":\"version_id\",\"value\":\"{{inputs.parameters.version_id}}\"}]},\"dependencies\":[\"evaluate-model\",\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\"],\"when\":\"{{tasks.evaluate-model.outputs.parameters.evaluate-model-metric_value}} \\\\u003e {{inputs.parameters.evaluation_metric_threshold}}\"},{\"name\":\"evaluate-model\",\"template\":\"evaluate-model\",\"arguments\":{\"parameters\":[{\"name\":\"bigquery-query-3-output_gcs_path\",\"value\":\"{{tasks.bigquery-query-3.outputs.parameters.bigquery-query-3-output_gcs_path}}\"},{\"name\":\"evaluation_metric_name\",\"value\":\"{{inputs.parameters.evaluation_metric_name}}\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"{{tasks.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2.outputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\"}]},\"dependencies\":[\"bigquery-query-3\",\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\"]},{\"name\":\"retrieve-best-run\",\"template\":\"retrieve-best-run\",\"arguments\":{\"parameters\":[{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\",\"value\":\"{{tasks.submitting-a-cloud-ml-training-job-as-a-pipeline-step.outputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id}}\"}]},\"dependencies\":[\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\"]},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\",\"template\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\",\"arguments\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\",\"value\":\"{{tasks.bigquery-query-2.outputs.parameters.bigquery-query-2-output_gcs_path}}\"},{\"name\":\"bigquery-query-output_gcs_path\",\"value\":\"{{tasks.bigquery-query.outputs.parameters.bigquery-query-output_gcs_path}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"hypertune_settings\",\"value\":\"{{inputs.parameters.hypertune_settings}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"region\",\"value\":\"{{inputs.parameters.region}}\"}]},\"dependencies\":[\"bigquery-query\",\"bigquery-query-2\"]},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\",\"template\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\",\"arguments\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\",\"value\":\"{{tasks.bigquery-query-2.outputs.parameters.bigquery-query-2-output_gcs_path}}\"},{\"name\":\"bigquery-query-output_gcs_path\",\"value\":\"{{tasks.bigquery-query.outputs.parameters.bigquery-query-output_gcs_path}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"region\",\"value\":\"{{inputs.parameters.region}}\"},{\"name\":\"retrieve-best-run-alpha\",\"value\":\"{{tasks.retrieve-best-run.outputs.parameters.retrieve-best-run-alpha}}\"},{\"name\":\"retrieve-best-run-max_iter\",\"value\":\"{{tasks.retrieve-best-run.outputs.parameters.retrieve-best-run-max_iter}}\"}]},\"dependencies\":[\"bigquery-query\",\"bigquery-query-2\",\"retrieve-best-run\"]}]}},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"model_id\"},{\"name\":\"project_id\"},{\"name\":\"replace_existing_version\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\"},{\"name\":\"version_id\"}]},\"outputs\":{\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\"},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine-model_name\",\"path\":\"/tmp/kfp/output/ml_engine/model_name.txt\"},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine-model_uri\",\"path\":\"/tmp/kfp/output/ml_engine/model_uri.txt\"},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine-version_name\",\"path\":\"/tmp/kfp/output/ml_engine/version_name.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to deploy a trained model from a Cloud Storage\\\\\\\\npath to a Cloud Machine Learning Engine service.\\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"Required. The Cloud Storage URI which contains a model file. Commonly used TF model search paths (export/exporter) will be used if they exist.\\\\\", \\\\\"name\\\\\": \\\\\"model_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"description\\\\\": \\\\\"Required.The ID of the parent project of the serving model.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The user-specified name of the model. If it is not provided, the operation uses a random name.\\\\\", \\\\\"name\\\\\": \\\\\"model_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The user-specified name of the version. If it is not provided, the operation uses a random name.\\\\\", \\\\\"name\\\\\": \\\\\"version_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The [Cloud ML Engine runtime version](https://cloud.google.com/ml-engine/docs/tensorflow/runtime-version-list) to use for this deployment. If it is not set, the Cloud ML Engine uses the default stable version, 1.0.\\\\\", \\\\\"name\\\\\": \\\\\"runtime_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The version of Python used in the prediction. If it is not set, the default version is `2.7`. Python `3.5` is available when the runtime_version is set to `1.4` and above. Python `2.7` works with all supported runtime versions.\\\\\", \\\\\"name\\\\\": \\\\\"python_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The JSON payload of the new [Model](https://cloud.google.com/ml-engine/reference/rest/v1/projects.models), if it does not exist.\\\\\", \\\\\"name\\\\\": \\\\\"model\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The JSON payload of the new [Version](https://cloud.google.com/ml-engine/reference/rest/v1/projects.models.versions).\\\\\", \\\\\"name\\\\\": \\\\\"version\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}, {\\\\\"default\\\\\": \\\\\"Fasle\\\\\", \\\\\"description\\\\\": \\\\\"A Boolean flag that indicates whether to replace existing version in case of conflict.\\\\\", \\\\\"name\\\\\": \\\\\"replace_existing_version\\\\\", \\\\\"type\\\\\": \\\\\"Bool\\\\\"}, {\\\\\"default\\\\\": \\\\\"False\\\\\", \\\\\"description\\\\\": \\\\\"A Boolean flag that indicates whether to set the new version as default version in the model.\\\\\", \\\\\"name\\\\\": \\\\\"set_default\\\\\", \\\\\"type\\\\\": \\\\\"Bool\\\\\"}, {\\\\\"default\\\\\": \\\\\"30\\\\\", \\\\\"description\\\\\": \\\\\"A time-interval to wait for in case the operation has a long run time.\\\\\", \\\\\"name\\\\\": \\\\\"wait_interval\\\\\", \\\\\"type\\\\\": \\\\\"Integer\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Deploying a trained model to Cloud Machine Learning Engine\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The Cloud Storage URI of the trained model.\\\\\", \\\\\"name\\\\\": \\\\\"model_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"description\\\\\": \\\\\"The name of the deployed model.\\\\\", \\\\\"name\\\\\": \\\\\"model_name\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The name of the deployed version.\\\\\", \\\\\"name\\\\\": \\\\\"version_name\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\"},\"labels\":{\"add-pod-env\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.ml_engine\",\"deploy\",\"--model_uri\",\"{{inputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--model_id\",\"{{inputs.parameters.model_id}}\",\"--version_id\",\"{{inputs.parameters.version_id}}\",\"--runtime_version\",\"1.15\",\"--python_version\",\"3.7\",\"--model\",\"\",\"--version\",\"\",\"--replace_existing_version\",\"{{inputs.parameters.replace_existing_version}}\",\"--set_default\",\"False\",\"--wait_interval\",\"30\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"evaluate-model\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"bigquery-query-3-output_gcs_path\"},{\"name\":\"evaluation_metric_name\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\"}]},\"outputs\":{\"parameters\":[{\"name\":\"evaluate-model-metric_value\",\"valueFrom\":{\"path\":\"/tmp/outputs/metric_value/data\"}}],\"artifacts\":[{\"name\":\"mlpipeline-metrics\",\"path\":\"/tmp/outputs/mlpipeline_metrics/data\"},{\"name\":\"evaluate-model-metric_name\",\"path\":\"/tmp/outputs/metric_name/data\"},{\"name\":\"evaluate-model-metric_value\",\"path\":\"/tmp/outputs/metric_value/data\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"Evaluates a trained sklearn model.\\\\\", \\\\\"inputs\\\\\": [{\\\\\"name\\\\\": \\\\\"dataset_path\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"model_path\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"metric_name\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}], \\\\\"name\\\\\": \\\\\"Evaluate model\\\\\", \\\\\"outputs\\\\\": [{\\\\\"name\\\\\": \\\\\"metric_name\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"metric_value\\\\\", \\\\\"type\\\\\": \\\\\"Float\\\\\"}, {\\\\\"name\\\\\": \\\\\"mlpipeline_metrics\\\\\", \\\\\"type\\\\\": \\\\\"Metrics\\\\\"}]}\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/dna-gcp-data/base_image:test\",\"command\":[\"python3\",\"-u\",\"-c\",\"def evaluate_model(\\\\n dataset_path , model_path , metric_name \\\\n): \\\\n \\\\n \\\\\"\\\\\"\\\\\"Evaluates a trained sklearn model.\\\\\"\\\\\"\\\\\"\\\\n #import joblib\\\\n import pickle\\\\n import json\\\\n import pandas as pd\\\\n import subprocess\\\\n import sys\\\\n\\\\n from sklearn.metrics import accuracy_score, recall_score\\\\n\\\\n df_test = pd.read_csv(dataset_path)\\\\n\\\\n X_test = df_test.drop(\\'Cover_Type\\', axis=1)\\\\n y_test = df_test[\\'Cover_Type\\']\\\\n\\\\n # Copy the model from GCS\\\\n model_filename = \\'model.pkl\\'\\\\n gcs_model_filepath = \\'{}/{}\\'.format(model_path, model_filename)\\\\n print(gcs_model_filepath)\\\\n subprocess.check_call([\\'gsutil\\', \\'cp\\', gcs_model_filepath, model_filename],\\\\n stderr=sys.stdout)\\\\n\\\\n with open(model_filename, \\'rb\\') as model_file:\\\\n model = pickle.load(model_file)\\\\n\\\\n y_hat = model.predict(X_test)\\\\n\\\\n if metric_name == \\'accuracy\\':\\\\n metric_value = accuracy_score(y_test, y_hat)\\\\n elif metric_name == \\'recall\\':\\\\n metric_value = recall_score(y_test, y_hat)\\\\n else:\\\\n metric_name = \\'N/A\\'\\\\n metric_value = 0\\\\n\\\\n # Export the metric\\\\n metrics = {\\\\n \\'metrics\\': [{\\\\n \\'name\\': metric_name,\\\\n \\'numberValue\\': float(metric_value)\\\\n }]\\\\n }\\\\n\\\\n return (metric_name, metric_value, json.dumps(metrics))\\\\n\\\\ndef _serialize_float(float_value: float) -\\\\u003e str:\\\\n if isinstance(float_value, str):\\\\n return float_value\\\\n if not isinstance(float_value, (float, int)):\\\\n raise TypeError(\\'Value \\\\\"{}\\\\\" has type \\\\\"{}\\\\\" instead of float.\\'.format(str(float_value), str(type(float_value))))\\\\n return str(float_value)\\\\n\\\\ndef _serialize_str(str_value: str) -\\\\u003e str:\\\\n if not isinstance(str_value, str):\\\\n raise TypeError(\\'Value \\\\\"{}\\\\\" has type \\\\\"{}\\\\\" instead of str.\\'.format(str(str_value), str(type(str_value))))\\\\n return str_value\\\\n\\\\nimport argparse\\\\n_parser = argparse.ArgumentParser(prog=\\'Evaluate model\\', description=\\'Evaluates a trained sklearn model.\\')\\\\n_parser.add_argument(\\\\\"--dataset-path\\\\\", dest=\\\\\"dataset_path\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"--model-path\\\\\", dest=\\\\\"model_path\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"--metric-name\\\\\", dest=\\\\\"metric_name\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"----output-paths\\\\\", dest=\\\\\"_output_paths\\\\\", type=str, nargs=3)\\\\n_parsed_args = vars(_parser.parse_args())\\\\n_output_files = _parsed_args.pop(\\\\\"_output_paths\\\\\", [])\\\\n\\\\n_outputs = evaluate_model(**_parsed_args)\\\\n\\\\nif not hasattr(_outputs, \\'__getitem__\\') or isinstance(_outputs, str):\\\\n _outputs = [_outputs]\\\\n\\\\n_output_serializers = [\\\\n _serialize_str,\\\\n _serialize_float,\\\\n str,\\\\n\\\\n]\\\\n\\\\nimport os\\\\nfor idx, output_file in enumerate(_output_files):\\\\n try:\\\\n os.makedirs(os.path.dirname(output_file))\\\\n except OSError:\\\\n pass\\\\n with open(output_file, \\'w\\') as f:\\\\n f.write(_output_serializers[idx](_outputs[idx]))\\\\n\"],\"args\":[\"--dataset-path\",\"{{inputs.parameters.bigquery-query-3-output_gcs_path}}\",\"--model-path\",\"{{inputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\",\"--metric-name\",\"{{inputs.parameters.evaluation_metric_name}}\",\"----output-paths\",\"/tmp/outputs/metric_name/data\",\"/tmp/outputs/metric_value/data\",\"/tmp/outputs/mlpipeline_metrics/data\"],\"resources\":{}}},{\"name\":\"retrieve-best-run\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"project_id\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\"}]},\"outputs\":{\"parameters\":[{\"name\":\"retrieve-best-run-alpha\",\"valueFrom\":{\"path\":\"/tmp/outputs/alpha/data\"}},{\"name\":\"retrieve-best-run-max_iter\",\"valueFrom\":{\"path\":\"/tmp/outputs/max_iter/data\"}}],\"artifacts\":[{\"name\":\"retrieve-best-run-alpha\",\"path\":\"/tmp/outputs/alpha/data\"},{\"name\":\"retrieve-best-run-max_iter\",\"path\":\"/tmp/outputs/max_iter/data\"},{\"name\":\"retrieve-best-run-metric_value\",\"path\":\"/tmp/outputs/metric_value/data\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"Retrieves the parameters of the best Hypertune run.\\\\\", \\\\\"inputs\\\\\": [{\\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"job_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}], \\\\\"name\\\\\": \\\\\"Retrieve best run\\\\\", \\\\\"outputs\\\\\": [{\\\\\"name\\\\\": \\\\\"metric_value\\\\\", \\\\\"type\\\\\": \\\\\"Float\\\\\"}, {\\\\\"name\\\\\": \\\\\"alpha\\\\\", \\\\\"type\\\\\": \\\\\"Float\\\\\"}, {\\\\\"name\\\\\": \\\\\"max_iter\\\\\", \\\\\"type\\\\\": \\\\\"Integer\\\\\"}]}\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/dna-gcp-data/base_image:test\",\"command\":[\"python3\",\"-u\",\"-c\",\"def retrieve_best_run(\\\\n project_id , job_id \\\\n): \\\\n \\\\n \\\\\"\\\\\"\\\\\"Retrieves the parameters of the best Hypertune run.\\\\\"\\\\\"\\\\\"\\\\n\\\\n from googleapiclient import discovery\\\\n from googleapiclient import errors\\\\n\\\\n ml = discovery.build(\\'ml\\', \\'v1\\')\\\\n\\\\n job_name = \\'projects/{}/jobs/{}\\'.format(project_id, job_id)\\\\n request = ml.projects().jobs().get(name=job_name)\\\\n\\\\n try:\\\\n response = request.execute()\\\\n except errors.HttpError as err:\\\\n print(err)\\\\n except:\\\\n print(\\'Unexpected error\\')\\\\n\\\\n print(response)\\\\n\\\\n best_trial = response[\\'trainingOutput\\'][\\'trials\\'][0]\\\\n\\\\n metric_value = best_trial[\\'finalMetric\\'][\\'objectiveValue\\']\\\\n alpha = float(best_trial[\\'hyperparameters\\'][\\'alpha\\'])\\\\n max_iter = int(best_trial[\\'hyperparameters\\'][\\'max_iter\\'])\\\\n\\\\n return (metric_value, alpha, max_iter)\\\\n\\\\ndef _serialize_float(float_value: float) -\\\\u003e str:\\\\n if isinstance(float_value, str):\\\\n return float_value\\\\n if not isinstance(float_value, (float, int)):\\\\n raise TypeError(\\'Value \\\\\"{}\\\\\" has type \\\\\"{}\\\\\" instead of float.\\'.format(str(float_value), str(type(float_value))))\\\\n return str(float_value)\\\\n\\\\ndef _serialize_int(int_value: int) -\\\\u003e str:\\\\n if isinstance(int_value, str):\\\\n return int_value\\\\n if not isinstance(int_value, int):\\\\n raise TypeError(\\'Value \\\\\"{}\\\\\" has type \\\\\"{}\\\\\" instead of int.\\'.format(str(int_value), str(type(int_value))))\\\\n return str(int_value)\\\\n\\\\nimport argparse\\\\n_parser = argparse.ArgumentParser(prog=\\'Retrieve best run\\', description=\\'Retrieves the parameters of the best Hypertune run.\\')\\\\n_parser.add_argument(\\\\\"--project-id\\\\\", dest=\\\\\"project_id\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"--job-id\\\\\", dest=\\\\\"job_id\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"----output-paths\\\\\", dest=\\\\\"_output_paths\\\\\", type=str, nargs=3)\\\\n_parsed_args = vars(_parser.parse_args())\\\\n_output_files = _parsed_args.pop(\\\\\"_output_paths\\\\\", [])\\\\n\\\\n_outputs = retrieve_best_run(**_parsed_args)\\\\n\\\\nif not hasattr(_outputs, \\'__getitem__\\') or isinstance(_outputs, str):\\\\n _outputs = [_outputs]\\\\n\\\\n_output_serializers = [\\\\n _serialize_float,\\\\n _serialize_float,\\\\n _serialize_int,\\\\n\\\\n]\\\\n\\\\nimport os\\\\nfor idx, output_file in enumerate(_output_files):\\\\n try:\\\\n os.makedirs(os.path.dirname(output_file))\\\\n except OSError:\\\\n pass\\\\n with open(output_file, \\'w\\') as f:\\\\n f.write(_output_serializers[idx](_outputs[idx]))\\\\n\"],\"args\":[\"--project-id\",\"{{inputs.parameters.project_id}}\",\"--job-id\",\"{{inputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id}}\",\"----output-paths\",\"/tmp/outputs/metric_value/data\",\"/tmp/outputs/alpha/data\",\"/tmp/outputs/max_iter/data\"],\"resources\":{}}},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\"},{\"name\":\"bigquery-query-output_gcs_path\"},{\"name\":\"gcs_root\"},{\"name\":\"hypertune_settings\"},{\"name\":\"project_id\"},{\"name\":\"region\"}]},\"outputs\":{\"parameters\":[{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/ml_engine/job_id.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_dir\",\"path\":\"/tmp/kfp/output/ml_engine/job_dir.txt\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\",\"path\":\"/tmp/kfp/output/ml_engine/job_id.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a Cloud Machine Learning (Cloud ML) \\\\\\\\nEngine training job as a step in a pipeline.\\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"Required. The ID of the parent project of the job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Python module name to run after installing the packages.\\\\\", \\\\\"name\\\\\": \\\\\"python_module\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Cloud Storage location of the packages (that contain the training program and any additional dependencies). The maximum number of package URIs is 100.\\\\\", \\\\\"name\\\\\": \\\\\"package_uris\\\\\", \\\\\"type\\\\\": \\\\\"List\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Compute Engine region in which the training job is run.\\\\\", \\\\\"name\\\\\": \\\\\"region\\\\\", \\\\\"type\\\\\": \\\\\"GCPRegion\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The command line arguments to pass to the program.\\\\\", \\\\\"name\\\\\": \\\\\"args\\\\\", \\\\\"type\\\\\": \\\\\"List\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"A Cloud Storage path in which to store the training outputs and other data needed for training. This path is passed to your TensorFlow program as the `job-dir` command-line argument. The benefit of specifying this field is that Cloud ML validates the path for use in training.\\\\\", \\\\\"name\\\\\": \\\\\"job_dir\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The version of Python used in training. If not set, the default version is `2.7`. Python `3.5` is available when runtimeVersion is set to `1.4` and above.\\\\\", \\\\\"name\\\\\": \\\\\"python_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Cloud ML Engine runtime version to use for training. If not set, Cloud ML Engine uses the default stable version, 1.0.\\\\\", \\\\\"name\\\\\": \\\\\"runtime_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Docker image to run on the master replica. This image must be in Container Registry.\\\\\", \\\\\"name\\\\\": \\\\\"master_image_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCRPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Docker image to run on the worker replica. This image must be in Container Registry.\\\\\", \\\\\"name\\\\\": \\\\\"worker_image_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCRPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The input parameters to create a training job. It is the JSON payload of a [TrainingInput](https://cloud.google.com/ml-engine/reference/rest/v1/projects.jobs#TrainingInput)\\\\\", \\\\\"name\\\\\": \\\\\"training_input\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The prefix of the generated job id.\\\\\", \\\\\"name\\\\\": \\\\\"job_id_prefix\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"30\\\\\", \\\\\"description\\\\\": \\\\\"Optional. A time-interval to wait for between calls to get the job status. Defaults to 30.\\'\\\\\", \\\\\"name\\\\\": \\\\\"wait_interval\\\\\", \\\\\"type\\\\\": \\\\\"Integer\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Submitting a Cloud ML training job as a pipeline step\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The ID of the created job.\\\\\", \\\\\"name\\\\\": \\\\\"job_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The output path in Cloud Storage of the trainning job, which contains the trained model files.\\\\\", \\\\\"name\\\\\": \\\\\"job_dir\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\"},\"labels\":{\"add-pod-env\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.ml_engine\",\"train\",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--python_module\",\"\",\"--package_uris\",\"\",\"--region\",\"{{inputs.parameters.region}}\",\"--args\",\"[\\\\\"--training_dataset_path\\\\\", \\\\\"{{inputs.parameters.bigquery-query-output_gcs_path}}\\\\\", \\\\\"--validation_dataset_path\\\\\", \\\\\"{{inputs.parameters.bigquery-query-2-output_gcs_path}}\\\\\", \\\\\"--hptune\\\\\", \\\\\"True\\\\\"]\",\"--job_dir\",\"{{inputs.parameters.gcs_root}}/jobdir/hypertune/{{workflow.uid}}\",\"--python_version\",\"\",\"--runtime_version\",\"\",\"--master_image_uri\",\"gcr.io/dna-gcp-data/trainer_image:test\",\"--worker_image_uri\",\"\",\"--training_input\",\"{{inputs.parameters.hypertune_settings}}\",\"--job_id_prefix\",\"\",\"--wait_interval\",\"30\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\"},{\"name\":\"bigquery-query-output_gcs_path\"},{\"name\":\"gcs_root\"},{\"name\":\"project_id\"},{\"name\":\"region\"},{\"name\":\"retrieve-best-run-alpha\"},{\"name\":\"retrieve-best-run-max_iter\"}]},\"outputs\":{\"parameters\":[{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/ml_engine/job_dir.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"path\":\"/tmp/kfp/output/ml_engine/job_dir.txt\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_id\",\"path\":\"/tmp/kfp/output/ml_engine/job_id.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a Cloud Machine Learning (Cloud ML) \\\\\\\\nEngine training job as a step in a pipeline.\\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"Required. The ID of the parent project of the job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Python module name to run after installing the packages.\\\\\", \\\\\"name\\\\\": \\\\\"python_module\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Cloud Storage location of the packages (that contain the training program and any additional dependencies). The maximum number of package URIs is 100.\\\\\", \\\\\"name\\\\\": \\\\\"package_uris\\\\\", \\\\\"type\\\\\": \\\\\"List\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Compute Engine region in which the training job is run.\\\\\", \\\\\"name\\\\\": \\\\\"region\\\\\", \\\\\"type\\\\\": \\\\\"GCPRegion\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The command line arguments to pass to the program.\\\\\", \\\\\"name\\\\\": \\\\\"args\\\\\", \\\\\"type\\\\\": \\\\\"List\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"A Cloud Storage path in which to store the training outputs and other data needed for training. This path is passed to your TensorFlow program as the `job-dir` command-line argument. The benefit of specifying this field is that Cloud ML validates the path for use in training.\\\\\", \\\\\"name\\\\\": \\\\\"job_dir\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The version of Python used in training. If not set, the default version is `2.7`. Python `3.5` is available when runtimeVersion is set to `1.4` and above.\\\\\", \\\\\"name\\\\\": \\\\\"python_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Cloud ML Engine runtime version to use for training. If not set, Cloud ML Engine uses the default stable version, 1.0.\\\\\", \\\\\"name\\\\\": \\\\\"runtime_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Docker image to run on the master replica. This image must be in Container Registry.\\\\\", \\\\\"name\\\\\": \\\\\"master_image_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCRPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Docker image to run on the worker replica. This image must be in Container Registry.\\\\\", \\\\\"name\\\\\": \\\\\"worker_image_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCRPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The input parameters to create a training job. It is the JSON payload of a [TrainingInput](https://cloud.google.com/ml-engine/reference/rest/v1/projects.jobs#TrainingInput)\\\\\", \\\\\"name\\\\\": \\\\\"training_input\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The prefix of the generated job id.\\\\\", \\\\\"name\\\\\": \\\\\"job_id_prefix\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"30\\\\\", \\\\\"description\\\\\": \\\\\"Optional. A time-interval to wait for between calls to get the job status. Defaults to 30.\\'\\\\\", \\\\\"name\\\\\": \\\\\"wait_interval\\\\\", \\\\\"type\\\\\": \\\\\"Integer\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Submitting a Cloud ML training job as a pipeline step\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The ID of the created job.\\\\\", \\\\\"name\\\\\": \\\\\"job_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The output path in Cloud Storage of the trainning job, which contains the trained model files.\\\\\", \\\\\"name\\\\\": \\\\\"job_dir\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\"},\"labels\":{\"add-pod-env\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.ml_engine\",\"train\",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--python_module\",\"\",\"--package_uris\",\"\",\"--region\",\"{{inputs.parameters.region}}\",\"--args\",\"[\\\\\"--training_dataset_path\\\\\", \\\\\"{{inputs.parameters.bigquery-query-output_gcs_path}}\\\\\", \\\\\"--validation_dataset_path\\\\\", \\\\\"{{inputs.parameters.bigquery-query-2-output_gcs_path}}\\\\\", \\\\\"--alpha\\\\\", \\\\\"{{inputs.parameters.retrieve-best-run-alpha}}\\\\\", \\\\\"--max_iter\\\\\", \\\\\"{{inputs.parameters.retrieve-best-run-max_iter}}\\\\\", \\\\\"--hptune\\\\\", \\\\\"False\\\\\"]\",\"--job_dir\",\"{{inputs.parameters.gcs_root}}/jobdir/{{workflow.uid}}\",\"--python_version\",\"\",\"--runtime_version\",\"\",\"--master_image_uri\",\"gcr.io/dna-gcp-data/trainer_image:test\",\"--worker_image_uri\",\"\",\"--training_input\",\"\",\"--job_id_prefix\",\"\",\"--wait_interval\",\"30\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}}],\"entrypoint\":\"covertype-classifier-training\",\"arguments\":{\"parameters\":[{\"name\":\"project_id\"},{\"name\":\"region\"},{\"name\":\"source_table_name\"},{\"name\":\"gcs_root\"},{\"name\":\"dataset_id\"},{\"name\":\"evaluation_metric_name\"},{\"name\":\"evaluation_metric_threshold\"},{\"name\":\"model_id\"},{\"name\":\"version_id\"},{\"name\":\"replace_existing_version\"},{\"name\":\"hypertune_settings\",\"value\":\"\\\\n{\\\\n \\\\\"hyperparameters\\\\\": {\\\\n \\\\\"goal\\\\\": \\\\\"MAXIMIZE\\\\\",\\\\n \\\\\"maxTrials\\\\\": 6,\\\\n \\\\\"maxParallelTrials\\\\\": 3,\\\\n \\\\\"hyperparameterMetricTag\\\\\": \\\\\"accuracy\\\\\",\\\\n \\\\\"enableTrialEarlyStopping\\\\\": True,\\\\n \\\\\"params\\\\\": [\\\\n {\\\\n \\\\\"parameterName\\\\\": \\\\\"max_iter\\\\\",\\\\n \\\\\"type\\\\\": \\\\\"DISCRETE\\\\\",\\\\n \\\\\"discreteValues\\\\\": [500, 1000]\\\\n },\\\\n {\\\\n \\\\\"parameterName\\\\\": \\\\\"alpha\\\\\",\\\\n \\\\\"type\\\\\": \\\\\"DOUBLE\\\\\",\\\\n \\\\\"minValue\\\\\": 0.0001,\\\\n \\\\\"maxValue\\\\\": 0.001,\\\\n \\\\\"scaleType\\\\\": \\\\\"UNIT_LINEAR_SCALE\\\\\"\\\\n }\\\\n ]\\\\n }\\\\n}\\\\n\"},{\"name\":\"dataset_location\",\"value\":\"US\"}]},\"serviceAccountName\":\"pipeline-runner\"},\"status\":{\"startedAt\":null,\"finishedAt\":null}}', 'pipeline_manifest': None, 'parameters': [{'name': 'project_id', 'value': 'dna-gcp-data'}, {'name': 'region', 'value': 'us-central1'}, {'name': 'source_table_name', 'value': 'covertype_dataset.covertype'}, {'name': 'gcs_root', 'value': 'gs://dna-gcp-data-kubeflowpipelines-default/staging'}, {'name': 'dataset_id', 'value': 'covertype_dataset'}, {'name': 'evaluation_metric_name', 'value': 'accuracy'}, {'name': 'evaluation_metric_threshold', 'value': '0.69'}, {'name': 'model_id', 'value': 'covertype_classifier'}, {'name': 'version_id', 'value': 'v01'}, {'name': 'replace_existing_version', 'value': 'True'}, {'name': 'hypertune_settings', 'value': '{\\n \"hyperparameters\": {\\n \"goal\": \"MAXIMIZE\",\\n \"maxTrials\": 6,\\n \"maxParallelTrials\": 3,\\n \"hyperparameterMetricTag\": \"accuracy\",\\n \"enableTrialEarlyStopping\": True,\\n \"params\": [\\n {\\n \"parameterName\": \"max_iter\",\\n \"type\": \"DISCRETE\",\\n \"discreteValues\": [500, 1000]\\n },\\n {\\n \"parameterName\": \"alpha\",\\n \"type\": \"DOUBLE\",\\n \"minValue\": 0.0001,\\n \"maxValue\": 0.001,\\n \"scaleType\": \"UNIT_LINEAR_SCALE\"\\n }\\n ]\\n }\\n}'}, {'name': 'dataset_location', 'value': 'US'}]}, 'resource_references': [{'key': {'type': 'EXPERIMENT', 'id': '889c1532-fee9-4b06-bc2b-10b1cd332c9a'}, 'name': 'Covertype_Classifier_Training', 'relationship': 'OWNER'}, {'key': {'type': 'PIPELINE_VERSION', 'id': 'ba22a9b4-0b4d-4a7c-beee-239c15f6ab1b'}, 'name': 'covertype_continuous_training_test', 'relationship': 'CREATOR'}], 'service_account': 'pipeline-runner', 'created_at': datetime.datetime(2021, 5, 21, 8, 46, 52, tzinfo=tzlocal()), 'scheduled_at': datetime.datetime(1970, 1, 1, 0, 0, tzinfo=tzlocal()), 'finished_at': datetime.datetime(2021, 5, 21, 9, 10, 48, tzinfo=tzlocal()), 'status': 'Succeeded', 'error': None, 'metrics': [{'name': 'accuracy', 'node_id': 'covertype-classifier-training-qmb6n-1971586539', 'number_value': 0.7225525168450257, 'format': None}]}, 'pipeline_runtime': {'pipeline_manifest': None, 'workflow_manifest': '{\"metadata\":{\"name\":\"covertype-classifier-training-qmb6n\",\"generateName\":\"covertype-classifier-training-\",\"namespace\":\"kfpdemo\",\"selfLink\":\"/apis/argoproj.io/v1alpha1/namespaces/kfpdemo/workflows/covertype-classifier-training-qmb6n\",\"uid\":\"03e3a16e-7930-4cd2-9bff-6f02c514a250\",\"resourceVersion\":\"10412508\",\"generation\":28,\"creationTimestamp\":\"2021-05-21T08:46:52Z\",\"labels\":{\"pipeline/runid\":\"d16368c5-24bb-4292-9060-315755f79b0b\",\"workflows.argoproj.io/completed\":\"true\",\"workflows.argoproj.io/phase\":\"Succeeded\"},\"annotations\":{\"pipelines.kubeflow.org/pipeline_spec\":\"{\\\\\"description\\\\\": \\\\\"The pipeline training and deploying the Covertype classifierpipeline_yaml\\\\\", \\\\\"inputs\\\\\": [{\\\\\"name\\\\\": \\\\\"project_id\\\\\"}, {\\\\\"name\\\\\": \\\\\"region\\\\\"}, {\\\\\"name\\\\\": \\\\\"source_table_name\\\\\"}, {\\\\\"name\\\\\": \\\\\"gcs_root\\\\\"}, {\\\\\"name\\\\\": \\\\\"dataset_id\\\\\"}, {\\\\\"name\\\\\": \\\\\"evaluation_metric_name\\\\\"}, {\\\\\"name\\\\\": \\\\\"evaluation_metric_threshold\\\\\"}, {\\\\\"name\\\\\": \\\\\"model_id\\\\\"}, {\\\\\"name\\\\\": \\\\\"version_id\\\\\"}, {\\\\\"name\\\\\": \\\\\"replace_existing_version\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\\\\n{\\\\\\\\n \\\\\\\\\\\\\"hyperparameters\\\\\\\\\\\\\": {\\\\\\\\n \\\\\\\\\\\\\"goal\\\\\\\\\\\\\": \\\\\\\\\\\\\"MAXIMIZE\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"maxTrials\\\\\\\\\\\\\": 6,\\\\\\\\n \\\\\\\\\\\\\"maxParallelTrials\\\\\\\\\\\\\": 3,\\\\\\\\n \\\\\\\\\\\\\"hyperparameterMetricTag\\\\\\\\\\\\\": \\\\\\\\\\\\\"accuracy\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"enableTrialEarlyStopping\\\\\\\\\\\\\": True,\\\\\\\\n \\\\\\\\\\\\\"params\\\\\\\\\\\\\": [\\\\\\\\n {\\\\\\\\n \\\\\\\\\\\\\"parameterName\\\\\\\\\\\\\": \\\\\\\\\\\\\"max_iter\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"type\\\\\\\\\\\\\": \\\\\\\\\\\\\"DISCRETE\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"discreteValues\\\\\\\\\\\\\": [500, 1000]\\\\\\\\n },\\\\\\\\n {\\\\\\\\n \\\\\\\\\\\\\"parameterName\\\\\\\\\\\\\": \\\\\\\\\\\\\"alpha\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"type\\\\\\\\\\\\\": \\\\\\\\\\\\\"DOUBLE\\\\\\\\\\\\\",\\\\\\\\n \\\\\\\\\\\\\"minValue\\\\\\\\\\\\\": 0.0001,\\\\\\\\n \\\\\\\\\\\\\"maxValue\\\\\\\\\\\\\": 0.001,\\\\\\\\n \\\\\\\\\\\\\"scaleType\\\\\\\\\\\\\": \\\\\\\\\\\\\"UNIT_LINEAR_SCALE\\\\\\\\\\\\\"\\\\\\\\n }\\\\\\\\n ]\\\\\\\\n }\\\\\\\\n}\\\\\\\\n\\\\\", \\\\\"name\\\\\": \\\\\"hypertune_settings\\\\\", \\\\\"optional\\\\\": true}, {\\\\\"default\\\\\": \\\\\"US\\\\\", \\\\\"name\\\\\": \\\\\"dataset_location\\\\\", \\\\\"optional\\\\\": true}], \\\\\"name\\\\\": \\\\\"Covertype Classifier Training\\\\\"}\",\"pipelines.kubeflow.org/run_name\":\"Run_001\"},\"managedFields\":[{\"manager\":\"apiserver\",\"operation\":\"Update\",\"apiVersion\":\"argoproj.io/v1alpha1\",\"time\":\"2021-05-21T08:46:52Z\",\"fieldsType\":\"FieldsV1\",\"fieldsV1\":{\"f:metadata\":{\"f:annotations\":{\".\":{},\"f:pipelines.kubeflow.org/pipeline_spec\":{},\"f:pipelines.kubeflow.org/run_name\":{}},\"f:generateName\":{},\"f:labels\":{\".\":{},\"f:pipeline/runid\":{}}},\"f:spec\":{\".\":{},\"f:arguments\":{\".\":{},\"f:parameters\":{}},\"f:entrypoint\":{},\"f:serviceAccountName\":{},\"f:templates\":{}},\"f:status\":{}}},{\"manager\":\"workflow-controller\",\"operation\":\"Update\",\"apiVersion\":\"argoproj.io/v1alpha1\",\"time\":\"2021-05-21T09:10:48Z\",\"fieldsType\":\"FieldsV1\",\"fieldsV1\":{\"f:metadata\":{\"f:labels\":{\"f:workflows.argoproj.io/completed\":{},\"f:workflows.argoproj.io/phase\":{}}},\"f:status\":{\"f:conditions\":{},\"f:finishedAt\":{},\"f:nodes\":{\".\":{},\"f:covertype-classifier-training-qmb6n\":{\".\":{},\"f:children\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outboundNodes\":{},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}},\"f:covertype-classifier-training-qmb6n-1057815578\":{\".\":{},\"f:boundaryID\":{},\"f:children\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outputs\":{\".\":{},\"f:artifacts\":{},\"f:parameters\":{}},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}},\"f:covertype-classifier-training-qmb6n-1075982665\":{\".\":{},\"f:boundaryID\":{},\"f:children\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outputs\":{\".\":{},\"f:artifacts\":{},\"f:parameters\":{}},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}},\"f:covertype-classifier-training-qmb6n-1089717477\":{\".\":{},\"f:boundaryID\":{},\"f:children\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outputs\":{\".\":{},\"f:artifacts\":{},\"f:parameters\":{}},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}},\"f:covertype-classifier-training-qmb6n-1858401567\":{\".\":{},\"f:boundaryID\":{},\"f:children\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outputs\":{\".\":{},\"f:artifacts\":{},\"f:parameters\":{}},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}},\"f:covertype-classifier-training-qmb6n-1879447244\":{\".\":{},\"f:boundaryID\":{},\"f:children\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outputs\":{\".\":{},\"f:artifacts\":{},\"f:parameters\":{}},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}},\"f:covertype-classifier-training-qmb6n-1896224863\":{\".\":{},\"f:boundaryID\":{},\"f:children\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outputs\":{\".\":{},\"f:artifacts\":{},\"f:parameters\":{}},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}},\"f:covertype-classifier-training-qmb6n-1971586539\":{\".\":{},\"f:boundaryID\":{},\"f:children\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outputs\":{\".\":{},\"f:artifacts\":{},\"f:parameters\":{}},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}},\"f:covertype-classifier-training-qmb6n-2274394652\":{\".\":{},\"f:boundaryID\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outputs\":{\".\":{},\"f:artifacts\":{}},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}},\"f:covertype-classifier-training-qmb6n-2585557679\":{\".\":{},\"f:boundaryID\":{},\"f:children\":{},\"f:displayName\":{},\"f:finishedAt\":{},\"f:id\":{},\"f:inputs\":{\".\":{},\"f:parameters\":{}},\"f:name\":{},\"f:outboundNodes\":{},\"f:phase\":{},\"f:startedAt\":{},\"f:templateName\":{},\"f:type\":{}}},\"f:phase\":{},\"f:startedAt\":{}}}}]},\"spec\":{\"templates\":[{\"name\":\"bigquery-query\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\"},{\"name\":\"dataset_location\"},{\"name\":\"gcs_root\"},{\"name\":\"project_id\"},{\"name\":\"source_table_name\"}]},\"outputs\":{\"parameters\":[{\"name\":\"bigquery-query-output_gcs_path\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"bigquery-query-output_gcs_path\",\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a query to Google Cloud Bigquery \\\\\\\\nservice and dump outputs to a Google Cloud Storage blob. \\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The query used by Bigquery service to fetch the results.\\\\\", \\\\\"name\\\\\": \\\\\"query\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The project to execute the query job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the persistent dataset to keep the results of the query.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the table to keep the results of the query. If absent, the operation will generate a random id for the table.\\\\\", \\\\\"name\\\\\": \\\\\"table_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket to store the query output.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"US\\\\\", \\\\\"description\\\\\": \\\\\"The location to create the dataset. Defaults to `US`.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_location\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The full config spec for the query job.See [QueryJobConfig](https://googleapis.github.io/google-cloud-python/latest/bigquery/generated/google.cloud.bigquery.job.QueryJobConfig.html#google.cloud.bigquery.job.QueryJobConfig) for details.\\\\\", \\\\\"name\\\\\": \\\\\"job_config\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Bigquery - Query\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket containing the query output in CSV format.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\",\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"add-pod-env\":\"true\",\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.bigquery\",\"query\",\"--query\",\"\\\\n SELECT *\\\\n FROM \\\\n `{{inputs.parameters.source_table_name}}` AS cover\\\\n WHERE \\\\n MOD(ABS(FARM_FINGERPRINT(TO_JSON_STRING(cover))), 10) IN (1, 2, 3, 4)\\\\n \",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--dataset_id\",\"{{inputs.parameters.dataset_id}}\",\"--table_id\",\"\",\"--dataset_location\",\"{{inputs.parameters.dataset_location}}\",\"--output_gcs_path\",\"{{inputs.parameters.gcs_root}}/datasets/training/data.csv\",\"--job_config\",\"\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"bigquery-query-2\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\"},{\"name\":\"dataset_location\"},{\"name\":\"gcs_root\"},{\"name\":\"project_id\"},{\"name\":\"source_table_name\"}]},\"outputs\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"bigquery-query-2-output_gcs_path\",\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a query to Google Cloud Bigquery \\\\\\\\nservice and dump outputs to a Google Cloud Storage blob. \\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The query used by Bigquery service to fetch the results.\\\\\", \\\\\"name\\\\\": \\\\\"query\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The project to execute the query job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the persistent dataset to keep the results of the query.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the table to keep the results of the query. If absent, the operation will generate a random id for the table.\\\\\", \\\\\"name\\\\\": \\\\\"table_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket to store the query output.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"US\\\\\", \\\\\"description\\\\\": \\\\\"The location to create the dataset. Defaults to `US`.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_location\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The full config spec for the query job.See [QueryJobConfig](https://googleapis.github.io/google-cloud-python/latest/bigquery/generated/google.cloud.bigquery.job.QueryJobConfig.html#google.cloud.bigquery.job.QueryJobConfig) for details.\\\\\", \\\\\"name\\\\\": \\\\\"job_config\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Bigquery - Query\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket containing the query output in CSV format.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\",\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"add-pod-env\":\"true\",\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.bigquery\",\"query\",\"--query\",\"\\\\n SELECT *\\\\n FROM \\\\n `{{inputs.parameters.source_table_name}}` AS cover\\\\n WHERE \\\\n MOD(ABS(FARM_FINGERPRINT(TO_JSON_STRING(cover))), 10) IN (8)\\\\n \",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--dataset_id\",\"{{inputs.parameters.dataset_id}}\",\"--table_id\",\"\",\"--dataset_location\",\"{{inputs.parameters.dataset_location}}\",\"--output_gcs_path\",\"{{inputs.parameters.gcs_root}}/datasets/validation/data.csv\",\"--job_config\",\"\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"bigquery-query-3\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\"},{\"name\":\"dataset_location\"},{\"name\":\"gcs_root\"},{\"name\":\"project_id\"},{\"name\":\"source_table_name\"}]},\"outputs\":{\"parameters\":[{\"name\":\"bigquery-query-3-output_gcs_path\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"bigquery-query-3-output_gcs_path\",\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a query to Google Cloud Bigquery \\\\\\\\nservice and dump outputs to a Google Cloud Storage blob. \\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The query used by Bigquery service to fetch the results.\\\\\", \\\\\"name\\\\\": \\\\\"query\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The project to execute the query job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the persistent dataset to keep the results of the query.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The ID of the table to keep the results of the query. If absent, the operation will generate a random id for the table.\\\\\", \\\\\"name\\\\\": \\\\\"table_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket to store the query output.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"US\\\\\", \\\\\"description\\\\\": \\\\\"The location to create the dataset. Defaults to `US`.\\\\\", \\\\\"name\\\\\": \\\\\"dataset_location\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The full config spec for the query job.See [QueryJobConfig](https://googleapis.github.io/google-cloud-python/latest/bigquery/generated/google.cloud.bigquery.job.QueryJobConfig.html#google.cloud.bigquery.job.QueryJobConfig) for details.\\\\\", \\\\\"name\\\\\": \\\\\"job_config\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Bigquery - Query\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The path to the Cloud Storage bucket containing the query output in CSV format.\\\\\", \\\\\"name\\\\\": \\\\\"output_gcs_path\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\",\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"add-pod-env\":\"true\",\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.bigquery\",\"query\",\"--query\",\"\\\\n SELECT *\\\\n FROM \\\\n `{{inputs.parameters.source_table_name}}` AS cover\\\\n WHERE \\\\n MOD(ABS(FARM_FINGERPRINT(TO_JSON_STRING(cover))), 10) IN (9)\\\\n \",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--dataset_id\",\"{{inputs.parameters.dataset_id}}\",\"--table_id\",\"\",\"--dataset_location\",\"{{inputs.parameters.dataset_location}}\",\"--output_gcs_path\",\"{{inputs.parameters.gcs_root}}/datasets/testing/data.csv\",\"--job_config\",\"\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"condition-1\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"model_id\"},{\"name\":\"project_id\"},{\"name\":\"replace_existing_version\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\"},{\"name\":\"version_id\"}]},\"outputs\":{},\"metadata\":{\"annotations\":{\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"dag\":{\"tasks\":[{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine\",\"template\":\"deploying-a-trained-model-to-cloud-machine-learning-engine\",\"arguments\":{\"parameters\":[{\"name\":\"model_id\",\"value\":\"{{inputs.parameters.model_id}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"replace_existing_version\",\"value\":\"{{inputs.parameters.replace_existing_version}}\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"{{inputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\"},{\"name\":\"version_id\",\"value\":\"{{inputs.parameters.version_id}}\"}]}}]}},{\"name\":\"covertype-classifier-training\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\"},{\"name\":\"dataset_location\"},{\"name\":\"evaluation_metric_name\"},{\"name\":\"evaluation_metric_threshold\"},{\"name\":\"gcs_root\"},{\"name\":\"hypertune_settings\"},{\"name\":\"model_id\"},{\"name\":\"project_id\"},{\"name\":\"region\"},{\"name\":\"replace_existing_version\"},{\"name\":\"source_table_name\"},{\"name\":\"version_id\"}]},\"outputs\":{},\"metadata\":{\"annotations\":{\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"dag\":{\"tasks\":[{\"name\":\"bigquery-query\",\"template\":\"bigquery-query\",\"arguments\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"{{inputs.parameters.dataset_id}}\"},{\"name\":\"dataset_location\",\"value\":\"{{inputs.parameters.dataset_location}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"source_table_name\",\"value\":\"{{inputs.parameters.source_table_name}}\"}]}},{\"name\":\"bigquery-query-2\",\"template\":\"bigquery-query-2\",\"arguments\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"{{inputs.parameters.dataset_id}}\"},{\"name\":\"dataset_location\",\"value\":\"{{inputs.parameters.dataset_location}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"source_table_name\",\"value\":\"{{inputs.parameters.source_table_name}}\"}]}},{\"name\":\"bigquery-query-3\",\"template\":\"bigquery-query-3\",\"arguments\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"{{inputs.parameters.dataset_id}}\"},{\"name\":\"dataset_location\",\"value\":\"{{inputs.parameters.dataset_location}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"source_table_name\",\"value\":\"{{inputs.parameters.source_table_name}}\"}]}},{\"name\":\"condition-1\",\"template\":\"condition-1\",\"arguments\":{\"parameters\":[{\"name\":\"model_id\",\"value\":\"{{inputs.parameters.model_id}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"replace_existing_version\",\"value\":\"{{inputs.parameters.replace_existing_version}}\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"{{tasks.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2.outputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\"},{\"name\":\"version_id\",\"value\":\"{{inputs.parameters.version_id}}\"}]},\"dependencies\":[\"evaluate-model\",\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\"],\"when\":\"{{tasks.evaluate-model.outputs.parameters.evaluate-model-metric_value}} \\\\u003e {{inputs.parameters.evaluation_metric_threshold}}\"},{\"name\":\"evaluate-model\",\"template\":\"evaluate-model\",\"arguments\":{\"parameters\":[{\"name\":\"bigquery-query-3-output_gcs_path\",\"value\":\"{{tasks.bigquery-query-3.outputs.parameters.bigquery-query-3-output_gcs_path}}\"},{\"name\":\"evaluation_metric_name\",\"value\":\"{{inputs.parameters.evaluation_metric_name}}\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"{{tasks.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2.outputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\"}]},\"dependencies\":[\"bigquery-query-3\",\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\"]},{\"name\":\"retrieve-best-run\",\"template\":\"retrieve-best-run\",\"arguments\":{\"parameters\":[{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\",\"value\":\"{{tasks.submitting-a-cloud-ml-training-job-as-a-pipeline-step.outputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id}}\"}]},\"dependencies\":[\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\"]},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\",\"template\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\",\"arguments\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\",\"value\":\"{{tasks.bigquery-query-2.outputs.parameters.bigquery-query-2-output_gcs_path}}\"},{\"name\":\"bigquery-query-output_gcs_path\",\"value\":\"{{tasks.bigquery-query.outputs.parameters.bigquery-query-output_gcs_path}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"hypertune_settings\",\"value\":\"{{inputs.parameters.hypertune_settings}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"region\",\"value\":\"{{inputs.parameters.region}}\"}]},\"dependencies\":[\"bigquery-query\",\"bigquery-query-2\"]},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\",\"template\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\",\"arguments\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\",\"value\":\"{{tasks.bigquery-query-2.outputs.parameters.bigquery-query-2-output_gcs_path}}\"},{\"name\":\"bigquery-query-output_gcs_path\",\"value\":\"{{tasks.bigquery-query.outputs.parameters.bigquery-query-output_gcs_path}}\"},{\"name\":\"gcs_root\",\"value\":\"{{inputs.parameters.gcs_root}}\"},{\"name\":\"project_id\",\"value\":\"{{inputs.parameters.project_id}}\"},{\"name\":\"region\",\"value\":\"{{inputs.parameters.region}}\"},{\"name\":\"retrieve-best-run-alpha\",\"value\":\"{{tasks.retrieve-best-run.outputs.parameters.retrieve-best-run-alpha}}\"},{\"name\":\"retrieve-best-run-max_iter\",\"value\":\"{{tasks.retrieve-best-run.outputs.parameters.retrieve-best-run-max_iter}}\"}]},\"dependencies\":[\"bigquery-query\",\"bigquery-query-2\",\"retrieve-best-run\"]}]}},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"model_id\"},{\"name\":\"project_id\"},{\"name\":\"replace_existing_version\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\"},{\"name\":\"version_id\"}]},\"outputs\":{\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine-model_name\",\"path\":\"/tmp/kfp/output/ml_engine/model_name.txt\"},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine-model_uri\",\"path\":\"/tmp/kfp/output/ml_engine/model_uri.txt\"},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine-version_name\",\"path\":\"/tmp/kfp/output/ml_engine/version_name.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to deploy a trained model from a Cloud Storage\\\\\\\\npath to a Cloud Machine Learning Engine service.\\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"Required. The Cloud Storage URI which contains a model file. Commonly used TF model search paths (export/exporter) will be used if they exist.\\\\\", \\\\\"name\\\\\": \\\\\"model_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"description\\\\\": \\\\\"Required.The ID of the parent project of the serving model.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The user-specified name of the model. If it is not provided, the operation uses a random name.\\\\\", \\\\\"name\\\\\": \\\\\"model_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The user-specified name of the version. If it is not provided, the operation uses a random name.\\\\\", \\\\\"name\\\\\": \\\\\"version_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The [Cloud ML Engine runtime version](https://cloud.google.com/ml-engine/docs/tensorflow/runtime-version-list) to use for this deployment. If it is not set, the Cloud ML Engine uses the default stable version, 1.0.\\\\\", \\\\\"name\\\\\": \\\\\"runtime_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The version of Python used in the prediction. If it is not set, the default version is `2.7`. Python `3.5` is available when the runtime_version is set to `1.4` and above. Python `2.7` works with all supported runtime versions.\\\\\", \\\\\"name\\\\\": \\\\\"python_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The JSON payload of the new [Model](https://cloud.google.com/ml-engine/reference/rest/v1/projects.models), if it does not exist.\\\\\", \\\\\"name\\\\\": \\\\\"model\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"Optional. The JSON payload of the new [Version](https://cloud.google.com/ml-engine/reference/rest/v1/projects.models.versions).\\\\\", \\\\\"name\\\\\": \\\\\"version\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}, {\\\\\"default\\\\\": \\\\\"Fasle\\\\\", \\\\\"description\\\\\": \\\\\"A Boolean flag that indicates whether to replace existing version in case of conflict.\\\\\", \\\\\"name\\\\\": \\\\\"replace_existing_version\\\\\", \\\\\"type\\\\\": \\\\\"Bool\\\\\"}, {\\\\\"default\\\\\": \\\\\"False\\\\\", \\\\\"description\\\\\": \\\\\"A Boolean flag that indicates whether to set the new version as default version in the model.\\\\\", \\\\\"name\\\\\": \\\\\"set_default\\\\\", \\\\\"type\\\\\": \\\\\"Bool\\\\\"}, {\\\\\"default\\\\\": \\\\\"30\\\\\", \\\\\"description\\\\\": \\\\\"A time-interval to wait for in case the operation has a long run time.\\\\\", \\\\\"name\\\\\": \\\\\"wait_interval\\\\\", \\\\\"type\\\\\": \\\\\"Integer\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Deploying a trained model to Cloud Machine Learning Engine\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The Cloud Storage URI of the trained model.\\\\\", \\\\\"name\\\\\": \\\\\"model_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"description\\\\\": \\\\\"The name of the deployed model.\\\\\", \\\\\"name\\\\\": \\\\\"model_name\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The name of the deployed version.\\\\\", \\\\\"name\\\\\": \\\\\"version_name\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\",\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"add-pod-env\":\"true\",\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.ml_engine\",\"deploy\",\"--model_uri\",\"{{inputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--model_id\",\"{{inputs.parameters.model_id}}\",\"--version_id\",\"{{inputs.parameters.version_id}}\",\"--runtime_version\",\"1.15\",\"--python_version\",\"3.7\",\"--model\",\"\",\"--version\",\"\",\"--replace_existing_version\",\"{{inputs.parameters.replace_existing_version}}\",\"--set_default\",\"False\",\"--wait_interval\",\"30\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"evaluate-model\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"bigquery-query-3-output_gcs_path\"},{\"name\":\"evaluation_metric_name\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\"}]},\"outputs\":{\"parameters\":[{\"name\":\"evaluate-model-metric_value\",\"valueFrom\":{\"path\":\"/tmp/outputs/metric_value/data\"}}],\"artifacts\":[{\"name\":\"mlpipeline-metrics\",\"path\":\"/tmp/outputs/mlpipeline_metrics/data\",\"optional\":true},{\"name\":\"evaluate-model-metric_name\",\"path\":\"/tmp/outputs/metric_name/data\"},{\"name\":\"evaluate-model-metric_value\",\"path\":\"/tmp/outputs/metric_value/data\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"Evaluates a trained sklearn model.\\\\\", \\\\\"inputs\\\\\": [{\\\\\"name\\\\\": \\\\\"dataset_path\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"model_path\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"metric_name\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}], \\\\\"name\\\\\": \\\\\"Evaluate model\\\\\", \\\\\"outputs\\\\\": [{\\\\\"name\\\\\": \\\\\"metric_name\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"metric_value\\\\\", \\\\\"type\\\\\": \\\\\"Float\\\\\"}, {\\\\\"name\\\\\": \\\\\"mlpipeline_metrics\\\\\", \\\\\"type\\\\\": \\\\\"Metrics\\\\\"}]}\",\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/dna-gcp-data/base_image:test\",\"command\":[\"python3\",\"-u\",\"-c\",\"def evaluate_model(\\\\n dataset_path , model_path , metric_name \\\\n): \\\\n \\\\n \\\\\"\\\\\"\\\\\"Evaluates a trained sklearn model.\\\\\"\\\\\"\\\\\"\\\\n #import joblib\\\\n import pickle\\\\n import json\\\\n import pandas as pd\\\\n import subprocess\\\\n import sys\\\\n\\\\n from sklearn.metrics import accuracy_score, recall_score\\\\n\\\\n df_test = pd.read_csv(dataset_path)\\\\n\\\\n X_test = df_test.drop(\\'Cover_Type\\', axis=1)\\\\n y_test = df_test[\\'Cover_Type\\']\\\\n\\\\n # Copy the model from GCS\\\\n model_filename = \\'model.pkl\\'\\\\n gcs_model_filepath = \\'{}/{}\\'.format(model_path, model_filename)\\\\n print(gcs_model_filepath)\\\\n subprocess.check_call([\\'gsutil\\', \\'cp\\', gcs_model_filepath, model_filename],\\\\n stderr=sys.stdout)\\\\n\\\\n with open(model_filename, \\'rb\\') as model_file:\\\\n model = pickle.load(model_file)\\\\n\\\\n y_hat = model.predict(X_test)\\\\n\\\\n if metric_name == \\'accuracy\\':\\\\n metric_value = accuracy_score(y_test, y_hat)\\\\n elif metric_name == \\'recall\\':\\\\n metric_value = recall_score(y_test, y_hat)\\\\n else:\\\\n metric_name = \\'N/A\\'\\\\n metric_value = 0\\\\n\\\\n # Export the metric\\\\n metrics = {\\\\n \\'metrics\\': [{\\\\n \\'name\\': metric_name,\\\\n \\'numberValue\\': float(metric_value)\\\\n }]\\\\n }\\\\n\\\\n return (metric_name, metric_value, json.dumps(metrics))\\\\n\\\\ndef _serialize_float(float_value: float) -\\\\u003e str:\\\\n if isinstance(float_value, str):\\\\n return float_value\\\\n if not isinstance(float_value, (float, int)):\\\\n raise TypeError(\\'Value \\\\\"{}\\\\\" has type \\\\\"{}\\\\\" instead of float.\\'.format(str(float_value), str(type(float_value))))\\\\n return str(float_value)\\\\n\\\\ndef _serialize_str(str_value: str) -\\\\u003e str:\\\\n if not isinstance(str_value, str):\\\\n raise TypeError(\\'Value \\\\\"{}\\\\\" has type \\\\\"{}\\\\\" instead of str.\\'.format(str(str_value), str(type(str_value))))\\\\n return str_value\\\\n\\\\nimport argparse\\\\n_parser = argparse.ArgumentParser(prog=\\'Evaluate model\\', description=\\'Evaluates a trained sklearn model.\\')\\\\n_parser.add_argument(\\\\\"--dataset-path\\\\\", dest=\\\\\"dataset_path\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"--model-path\\\\\", dest=\\\\\"model_path\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"--metric-name\\\\\", dest=\\\\\"metric_name\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"----output-paths\\\\\", dest=\\\\\"_output_paths\\\\\", type=str, nargs=3)\\\\n_parsed_args = vars(_parser.parse_args())\\\\n_output_files = _parsed_args.pop(\\\\\"_output_paths\\\\\", [])\\\\n\\\\n_outputs = evaluate_model(**_parsed_args)\\\\n\\\\nif not hasattr(_outputs, \\'__getitem__\\') or isinstance(_outputs, str):\\\\n _outputs = [_outputs]\\\\n\\\\n_output_serializers = [\\\\n _serialize_str,\\\\n _serialize_float,\\\\n str,\\\\n\\\\n]\\\\n\\\\nimport os\\\\nfor idx, output_file in enumerate(_output_files):\\\\n try:\\\\n os.makedirs(os.path.dirname(output_file))\\\\n except OSError:\\\\n pass\\\\n with open(output_file, \\'w\\') as f:\\\\n f.write(_output_serializers[idx](_outputs[idx]))\\\\n\"],\"args\":[\"--dataset-path\",\"{{inputs.parameters.bigquery-query-3-output_gcs_path}}\",\"--model-path\",\"{{inputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir}}\",\"--metric-name\",\"{{inputs.parameters.evaluation_metric_name}}\",\"----output-paths\",\"/tmp/outputs/metric_name/data\",\"/tmp/outputs/metric_value/data\",\"/tmp/outputs/mlpipeline_metrics/data\"],\"resources\":{}}},{\"name\":\"retrieve-best-run\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"project_id\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\"}]},\"outputs\":{\"parameters\":[{\"name\":\"retrieve-best-run-alpha\",\"valueFrom\":{\"path\":\"/tmp/outputs/alpha/data\"}},{\"name\":\"retrieve-best-run-max_iter\",\"valueFrom\":{\"path\":\"/tmp/outputs/max_iter/data\"}}],\"artifacts\":[{\"name\":\"retrieve-best-run-alpha\",\"path\":\"/tmp/outputs/alpha/data\"},{\"name\":\"retrieve-best-run-max_iter\",\"path\":\"/tmp/outputs/max_iter/data\"},{\"name\":\"retrieve-best-run-metric_value\",\"path\":\"/tmp/outputs/metric_value/data\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"Retrieves the parameters of the best Hypertune run.\\\\\", \\\\\"inputs\\\\\": [{\\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"name\\\\\": \\\\\"job_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}], \\\\\"name\\\\\": \\\\\"Retrieve best run\\\\\", \\\\\"outputs\\\\\": [{\\\\\"name\\\\\": \\\\\"metric_value\\\\\", \\\\\"type\\\\\": \\\\\"Float\\\\\"}, {\\\\\"name\\\\\": \\\\\"alpha\\\\\", \\\\\"type\\\\\": \\\\\"Float\\\\\"}, {\\\\\"name\\\\\": \\\\\"max_iter\\\\\", \\\\\"type\\\\\": \\\\\"Integer\\\\\"}]}\",\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/dna-gcp-data/base_image:test\",\"command\":[\"python3\",\"-u\",\"-c\",\"def retrieve_best_run(\\\\n project_id , job_id \\\\n): \\\\n \\\\n \\\\\"\\\\\"\\\\\"Retrieves the parameters of the best Hypertune run.\\\\\"\\\\\"\\\\\"\\\\n\\\\n from googleapiclient import discovery\\\\n from googleapiclient import errors\\\\n\\\\n ml = discovery.build(\\'ml\\', \\'v1\\')\\\\n\\\\n job_name = \\'projects/{}/jobs/{}\\'.format(project_id, job_id)\\\\n request = ml.projects().jobs().get(name=job_name)\\\\n\\\\n try:\\\\n response = request.execute()\\\\n except errors.HttpError as err:\\\\n print(err)\\\\n except:\\\\n print(\\'Unexpected error\\')\\\\n\\\\n print(response)\\\\n\\\\n best_trial = response[\\'trainingOutput\\'][\\'trials\\'][0]\\\\n\\\\n metric_value = best_trial[\\'finalMetric\\'][\\'objectiveValue\\']\\\\n alpha = float(best_trial[\\'hyperparameters\\'][\\'alpha\\'])\\\\n max_iter = int(best_trial[\\'hyperparameters\\'][\\'max_iter\\'])\\\\n\\\\n return (metric_value, alpha, max_iter)\\\\n\\\\ndef _serialize_float(float_value: float) -\\\\u003e str:\\\\n if isinstance(float_value, str):\\\\n return float_value\\\\n if not isinstance(float_value, (float, int)):\\\\n raise TypeError(\\'Value \\\\\"{}\\\\\" has type \\\\\"{}\\\\\" instead of float.\\'.format(str(float_value), str(type(float_value))))\\\\n return str(float_value)\\\\n\\\\ndef _serialize_int(int_value: int) -\\\\u003e str:\\\\n if isinstance(int_value, str):\\\\n return int_value\\\\n if not isinstance(int_value, int):\\\\n raise TypeError(\\'Value \\\\\"{}\\\\\" has type \\\\\"{}\\\\\" instead of int.\\'.format(str(int_value), str(type(int_value))))\\\\n return str(int_value)\\\\n\\\\nimport argparse\\\\n_parser = argparse.ArgumentParser(prog=\\'Retrieve best run\\', description=\\'Retrieves the parameters of the best Hypertune run.\\')\\\\n_parser.add_argument(\\\\\"--project-id\\\\\", dest=\\\\\"project_id\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"--job-id\\\\\", dest=\\\\\"job_id\\\\\", type=str, required=True, default=argparse.SUPPRESS)\\\\n_parser.add_argument(\\\\\"----output-paths\\\\\", dest=\\\\\"_output_paths\\\\\", type=str, nargs=3)\\\\n_parsed_args = vars(_parser.parse_args())\\\\n_output_files = _parsed_args.pop(\\\\\"_output_paths\\\\\", [])\\\\n\\\\n_outputs = retrieve_best_run(**_parsed_args)\\\\n\\\\nif not hasattr(_outputs, \\'__getitem__\\') or isinstance(_outputs, str):\\\\n _outputs = [_outputs]\\\\n\\\\n_output_serializers = [\\\\n _serialize_float,\\\\n _serialize_float,\\\\n _serialize_int,\\\\n\\\\n]\\\\n\\\\nimport os\\\\nfor idx, output_file in enumerate(_output_files):\\\\n try:\\\\n os.makedirs(os.path.dirname(output_file))\\\\n except OSError:\\\\n pass\\\\n with open(output_file, \\'w\\') as f:\\\\n f.write(_output_serializers[idx](_outputs[idx]))\\\\n\"],\"args\":[\"--project-id\",\"{{inputs.parameters.project_id}}\",\"--job-id\",\"{{inputs.parameters.submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id}}\",\"----output-paths\",\"/tmp/outputs/metric_value/data\",\"/tmp/outputs/alpha/data\",\"/tmp/outputs/max_iter/data\"],\"resources\":{}}},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\"},{\"name\":\"bigquery-query-output_gcs_path\"},{\"name\":\"gcs_root\"},{\"name\":\"hypertune_settings\"},{\"name\":\"project_id\"},{\"name\":\"region\"}]},\"outputs\":{\"parameters\":[{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/ml_engine/job_id.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_dir\",\"path\":\"/tmp/kfp/output/ml_engine/job_dir.txt\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\",\"path\":\"/tmp/kfp/output/ml_engine/job_id.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a Cloud Machine Learning (Cloud ML) \\\\\\\\nEngine training job as a step in a pipeline.\\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"Required. The ID of the parent project of the job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Python module name to run after installing the packages.\\\\\", \\\\\"name\\\\\": \\\\\"python_module\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Cloud Storage location of the packages (that contain the training program and any additional dependencies). The maximum number of package URIs is 100.\\\\\", \\\\\"name\\\\\": \\\\\"package_uris\\\\\", \\\\\"type\\\\\": \\\\\"List\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Compute Engine region in which the training job is run.\\\\\", \\\\\"name\\\\\": \\\\\"region\\\\\", \\\\\"type\\\\\": \\\\\"GCPRegion\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The command line arguments to pass to the program.\\\\\", \\\\\"name\\\\\": \\\\\"args\\\\\", \\\\\"type\\\\\": \\\\\"List\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"A Cloud Storage path in which to store the training outputs and other data needed for training. This path is passed to your TensorFlow program as the `job-dir` command-line argument. The benefit of specifying this field is that Cloud ML validates the path for use in training.\\\\\", \\\\\"name\\\\\": \\\\\"job_dir\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The version of Python used in training. If not set, the default version is `2.7`. Python `3.5` is available when runtimeVersion is set to `1.4` and above.\\\\\", \\\\\"name\\\\\": \\\\\"python_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Cloud ML Engine runtime version to use for training. If not set, Cloud ML Engine uses the default stable version, 1.0.\\\\\", \\\\\"name\\\\\": \\\\\"runtime_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Docker image to run on the master replica. This image must be in Container Registry.\\\\\", \\\\\"name\\\\\": \\\\\"master_image_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCRPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Docker image to run on the worker replica. This image must be in Container Registry.\\\\\", \\\\\"name\\\\\": \\\\\"worker_image_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCRPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The input parameters to create a training job. It is the JSON payload of a [TrainingInput](https://cloud.google.com/ml-engine/reference/rest/v1/projects.jobs#TrainingInput)\\\\\", \\\\\"name\\\\\": \\\\\"training_input\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The prefix of the generated job id.\\\\\", \\\\\"name\\\\\": \\\\\"job_id_prefix\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"30\\\\\", \\\\\"description\\\\\": \\\\\"Optional. A time-interval to wait for between calls to get the job status. Defaults to 30.\\'\\\\\", \\\\\"name\\\\\": \\\\\"wait_interval\\\\\", \\\\\"type\\\\\": \\\\\"Integer\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Submitting a Cloud ML training job as a pipeline step\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The ID of the created job.\\\\\", \\\\\"name\\\\\": \\\\\"job_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The output path in Cloud Storage of the trainning job, which contains the trained model files.\\\\\", \\\\\"name\\\\\": \\\\\"job_dir\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\",\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"add-pod-env\":\"true\",\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.ml_engine\",\"train\",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--python_module\",\"\",\"--package_uris\",\"\",\"--region\",\"{{inputs.parameters.region}}\",\"--args\",\"[\\\\\"--training_dataset_path\\\\\", \\\\\"{{inputs.parameters.bigquery-query-output_gcs_path}}\\\\\", \\\\\"--validation_dataset_path\\\\\", \\\\\"{{inputs.parameters.bigquery-query-2-output_gcs_path}}\\\\\", \\\\\"--hptune\\\\\", \\\\\"True\\\\\"]\",\"--job_dir\",\"{{inputs.parameters.gcs_root}}/jobdir/hypertune/d16368c5-24bb-4292-9060-315755f79b0b\",\"--python_version\",\"\",\"--runtime_version\",\"\",\"--master_image_uri\",\"gcr.io/dna-gcp-data/trainer_image:test\",\"--worker_image_uri\",\"\",\"--training_input\",\"{{inputs.parameters.hypertune_settings}}\",\"--job_id_prefix\",\"\",\"--wait_interval\",\"30\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\",\"arguments\":{},\"inputs\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\"},{\"name\":\"bigquery-query-output_gcs_path\"},{\"name\":\"gcs_root\"},{\"name\":\"project_id\"},{\"name\":\"region\"},{\"name\":\"retrieve-best-run-alpha\"},{\"name\":\"retrieve-best-run-max_iter\"}]},\"outputs\":{\"parameters\":[{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/ml_engine/job_dir.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"path\":\"/tmp/kfp/output/ml_engine/job_dir.txt\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_id\",\"path\":\"/tmp/kfp/output/ml_engine/job_id.txt\"}]},\"metadata\":{\"annotations\":{\"pipelines.kubeflow.org/component_spec\":\"{\\\\\"description\\\\\": \\\\\"A Kubeflow Pipeline component to submit a Cloud Machine Learning (Cloud ML) \\\\\\\\nEngine training job as a step in a pipeline.\\\\\\\\n\\\\\", \\\\\"inputs\\\\\": [{\\\\\"description\\\\\": \\\\\"Required. The ID of the parent project of the job.\\\\\", \\\\\"name\\\\\": \\\\\"project_id\\\\\", \\\\\"type\\\\\": \\\\\"GCPProjectID\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Python module name to run after installing the packages.\\\\\", \\\\\"name\\\\\": \\\\\"python_module\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Cloud Storage location of the packages (that contain the training program and any additional dependencies). The maximum number of package URIs is 100.\\\\\", \\\\\"name\\\\\": \\\\\"package_uris\\\\\", \\\\\"type\\\\\": \\\\\"List\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Compute Engine region in which the training job is run.\\\\\", \\\\\"name\\\\\": \\\\\"region\\\\\", \\\\\"type\\\\\": \\\\\"GCPRegion\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The command line arguments to pass to the program.\\\\\", \\\\\"name\\\\\": \\\\\"args\\\\\", \\\\\"type\\\\\": \\\\\"List\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"A Cloud Storage path in which to store the training outputs and other data needed for training. This path is passed to your TensorFlow program as the `job-dir` command-line argument. The benefit of specifying this field is that Cloud ML validates the path for use in training.\\\\\", \\\\\"name\\\\\": \\\\\"job_dir\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The version of Python used in training. If not set, the default version is `2.7`. Python `3.5` is available when runtimeVersion is set to `1.4` and above.\\\\\", \\\\\"name\\\\\": \\\\\"python_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Cloud ML Engine runtime version to use for training. If not set, Cloud ML Engine uses the default stable version, 1.0.\\\\\", \\\\\"name\\\\\": \\\\\"runtime_version\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Docker image to run on the master replica. This image must be in Container Registry.\\\\\", \\\\\"name\\\\\": \\\\\"master_image_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCRPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The Docker image to run on the worker replica. This image must be in Container Registry.\\\\\", \\\\\"name\\\\\": \\\\\"worker_image_uri\\\\\", \\\\\"type\\\\\": \\\\\"GCRPath\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The input parameters to create a training job. It is the JSON payload of a [TrainingInput](https://cloud.google.com/ml-engine/reference/rest/v1/projects.jobs#TrainingInput)\\\\\", \\\\\"name\\\\\": \\\\\"training_input\\\\\", \\\\\"type\\\\\": \\\\\"Dict\\\\\"}, {\\\\\"default\\\\\": \\\\\"\\\\\", \\\\\"description\\\\\": \\\\\"The prefix of the generated job id.\\\\\", \\\\\"name\\\\\": \\\\\"job_id_prefix\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"default\\\\\": \\\\\"30\\\\\", \\\\\"description\\\\\": \\\\\"Optional. A time-interval to wait for between calls to get the job status. Defaults to 30.\\'\\\\\", \\\\\"name\\\\\": \\\\\"wait_interval\\\\\", \\\\\"type\\\\\": \\\\\"Integer\\\\\"}], \\\\\"metadata\\\\\": {\\\\\"labels\\\\\": {\\\\\"add-pod-env\\\\\": \\\\\"true\\\\\"}}, \\\\\"name\\\\\": \\\\\"Submitting a Cloud ML training job as a pipeline step\\\\\", \\\\\"outputs\\\\\": [{\\\\\"description\\\\\": \\\\\"The ID of the created job.\\\\\", \\\\\"name\\\\\": \\\\\"job_id\\\\\", \\\\\"type\\\\\": \\\\\"String\\\\\"}, {\\\\\"description\\\\\": \\\\\"The output path in Cloud Storage of the trainning job, which contains the trained model files.\\\\\", \\\\\"name\\\\\": \\\\\"job_dir\\\\\", \\\\\"type\\\\\": \\\\\"GCSPath\\\\\"}, {\\\\\"name\\\\\": \\\\\"MLPipeline UI metadata\\\\\", \\\\\"type\\\\\": \\\\\"UI metadata\\\\\"}]}\",\"sidecar.istio.io/inject\":\"false\"},\"labels\":{\"add-pod-env\":\"true\",\"pipelines.kubeflow.org/cache_enabled\":\"true\"}},\"container\":{\"name\":\"\",\"image\":\"gcr.io/ml-pipeline/ml-pipeline-gcp:e66dcb18607406330f953bf99b04fe7c3ed1a4a8\",\"args\":[\"--ui_metadata_path\",\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"kfp_component.google.ml_engine\",\"train\",\"--project_id\",\"{{inputs.parameters.project_id}}\",\"--python_module\",\"\",\"--package_uris\",\"\",\"--region\",\"{{inputs.parameters.region}}\",\"--args\",\"[\\\\\"--training_dataset_path\\\\\", \\\\\"{{inputs.parameters.bigquery-query-output_gcs_path}}\\\\\", \\\\\"--validation_dataset_path\\\\\", \\\\\"{{inputs.parameters.bigquery-query-2-output_gcs_path}}\\\\\", \\\\\"--alpha\\\\\", \\\\\"{{inputs.parameters.retrieve-best-run-alpha}}\\\\\", \\\\\"--max_iter\\\\\", \\\\\"{{inputs.parameters.retrieve-best-run-max_iter}}\\\\\", \\\\\"--hptune\\\\\", \\\\\"False\\\\\"]\",\"--job_dir\",\"{{inputs.parameters.gcs_root}}/jobdir/d16368c5-24bb-4292-9060-315755f79b0b\",\"--python_version\",\"\",\"--runtime_version\",\"\",\"--master_image_uri\",\"gcr.io/dna-gcp-data/trainer_image:test\",\"--worker_image_uri\",\"\",\"--training_input\",\"\",\"--job_id_prefix\",\"\",\"--wait_interval\",\"30\"],\"env\":[{\"name\":\"KFP_POD_NAME\",\"value\":\"{{pod.name}}\"},{\"name\":\"KFP_POD_NAME\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.name\"}}},{\"name\":\"KFP_NAMESPACE\",\"valueFrom\":{\"fieldRef\":{\"fieldPath\":\"metadata.namespace\"}}}],\"resources\":{}}}],\"entrypoint\":\"covertype-classifier-training\",\"arguments\":{\"parameters\":[{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"region\",\"value\":\"us-central1\"},{\"name\":\"source_table_name\",\"value\":\"covertype_dataset.covertype\"},{\"name\":\"gcs_root\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging\"},{\"name\":\"dataset_id\",\"value\":\"covertype_dataset\"},{\"name\":\"evaluation_metric_name\",\"value\":\"accuracy\"},{\"name\":\"evaluation_metric_threshold\",\"value\":\"0.69\"},{\"name\":\"model_id\",\"value\":\"covertype_classifier\"},{\"name\":\"version_id\",\"value\":\"v01\"},{\"name\":\"replace_existing_version\",\"value\":\"True\"},{\"name\":\"hypertune_settings\",\"value\":\"{\\\\n \\\\\"hyperparameters\\\\\": {\\\\n \\\\\"goal\\\\\": \\\\\"MAXIMIZE\\\\\",\\\\n \\\\\"maxTrials\\\\\": 6,\\\\n \\\\\"maxParallelTrials\\\\\": 3,\\\\n \\\\\"hyperparameterMetricTag\\\\\": \\\\\"accuracy\\\\\",\\\\n \\\\\"enableTrialEarlyStopping\\\\\": True,\\\\n \\\\\"params\\\\\": [\\\\n {\\\\n \\\\\"parameterName\\\\\": \\\\\"max_iter\\\\\",\\\\n \\\\\"type\\\\\": \\\\\"DISCRETE\\\\\",\\\\n \\\\\"discreteValues\\\\\": [500, 1000]\\\\n },\\\\n {\\\\n \\\\\"parameterName\\\\\": \\\\\"alpha\\\\\",\\\\n \\\\\"type\\\\\": \\\\\"DOUBLE\\\\\",\\\\n \\\\\"minValue\\\\\": 0.0001,\\\\n \\\\\"maxValue\\\\\": 0.001,\\\\n \\\\\"scaleType\\\\\": \\\\\"UNIT_LINEAR_SCALE\\\\\"\\\\n }\\\\n ]\\\\n }\\\\n}\"},{\"name\":\"dataset_location\",\"value\":\"US\"}]},\"serviceAccountName\":\"pipeline-runner\"},\"status\":{\"phase\":\"Succeeded\",\"startedAt\":\"2021-05-21T08:46:52Z\",\"finishedAt\":\"2021-05-21T09:10:48Z\",\"nodes\":{\"covertype-classifier-training-qmb6n\":{\"id\":\"covertype-classifier-training-qmb6n\",\"name\":\"covertype-classifier-training-qmb6n\",\"displayName\":\"covertype-classifier-training-qmb6n\",\"type\":\"DAG\",\"templateName\":\"covertype-classifier-training\",\"phase\":\"Succeeded\",\"startedAt\":\"2021-05-21T08:46:52Z\",\"finishedAt\":\"2021-05-21T09:10:48Z\",\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"covertype_dataset\"},{\"name\":\"dataset_location\",\"value\":\"US\"},{\"name\":\"evaluation_metric_name\",\"value\":\"accuracy\"},{\"name\":\"evaluation_metric_threshold\",\"value\":\"0.69\"},{\"name\":\"gcs_root\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging\"},{\"name\":\"hypertune_settings\",\"value\":\"{\\\\n \\\\\"hyperparameters\\\\\": {\\\\n \\\\\"goal\\\\\": \\\\\"MAXIMIZE\\\\\",\\\\n \\\\\"maxTrials\\\\\": 6,\\\\n \\\\\"maxParallelTrials\\\\\": 3,\\\\n \\\\\"hyperparameterMetricTag\\\\\": \\\\\"accuracy\\\\\",\\\\n \\\\\"enableTrialEarlyStopping\\\\\": True,\\\\n \\\\\"params\\\\\": [\\\\n {\\\\n \\\\\"parameterName\\\\\": \\\\\"max_iter\\\\\",\\\\n \\\\\"type\\\\\": \\\\\"DISCRETE\\\\\",\\\\n \\\\\"discreteValues\\\\\": [500, 1000]\\\\n },\\\\n {\\\\n \\\\\"parameterName\\\\\": \\\\\"alpha\\\\\",\\\\n \\\\\"type\\\\\": \\\\\"DOUBLE\\\\\",\\\\n \\\\\"minValue\\\\\": 0.0001,\\\\n \\\\\"maxValue\\\\\": 0.001,\\\\n \\\\\"scaleType\\\\\": \\\\\"UNIT_LINEAR_SCALE\\\\\"\\\\n }\\\\n ]\\\\n }\\\\n}\"},{\"name\":\"model_id\",\"value\":\"covertype_classifier\"},{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"region\",\"value\":\"us-central1\"},{\"name\":\"replace_existing_version\",\"value\":\"True\"},{\"name\":\"source_table_name\",\"value\":\"covertype_dataset.covertype\"},{\"name\":\"version_id\",\"value\":\"v01\"}]},\"children\":[\"covertype-classifier-training-qmb6n-1896224863\",\"covertype-classifier-training-qmb6n-1858401567\",\"covertype-classifier-training-qmb6n-1879447244\"],\"outboundNodes\":[\"covertype-classifier-training-qmb6n-2274394652\"]},\"covertype-classifier-training-qmb6n-1057815578\":{\"id\":\"covertype-classifier-training-qmb6n-1057815578\",\"name\":\"covertype-classifier-training-qmb6n.submitting-a-cloud-ml-training-job-as-a-pipeline-step\",\"displayName\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\",\"type\":\"Pod\",\"templateName\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step\",\"phase\":\"Succeeded\",\"boundaryID\":\"covertype-classifier-training-qmb6n\",\"startedAt\":\"2021-05-21T08:47:07Z\",\"finishedAt\":\"2021-05-21T09:01:46Z\",\"inputs\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/datasets/validation/data.csv\"},{\"name\":\"bigquery-query-output_gcs_path\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/datasets/training/data.csv\"},{\"name\":\"gcs_root\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging\"},{\"name\":\"hypertune_settings\",\"value\":\"{\\\\n \\\\\"hyperparameters\\\\\": {\\\\n \\\\\"goal\\\\\": \\\\\"MAXIMIZE\\\\\",\\\\n \\\\\"maxTrials\\\\\": 6,\\\\n \\\\\"maxParallelTrials\\\\\": 3,\\\\n \\\\\"hyperparameterMetricTag\\\\\": \\\\\"accuracy\\\\\",\\\\n \\\\\"enableTrialEarlyStopping\\\\\": True,\\\\n \\\\\"params\\\\\": [\\\\n {\\\\n \\\\\"parameterName\\\\\": \\\\\"max_iter\\\\\",\\\\n \\\\\"type\\\\\": \\\\\"DISCRETE\\\\\",\\\\n \\\\\"discreteValues\\\\\": [500, 1000]\\\\n },\\\\n {\\\\n \\\\\"parameterName\\\\\": \\\\\"alpha\\\\\",\\\\n \\\\\"type\\\\\": \\\\\"DOUBLE\\\\\",\\\\n \\\\\"minValue\\\\\": 0.0001,\\\\n \\\\\"maxValue\\\\\": 0.001,\\\\n \\\\\"scaleType\\\\\": \\\\\"UNIT_LINEAR_SCALE\\\\\"\\\\n }\\\\n ]\\\\n }\\\\n}\"},{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"region\",\"value\":\"us-central1\"}]},\"outputs\":{\"parameters\":[{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\",\"value\":\"job_6432b8d633522e4d387b6335bfbf3639\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/ml_engine/job_id.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_dir\",\"path\":\"/tmp/kfp/output/ml_engine/job_dir.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1057815578/submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_dir.tgz\"}},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\",\"path\":\"/tmp/kfp/output/ml_engine/job_id.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1057815578/submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id.tgz\"}},{\"name\":\"main-logs\",\"archiveLogs\":true,\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1057815578/main.log\"}}]},\"children\":[\"covertype-classifier-training-qmb6n-1075982665\"]},\"covertype-classifier-training-qmb6n-1075982665\":{\"id\":\"covertype-classifier-training-qmb6n-1075982665\",\"name\":\"covertype-classifier-training-qmb6n.retrieve-best-run\",\"displayName\":\"retrieve-best-run\",\"type\":\"Pod\",\"templateName\":\"retrieve-best-run\",\"phase\":\"Succeeded\",\"boundaryID\":\"covertype-classifier-training-qmb6n\",\"startedAt\":\"2021-05-21T09:01:47Z\",\"finishedAt\":\"2021-05-21T09:01:52Z\",\"inputs\":{\"parameters\":[{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-job_id\",\"value\":\"job_6432b8d633522e4d387b6335bfbf3639\"}]},\"outputs\":{\"parameters\":[{\"name\":\"retrieve-best-run-alpha\",\"value\":\"0.00012415813718618125\",\"valueFrom\":{\"path\":\"/tmp/outputs/alpha/data\"}},{\"name\":\"retrieve-best-run-max_iter\",\"value\":\"500\",\"valueFrom\":{\"path\":\"/tmp/outputs/max_iter/data\"}}],\"artifacts\":[{\"name\":\"retrieve-best-run-alpha\",\"path\":\"/tmp/outputs/alpha/data\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1075982665/retrieve-best-run-alpha.tgz\"}},{\"name\":\"retrieve-best-run-max_iter\",\"path\":\"/tmp/outputs/max_iter/data\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1075982665/retrieve-best-run-max_iter.tgz\"}},{\"name\":\"retrieve-best-run-metric_value\",\"path\":\"/tmp/outputs/metric_value/data\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1075982665/retrieve-best-run-metric_value.tgz\"}},{\"name\":\"main-logs\",\"archiveLogs\":true,\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1075982665/main.log\"}}]},\"children\":[\"covertype-classifier-training-qmb6n-1089717477\"]},\"covertype-classifier-training-qmb6n-1089717477\":{\"id\":\"covertype-classifier-training-qmb6n-1089717477\",\"name\":\"covertype-classifier-training-qmb6n.submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\",\"displayName\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\",\"type\":\"Pod\",\"templateName\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2\",\"phase\":\"Succeeded\",\"boundaryID\":\"covertype-classifier-training-qmb6n\",\"startedAt\":\"2021-05-21T09:01:53Z\",\"finishedAt\":\"2021-05-21T09:08:59Z\",\"inputs\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/datasets/validation/data.csv\"},{\"name\":\"bigquery-query-output_gcs_path\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/datasets/training/data.csv\"},{\"name\":\"gcs_root\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging\"},{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"region\",\"value\":\"us-central1\"},{\"name\":\"retrieve-best-run-alpha\",\"value\":\"0.00012415813718618125\"},{\"name\":\"retrieve-best-run-max_iter\",\"value\":\"500\"}]},\"outputs\":{\"parameters\":[{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/jobdir/d16368c5-24bb-4292-9060-315755f79b0b\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/ml_engine/job_dir.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"path\":\"/tmp/kfp/output/ml_engine/job_dir.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1089717477/submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir.tgz\"}},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_id\",\"path\":\"/tmp/kfp/output/ml_engine/job_id.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1089717477/submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_id.tgz\"}},{\"name\":\"main-logs\",\"archiveLogs\":true,\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1089717477/main.log\"}}]},\"children\":[\"covertype-classifier-training-qmb6n-1971586539\",\"covertype-classifier-training-qmb6n-2585557679\"]},\"covertype-classifier-training-qmb6n-1858401567\":{\"id\":\"covertype-classifier-training-qmb6n-1858401567\",\"name\":\"covertype-classifier-training-qmb6n.bigquery-query\",\"displayName\":\"bigquery-query\",\"type\":\"Pod\",\"templateName\":\"bigquery-query\",\"phase\":\"Succeeded\",\"boundaryID\":\"covertype-classifier-training-qmb6n\",\"startedAt\":\"2021-05-21T08:46:52Z\",\"finishedAt\":\"2021-05-21T08:47:05Z\",\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"covertype_dataset\"},{\"name\":\"dataset_location\",\"value\":\"US\"},{\"name\":\"gcs_root\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging\"},{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"source_table_name\",\"value\":\"covertype_dataset.covertype\"}]},\"outputs\":{\"parameters\":[{\"name\":\"bigquery-query-output_gcs_path\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/datasets/training/data.csv\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"bigquery-query-output_gcs_path\",\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1858401567/bigquery-query-output_gcs_path.tgz\"}},{\"name\":\"main-logs\",\"archiveLogs\":true,\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1858401567/main.log\"}}]},\"children\":[\"covertype-classifier-training-qmb6n-1057815578\",\"covertype-classifier-training-qmb6n-1089717477\"]},\"covertype-classifier-training-qmb6n-1879447244\":{\"id\":\"covertype-classifier-training-qmb6n-1879447244\",\"name\":\"covertype-classifier-training-qmb6n.bigquery-query-2\",\"displayName\":\"bigquery-query-2\",\"type\":\"Pod\",\"templateName\":\"bigquery-query-2\",\"phase\":\"Succeeded\",\"boundaryID\":\"covertype-classifier-training-qmb6n\",\"startedAt\":\"2021-05-21T08:46:52Z\",\"finishedAt\":\"2021-05-21T08:47:05Z\",\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"covertype_dataset\"},{\"name\":\"dataset_location\",\"value\":\"US\"},{\"name\":\"gcs_root\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging\"},{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"source_table_name\",\"value\":\"covertype_dataset.covertype\"}]},\"outputs\":{\"parameters\":[{\"name\":\"bigquery-query-2-output_gcs_path\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/datasets/validation/data.csv\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"bigquery-query-2-output_gcs_path\",\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1879447244/bigquery-query-2-output_gcs_path.tgz\"}},{\"name\":\"main-logs\",\"archiveLogs\":true,\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1879447244/main.log\"}}]},\"children\":[\"covertype-classifier-training-qmb6n-1057815578\",\"covertype-classifier-training-qmb6n-1089717477\"]},\"covertype-classifier-training-qmb6n-1896224863\":{\"id\":\"covertype-classifier-training-qmb6n-1896224863\",\"name\":\"covertype-classifier-training-qmb6n.bigquery-query-3\",\"displayName\":\"bigquery-query-3\",\"type\":\"Pod\",\"templateName\":\"bigquery-query-3\",\"phase\":\"Succeeded\",\"boundaryID\":\"covertype-classifier-training-qmb6n\",\"startedAt\":\"2021-05-21T08:46:52Z\",\"finishedAt\":\"2021-05-21T08:47:05Z\",\"inputs\":{\"parameters\":[{\"name\":\"dataset_id\",\"value\":\"covertype_dataset\"},{\"name\":\"dataset_location\",\"value\":\"US\"},{\"name\":\"gcs_root\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging\"},{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"source_table_name\",\"value\":\"covertype_dataset.covertype\"}]},\"outputs\":{\"parameters\":[{\"name\":\"bigquery-query-3-output_gcs_path\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/datasets/testing/data.csv\",\"valueFrom\":{\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\"}}],\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"bigquery-query-3-output_gcs_path\",\"path\":\"/tmp/kfp/output/bigquery/query-output-path.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1896224863/bigquery-query-3-output_gcs_path.tgz\"}},{\"name\":\"main-logs\",\"archiveLogs\":true,\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1896224863/main.log\"}}]},\"children\":[\"covertype-classifier-training-qmb6n-1971586539\"]},\"covertype-classifier-training-qmb6n-1971586539\":{\"id\":\"covertype-classifier-training-qmb6n-1971586539\",\"name\":\"covertype-classifier-training-qmb6n.evaluate-model\",\"displayName\":\"evaluate-model\",\"type\":\"Pod\",\"templateName\":\"evaluate-model\",\"phase\":\"Succeeded\",\"boundaryID\":\"covertype-classifier-training-qmb6n\",\"startedAt\":\"2021-05-21T09:09:00Z\",\"finishedAt\":\"2021-05-21T09:09:08Z\",\"inputs\":{\"parameters\":[{\"name\":\"bigquery-query-3-output_gcs_path\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/datasets/testing/data.csv\"},{\"name\":\"evaluation_metric_name\",\"value\":\"accuracy\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/jobdir/d16368c5-24bb-4292-9060-315755f79b0b\"}]},\"outputs\":{\"parameters\":[{\"name\":\"evaluate-model-metric_value\",\"value\":\"0.7225525168450257\",\"valueFrom\":{\"path\":\"/tmp/outputs/metric_value/data\"}}],\"artifacts\":[{\"name\":\"mlpipeline-metrics\",\"path\":\"/tmp/outputs/mlpipeline_metrics/data\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1971586539/mlpipeline-metrics.tgz\"},\"optional\":true},{\"name\":\"evaluate-model-metric_name\",\"path\":\"/tmp/outputs/metric_name/data\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1971586539/evaluate-model-metric_name.tgz\"}},{\"name\":\"evaluate-model-metric_value\",\"path\":\"/tmp/outputs/metric_value/data\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1971586539/evaluate-model-metric_value.tgz\"}},{\"name\":\"main-logs\",\"archiveLogs\":true,\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-1971586539/main.log\"}}]},\"children\":[\"covertype-classifier-training-qmb6n-2585557679\"]},\"covertype-classifier-training-qmb6n-2274394652\":{\"id\":\"covertype-classifier-training-qmb6n-2274394652\",\"name\":\"covertype-classifier-training-qmb6n.condition-1.deploying-a-trained-model-to-cloud-machine-learning-engine\",\"displayName\":\"deploying-a-trained-model-to-cloud-machine-learning-engine\",\"type\":\"Pod\",\"templateName\":\"deploying-a-trained-model-to-cloud-machine-learning-engine\",\"phase\":\"Succeeded\",\"boundaryID\":\"covertype-classifier-training-qmb6n-2585557679\",\"startedAt\":\"2021-05-21T09:09:10Z\",\"finishedAt\":\"2021-05-21T09:10:46Z\",\"inputs\":{\"parameters\":[{\"name\":\"model_id\",\"value\":\"covertype_classifier\"},{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"replace_existing_version\",\"value\":\"True\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/jobdir/d16368c5-24bb-4292-9060-315755f79b0b\"},{\"name\":\"version_id\",\"value\":\"v01\"}]},\"outputs\":{\"artifacts\":[{\"name\":\"mlpipeline-ui-metadata\",\"path\":\"/tmp/outputs/MLPipeline_UI_metadata/data\",\"optional\":true},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine-model_name\",\"path\":\"/tmp/kfp/output/ml_engine/model_name.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-2274394652/deploying-a-trained-model-to-cloud-machine-learning-engine-model_name.tgz\"}},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine-model_uri\",\"path\":\"/tmp/kfp/output/ml_engine/model_uri.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-2274394652/deploying-a-trained-model-to-cloud-machine-learning-engine-model_uri.tgz\"}},{\"name\":\"deploying-a-trained-model-to-cloud-machine-learning-engine-version_name\",\"path\":\"/tmp/kfp/output/ml_engine/version_name.txt\",\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-2274394652/deploying-a-trained-model-to-cloud-machine-learning-engine-version_name.tgz\"}},{\"name\":\"main-logs\",\"archiveLogs\":true,\"s3\":{\"endpoint\":\"minio-service.kfpdemo:9000\",\"bucket\":\"mlpipeline\",\"insecure\":true,\"accessKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"accesskey\"},\"secretKeySecret\":{\"name\":\"mlpipeline-minio-artifact\",\"key\":\"secretkey\"},\"key\":\"artifacts/covertype-classifier-training-qmb6n/covertype-classifier-training-qmb6n-2274394652/main.log\"}}]}},\"covertype-classifier-training-qmb6n-2585557679\":{\"id\":\"covertype-classifier-training-qmb6n-2585557679\",\"name\":\"covertype-classifier-training-qmb6n.condition-1\",\"displayName\":\"condition-1\",\"type\":\"DAG\",\"templateName\":\"condition-1\",\"phase\":\"Succeeded\",\"boundaryID\":\"covertype-classifier-training-qmb6n\",\"startedAt\":\"2021-05-21T09:09:10Z\",\"finishedAt\":\"2021-05-21T09:10:48Z\",\"inputs\":{\"parameters\":[{\"name\":\"model_id\",\"value\":\"covertype_classifier\"},{\"name\":\"project_id\",\"value\":\"dna-gcp-data\"},{\"name\":\"replace_existing_version\",\"value\":\"True\"},{\"name\":\"submitting-a-cloud-ml-training-job-as-a-pipeline-step-2-job_dir\",\"value\":\"gs://dna-gcp-data-kubeflowpipelines-default/staging/jobdir/d16368c5-24bb-4292-9060-315755f79b0b\"},{\"name\":\"version_id\",\"value\":\"v01\"}]},\"children\":[\"covertype-classifier-training-qmb6n-2274394652\"],\"outboundNodes\":[\"covertype-classifier-training-qmb6n-2274394652\"]}},\"conditions\":[{\"type\":\"Completed\",\"status\":\"True\"}]}}'}}\n0.7225525168450257\n" ], [ "from googleapiclient import discovery\nml = discovery.build('ml', 'v1')\njob_name = 'projects/{}/jobs/{}'.format('dna-gcp-data', 'job_1dae51e7dd77989943e0aaf271f1effd')\nrequest = ml.projects().jobs().get(name=job_name)\nprint(type(request.execute())", "_____no_output_____" ] ], [ [ "### Monitoring the run\n\nYou can monitor the run using KFP UI. Follow the instructor who will walk you through the KFP UI and monitoring techniques.\n\nTo access the KFP UI in your environment use the following URI:\n\nhttps://[ENDPOINT]\n\n\n**NOTE that your pipeline run may fail due to the bug in a BigQuery component that does not handle certain race conditions. If you observe the pipeline failure, re-run the last cell of the notebook to submit another pipeline run or retry the run from the KFP UI**\n", "_____no_output_____" ], [ "<font size=-1>Licensed under the Apache License, Version 2.0 (the \\\"License\\\");\nyou may not use this file except in compliance with the License.\nYou may obtain a copy of the License at [https://www.apache.org/licenses/LICENSE-2.0](https://www.apache.org/licenses/LICENSE-2.0)\n\nUnless required by applicable law or agreed to in writing, software distributed under the License is distributed on an \\\"AS IS\\\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.</font>", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Feature: TF-IDF Distances", "_____no_output_____" ], [ "Create TF-IDF vectors from question texts and compute vector distances between them.", "_____no_output_____" ], [ "## Imports", "_____no_output_____" ], [ "This utility package imports `numpy`, `pandas`, `matplotlib` and a helper `kg` module into the root namespace.", "_____no_output_____" ] ], [ [ "from pygoose import *", "_____no_output_____" ], [ "from sklearn.feature_extraction.text import TfidfVectorizer\nfrom sklearn.metrics.pairwise import cosine_distances, euclidean_distances", "_____no_output_____" ] ], [ [ "## Config", "_____no_output_____" ], [ "Automatically discover the paths to various data folders and compose the project structure.", "_____no_output_____" ] ], [ [ "project = kg.Project.discover()", "_____no_output_____" ] ], [ [ "Identifier for storing these features on disk and referring to them later.", "_____no_output_____" ] ], [ [ "feature_list_id = 'tfidf'", "_____no_output_____" ] ], [ [ "## Read Data", "_____no_output_____" ], [ "Preprocessed and tokenized questions.", "_____no_output_____" ] ], [ [ "tokens_train = kg.io.load(project.preprocessed_data_dir + 'tokens_lowercase_spellcheck_no_stopwords_train.pickle')\ntokens_test = kg.io.load(project.preprocessed_data_dir + 'tokens_lowercase_spellcheck_no_stopwords_test.pickle')", "_____no_output_____" ], [ "tokens = tokens_train + tokens_test", "_____no_output_____" ] ], [ [ "Extract a set of unique question texts (document corpus).", "_____no_output_____" ] ], [ [ "all_questions_flat = np.array(tokens).ravel()", "_____no_output_____" ], [ "documents = list(set(' '.join(question) for question in all_questions_flat))", "_____no_output_____" ], [ "del all_questions_flat", "_____no_output_____" ] ], [ [ "## Train TF-IDF vectorizer", "_____no_output_____" ], [ "Create a bag-of-token-unigrams vectorizer.", "_____no_output_____" ] ], [ [ "vectorizer = TfidfVectorizer(\n encoding='utf-8',\n analyzer='word',\n strip_accents='unicode',\n ngram_range=(1, 1),\n lowercase=True,\n norm='l2',\n use_idf=True,\n smooth_idf=True,\n sublinear_tf=True,\n)", "_____no_output_____" ], [ "vectorizer.fit(documents)", "_____no_output_____" ], [ "model_filename = 'tfidf_vectorizer_{}_ngrams_{}_{}_penalty_{}.pickle'.format(\n vectorizer.analyzer,\n vectorizer.ngram_range[0],\n vectorizer.ngram_range[1],\n vectorizer.norm,\n)", "_____no_output_____" ], [ "kg.io.save(vectorizer, project.trained_model_dir + model_filename)", "_____no_output_____" ] ], [ [ "## Vectorize train and test sets, compute distances", "_____no_output_____" ] ], [ [ "def compute_pair_distances(pair):\n q1_doc = ' '.join(pair[0])\n q2_doc = ' '.join(pair[1])\n \n pair_dtm = vectorizer.transform([q1_doc, q2_doc])\n q1_doc_vec = pair_dtm[0]\n q2_doc_vec = pair_dtm[1]\n \n return [\n cosine_distances(q1_doc_vec, q2_doc_vec)[0][0],\n euclidean_distances(q1_doc_vec, q2_doc_vec)[0][0],\n ]", "_____no_output_____" ], [ "features = kg.jobs.map_batch_parallel(\n tokens,\n item_mapper=compute_pair_distances,\n batch_size=1000,\n)", "Batches: 100%|██████████| 2751/2751 [16:11<00:00, 3.01it/s]\n" ], [ "X_train = np.array(features[:len(tokens_train)], dtype='float64')\nX_test = np.array(features[len(tokens_train):], dtype='float64')", "_____no_output_____" ], [ "print('X_train:', X_train.shape)\nprint('X_test: ', X_test.shape)", "X_train: (404290, 2)\nX_test: (2345796, 2)\n" ] ], [ [ "## Save features", "_____no_output_____" ] ], [ [ "feature_names = [\n 'tfidf_cosine',\n 'tfidf_euclidean',\n]", "_____no_output_____" ], [ "project.save_features(X_train, X_test, feature_names, feature_list_id)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "%matplotlib inline\nfrom matplotlib import pyplot as plt\nimport numpy as np\nimport seaborn as sns\nimport os\nif not os.path.exists('figures'):\n os.makedirs('figures')", "_____no_output_____" ], [ "from sklearn.linear_model import LinearRegression\n# Create some data for the regression\nrng = np.random.RandomState(1)\n\nX = rng.randn(200, 2)\ny = np.dot(X, [-2, 1]) + 0.1 * rng.randn(X.shape[0])\n\n# fit the regressions model\nmodel = LinearRegression()\nmodel.fit(X, y)\n\n# create some new points to predict\nX2 = rng.randn(100, 2)\n\n# predict the labels\ny2 = model.predict(X2)", "_____no_output_____" ], [ "def format_plot(ax, title):\n ax.xaxis.set_major_formatter(plt.NullFormatter())\n ax.yaxis.set_major_formatter(plt.NullFormatter())\n ax.set_xlabel('feature 1', color='gray')\n ax.set_ylabel('feature 2', color='gray')\n ax.set_title(title, color='gray')", "_____no_output_____" ], [ "fig, ax = plt.subplots()\npoints = ax.scatter(X[:, 0], X[:, 1], c=y, s=50,\n cmap='viridis')\n\n# format plot\nformat_plot(ax, 'Input Data')\nax.axis([-4, 4, -3, 3])\nfig.savefig('figures/05.01-regression-1.png')", "_____no_output_____" ], [ "from mpl_toolkits.mplot3d.art3d import Line3DCollection\n\npoints = np.hstack([X, y[:, None]]).reshape(-1, 1, 3)\nsegments = np.hstack([points, points])\nsegments[:, 0, 2] = -8\n\n# plot points in 3D\nfig = plt.figure()\nax = fig.add_subplot(111, projection='3d')\nax.scatter(X[:, 0], X[:, 1], y, c=y, s=35,\n cmap='viridis')\nax.add_collection3d(Line3DCollection(segments, colors='gray', alpha=0.2))\nax.scatter(X[:, 0], X[:, 1], -8 + np.zeros(X.shape[0]), c=y, s=10, cmap='viridis')\n\n# format \nax.patch.set_facecolor('white')\nax.view_init(elev=20, azim=-70)\nax.set_zlim3d(-8, 8)\nax.xaxis.set_major_formatter(plt.NullFormatter())\nax.yaxis.set_major_formatter(plt.NullFormatter())\nax.zaxis.set_major_formatter(plt.NullFormatter())\nax.set(xlabel='feature 1', ylabel='feature 2', zlabel='label')\n\n# Hide axes\n\nax.w_xaxis.line.set_visible(False)\nax.w_yaxis.line.set_visible(False)\nax.w_zaxis.line.set_visible(False)\n\nfor tick in ax.w_xaxis.get_ticklines():\n tick.set_visible(False)\n \nfor tick in ax.w_yaxis.get_ticklines():\n tick.set_visible(False)\n \nfor tick in ax.w_zaxis.get_ticklines():\n tick.set_visible(False)\n \nfig.savefig('figures/05.01-regression-2.png')", "_____no_output_____" ], [ "from matplotlib.collections import LineCollection\n\n# plot data points\nfig, ax = plt.subplots()\npts = ax.scatter(X[:, 0], X[:, 1], c=y, s=50,\n cmap='viridis', zorder=2)\n\nxx, yy = np.meshgrid(np.linspace(-4, 4),\n np.linspace(-3, 3))\nXfit = np.vstack([xx.ravel(), yy.ravel()]).T\nyfit = model.predict(Xfit)\nzz = yfit.reshape(xx.shape)\nax.pcolorfast([-4, 4], [-3, 3], zz, alpha=0.5,\n cmap='viridis', norm=pts.norm, zorder=1)\n\n# format plot\nformat_plot(ax, 'Input Data with Linear Fit')\nax.axis([-4, 4, -3, 3])\nfig.savefig('figures/05.01-regression-3.png')", "_____no_output_____" ], [ "# plot the model fit\nfig, ax = plt.subplots(1, 2, figsize=(16, 6))\nfig.subplots_adjust(left=0.0625, right=0.95, wspace=0.1)\n\nax[0].scatter(X2[:, 0], X2[:, 1], c='gray', s=50)\nax[0].axis([-4, 4, -3, 3])\n\nax[1].scatter(X2[:, 0], X2[:, 1], c=y2, s=50,\n cmap='viridis', norm=pts.norm)\n\nax[1].axis([-4, 4, -3, 3])\n\n# format plots\nformat_plot(ax[0], 'Unknown Data')\nformat_plot(ax[1], 'Predicted Labels')\n\nfig.savefig('figures/05.01-regression-4.png')", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "#예제 1-1 딕셔너리 > 시리즈 변환\n# -*- coding: utf-8 -*-\n\n#pandas 불러오기\nimport pandas as pd\n\n#key:value 쌍으로 딕셔너리를 만들고, 변수 dict_data에 저장\ndict_data = {'a':1, 'b':2, 'c':3}\n\n#판다스 Series() 함수로 dictionary를 Series로 변환, 변수 sr에 저장\nsr = pd.Series(dict_data)\n\n#sr 의 자료형 출력\nprint(type(sr))\nprint('\\n')\n#변수 sr에 저장되어 있는 시리즈 객체를 출력\nprint(sr)", "<class 'pandas.core.series.Series'>\n\n\na 1\nb 2\nc 3\ndtype: int64\n" ], [ "#예제 1-2 시리즈 인덱스 \n# -*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#리스트를 시리즈로 변환하여 변수 sr 에 저장.\nlist_data = ['2019-01-02', 3.14, 'abc', 100, True]\n\n#딕셔너리의 키 처럼 인덱스로 변환될 값이 없다.\n#인덱스를 별도로 정하지 않으면 디폴트로 정수형 위치 인덱스가 자동지정된다\n\nsr = pd.Series(list_data)\nprint(sr)", "0 2019-01-02\n1 3.14\n2 abc\n3 100\n4 True\ndtype: object\n" ], [ "#인덱스 배열은 idx에 저장, 값은 val에 저장\nidx = sr.index\nval = sr.values\nprint(idx)\nprint()\nprint(val)", "RangeIndex(start=0, stop=5, step=1)\n\n['2019-01-02' 3.14 'abc' 100 True]\n" ], [ "dict_data = {'a' :10, 'b':20, 'c':30, 'd':40, 'e':50, 'f':60}\nsr = pd.Series(dict_data)", "_____no_output_____" ], [ "#인덱스 이름 으로 접근\nprint(sr['a'])\nprint()\n#인덱스 번호로 접근\nprint(sr[-1])\nprint()\n", "10\n\n60\n\n" ], [ "print(sr[['d','e']])\nprint(sr['d':'e'])\nprint(sr[3:5])\nprint(sr[[3,4]])", "d 40\ne 50\ndtype: int64\nd 40\ne 50\ndtype: int64\nd 40\ne 50\ndtype: int64\nd 40\ne 50\ndtype: int64\n" ], [ "#예제 1-3 시리즈 원소선택\n\n# -*- coding: utf-8 -*-\nimport pandas as pd\n\n#튜플을 시리즈로 변환(인덱스 옵션 지정)\ntup_data = ('영인', '2010-05-01', '여', True)\nsr = pd.Series(tup_data, index=['이름','생년월일','성별','학생여부'])\nprint(sr)", "이름 영인\n생년월일 2010-05-01\n성별 여\n학생여부 True\ndtype: object\n" ], [ "#원소를 1개 선택\nprint(sr[0])\nprint(sr['이름'])", "영인\n영인\n" ], [ "#여러 개의 원소를 선택 (인덱스 리스트 활용)\nprint(sr[[0,3]])\nprint()\nprint(sr[['이름', '학생여부']])", "이름 영인\n학생여부 True\ndtype: object\n\n이름 영인\n학생여부 True\ndtype: object\n" ], [ "#여러 개의 원소를 선택(인덱스 범위 지정)\nprint(sr[1:2])\nprint()\nprint(sr['생년월일':'성별'])\n#인덱스로 선택시에는 1부터 2(n-1) 까지 출력\n#인덱스 이름으로 슬라이싱 시에는 생년월일 부터 성별 까지", "생년월일 2010-05-01\ndtype: object\n\n생년월일 2010-05-01\n성별 여\ndtype: object\n" ], [ "#데이터 프레임은 여러개의 시리즈를 모아둔 집합으로 생각하면된다.", "_____no_output_____" ], [ "#예제 1-4 딕셔너리 > 데이터 프레임 변환\n# -*- coding: utf-8 -*-\nimport pandas as pd\nimport numpy as np\n#열 이름을 key로 하고, 리스트를 value로 갖는 딕셔너리 정의 (2차원 배열)\ndict_data = {'c0':[1,2,3], 'c1':[4,5,6], 'c2':[7,8,9], 'c3':[10,11,12], 'c4':[13,14,15]}\n\ndf = pd.DataFrame(dict_data)\n\n#df의 자료형 출력\nprint(type(df))\nprint('\\n')\nprint(df)\n", "<class 'pandas.core.frame.DataFrame'>\n\n\n c0 c1 c2 c3 c4\n0 1 4 7 10 13\n1 2 5 8 11 14\n2 3 6 9 12 15\n" ], [ "#예제 1-5 행 인덱스/열 이름설정\n# -*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#행 인덱스/ 열 이름 지정하여 데이터 프레임 만들기\ndf= pd.DataFrame([[15,'남','덕영중'],\n [17,'여','수리중']],\n index=['준서','예은'],\n columns=['나이','성별','학교'])\n", "_____no_output_____" ], [ "#행 인덱스, 열 이름 확인하기\nprint(df) #데이터 프레임\nprint()\nprint(df.index)# 행 인덱스\nprint()\nprint(df.columns)#열 이름\n\n#실행 결과에서 리스트가 행으로 변환 되는 점에 유의한다.\n", " 나이 성별 학교\n준서 15 남 덕영중\n예은 17 여 수리중\n\nIndex(['준서', '예은'], dtype='object')\n\nIndex(['나이', '성별', '학교'], dtype='object')\n" ], [ "#예제 1-5 행 인덱스/열 이름 설정\ndf.index=['학생1','학생2']\ndf.columns=['연령','남녀','소속']", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "#데이터 프레임에 rename()메소드를 적용하면 \n#행 인덱스 또는 열 이름의 일부를 선택하여 변경가능\n#단, 원본 객체를 직접 수정하는것이 아니라\n#새로운 데이터 프레임 객체를 반환한다.\n#원본 객체를 변경하려면 inplace=True 옵션을 사용한다.", "_____no_output_____" ], [ "#예제 1-6 행 인덱스/ 열 이름 변경\n# -*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#행 인덱스/ 열 이름 지정하여 데이터 프레임 만들기\ndf = pd.DataFrame([[15,'남','덕영중'], [17,'여','수리중']],\n index=['준서','예은'],\n columns=['나이', '성별', '학교'])\n#데이터프레임 df 출력\nprint(df)", " 나이 성별 학교\n준서 15 남 덕영중\n예은 17 여 수리중\n" ], [ "#열 이름 중, '나이' 를 '연령'으로, '성별'을 '남녀'로,'학교'를''소속'으로\ndf.rename(columns={'나이':'연령', '성별':'남녀','학교':'소속'},inplace=True)\n\n#df 의 행 인덱스 중에서, '준서'를 '학생1'로, '예은'을 '학생2'로 바꾸기\ndf.rename(index={'준서':'학생1','예은':'학생2'}, inplace=True)", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "df = pd.DataFrame([[90, 98, 85, 100],\n [90, 98, 85, 100],\n [90, 98, 85, 100]],\n index=['서준','우현','인아'],\n columns=['수학','영어','음악','체육']\n \n)\ndf", "_____no_output_____" ], [ "#DataFrame()함수로 데이터 프레임 변환, 변수 df에 저장\n\nexam_data = {'수학':[90,80,70],'영어':[98,98,98],'음악':[85,85,85],'체육':[100,100,100]}\ndf = pd.DataFrame(exam_data, index=['서준','우현','인하'])", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "#데이터프레임 df를 복제하여 변수 df2에 저장, df2의 1개 행(row)삭제\n\ndf2=df[:]\ndf2.drop('우현',inplace=True)\nprint(df2)", " 수학 영어 음악 체육\n서준 90 98 85 100\n인하 70 98 85 100\n" ], [ "df", "_____no_output_____" ], [ "#데이터 프레임 df 를 복제하여 변수 df3에 저장, df3의 2개행(row) 삭제\ndf3 = df[:]\ndf3.drop(['우현','인하'], axis=0,inplace=True)\ndf3", "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\pandas\\core\\frame.py:4163: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n return super().drop(\n" ], [ "#예제 1-8 열 삭제\ndf", "_____no_output_____" ], [ "#데이터 프레임 df 를 복제하여 변수 df4에 저장, df4의 1개열(column)삭제\ndf4 =df[:]\ndf4.drop('수학', axis=1, inplace=True)\ndf4", "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\pandas\\core\\frame.py:4163: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n return super().drop(\n" ], [ "#데이터프레임 df를 복제하여 변수 df5에 저장, df5의 2개 열(column)삭제\ndf5=df[:]\ndf5.drop(['영어','음악'],axis=1, inplace=True)\ndf5", "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\pandas\\core\\frame.py:4163: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame\n\nSee the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n return super().drop(\n" ], [ "#예제 1-9 행 선택\ndf", "_____no_output_____" ], [ "label1 = df.loc['서준']\nposition1 = df.iloc[0]\nprint(label1)\nprint()\nprint(position1)", "수학 90\n영어 98\n음악 85\n체육 100\nName: 서준, dtype: int64\n\n수학 90\n영어 98\n음악 85\n체육 100\nName: 서준, dtype: int64\n" ], [ "#행 여러개 선택\nlabel2 = df.loc[['서준','우현']]\nposition2 = df.iloc[[0,1]]\nprint(label2)\nprint()\nprint(position2)", " 수학 영어 음악 체육\n서준 90 98 85 100\n우현 80 98 85 100\n\n 수학 영어 음악 체육\n서준 90 98 85 100\n우현 80 98 85 100\n" ], [ "#행 슬라이싱\nlabel3 = df.loc['서준':'우현']\nposition3 = df.iloc[0:1]\nprint(label3)\nprint()\nprint(position3)\n#슬라이싱은 둘의 결과가 다름을 주의해야한다", " 수학 영어 음악 체육\n서준 90 98 85 100\n우현 80 98 85 100\n\n 수학 영어 음악 체육\n서준 90 98 85 100\n" ], [ "#예제 1-10 열 선택\n\n# -*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#DataFrame() 함수로 데이터프레임 변환, 변수 df에 저장\nexam_data = {'이름':['서준', '우현', '인아'],\n '수학':[90,80,70],\n '영어':[98,89,95],\n '음악':[85,95,100],\n '체육':[100,90,90]\n }\ndf = pd.DataFrame(exam_data)\ndf", "_____no_output_____" ], [ "#수학 점수 데이터만 선택, 변수 math1에 저장\nmath1 = df['수학']\nprint(math1)", "0 90\n1 80\n2 70\nName: 수학, dtype: int64\n" ], [ "#영어 점수 데이터만 선택, 변수english에 저장\nenglish = df['영어']\nprint(english)", "0 98\n1 89\n2 95\nName: 영어, dtype: int64\n" ], [ "#'음악', '체육' 점수 데이터를 선택, 변수 music_gym에 저장\nmusic_gym = df[['음악','체육']]\nprint(music_gym)", " 음악 체육\n0 85 100\n1 95 90\n2 100 90\n" ], [ "#'수학' 점수 데이터만 선택 변수 math2에 저장\nmath2 = df[['수학']]\nprint(math2)\nprint(type(math2))", " 수학\n0 90\n1 80\n2 70\n<class 'pandas.core.frame.DataFrame'>\n" ], [ "df.iloc[::2]", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "df.iloc[0:3:2]", "_____no_output_____" ], [ "#역순으로 인덱싱하려면 슬라이싱 간격에 -1을 입력", "_____no_output_____" ], [ "df.iloc[::-1]", "_____no_output_____" ], [ "df.iloc[3::-1]", "_____no_output_____" ], [ "#예제 1-11 원소 선택\n# -*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#DataFrame() 함수로 데이터프레임 변환, 변수 df에 저장\nexam_data = {'이름':['서준','우현','인아'],\n '수학':[90,80,70],\n '영어':[98,89,95],\n '음악':[85,95,100],\n '체육':[100,90,90]\n }\ndf = pd.DataFrame(exam_data)\ndf", "_____no_output_____" ], [ "#'이름' 열을 새로운 인덱스로 지정하고, df 객체에 변경 사항 반영\ndf.set_index('이름', inplace=True)\ndf", "_____no_output_____" ], [ "df.loc['서준','음악']", "_____no_output_____" ], [ "df.iloc[0,2]", "_____no_output_____" ], [ "df.loc['서준',['음악','체육']]", "_____no_output_____" ], [ "df.iloc[0,[2,3]]", "_____no_output_____" ], [ "df.iloc[0, 2:]", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "df.loc[['서준','우현'],['음악','체육']]", "_____no_output_____" ], [ "df.iloc[[0,1],[2,3]]", "_____no_output_____" ], [ "#슬라이싱으로 동일하게\ndf.loc['서준':'우현', '음악':'체육']", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "df.iloc[0:2, 2:]", "_____no_output_____" ], [ "#예제 1-12 열 추가\ndf", "_____no_output_____" ], [ "df['국어']=80\ndf", "_____no_output_____" ], [ "#예제 1-13 행추가, 기존 행 인덱스와 겹칠경우, 기존행의 원소값을 변경\ndf.loc[3]=0\ndf", "_____no_output_____" ], [ "#새로운 행 추가- 원소값 여러 개의 배열 입력\ndf.loc[4] = ['동규', 90, 80, 70, 60]\ndf", "_____no_output_____" ], [ "#새로운 행 추가- 기존 행 복사\ndf.loc['행5'] = df.loc[3]\ndf", "_____no_output_____" ], [ "#예제 1-14 원소 값 변경\n# -*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#DataFrame() 함수로 데이터프레임 변환, 변수df에 저장\nexam_data = {\n '이름':['서준','우현','인아'],\n '수학':[90,80,70],\n '영어':[98,89,95],\n '음악':[85,95,100],\n '체육':[100,90,90]\n}\n\ndf = pd.DataFrame(exam_data)\ndf", "_____no_output_____" ], [ "#'이름' 열을 새로운 인덱스로 지정하고, df 객체에 변경사항 반영\ndf.set_index('이름', inplace=True)\ndf", "_____no_output_____" ], [ "#데이터프레임 df의 특정 원소를 변경하는 방법: '서준'의 '체육' 점수\ndf.iloc[0,3] =60\ndf", "_____no_output_____" ], [ "df.loc['서준','체육']=100\ndf", "_____no_output_____" ], [ "df.loc['서준']['체육'] = 80\ndf", "_____no_output_____" ], [ "#예제 1-14 여러개의 원소를 선택해 새로운 값을 할당\ndf.loc['서준',['음악','체육']] = 50\ndf", "_____no_output_____" ], [ "df.loc['서준']['음악','체육']=100\ndf", "_____no_output_____" ], [ "#예제 1-15 행, 열 바꾸기\n#-*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#DataFrame() 함수로 데이터프레임 변환, 변수df에 저장\nexam_data = {\n '이름':['서준','우현','인아'],\n '수학':[90,80,70],\n '영어':[98,89,95],\n '음악':[85,95,100],\n '체육':[100,90,90]\n}\n\ndf = pd.DataFrame(exam_data)\ndf", "_____no_output_____" ], [ "#데이터프레임 df를 전치하기(메소드 활용)\ndf = df.transpose()\ndf", "_____no_output_____" ], [ "#데이터프레임 df를 다시 전치하기(클래스 속성 활용)\ndf = df.T\ndf", "_____no_output_____" ], [ "#예제 1-16 특정 열을 행 인덱스로 설정\ndf", "_____no_output_____" ], [ "#특정 열(column)을 데이터프레임의 행 인덱스(index)로 설정\nndf = df.set_index(['이름'])\nndf", "_____no_output_____" ], [ "ndf2 = ndf.set_index(['음악'])\nndf2", "_____no_output_____" ], [ "ndf3 = ndf.set_index(['수학', '음악'])\nndf3", "_____no_output_____" ], [ "#예제 1-17 새로운 배열로 행 인덱스 재지정\n#-*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#딕셔너리 정의\ndict_data = {'c0':[1,2,3], 'c1':[4,5,6], 'c2':[7,8,9],'c3':[10,11,12], 'c4':[13,14,15]}\n\n#딕셔너리를 데이터프레임으로 변환, 익덱스를 [r0,r1,r2]로 지정\ndf = pd.DataFrame(dict_data, index=['r0','r1','r2'])\ndf", "_____no_output_____" ], [ "#인덱스를 [r0,r1,r2,r3,r4]로 재지정\nnew_index = ['r0','r1','r2','r3','r4']\nndf = df.reindex(['r0','r1','r2','r3','r4'])\nndf", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "new_index = ['r0','r1','r2','r3','r4']\nndf = df.reindex(new_index)\nndf", "_____no_output_____" ], [ "#reindex로 발생한 NaN값을 숫자 0으로 채우기\nnew_index = ['r0','r1','r2','r3','r4']\nndf = df.reindex(new_index, fill_value=0)\nndf", "_____no_output_____" ], [ "#예제 1-18 정수형 위치 인덱스로 초기화\n#-*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#딕셔너리 정의\ndict_data = {'c0':[1,2,3], 'c1':[4,5,6], 'c2':[7,8,9],'c3':[10,11,12], 'c4':[13,14,15]}\n\n#딕셔너리를 데이터프레임으로 변환, 익덱스를 [r0,r1,r2]로 지정\ndf = pd.DataFrame(dict_data, index=['r0','r1','r2'])\ndf", "_____no_output_____" ], [ "#행 인덱스를 정수형으로 초기화\nndf = df.reset_index()\nndf", "_____no_output_____" ], [ "#예제 1-19 데이터프레임 정렬\n#-*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#딕셔너리 정의\ndict_data = {'c0':[1,2,3], 'c1':[4,5,6], 'c2':[7,8,9],'c3':[10,11,12], 'c4':[13,14,15]}\n\n#딕셔너리를 데이터프레임으로 변환, 익덱스를 [r0,r1,r2]로 지정\ndf = pd.DataFrame(dict_data, index=['r0','r1','r2'])\ndf", "_____no_output_____" ], [ "#내림차순으로 행 인덱스 정렬\nndf = df.sort_index(ascending=False)\nndf", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "#c1열을 기준으로 내림차순 정렬\nndf = df.sort_values(by='c1', ascending=False)\nndf", "_____no_output_____" ], [ "#예제 1-21 시리즈를 숫자로 나누기\n# -*- coding: utf-8 -*-\n\n#라이브러리 불러오기\nimport pandas as pd\n\n#딕셔너리 데이터로 판다스 시리즈 만들기\nstudent1 = pd.Series({'국어':100,'영어': 80, '수학':90})\nstudent1", "_____no_output_____" ], [ "#학생의 과목별 점수를 100으로 나누기\npercentage = student1/100\npercentage", "_____no_output_____" ], [ "#예제 1-22 시리즈 사칙연산\n# -*- coding: utf-8 -*-\n\n#라이브러리 불러오기\nimport pandas as pd\n\n#딕셔너리 데이터로 판다스 시리즈 만들기\nstudent1 = pd.Series({'국어':100, '영어':80, '수학':90})\nstudent2 = pd.Series({'수학':80,'국어':90,'영어':80})\nprint(student1)\nprint(student2)", "국어 100\n영어 80\n수학 90\ndtype: int64\n수학 80\n국어 90\n영어 80\ndtype: int64\n" ], [ "#두 학생의 과목별 점수로 사칙연산 수행\nadd = student1 + student2\nsub = student1 - student2\nmul = student1 * student2\ndiv = student1 / student2\nprint(type(div))", "<class 'pandas.core.series.Series'>\n" ], [ "#사칙연산 결과를 데이터 프레임으로 합치기(시리즈 > 데이터프레임)\n\nresult = pd.DataFrame([add, sub, mul, div],\n index=['덧셈','뺄셈','곱셈','나눗셈'])\nresult", "_____no_output_____" ], [ "#예제 1-23 NaN 값이 있는 시리즈 연산\n# -*- coding: utf-8 -*-\n\n#라이브러리 불러오기\nimport pandas as pd\nimport numpy as np\n\n#딕셔너리 데이터로 판다스 시리즈 만들기\nstudent1 = pd.Series({'국어':np.nan, '영어':80, '수학':90})\nstudent2 = pd.Series({'수학':80, '국어':90})\n\nprint(student1)\nprint()\nprint(student2)", "국어 NaN\n영어 80.0\n수학 90.0\ndtype: float64\n\n수학 80\n국어 90\ndtype: int64\n" ], [ "#두 학생의 과목별 점수로 사칙연산 수행(시리즈 vs 시리즈)\nadd = student1 + student2\nsub = student1 - student2\nmul = student1 * student2\ndiv = student1 / student2", "_____no_output_____" ], [ "#사칙연산 결과를 데이터프레임으로 합치기\nresult = pd.DataFrame([add, sub, mul, div],\n index=['덧셈','뺄셈','곱샘','나눗셈'])\nresult", "_____no_output_____" ], [ "#예제 1-24 연산 메소드 사용- 시리즈 연산\n# -*- coding: utf-8 -*-\n\n#라이브러리 불러오기\nimport pandas as pd\nimport numpy as np\n\n#딕셔너리 데이터로 판다스 시리즈 만들기\nstudent1 = pd.Series({'국어':np.nan, '영어':80, '수학':90})\nstudent2 = pd.Series({'수학':80, '국어':90})\n\nprint(student1)\nprint()\nprint(student2)\n", "국어 NaN\n영어 80.0\n수학 90.0\ndtype: float64\n\n수학 80\n국어 90\ndtype: int64\n" ], [ "#두 학생의 과목별 점수로 사칙연산 수행(연산 메소드 사용)\nsr_add = student1.add(student2, fill_value=0)\nsr_sub = student1.sub(student2, fill_value=0)\nsr_mul = student1.mul(student2, fill_value=0)\nsr_div = student1.div(student2, fill_value=0)", "_____no_output_____" ], [ "#사칙연산 결과를 데이터 프레임으로 합치기\nresult = pd.DataFrame([sr_add, sr_sub, sr_mul, sr_div],\n index =['덧셈','뺄셈','곱셈','나눗셈'])\nprint(result)", " 국어 수학 영어\n덧셈 90.0 170.000 80.0\n뺄셈 -90.0 10.000 80.0\n곱셈 0.0 7200.000 0.0\n나눗셈 0.0 1.125 inf\n" ], [ "#예제 1-25 데이터프레임에 숫자 더하기\n# -*- coding: utf-8 -*-\n#라이브러리 불러오기\nimport pandas as pd\nimport seaborn as sns", "_____no_output_____" ], [ "#titanic 데이터셋에서 age, fare 2개 열을 선택하여 데이터프레임 만들기\ntitanic = sns.load_dataset('titanic')\ndf = titanic.loc[:, ['age','fare']]\nprint(df.head()) #첫 5행만 표시\nprint()\nprint(type(df))", " age fare\n0 22.0 7.2500\n1 38.0 71.2833\n2 26.0 7.9250\n3 35.0 53.1000\n4 35.0 8.0500\n\n<class 'pandas.core.frame.DataFrame'>\n" ], [ "#데이터프레임에 숫자 10 더하기\naddition = df + 10\nprint(addition.head())", " age fare\n0 32.0 17.2500\n1 48.0 81.2833\n2 36.0 17.9250\n3 45.0 63.1000\n4 45.0 18.0500\n" ], [ "#예제 1-26 데이터프레임끼리 더하기\n# -*- coding: utf-8 -*-\n\n#라이브러리 불러오기\nimport pandas as pd\nimport seaborn as sns\n\n#titanic 데이터셋에서 age, fare 2개 열을 선택하여 데이터프레임 만들기\ntitanic = sns.load_dataset('titanic')\ndf =titanic.loc[:,['age','fare']]\nprint(df.tail()) #마지막 5행 표시\nprint()\nprint(type(df))", " age fare\n886 27.0 13.00\n887 19.0 30.00\n888 NaN 23.45\n889 26.0 30.00\n890 32.0 7.75\n\n<class 'pandas.core.frame.DataFrame'>\n" ], [ "#데이터프레임에 숫자 10 더하기\naddition = df + 10\nprint(addition.tail())", " age fare\n886 37.0 23.00\n887 29.0 40.00\n888 NaN 33.45\n889 36.0 40.00\n890 42.0 17.75\n" ], [ "#데이터프레임끼리 연산하기(addition-df)\nsubtraction = addition - df\nprint(subtraction.tail())", " age fare\n886 10.0 10.0\n887 10.0 10.0\n888 NaN 10.0\n889 10.0 10.0\n890 10.0 10.0\n" ], [ "#예제 2-1 csv파일 읽기\n# -*- coding: utf-8 -*-\n\n#라이브러리 불러오기\nimport pandas as pd\n\n#파일 경로(파이썬 파일과 같은 폴더)를 찾고, 변수 file_path에 저장\nfile_path = './read_csv_sample.csv'", "_____no_output_____" ], [ "#read_csv()함수로 데이터프레임 변환, 변수 df1에 저장\ndf1 = pd.read_csv(file_path)\ndf1", "_____no_output_____" ], [ "#read_csv() 함수로 데이터 프레임변환, 변수df2 에 저장\n#header=None 옵션\ndf2=pd.read_csv(file_path, header=None)\ndf2", "_____no_output_____" ], [ "#read_csv()함수로 데이터프레임 변환, df3에 저장 \n#index_col=None 옵션\ndf3 = pd.read_csv(file_path, index_col=None)\ndf3", "_____no_output_____" ], [ "#read_csv()함수로 데이터프레임 변환, 변수 df4에 저장\n#index_col='c0'옵션\ndf4 = pd.read_csv(file_path, index_col='c0')\nprint(df4)", " c1 c2 c3\nc0 \n0 1 4 7\n1 2 5 8\n2 3 6 9\n" ], [ "#예제 2-2 Excel 파일 읽기\n# -*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#read_excel() 함수로 데이터프레임 변환\ndf1 = pd.read_excel('./남북한발전전력량.xlsx')\ndf2 = pd.read_excel('./남북한발전전력량.xlsx', header=None)\n\nprint(df1)\nprint('\\n')\nprint(df2)", " 전력량 (억㎾h) 발전 전력별 1990 1991 1992 1993 1994 1995 1996 1997 ... 2007 \\\n0 남한 합계 1077 1186 1310 1444 1650 1847 2055 2244 ... 4031 \n1 NaN 수력 64 51 49 60 41 55 52 54 ... 50 \n2 NaN 화력 484 573 696 803 1022 1122 1264 1420 ... 2551 \n3 NaN 원자력 529 563 565 581 587 670 739 771 ... 1429 \n4 NaN 신재생 - - - - - - - - ... - \n5 북한 합계 277 263 247 221 231 230 213 193 ... 236 \n6 NaN 수력 156 150 142 133 138 142 125 107 ... 133 \n7 NaN 화력 121 113 105 88 93 88 88 86 ... 103 \n8 NaN 원자력 - - - - - - - - ... - \n\n 2008 2009 2010 2011 2012 2013 2014 2015 2016 \n0 4224 4336 4747 4969 5096 5171 5220 5281 5404 \n1 56 56 65 78 77 84 78 58 66 \n2 2658 2802 3196 3343 3430 3581 3427 3402 3523 \n3 1510 1478 1486 1547 1503 1388 1564 1648 1620 \n4 - - - - 86 118 151 173 195 \n5 255 235 237 211 215 221 216 190 239 \n6 141 125 134 132 135 139 130 100 128 \n7 114 110 103 79 80 82 86 90 111 \n8 - - - - - - - - - \n\n[9 rows x 29 columns]\n\n\n 0 1 2 3 4 5 6 7 8 9 ... \\\n0 전력량 (억㎾h) 발전 전력별 1990 1991 1992 1993 1994 1995 1996 1997 ... \n1 남한 합계 1077 1186 1310 1444 1650 1847 2055 2244 ... \n2 NaN 수력 64 51 49 60 41 55 52 54 ... \n3 NaN 화력 484 573 696 803 1022 1122 1264 1420 ... \n4 NaN 원자력 529 563 565 581 587 670 739 771 ... \n5 NaN 신재생 - - - - - - - - ... \n6 북한 합계 277 263 247 221 231 230 213 193 ... \n7 NaN 수력 156 150 142 133 138 142 125 107 ... \n8 NaN 화력 121 113 105 88 93 88 88 86 ... \n9 NaN 원자력 - - - - - - - - ... \n\n 19 20 21 22 23 24 25 26 27 28 \n0 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 \n1 4031 4224 4336 4747 4969 5096 5171 5220 5281 5404 \n2 50 56 56 65 78 77 84 78 58 66 \n3 2551 2658 2802 3196 3343 3430 3581 3427 3402 3523 \n4 1429 1510 1478 1486 1547 1503 1388 1564 1648 1620 \n5 - - - - - 86 118 151 173 195 \n6 236 255 235 237 211 215 221 216 190 239 \n7 133 141 125 134 132 135 139 130 100 128 \n8 103 114 110 103 79 80 82 86 90 111 \n9 - - - - - - - - - - \n\n[10 rows x 29 columns]\n" ], [ "#예제 2-3 json 파일 읽기\n# -*- coding: utf-8 -*-\n\nimport pandas as pd\n\n#read_json() 함수로 데이터프레임 변환\ndf = pd.read_json('./read_json_sample.json')\nprint(df)\nprint('\\n')\nprint(df.index)", " name year developer opensource\npandas 2008 Wes Mckinneye True\nNumPy 2006 Travis Oliphant True\nmatplotlib 2003 John D. Hunter True\n\n\nIndex(['pandas', 'NumPy', 'matplotlib'], dtype='object')\n" ], [ "#예제 2-4 웹에서 표 정보 읽기\n# -*- coding: utf-8 -*-\nimport pandas as pd\n\n#HTML 파일 경로 or 웹 페이지 주소를 url 변수에 저장\nurl='./sample.html'\n\n#HTML 웹페이지의 표(table) 를 가져와서 데이터프레임으로 변환\ntables = pd.read_html(url)", "_____no_output_____" ], [ "#표(table)의 개수 확인\nprint(len(tables))\nprint('\\n')", "2\n\n\n" ], [ "#tables 리스트의 원소를 iteration 하면서 각각 화면 출력\nfor i in range(len(tables)):\n print(\"tables[%s]\"%i)\n print(tables[i])\n print('\\n')\n ", "tables[0]\n Unnamed: 0 c0 c1 c2 c3\n0 0 0 1 4 7\n1 1 1 2 5 8\n2 2 2 3 6 9\n\n\ntables[1]\n name year developer opensource\n0 NumPy 2006 Travis Oliphant True\n1 matplotlib 2003 John D. Hunter True\n2 pandas 2008 Wes Mckinneye True\n\n\n" ], [ "#파이썬 패키지 정보가 들어 있는 두 번째 데이터프레임을 선택하여 df 변수에 저장\ndf = tables[1]\nprint(df)", " name year developer opensource\n0 NumPy 2006 Travis Oliphant True\n1 matplotlib 2003 John D. Hunter True\n2 pandas 2008 Wes Mckinneye True\n" ], [ "#'name' 열을 인덱스로 지정\ndf.set_index(['name'], inplace=True)\nprint(df)", " year developer opensource\nname \nNumPy 2006 Travis Oliphant True\nmatplotlib 2003 John D. Hunter True\npandas 2008 Wes Mckinneye True\n" ], [ "#예제 2-5 미국 etf 리스트 가져오기\n\n# -*- coding: utf-8 -*-\n\n#라이브러리 불러오기\n\nfrom bs4 import BeautifulSoup\nimport requests\nimport re\nimport pandas as pd", "_____no_output_____" ], [ "#위키피디아 미국 ETF 웹 페이지에서 필요한 정보를 스크래핑하여 딕셔너리 형태로 변수 etfs에 저장\nurl = \"https://en.wikipedia.org/wiki/List_of_American_exchange-traded_funds\"\nresp = requests.get(url)\nsoup = BeautifulSoup(resp.text, 'lxml')\nrows = soup.select('div > ul > li')", "_____no_output_____" ], [ "etfs = {}\nfor row in rows:\n try:\n etf_name = re.findall('^(.*) \\(NYSE', row.text)\n etf_market = re.findall('\\((.*)\\|',row.text)\n etf_ticker = re.findall('NYSE Arca\\|(.*)\\)',row.text)\n \n if (len(etf_ticker) > 0) & (len(etf_market) > 0) & (len(etf_name) > 0):\n etfs[etf_ticker[0]] = [etf_market[0], etf_name[0]]\n \n except AttributeError as err:\n pass\n ", "_____no_output_____" ], [ "#etfs 딕셔너리 출력\nprint(etfs)", "{'DIA': ['NYSE Arca', 'DIAMONDS Trust, Series 1'], 'RSP': ['NYSE Arca', 'Guggenheim S&P 500 Equal Weight'], 'IOO': ['NYSE Arca', 'iShares S&P Global 100 Index'], 'IVV': ['NYSE Arca', 'iShares S&P 500 Index'], 'SPY': ['NYSE Arca', 'SPDR S&P 500'], 'VOO': ['NYSE Arca', 'Vanguard S&P 500'], 'IWM': ['NYSE Arca', 'iShares Russell 2000 Index'], 'OEF': ['NYSE Arca', 'iShares S&P 100 Index'], 'CVY': ['NYSE Arca', 'Guggenheim Multi-Asset Income'], 'RPG': ['NYSE Arca', 'Guggenheim S&P 500 Pure Growth ETF'], 'RPV': ['NYSE Arca', 'Guggenheim S&P 500 Pure Value ETF'], 'IWB': ['NYSE Arca', 'iShares Russell 1000 Index'], 'PKW': ['NYSE Arca', 'PowerShares Buyback Achievers'], 'PRF': ['NYSE Arca', 'PowerShares FTSE RAFI US 1000'], 'SPLV': ['NYSE Arca', 'PowerShares S&P 500 Low Volatility'], 'SCHX': ['NYSE Arca', 'Schwab US Large-Cap ETF'], 'SCHD': ['NYSE Arca', 'Schwab US Dividend Equity ETF'], 'FNDX': ['NYSE Arca', 'Schwab Fundamental U.S. Large Company Index ETF'], 'SDY': ['NYSE Arca', 'SPDR S&P Dividend ETF'], 'VV': ['NYSE Arca', 'Vanguard Large-Cap'], 'MGC': ['NYSE Arca', 'Vanguard Mega-Cap 300'], 'VONE': ['NYSE Arca', 'Vanguard Russell 1000'], 'VIG': ['NYSE Arca', 'Vanguard Dividend Appreciation'], 'VYM': ['NYSE Arca', 'Vanguard High Dividend Yield'], 'DTN': ['NYSE Arca', 'WisdomTree Dividend ex-Financials'], 'DLN': ['NYSE Arca', 'WisdomTree LargeCap Dividend'], 'MDY': ['NYSE Arca', 'MidCap SPDR'], 'DVY': ['NYSE Arca', 'iShares Select Dividend'], 'IWR': ['NYSE Arca', 'iShares Russell Midcap Index'], 'IJH': ['NYSE Arca', 'iShares S&P MidCap 400 Index'], 'PDP': ['NYSE Arca', 'PowerShares DWA Mom Port'], 'SCHM': ['NYSE Arca', 'Schwab US Mid-Cap'], 'IVOO': ['NYSE Arca', 'Vanguard S&P Mid-Cap 400'], 'VO': ['NYSE Arca', 'Vanguard Mid-Cap'], 'VXF': ['NYSE Arca', 'Vanguard Extended Market'], 'DON': ['NYSE Arca', 'WisdomTree MidCap Dividend ETF'], 'IWC': ['NYSE Arca', 'iShares Micro-Cap'], 'IJR': ['NYSE Arca', 'iShares S&P SmallCap 600 Index'], 'SCHA': ['NYSE Arca', 'Schwab US Small-Cap ETF'], 'FNDA': ['NYSE Arca', 'Schwab Fundamental U.S. Small Company Index ETF'], 'VIOO': ['NYSE Arca', 'Vanguard S&P Small-Cap 600'], 'VB': ['NYSE Arca', 'Vanguard Small-Cap'], 'VTWO': ['NYSE Arca', 'Vanguard Russell 2000'], 'EEB': ['NYSE Arca', 'Claymore/BNY BRIC'], 'ECON': ['NYSE Arca', 'EGShares Emerging Markets Consumer'], 'IDV': ['NYSE Arca', 'iShares International Select Div'], 'BKF': ['NYSE Arca', 'iShares MSCI BRIC Index'], 'EFA': ['NYSE Arca', 'iShares MSCI EAFE Index'], 'SCZ': ['NYSE Arca', 'iShares MSCI EAFE Small-Cap'], 'EEM': ['NYSE Arca', 'iShares MSCI Emerging Markets Index'], 'PID': ['NYSE Arca', 'PowerShares Intl Dividend Achievers'], 'SCHC': ['NYSE Arca', 'Schwab International Small-Cap Equity'], 'SCHE': ['NYSE Arca', 'Schwab Emerging Markets Equity ETF'], 'SCHF': ['NYSE Arca', 'Schwab International Equity ETF'], 'FNDF': ['NYSE Arca', 'Schwab Fundamental International Large Company Index ETF'], 'FNDC': ['NYSE Arca', 'Schwab Fundamental International Small Company Index ETF'], 'FNDE': ['NYSE Arca', 'Schwab Fundamental Emerging Markets Large Company Index ETF'], 'DWX': ['NYSE Arca', 'SPDR S&P International Dividend'], 'VEA': ['NYSE Arca', 'Vanguard MSCI EAFE'], 'VWO': ['NYSE Arca', 'Vanguard MSCI Emerging Markets'], 'VXUS': ['NYSE Arca', 'Vanguard Total International Stock'], 'VEU': ['NYSE Arca', 'Vanguard FTSE All-World ex-US'], 'VSS': ['NYSE Arca', 'Vanguard FTSE All-World ex-US Small-Cap'], 'DEM': ['NYSE Arca', 'WisdomTree Emerging Markets Equity Inc'], 'DGS': ['NYSE Arca', 'WisdomTree Emerging Mkts SmallCap Div'], 'EZU': ['NYSE Arca', 'iShares MSCI EMU Index'], 'EPP': ['NYSE Arca', 'iShares MSCI Pacific ex-Japan'], 'IEV': ['NYSE Arca', 'iShares S&P Europe 350 Index'], 'ILF': ['NYSE Arca', 'iShares S&P Latin America 40 Index'], 'FEZ': ['NYSE Arca', 'SPDR EURO STOXX 50'], 'VGK': ['NYSE Arca', 'Vanguard MSCI Europe'], 'VPL': ['NYSE Arca', 'Vanguard MSCI Pacific'], 'HEDJ': ['NYSE Arca', 'WisdomTree Europe Hedged Equity ETF'], 'DFE': ['NYSE Arca', 'WisdomTree Europe SmallCap Dividend'], 'AND': ['NYSE Arca', 'Global X FTSE Andean 40 ETF'], 'GXF': ['NYSE Arca', 'Global X FTSE Nordic Region ETF'], 'EWA': ['NYSE Arca', 'iShares MSCI Australia Index'], 'EWC': ['NYSE Arca', 'iShares MSCI Canada Index'], 'EWG': ['NYSE Arca', 'iShares MSCI German Index'], 'EIS': ['NYSE Arca', 'iShares MSCI Israel ETF'], 'EWI': ['NYSE Arca', 'iShares MSCI Italy Capped'], 'EWJ': ['NYSE Arca', 'iShares MSCI Japan Index'], 'EWY': ['NYSE Arca', 'iShares MSCI Korea Index'], 'EWD': ['NYSE Arca', 'iShares MSCI Sweden Index'], 'EWL': ['NYSE Arca', 'iShares MSCI Switzerland Capped'], 'EWP': ['NYSE Arca', 'iShares MSCI Spain Capped'], 'EWU': ['NYSE Arca', 'iShares MSCI United Kingdom Index'], 'DXJ': ['NYSE Arca', 'WisdomTree Japan Hedged Equity'], 'NORW': ['NYSE Arca', 'Global X FTSE Norway 30 ETF'], 'INDF': ['NYSE Arca', 'Nifty India Financials ETF'], 'EWZ': ['NYSE Arca', 'iShares MSCI Brazil Index'], 'FXI': ['NYSE Arca', 'iShares FTSE/Xinhua China 25 Index'], 'EWH': ['NYSE Arca', 'iShares MSCI Hong Kong Index'], 'EWW': ['NYSE Arca', 'iShares MSCI Mexico Index'], 'EPHE': ['NYSE Arca', 'iShares MSCI Philippines Index'], 'RSX': ['NYSE Arca', 'Market Vectors Russia ETF'], 'EWS': ['NYSE Arca', 'iShares MSCI Singapore Index'], 'EWM': ['NYSE Arca', 'iShares MSCI Malaysia Index'], 'EWT': ['NYSE Arca', 'iShares MSCI Taiwan Index'], 'EPI': ['NYSE Arca', 'WisdomTree India Earnings ETF'], 'ARGT': ['NYSE Arca', 'Global X FTSE Argentina 20 ETF'], 'BRAF': ['NYSE Arca', 'Global X Brazil Financials ETF'], 'BRAQ': ['NYSE Arca', 'Global X Brazil Consumer ETF'], 'BRAZ': ['NYSE Arca', 'Global X Brazil Mid Cap ETF'], 'GXG': ['NYSE Arca', 'Global X FTSE Colombia 20 ETF'], 'XLY': ['NYSE Arca', 'Consumer Discretionary Select Sector SPDR'], 'IYC': ['NYSE Arca', 'iShares Dow Jones US Consumer Services'], 'ITB': ['NYSE Arca', 'iShares US Home Construction'], 'XHB': ['NYSE Arca', 'SPDR S&P Homebuilders ETF'], 'VCR': ['NYSE Arca', 'Vanguard Consumer Discretionary'], 'XLP': ['NYSE Arca', 'Consumer Staples Select Sector SPDR'], 'IYK': ['NYSE Arca', 'iShares Dow Jones US Consumer Goods'], 'VDC': ['NYSE Arca', 'Vanguard Consumer Staples'], 'AMLP': ['NYSE Arca', 'ALPS Alerian MLP ETF'], 'XLE': ['NYSE Arca', 'Energy Select Sector SPDR'], 'IYE': ['NYSE Arca', 'iShares Dow Jones US Energy'], 'IGE': ['NYSE Arca', 'iShares North American Natural Resources'], 'OIH': ['NYSE Arca', 'Market Vectors Oil Services ETF'], 'XOP': ['NYSE Arca', 'SPDR S&P Oil & Gas Explor & Prod ETF'], 'VDE': ['NYSE Arca', 'Vanguard Energy'], 'ESPO': ['NYSE Arca', 'VanEck Vectors Video Gaming and eSports ETF'], 'XLF': ['NYSE Arca', 'Financial Select Sector SPDR'], 'IYF': ['NYSE Arca', 'iShares Dow Jones US Financial'], 'KBE': ['NYSE Arca', 'SPDR S&P Bank ETF'], 'KRE': ['NYSE Arca', 'SPDR S&P Regional Banking ETF'], 'VFH': ['NYSE Arca', 'Vanguard Financials'], 'FXH': ['NYSE Arca', 'First Trust Health Care AlphaDEX'], 'FBT': ['NYSE Arca', 'First Trust NYSE Arca Biotech Index'], 'XLV': ['NYSE Arca', 'Health Care Select Sector SPDR'], 'IYH': ['NYSE Arca', 'iShares Dow Jones US Health Care'], 'PJP': ['NYSE Arca', 'PowerShares Dynamic Pharmaceuticals'], 'XBI': ['NYSE Arca', 'SPDR S&P Biotech ETF'], 'VHT': ['NYSE Arca', 'Vanguard Health Care'], 'XLI': ['NYSE Arca', 'Industrial Select Sector SPDR'], 'IYJ': ['NYSE Arca', 'iShares Dow Jones US Industrial'], 'VIS': ['NYSE Arca', 'Vanguard Industrials'], 'XLB': ['NYSE Arca', 'Materials Select Sector SPDR'], 'IYM': ['NYSE Arca', 'iShares Dow Jones US Materials'], 'GDX': ['NYSE Arca', 'Market Vectors Gold Miners ETF'], 'GDXJ': ['NYSE Arca', 'Market Vectors Junior Gold Miners ETF'], 'VAW': ['NYSE Arca', 'Vanguard Materials'], 'FDN': ['NYSE Arca', 'First Trust Dow Jones Internet Index'], 'XLK': ['NYSE Arca', 'Technology Select Sector SPDR'], 'IYW': ['NYSE Arca', 'iShares Dow Jones US Technology'], 'IGV': ['NYSE Arca', 'iShares North American Tech-Software'], 'VGT': ['NYSE Arca', 'Vanguard Information Technology'], 'IYZ': ['NYSE Arca', 'iShares Dow Jones US Telecommunications'], 'VOX': ['NYSE Arca', 'Vanguard Telecommunication Services'], 'XLU': ['NYSE Arca', 'Utilities Select Sector SPDR'], 'IDU': ['NYSE Arca', 'iShares Dow Jones US Utilities'], 'VPU': ['NYSE Arca', 'Vanguard Utilities'], 'IPD': ['NYSE Arca', 'SPDR S&P International Consumer Discretionary'], 'RXI': ['NYSE Arca', 'iShares S&P Global Consumer Discretionary'], 'IPS': ['NYSE Arca', 'SPDR S&P International Consumer Staples'], 'KXI': ['NYSE Arca', 'iShares S&P Global Consumer Staples'], 'IPW': ['NYSE Arca', 'SPDR S&P International Energy'], 'IXC': ['NYSE Arca', 'iShares S&P Global Energy'], 'IPF': ['NYSE Arca', 'SPDR S&P International Financial'], 'IXG': ['NYSE Arca', 'iShares S&P Global Financial'], 'IRY': ['NYSE Arca', 'SPDR S&P International Health Care'], 'IXJ': ['NYSE Arca', 'iShares S&P Global Healthcare'], 'IPN': ['NYSE Arca', 'SPDR S&P International Industrial'], 'EXI': ['NYSE Arca', 'iShares S&P Global Industrials'], 'GUNR': ['NYSE Arca', 'FlexShares Mstar Gbl Upstrm Nat Res ETF'], 'IRV': ['NYSE Arca', 'SPDR S&P International Materials'], 'MXI': ['NYSE Arca', 'iShares S7P Global Materials'], 'IPK': ['NYSE Arca', 'SPDR S&P International Technology'], 'IXN': ['NYSE Arca', 'iShares S&P Global Technology'], 'IST': ['NYSE Arca', 'SPDR S&P International Telecommunications'], 'IXP': ['NYSE Arca', 'iShares S&P Global Telecommunications'], 'IPU': ['NYSE Arca', 'SPDR S&P International Utilities'], 'JXI': ['NYSE Arca', 'iShares S&P Global Utilities'], 'HYLD': ['NYSE Arca', 'AdvisorShares Peritus High Yield ETF'], 'TDTT': ['NYSE Arca', 'FlexShares iBoxx 3Yr Target Dur TIPS ETF'], 'CSJ': ['NYSE Arca', 'iShares 1-3 Year Credit Bond'], 'IEI': ['NYSE Arca', 'iShares 3-7 Year Treasury Bond'], 'AGG': ['NYSE Arca', 'iShares Core U.S. Aggregate Bond'], 'SHY': ['NYSE Arca', 'iShares Barclays 1-3 Year Treasury Bond'], 'TIP': ['NYSE Arca', 'iShares Barclays TIPS Bond'], 'HYG': ['NYSE Arca', 'iShares iBoxx $ High Yield Corp Bond'], 'LQD': ['NYSE Arca', 'iShares iBoxx $ Invest Grade Corp Bond'], 'IEF': ['NYSE Arca', 'iShares Barclays 7-10 Year Treasury'], 'TLT': ['NYSE Arca', 'iShares Barclays 20+ Year Treas Bond'], 'FLOT': ['NYSE Arca', 'iShares Floating Rate Bond'], 'CIU': ['NYSE Arca', 'iShares Intermediate Credit Bd'], 'GVI': ['NYSE Arca', 'iShares Intm Government/Credit Bond'], 'EMB': ['NYSE Arca', 'iShares JPMorgan USD Emerg Markets Bond'], 'MBB': ['NYSE Arca', 'iShares MBS'], 'MUB': ['NYSE Arca', 'iShares National AMT-Free Muni Bond'], 'SHV': ['NYSE Arca', 'iShares Short Treasury Bond'], 'HYD': ['NYSE Arca', 'Market Vectors High-Yield Muni ETF'], 'HYS': ['NYSE Arca', 'PIMCO 0-5 Year High Yld Corp Bd Idx ETF'], 'STPZ': ['NYSE Arca', 'PIMCO 1-5 Year US TIPS Index ETF'], 'MINT': ['NYSE Arca', 'PIMCO Enhanced Short Duration ETF'], 'BOND': ['NYSE Arca', 'PIMCO Total Return ETF'], 'PCY': ['NYSE Arca', 'PowerShares Emerging Mkts Sovereign Debt'], 'BKLN': ['NYSE Arca', 'PowerShares Senior Loan Port'], 'SCHZ': ['NYSE Arca', 'Schwab US Aggregate Bond'], 'SCHP': ['NYSE Arca', 'Schwab US TIPS'], 'SCHO': ['NYSE Arca', 'Schwab Short-Term US Treasury'], 'SCHR': ['NYSE Arca', 'Schwab Intermediate-Term US Treasury'], 'JNK': ['NYSE Arca', 'SPDR Barclays Capital High Yield Bond ETF'], 'BIL': ['NYSE Arca', 'SPDR Barclays 1-3 Month T-Bill'], 'SCPB': ['NYSE Arca', 'SPDR Barclays Capital Short Term Corp Bd'], 'BWX': ['NYSE Arca', 'SPDR Barclays International Treasury Bd'], 'SJNK': ['NYSE Arca', 'SPDR Barclays Short Term Hi Yld Bd ETF'], 'TFI': ['NYSE Arca', 'SPDR Nuveen Barclays Capital Muni Bond'], 'SHM': ['NYSE Arca', 'SPDR Nuveen Barclays Capital S/T Muni Bd'], 'EDV': ['NYSE Arca', 'Vanguard Extended Duration Treasury'], 'BIV': ['NYSE Arca', 'Vanguard Intermediate-Term Bond'], 'VCIT': ['NYSE Arca', 'Vanguard Intermediate-Term Corporate Bond'], 'VGIT': ['NYSE Arca', 'Vanguard Intermediate-Term Government Bond'], 'BLV': ['NYSE Arca', 'Vanguard Long-Term Bond'], 'VCLT': ['NYSE Arca', 'Vanguard Long-Term Corporate Bond'], 'VGLT': ['NYSE Arca', 'Vanguard Long-Term Government Bond'], 'VMBS': ['NYSE Arca', 'Vanguard Mortgage-Backed Securities'], 'BSV': ['NYSE Arca', 'Vanguard Short-Term Bond'], 'VCSH': ['NYSE Arca', 'Vanguard Short-Term Corporate Bond'], 'VGSH': ['NYSE Arca', 'Vanguard Short-Term Government Bond'], 'BND': ['NYSE Arca', 'Vanguard Total Bond Market'], 'RJI': ['NYSE Arca', 'ELEMENTS Rogers International Commodity Index ETN'], 'DJP': ['NYSE Arca', 'iPath Dow Jones-UBS Commodity Idx TR ETN'], 'GSG': ['NYSE Arca', 'iShares S&P GSCI Commodity-Indexed Trust'], 'DBC': ['NYSE Arca', 'PowerShares DB Commodity Idx Trking Fund'], 'RJA': ['NYSE Arca', 'ELEMENTS Rogers Agriculture ETN'], 'JJA': ['NYSE Arca', 'iPath Dow Jones-UBS Agriculture ETN'], 'DBA': ['NYSE Arca', 'PowerShares DB Agriculture'], 'RJN': ['NYSE Arca', 'ELEMENTS Rogers Energy ETN'], 'OIL': ['NYSE Arca', 'iPath Dow Jones-UBS Crude Oil ETN'], 'GAZ': ['NYSE Arca', 'iPath Dow Jones-UBS Natural Gas ETN'], 'UNG': ['NYSE Arca', 'United States Natural Gas Fund'], 'USO': ['NYSE Arca', 'United States Oil Fund'], 'RJZ': ['NYSE Arca', 'ELEMENTS Rogers Metal ETN'], 'JJM': ['NYSE Arca', 'iPath Dow Jones-UBS Industrial Metals ETN'], 'DBB': ['NYSE Arca', 'PowerShares DB Base Metals'], 'SGOL': ['NYSE Arca', 'ETFS Physical Swiss Gold Shares'], 'IAU': ['NYSE Arca', 'iShares Gold Trust'], 'GLD': ['NYSE Arca', 'SPDR Gold Shares'], 'SIVR': ['NYSE Arca', 'ETFS Physical Silver Shares'], 'SLV': ['NYSE Arca', 'iShares Silver Trust'], 'PALL': ['NYSE Arca', 'ETFS Physical Palladium Shares'], 'PPLT': ['NYSE Arca', 'ETFS Physical Platinum Shares'], 'ICF': ['NYSE Arca', 'iShares Cohen & Steers Realty Majors'], 'IFAS': ['NYSE Arca', 'iShares Dow Jones Asia Real Estate'], 'IFEU': ['NYSE Arca', 'iShares Dow Jones Europe Real Estate'], 'IYR': ['NYSE Arca', 'iShares Dow Jones US Real Estate'], 'REM': ['NYSE Arca', 'iShares Mortgage Real Estate Capped'], 'SCHH': ['NYSE Arca', 'Schwab US REIT'], 'RWO': ['NYSE Arca', 'SPDR Dow Jones Global Real Estate'], 'RWX': ['NYSE Arca', 'SPDR Dow Jones Intl Real Estate'], 'RWR': ['NYSE Arca', 'SPDR Dow Jones REIT ETF'], 'WREI': ['NYSE Arca', 'Wilshire US REIT ETF'], 'VNQ': ['NYSE Arca', 'Vanguard Real Estate'], 'VNQI': ['NYSE Arca', 'Vanguard Global ex-US Real Estate'], 'HDGE': ['NYSE Arca', 'AdvisorShares Ranger Equity Bear ETF'], 'HDGI': ['NYSE Arca', 'AdvisorShares Athena International Bear ETF'], 'DOG': ['NYSE Arca', 'ProShares Short Dow 30'], 'SH': ['NYSE Arca', 'ProShares Short S&P 500'], 'MYY': ['NYSE Arca', 'ProShares Short S&P MidCap 400'], 'SBB': ['NYSE Arca', 'ProShares Short S&P SmallCap 600'], 'PSQ': ['NYSE Arca', 'ProShares Short Nasdaq 100'], 'RWM': ['NYSE Arca', 'ProShares Short Russell 2000'], 'EFZ': ['NYSE Arca', 'ProShares Short MSCI EAFE'], 'FBGX': ['NYSE Arca', 'UBS AG FI Enhanced Large Cap Growth 2x ETF'], 'FLGE': ['NYSE Arca', 'Credit Suisse FI Large Cap Growth Enhanced ETF'], 'MIDU': ['NYSE Arca', 'Direxion Daily Mid Cap Bull 3x ETF'], 'SPUU': ['NYSE Arca', 'Direxion Daily S&P 500 Bull 2x ETF'], 'SPXL': ['NYSE Arca', 'Direxion Daily S&P 500 Bull 3x ETF'], 'ERX': ['NYSE Arca', 'Direxion Energy Bull 3x ETF'], 'FAS': ['NYSE Arca', 'Direxion Financials Bull 3x ETF'], 'BGU': ['NYSE Arca', 'Direxion Large Cap Bull 3x'], 'TNA': ['NYSE Arca', 'Direxion Small Cap Bull 3x'], 'DDM': ['NYSE Arca', 'ProShares Ultra Dow 30'], 'QLD': ['NYSE Arca', 'ProShares Ultra NASDAQ-100'], 'UWM': ['NYSE Arca', 'ProShares Ultra Russell 2000'], 'SSO': ['NYSE Arca', 'ProShares Ultra S&P 500'], 'UPRO': ['NYSE Arca', 'ProShares S&P 500 3x'], 'SDS': ['NYSE Arca', 'UltraShort S&P 500 ProShares 2x'], 'SPXU': ['NYSE Arca', 'ProShares S&P 500 Direxionshares Bear 3x ETF'], 'TZA': ['NYSE Arca', 'Direxion Russell 2000 Direxionshares Bear 3x ETF'], 'SQQQ': ['NYSE Arca', 'UltraPro Short QQQ'], 'QID': ['NYSE Arca', 'UltraShort NASDAQ-100 ProShares 2X'], 'SKF': ['NYSE Arca', 'UltraShort Financials ProShares'], 'TWM': ['NYSE Arca', 'UltraShort Russell 2000 ProShares'], 'DXD': ['NYSE Arca', 'UltraShort Dow 30 ProShares'], 'SRS': ['NYSE Arca', 'UltraShort Real Estate ProShares'], 'MZZ': ['NYSE Arca', 'UltraShort MidCap 400 ProShares'], 'DUG': ['NYSE Arca', 'UltraShort Oil & Gas ProShares'], 'BGZ': ['NYSE Arca', 'Direxion Large Cap Bear 3x ETF'], 'ERY': ['NYSE Arca', 'Direxion Energy Bear 3x ETF'], 'FAZ': ['NYSE Arca', 'Direxion Financials Bear 3x ETF'], 'AADR': ['NYSE Arca', 'AdvisorShares WCM/BNY Mellon Focused Growth ADR ETF'], 'ACCU': ['NYSE Arca', 'AdvisorShares Accuvest Global Opportunities ETF'], 'DBIZ': ['NYSE Arca', 'AdvisorShares Pring Turner Business Cycle ETF'], 'EPRO': ['NYSE Arca', 'AdvisorShares EquityPro ETF'], 'FWDB': ['NYSE Arca', 'AdvisorShares Madrona Global Bond ETF'], 'FWDD': ['NYSE Arca', 'AdvisorShares Madrona Domestic ETF'], 'FWDI': ['NYSE Arca', 'AdvisorShares Madrona International ETF'], 'GEUR': ['NYSE Arca', 'AdvisorShares Gartman Gold/Euro ETF'], 'GGBP': ['NYSE Arca', 'AdvisorShares Gartman Gold/British Pound ETF'], 'GIVE': ['NYSE Arca', 'AdvisorShares Global Echo ETF'], 'GLDE': ['NYSE Arca', 'AdvisorShares International Gold ETF'], 'GYEN': ['NYSE Arca', 'AdvisorShares Gartman Gold/Yen ETF'], 'GTAA': ['NYSE Arca', 'AdvisorShares Cambria Global Tactical ETF'], 'HOLD': ['NYSE Arca', 'AdvisorShares Sage Core Reserves ETF'], 'MATH': ['NYSE Arca', 'AdvisorShares Meidell Tactical Advantage ETF'], 'MINC': ['NYSE Arca', 'AdvisorShares Newfleet Multi-Sector Income ETF'], 'QEH': ['NYSE Arca', 'AdvisorShares QAM Equity Hedge ETF'], 'TTFS': ['NYSE Arca', 'AdvisorShares TrimTabs Float Shrink ETF'], 'VEGA': ['NYSE Arca', 'AdvisorShares STAR Global Buy-Write ETF'], 'YPRO': ['NYSE Arca', 'AdvisorShares YieldPro ETF'], 'RIGS': ['NYSE Arca', 'Riverfront Strategic Income Fund'], 'ARKG': ['NYSE Arca', 'ARK Genomic Revolution Multi-Sector ETF'], 'ARKQ': ['NYSE Arca', 'ARK Industrial Innovation ETF'], 'ARKW': ['NYSE Arca', 'ARK Web x.0 ETF'], 'ARKK': ['NYSE Arca', 'ARK Innovation ETF'], 'SYLD': ['NYSE Arca', 'Cambria Shareholder Yield ETF'], 'GMMB': ['NYSE Arca', 'Columbia Intermediate Municipal Bond ETF'], 'GMTB': ['NYSE Arca', 'Columbia Core Bond ETF'], 'GVT': ['NYSE Arca', 'Columbia Select Large Cap Value ETF'], 'RPX': ['NYSE Arca', 'Columbia Large Cap Growth ETF'], 'RWG': ['NYSE Arca', 'Columbia Select Large Cap Growth ETF'], 'EMLP': ['NYSE Arca', 'First Trust North American Energy Infrastructure Fund'], 'FMB': ['NYSE Arca', 'First Trust Managed Municipal ETF'], 'FMF': ['NYSE Arca', 'First Trust Morningstar Managed Futures Strategy Fund'], 'FPE': ['NYSE Arca', 'First Trust Preferred Securities and Income ETF'], 'FTGS': ['NYSE Arca', 'First Trust Global Tactical Commodity Strategy Fund'], 'FTHI': ['NYSE Arca', 'First Trust High Income ETF'], 'FTLB': ['NYSE Arca', 'First Trust Low Beta Income ETF'], 'FTSL': ['NYSE Arca', 'First Trust Senior Loan ETF'], 'HYLS': ['NYSE Arca', 'First Trust Tactical High Yield ETF'], 'RAVI': ['NYSE Arca', 'Flexshares Ready Access Variable Income Fund'], 'FTSD': ['NYSE Arca', 'Franklin Short Duration US Government ETF'], 'GSY': ['NYSE Arca', 'Guggenheim Enhanced Short Duration Bond ETF'], 'HECO': ['NYSE Arca', 'Huntington EcoLogical Strategy ETF'], 'HUSE': ['NYSE Arca', 'Huntington U.S. Equity Rotation Strategy ETF'], 'ICSH': ['NYSE Arca', 'iShares Liquidity Income ETF'], 'IEIL': ['NYSE Arca', 'iShares Enhanced International Large-Cap ETF'], 'IEIS': ['NYSE Arca', 'iShares Enhanced International Small-Cap ETF'], 'IELG': ['NYSE Arca', 'iShares Enhanced U.S. Large-Cap ETF'], 'IESM': ['NYSE Arca', 'iShares Enhanced U.S. Small-Cap ETF'], 'NEAR': ['NYSE Arca', 'iShares Short Maturity Bond ETF'], 'BABZ': ['NYSE Arca', 'PIMCO Build America Bond Strategy'], 'DI': ['NYSE Arca', 'PIMCO Diversified Income ETF'], 'FORX': ['NYSE Arca', 'PIMCO Foreign Currency Strategy ETF'], 'ILB': ['NYSE Arca', 'PIMCO Global Advantage Inflation-Linked Bond Strategy'], 'LDUR': ['NYSE Arca', 'PIMCO Low Duration ETF'], 'MUNI': ['NYSE Arca', 'PIMCO Intermediate Muni Bond Strategy ETF'], 'SMMU': ['NYSE Arca', 'PIMCO Short Term Muni Bond Strategy ETF'], 'CHNA': ['NYSE Arca', 'PowerShares China-A Share Portfolio'], 'LALT': ['NYSE Arca', 'PowerShares Multi-Strategy Alternative Portfolio'], 'PHDG': ['NYSE Arca', 'S&P 500 Downside Hedged Portfolio'], 'PSR': ['NYSE Arca', 'Active U.S. Real Estate Fund ETF'], 'ONEF': ['NYSE Arca', 'Russell Equity ETF'], 'GAL': ['NYSE Arca', 'SPDR SSgA Global Allocation'], 'INKM': ['NYSE Arca', 'SPDR SSgA Income Allocation'], 'RLY': ['NYSE Arca', 'SPDR SSgA Multi-Asset Real Return'], 'SYE': ['NYSE Arca', 'SPDR MFS Systematic Core Equity ETF'], 'SYG': ['NYSE Arca', 'SPDR MFS Systematic Growth Equity ETF'], 'SYV': ['NYSE Arca', 'SPDR MFS Systematic Value Equity ETF'], 'SRLN': ['NYSE Arca', 'SPDR Blackstone/GSO Senior Loan ETF'], 'ULST': ['NYSE Arca', 'SPDR SSgA Ultra Short Term Bond ETF'], 'ALD': ['NYSE Arca', 'WisdomTree Asia Local Debt'], 'AUNZ': ['NYSE Arca', 'WisdomTree Australia & New Zealand Debt Fund'], 'BZF': ['NYSE Arca', 'WisdomTree Dreyfus Brazilian Real Fund'], 'CCX': ['NYSE Arca', 'WisdomTree Commodity Currency'], 'CEW': ['NYSE Arca', 'WisdomTree Dreyfus Emerging Currency'], 'CRDT': ['NYSE Arca', 'WisdomTree Strategic Corporate Bond Fund'], 'CYB': ['NYSE Arca', 'WisdomTree Dreyfus Chinese Yuan'], 'ELD': ['NYSE Arca', 'WisdomTree Emerging Markets Local Debts Fund'], 'EMCB': ['NYSE Arca', 'WisdomTree Emerging Markets Corporate Bond Fund'], 'EU': ['NYSE Arca', 'WisdomTree Euro Debt Fund'], 'ICB': ['NYSE Arca', 'WisdomTree Dreyfus Indian Rupee'], 'RRF': ['NYSE Arca', 'WisdomTree Global Real Return'], 'USDU': ['NYSE Arca', 'WisdomTree Bloomberg U.S. Dollar Bullish Fund'], 'WDTI': ['NYSE Arca', 'WisdomTree Managed Futures Strategy Fund']}\n" ], [ "#etfs 딕셔너리를 데이터 프레임으로 변환\ndf = pd.DataFrame(etfs)\ndf", "_____no_output_____" ], [ "#예제 2-7 csv 파일로 저장\n# -*- coding: utf-8 -*-\nimport pandas as pd\n\n#판다스 DataFrame() 함수로 데이터프레임 변환. df에 저장\ndata = {\n 'name':['Jerry', 'Riah', 'Paul'],\n 'algol' :[\"A\", \"A+\", \"B\"],\n 'basic': [\"C\", \"B\", \"B+\"],\n 'c++' :[\"B+\", \"C\", \"C+\"]\n}\ndf = pd.DataFrame(data)\ndf.set_index('name', inplace=True) #name 열을 인덱스로 지정\nprint(df)", " algol basic c++\nname \nJerry A C B+\nRiah A+ B C\nPaul B B+ C+\n" ], [ "#to_csv() 메소드를 사용하여 csv파일로 내보내기.\ndf.to_csv(\"./df_sample.csv\")", "_____no_output_____" ], [ "#to_json() 메소드를 사용하여 json 파일로 내보내기\ndf.to_json(\"./df_sample.json\")", "_____no_output_____" ], [ "#to_excel() 메소드를 사용하여 excel 파일로 내보내기\ndf.to_excel(\"./df_sample.xlsx\")", "_____no_output_____" ], [ "#예제 2-10 ExcelWriter() 활용 ExcelWriter 함수는 엑셀 워크북 객체를 생성\n#파일이라고 보면된다. \n#시트 1, 2 만들기\n\n# -*- coding: utf-8 -*-\nimport pandas as pd\n\n#판다스 DataFrame() 함수로 데이터프레임 변환. 변수 df1,df2에 저장\ndata1 = {\n 'name':['Jerry', 'Riah', 'Paul'],\n 'algol' :[\"A\", \"A+\", \"B\"],\n 'basic': [\"C\", \"B\", \"B+\"],\n 'c++' :[\"B+\", \"C\", \"C+\"]\n}\n\ndata2 = {\n 'c0':[1,2,3],\n 'c1':[4,5,6],\n 'c2':[7,8,9],\n 'c3':[10,11,12],\n 'c4':[13,14,15]\n}\n\ndf1 = pd.DataFrame(data1)\ndf1.set_index('name',inplace=True) # name 열을 인덱스로 지정\nprint(df1)", " algol basic c++\nname \nJerry A C B+\nRiah A+ B C\nPaul B B+ C+\n" ], [ "df2 = pd.DataFrame(data2)\ndf2.set_index('c0', inplace=True) #c0 열을 인덱스로 지정\nprint(df2)", " c1 c2 c3 c4\nc0 \n1 4 7 10 13\n2 5 8 11 14\n3 6 9 12 15\n" ], [ "#df1을 'sheet1'으로, df2를 'sheet2'로 저장(Excel 파일명은 \"df_excelwriter.xlsx\")\nwriter = pd.ExcelWriter(\"./df_excelwriter.xlsx\")\ndf1.to_excel(writer, sheet_name=\"sheet1\")\ndf2.to_excel(writer, sheet_name=\"sheet2\")\nwriter.save()", "_____no_output_____" ], [ "#예제 3-1 데이터 살펴보기\n# -*- coding: utf-8 -*-\nimport pandas as pd\n\n#read_csv() 함수로 df 생성\ndf = pd.read_csv('./auto-mpg.csv', header=None)\ndf.head()\n\n#열 이름 지정\ndf.columns = ['mpg','cylinders', 'displacement','horsepower','weight',\n 'acceleration', 'model year', 'origin', 'name']\n\n#데이터프레임 df 의 내용을 일부 확인\nprint(df.head())", " mpg cylinders displacement horsepower weight acceleration model year \\\n0 18.0 8 307.0 130.0 3504.0 12.0 70 \n1 15.0 8 350.0 165.0 3693.0 11.5 70 \n2 18.0 8 318.0 150.0 3436.0 11.0 70 \n3 16.0 8 304.0 150.0 3433.0 12.0 70 \n4 17.0 8 302.0 140.0 3449.0 10.5 70 \n\n origin name \n0 1 chevrolet chevelle malibu \n1 1 buick skylark 320 \n2 1 plymouth satellite \n3 1 amc rebel sst \n4 1 ford torino \n" ], [ "df.tail()", "_____no_output_____" ], [ "#df의 모양과 크기 확인: (행의 개수, 열의 개수)를 튜플로 반환\nprint(df.shape)", "(398, 9)\n" ], [ "#데이터프레임 df의 내용 확인\nprint(df.info())", "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 398 entries, 0 to 397\nData columns (total 9 columns):\n # Column Non-Null Count Dtype \n--- ------ -------------- ----- \n 0 mpg 398 non-null float64\n 1 cylinders 398 non-null int64 \n 2 displacement 398 non-null float64\n 3 horsepower 398 non-null object \n 4 weight 398 non-null float64\n 5 acceleration 398 non-null float64\n 6 model year 398 non-null int64 \n 7 origin 398 non-null int64 \n 8 name 398 non-null object \ndtypes: float64(4), int64(3), object(2)\nmemory usage: 28.1+ KB\nNone\n" ], [ "#데이터프레임 df의 자료형 확인\nprint(df.dtypes)", "mpg float64\ncylinders int64\ndisplacement float64\nhorsepower object\nweight float64\nacceleration float64\nmodel year int64\norigin int64\nname object\ndtype: object\n" ], [ "#시리즈 (mpg 열)의 자료형 확인\nprint(df.mpg.dtypes)", "float64\n" ], [ "#데이터프레임 df의 기술 통계정보 확인\nprint(df.describe())", " mpg cylinders displacement weight acceleration \\\ncount 398.000000 398.000000 398.000000 398.000000 398.000000 \nmean 23.514573 5.454774 193.425879 2970.424623 15.568090 \nstd 7.815984 1.701004 104.269838 846.841774 2.757689 \nmin 9.000000 3.000000 68.000000 1613.000000 8.000000 \n25% 17.500000 4.000000 104.250000 2223.750000 13.825000 \n50% 23.000000 4.000000 148.500000 2803.500000 15.500000 \n75% 29.000000 8.000000 262.000000 3608.000000 17.175000 \nmax 46.600000 8.000000 455.000000 5140.000000 24.800000 \n\n model year origin \ncount 398.000000 398.000000 \nmean 76.010050 1.572864 \nstd 3.697627 0.802055 \nmin 70.000000 1.000000 \n25% 73.000000 1.000000 \n50% 76.000000 1.000000 \n75% 79.000000 2.000000 \nmax 82.000000 3.000000 \n" ], [ "print(df.describe(include='all')) \n#산술 데이터가 아닌 정보까지 포함 include='all'", " mpg cylinders displacement horsepower weight \\\ncount 398.000000 398.000000 398.000000 398 398.000000 \nunique NaN NaN NaN 94 NaN \ntop NaN NaN NaN 150.0 NaN \nfreq NaN NaN NaN 22 NaN \nmean 23.514573 5.454774 193.425879 NaN 2970.424623 \nstd 7.815984 1.701004 104.269838 NaN 846.841774 \nmin 9.000000 3.000000 68.000000 NaN 1613.000000 \n25% 17.500000 4.000000 104.250000 NaN 2223.750000 \n50% 23.000000 4.000000 148.500000 NaN 2803.500000 \n75% 29.000000 8.000000 262.000000 NaN 3608.000000 \nmax 46.600000 8.000000 455.000000 NaN 5140.000000 \n\n acceleration model year origin name \ncount 398.000000 398.000000 398.000000 398 \nunique NaN NaN NaN 305 \ntop NaN NaN NaN ford pinto \nfreq NaN NaN NaN 6 \nmean 15.568090 76.010050 1.572864 NaN \nstd 2.757689 3.697627 0.802055 NaN \nmin 8.000000 70.000000 1.000000 NaN \n25% 13.825000 73.000000 1.000000 NaN \n50% 15.500000 76.000000 1.000000 NaN \n75% 17.175000 79.000000 2.000000 NaN \nmax 24.800000 82.000000 3.000000 NaN \n" ], [ "#데이터 개수확인\ndf", "_____no_output_____" ], [ "#데이터프레임 df의 각 열이 가지고 있는 원소 개수 확인\nprint(df.count())", "mpg 398\ncylinders 398\ndisplacement 398\nhorsepower 398\nweight 398\nacceleration 398\nmodel year 398\norigin 398\nname 398\ndtype: int64\n" ], [ "#df.count() 가 반환하는 객체 타입 출력\nprint(type(df.count()))", "<class 'pandas.core.series.Series'>\n" ], [ "#데이터 개수 확인\n#데이터프레임 df의 특정 열이 가지고 있는 고유값 확인\nunique_values = df['origin'].value_counts()\nprint(unique_values)\nprint()", "1 249\n3 79\n2 70\nName: origin, dtype: int64\n\n" ], [ "#평균값\nprint(df.mean())", "mpg 23.514573\ncylinders 5.454774\ndisplacement 193.425879\nweight 2970.424623\nacceleration 15.568090\nmodel year 76.010050\norigin 1.572864\ndtype: float64\n" ], [ "print(df['mpg'].mean())", "23.514572864321615\n" ], [ "print(df[['mpg','weight']].mean())", "mpg 23.514573\nweight 2970.424623\ndtype: float64\n" ], [ "#통계함수\n#중간값\nprint(df.median())", "mpg 23.0\ncylinders 4.0\ndisplacement 148.5\nweight 2803.5\nacceleration 15.5\nmodel year 76.0\norigin 1.0\ndtype: float64\n" ], [ "print(df['mpg'].median())", "23.0\n" ], [ "#최대값\nprint(df.max())", "mpg 46.6\ncylinders 8\ndisplacement 455\nhorsepower ?\nweight 5140\nacceleration 24.8\nmodel year 82\norigin 3\nname vw rabbit custom\ndtype: object\n" ], [ "print(df['mpg'].max())", "46.6\n" ], [ "#최소값\nprint(df.min())", "mpg 9\ncylinders 3\ndisplacement 68\nhorsepower 100.0\nweight 1613\nacceleration 8\nmodel year 70\norigin 1\nname amc ambassador brougham\ndtype: object\n" ], [ "print(df['mpg'].min())", "9.0\n" ], [ "#표준편차\nprint(df.std())", "mpg 7.815984\ncylinders 1.701004\ndisplacement 104.269838\nweight 846.841774\nacceleration 2.757689\nmodel year 3.697627\norigin 0.802055\ndtype: float64\n" ], [ "print(df['mpg'].std())", "7.815984312565782\n" ], [ "#상관계수\nprint(df.corr())\n", " mpg cylinders displacement weight acceleration \\\nmpg 1.000000 -0.775396 -0.804203 -0.831741 0.420289 \ncylinders -0.775396 1.000000 0.950721 0.896017 -0.505419 \ndisplacement -0.804203 0.950721 1.000000 0.932824 -0.543684 \nweight -0.831741 0.896017 0.932824 1.000000 -0.417457 \nacceleration 0.420289 -0.505419 -0.543684 -0.417457 1.000000 \nmodel year 0.579267 -0.348746 -0.370164 -0.306564 0.288137 \norigin 0.563450 -0.562543 -0.609409 -0.581024 0.205873 \n\n model year origin \nmpg 0.579267 0.563450 \ncylinders -0.348746 -0.562543 \ndisplacement -0.370164 -0.609409 \nweight -0.306564 -0.581024 \nacceleration 0.288137 0.205873 \nmodel year 1.000000 0.180662 \norigin 0.180662 1.000000 \n" ], [ "print(df[['mpg', 'weight']].corr())", " mpg weight\nmpg 1.000000 -0.831741\nweight -0.831741 1.000000\n" ], [ "#예제 3-4 선 그래프 그리기 \n\n# -*- coding: utf-8 -*-\n\nimport pandas as pd\n\ndf = pd.read_excel('./남북한발전전력량.xlsx') #데이터프레임 변환\n\ndf_ns = df.iloc[[0,5],3:] #남한, 북한 발전량 합계 데이터만 추출\ndf_ns", "_____no_output_____" ], [ "df_ns.index=['South', 'North'] #행 인덱스 변경\ndf_ns.columns = df_ns.columns.map(int) #열 이름의 자료형을 정수형으로 변경\nprint(df_ns.head())", " 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 ... 2007 \\\nSouth 1186 1310 1444 1650 1847 2055 2244 2153 2393 2664 ... 4031 \nNorth 263 247 221 231 230 213 193 170 186 194 ... 236 \n\n 2008 2009 2010 2011 2012 2013 2014 2015 2016 \nSouth 4224 4336 4747 4969 5096 5171 5220 5281 5404 \nNorth 255 235 237 211 215 221 216 190 239 \n\n[2 rows x 26 columns]\n" ], [ "df_ns", "_____no_output_____" ], [ "#선 그래프 그리기\ndf_ns.plot()", "_____no_output_____" ], [ "#행 ,열 전치하여 다시 그리기\ntdf_ns = df_ns.T\nprint(tdf_ns.head())", " South North\n1991 1186 263\n1992 1310 247\n1993 1444 221\n1994 1650 231\n1995 1847 230\n" ], [ "tdf_ns.plot()", "_____no_output_____" ], [ "tdf_ns.plot(kind='bar')", "_____no_output_____" ], [ "tdf_ns.plot(kind='barh')", "_____no_output_____" ], [ "tdf_ns.plot(kind='hist')", "_____no_output_____" ], [ "# -*- coding: utf-8 -*-\nimport pandas as pd\n\n#read_csv() 함수로 df 생성\ndf = pd.read_csv('./auto-mpg.csv', header=None)\n\n\n#열 이름 지정\ndf.columns = ['mpg','cylinders', 'displacement','horsepower','weight',\n 'acceleration', 'model year', 'origin', 'name']\n\n#2개의 열을 선택하여 산저도 그리기\ndf.plot(x='weight', y='mpg', kind='scatter')", "_____no_output_____" ], [ "#열을 선택하여 박스 플롯 그리기\ndf[['mpg','cylinders']].plot(kind='box')", "_____no_output_____" ], [ "#예제 4-1 선 그래프\n\n# -*- coding: utf-8 -*-\n\n#라이브러리 불러오기\nimport pandas as pd\nimport matplotlib.pyplot as plt\n\n#Excel 데이터를 데이터프레임으로 변환\ndf = pd.read_excel('시도별 전출입 인구수.xlsx', header=0)\ndf.head()", "_____no_output_____" ], [ "#누락된값(NaN)을 앞 데이터로 채움(엑셀 양식 병합 부분)\ndf = df.fillna(method='ffill')\ndf.head()", "_____no_output_____" ], [ "#서울에서 다른지역으로 이동한 데이터만 추출하여 정리\nmask = (df['전출지별']=='서울특별시') & (df['전입지별'] != '서울특별시')\ndf_seoul = df[mask]\ndf_seoul = df_seoul.drop(['전출지별'],axis = 1)\ndf_seoul.rename({'전입지별':'전입지'}, axis=1,inplace=True)\ndf_seoul.set_index('전입지', inplace=True)", "_____no_output_____" ], [ "df_seoul.head()", "_____no_output_____" ], [ "#서울에서 경기도로 이동한 인구 데이터 값만 선택\nsr_one = df_seoul.loc['경기도']\nsr_one.tail()", "_____no_output_____" ], [ "#x, y 축 데이터를 plot 함수에 입력\nplt.plot(sr_one.index, sr_one.values)", "_____no_output_____" ], [ "plt.plot(sr_one)", "_____no_output_____" ], [ "#x, y축 데이터를 plot 함수에 입력\nplt.plot(sr_one.index, sr_one.values)\n\n#차트 제목 추가\nplt.title('서울-> 경기 인구 이동')\n\n#축 이름 추가\nplt.xlabel('기간')\nplt.ylabel('이동 인구수')\n\nplt.show() #변경사항 저장하고 그래프 출력", "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 49436 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 50872 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 44221 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 44592 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 51064 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 44396 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 51060 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 46041 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 44036 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:238: RuntimeWarning: Glyph 49688 missing from current font.\n font.set_text(s, 0.0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 44592 missing from current font.\n font.set_text(s, 0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 44036 missing from current font.\n font.set_text(s, 0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 51060 missing from current font.\n font.set_text(s, 0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 46041 missing from current font.\n font.set_text(s, 0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 51064 missing from current font.\n font.set_text(s, 0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 44396 missing from current font.\n font.set_text(s, 0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 49688 missing from current font.\n font.set_text(s, 0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 49436 missing from current font.\n font.set_text(s, 0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 50872 missing from current font.\n font.set_text(s, 0, flags=flags)\nC:\\ProgramData\\Anaconda3\\lib\\site-packages\\matplotlib\\backends\\backend_agg.py:201: RuntimeWarning: Glyph 44221 missing from current font.\n font.set_text(s, 0, flags=flags)\n" ], [ "#matplotlib 한글 폰트 오류 문제해결\nfrom matplotlib import font_manager, rc\nfont_path = \"./malgun.ttf\" #폰트 파일 위치\nfont_name = font_manager.FontProperties(fname=font_path).get_name()\nrc('font', family=font_name)", "_____no_output_____" ], [ "#x, y축 데이터를 plot 함수에 입력\n#plt.plot(sr_one.index, sr_one.values)\n\n#차트 제목 추가\n#plt.title('서울-> 경기 인구 이동')\n\n#축 이름 추가\n#plt.xlabel('기간')\n#plt.ylabel('이동 인구수')\n\n#그래프 x축 글씨 문제 해결\n#그림사이즈 지정(가로 14인치, 세로 5인치)\nplt.figure(figsize=(14,5))\n\n#x축 눈금 라벨 회전하기\nplt.xticks(rotation='vertical')\n\n#x,y축 데이터를 plot 함수에 입력\nplt.plot(sr_one.index, sr_one.values)\n\nplt.title('서울 -> 경기 인구 이동') #차트 제목\nplt.xlabel('기간') # x축 이름\nplt.ylabel('이동 인구수') # y축 이름\n\n\nplt.legend(labels=['서울-> 경기'], loc='best')\n\nplt.show() #변경사항 저장하고 그래프 출력", "_____no_output_____" ], [ "#예제 4-5 스타일 서식 지정 등\n\n#서울에서 경기도로 이동한 인구 데이터 값만 선택\nsr_one = df_seoul.loc['경기도']\n\n#스타일 서식 지정\nplt.style.use('ggplot')\n\n#그림 사이즈 지정\nplt.figure(figsize=(14,5))\n\n#x축 눈금 라벨 회전하기\nplt.xticks(size=10, rotation='vertical')\n\n#x,y축 데이터를 plot 함수에 입력\nplt.plot(sr_one.index, sr_one.values, marker='o', markersize=10)\n\nplt.title('서울 -> 경기 인구 이동', size=30) #차트제목\nplt.xlabel('기간', size=20) #x축 이름\nplt.ylabel('이동 인구수', size=20) #y축 이름\n\nplt.legend(labels=['서울 -> 경기'], loc='best', fontsize=15)\n\nplt.show()", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# 各種パッケージのインストールとバージョン\n!pip install transformers\n!pip install tokenizers\n!pip install sentencepiece\n!pip list | grep torch\n!pip list | grep transformers\n!pip list | grep tokenizers\n!pip list | grep sentencepiece", "_____no_output_____" ], [ "# ディレクトリの指定\ndir = \"./\"\n\n", "_____no_output_____" ], [ "# 事前学習用コーパスの準備\n# 1行に1文章となるようなテキストを準備する\n\ndf_header = pd.read_csv('XXX.csv')\nprint(df_header)\n\n", "_____no_output_____" ], [ "# Tokenization\nfrom sentencepiece import SentencePieceTrainer\n\n# sentencepieceの学習\nSentencePieceTrainer.Train(\n '--input='+dir+'corpus/corpus.txt, --model_prefix='+dir+'model/sentencepiece --character_coverage=0.9995 --vocab_size=100'\n)\n\n# sentencepieceのパラメータ\n# https://github.com/google/sentencepiece#train-sentencepiece-model\n# training options\n# https://github.com/google/sentencepiece/blob/master/doc/options.md\n\n\n", "_____no_output_____" ], [ "# sentencepieceのモデルをTokenizerで読み込み\n\n# sentencepieceを使ったTokenizerは現時点では以下。\n# >All transformers models in the library that use SentencePiece use it \n# in combination with unigram. Examples of models using SentencePiece are ALBERT, XLNet, Marian, and T5.\n# https://huggingface.co/transformers/tokenizer_summary.html\n\nfrom transformers import AlbertTokenizer\n\n# ALBERTのトークナイザを定義\ntokenizer = AlbertTokenizer.from_pretrained(dir+'model/sentencepiece.model', keep_accents=True)\n\n# textをトークナイズ\ntext = \"吾輩は猫である。名前はまだ無い。\"\nprint(tokenizer.tokenize(text))", "/usr/local/lib/python3.7/dist-packages/transformers/tokenization_utils_base.py:1641: FutureWarning: Calling AlbertTokenizer.from_pretrained() with the path to a single file or url is deprecated and won't be possible anymore in v5. Use a model identifier or the path to a directory instead.\n FutureWarning,\nSpecial tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\n" ], [ "# BERTモデルのconfigを設定\nfrom transformers import BertConfig\nfrom transformers import BertForMaskedLM\n\n# BERTconfigを定義\nconfig = BertConfig(vocab_size=32003, num_hidden_layers=12, intermediate_size=768, num_attention_heads=12)\n\n# BERT MLMのインスタンスを生成\nmodel = BertForMaskedLM(config)\n\n# パラメータ数を表示\nprint('No of parameters: ', model.num_parameters())", "No of parameters: 68158211\n" ], [ "# 事前学習用のデータセットを準備\nfrom transformers import LineByLineTextDataset\nfrom transformers import DataCollatorForLanguageModeling\n\n# textを1行ずつ読み込んでトークンへ変換\ndataset = LineByLineTextDataset(\n tokenizer=tokenizer,\n file_path=dir + 'corpus/corpus.txt',\n block_size=256, # tokenizerのmax_length\n)\n\n# データセットからサンプルのリストを受け取り、それらをテンソルの辞書としてバッチに照合するための関数\ndata_collator = DataCollatorForLanguageModeling(\n tokenizer=tokenizer, \n mlm=True,\n mlm_probability= 0.15\n)\n\n", "/usr/local/lib/python3.7/dist-packages/transformers/data/datasets/language_modeling.py:124: FutureWarning: This dataset will be removed from the library soon, preprocessing should be handled with the 🤗 Datasets library. You can have a look at this example script for pointers: https://github.com/huggingface/transformers/blob/master/examples/pytorch/language-modeling/run_mlm.py\n FutureWarning,\n" ], [ "# 事前学習を行う\nfrom transformers import TrainingArguments\nfrom transformers import Trainer\n\n# 事前学習のパラメータを定義\ntraining_args = TrainingArguments(\n output_dir= drive_dir + 'outputBERT/',\n overwrite_output_dir=True,\n num_train_epochs=10,\n per_device_train_batch_size=32,\n save_steps=10000,\n save_total_limit=2,\n prediction_loss_only=True\n)\n\n# trainerインスタンスの生成\ntrainer = Trainer(\n model=model,\n args=training_args,\n data_collator=data_collator,\n train_dataset=dataset\n)\n\n#　学習\ntrainer.train()\n\n# 学習したモデルの保存\ntrainer.save_model(dir + 'outputBERT/')", "PyTorch: setting up devices\nThe default value for the training argument `--report_to` will change in v5 (from all installed integrations to none). In v5, you will need to use `--report_to all` to get the same behavior as now. You should start updating your code and make this info disappear :-).\n***** Running training *****\n Num examples = 5\n Num Epochs = 10\n Instantaneous batch size per device = 32\n Total train batch size (w. parallel, distributed & accumulation) = 32\n Gradient Accumulation steps = 1\n Total optimization steps = 10\n" ], [ "# 言語モデルの確認\nfrom transformers import pipeline\n\n# tokenizerとmodel\ntokenizer = AlbertTokenizer.from_pretrained(drive_dir+'model/sentencepiece.model', keep_accents=True)\nmodel = BertForMaskedLM.from_pretrained(drive_dir + 'outputBERT')\n\nfill_mask = pipeline(\n \"fill-mask\",\n model=model,\n tokenizer=tokenizer\n)\n\nMASK_TOKEN = tokenizer.mask_token\n\n# コーパスに応じた文章から穴埋めをとく\n\ntext = \"XXX{}XXX\".format(MASK_TOKEN)\nfill_mask(text)", "/usr/local/lib/python3.7/dist-packages/transformers/tokenization_utils_base.py:1641: FutureWarning: Calling AlbertTokenizer.from_pretrained() with the path to a single file or url is deprecated and won't be possible anymore in v5. Use a model identifier or the path to a directory instead.\n FutureWarning,\nloading file /content/drive/MyDrive/BERT-pretrained-transformers/model/sentencepiece.model\nAdding [CLS] to the vocabulary\nAdding [SEP] to the vocabulary\nAdding <pad> to the vocabulary\nAdding [MASK] to the vocabulary\nSpecial tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\nloading configuration file /content/drive/MyDrive/BERT-pretrained-transformers/outputBERT/config.json\nModel config BertConfig {\n \"architectures\": [\n \"BertForMaskedLM\"\n ],\n \"attention_probs_dropout_prob\": 0.1,\n \"gradient_checkpointing\": false,\n \"hidden_act\": \"gelu\",\n \"hidden_dropout_prob\": 0.1,\n \"hidden_size\": 768,\n \"initializer_range\": 0.02,\n \"intermediate_size\": 768,\n \"layer_norm_eps\": 1e-12,\n \"max_position_embeddings\": 512,\n \"model_type\": \"bert\",\n \"num_attention_heads\": 12,\n \"num_hidden_layers\": 12,\n \"pad_token_id\": 0,\n \"position_embedding_type\": \"absolute\",\n \"torch_dtype\": \"float32\",\n \"transformers_version\": \"4.9.1\",\n \"type_vocab_size\": 2,\n \"use_cache\": true,\n \"vocab_size\": 32003\n}\n\nloading weights file /content/drive/MyDrive/BERT-pretrained-transformers/outputBERT/pytorch_model.bin\nAll model checkpoint weights were used when initializing BertForMaskedLM.\n\nAll the weights of BertForMaskedLM were initialized from the model checkpoint at /content/drive/MyDrive/BERT-pretrained-transformers/outputBERT.\nIf your task is similar to the task the model of the checkpoint was trained on, you can already use BertForMaskedLM for predictions without further training.\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "##### PyMC3 Examples\n\n# GLM Robust Regression with Outlier Detection\n\n**A minimal reproducable example of Robust Regression with Outlier Detection using Hogg 2010 Signal vs Noise method.**\n\n+ This is a complementary approach to the Student-T robust regression as illustrated in Thomas Wiecki's notebook in the [PyMC3 documentation](http://pymc-devs.github.io/pymc3/GLM-robust/), that approach is also compared here.\n+ This model returns a robust estimate of linear coefficients and an indication of which datapoints (if any) are outliers.\n+ The likelihood evaluation is essentially a copy of eqn 17 in \"Data analysis recipes: Fitting a model to data\" - [Hogg 2010](http://arxiv.org/abs/1008.4686).\n+ The model is adapted specifically from Jake Vanderplas' [implementation](http://www.astroml.org/book_figures/chapter8/fig_outlier_rejection.html) (3rd model tested).\n+ The dataset is tiny and hardcoded into this Notebook. It contains errors in both the x and y, but we will deal here with only errors in y.\n\n\n**Note:**\n\n+ Python 3.4 project using latest available [PyMC3](https://github.com/pymc-devs/pymc3)\n+ Developed using [ContinuumIO Anaconda](https://www.continuum.io/downloads) distribution on a Macbook Pro 3GHz i7, 16GB RAM, OSX 10.10.5.\n+ During development I've found that 3 data points are always indicated as outliers, but the remaining ordering of datapoints by decreasing outlier-hood is slightly unstable between runs: the posterior surface appears to have a small number of solutions with similar probability. \n+ Finally, if runs become unstable or Theano throws weird errors, try clearing the cache `$> theano-cache clear` and rerunning the notebook.\n\n\n**Package Requirements (shown as a conda-env YAML):**\n```\n$> less conda_env_pymc3_examples.yml\n\nname: pymc3_examples\n channels:\n - defaults\n dependencies:\n - python=3.4\n - ipython\n - ipython-notebook\n - ipython-qtconsole\n - numpy\n - scipy\n - matplotlib\n - pandas\n - seaborn\n - patsy \n - pip\n\n$> conda env create --file conda_env_pymc3_examples.yml\n\n$> source activate pymc3_examples\n\n$> pip install --process-dependency-links git+https://github.com/pymc-devs/pymc3\n\n```", "_____no_output_____" ], [ "# Setup", "_____no_output_____" ] ], [ [ "%matplotlib inline\n%qtconsole --colors=linux\n\nimport warnings\nwarnings.filterwarnings('ignore')", "_____no_output_____" ], [ "import numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nimport seaborn as sns\n\nfrom scipy import optimize\nimport pymc3 as pm\nimport theano as thno\nimport theano.tensor as T \n\n# configure some basic options\nsns.set(style=\"darkgrid\", palette=\"muted\")\npd.set_option('display.notebook_repr_html', True)\nplt.rcParams['figure.figsize'] = 12, 8\nnp.random.seed(0)", "_____no_output_____" ] ], [ [ "## Load and Prepare Data", "_____no_output_____" ], [ "We'll use the Hogg 2010 data available at https://github.com/astroML/astroML/blob/master/astroML/datasets/hogg2010test.py\n\nIt's a very small dataset so for convenience, it's hardcoded below", "_____no_output_____" ] ], [ [ "#### cut & pasted directly from the fetch_hogg2010test() function\n## identical to the original dataset as hardcoded in the Hogg 2010 paper\n\ndfhogg = pd.DataFrame(np.array([[1, 201, 592, 61, 9, -0.84],\n [2, 244, 401, 25, 4, 0.31],\n [3, 47, 583, 38, 11, 0.64],\n [4, 287, 402, 15, 7, -0.27],\n [5, 203, 495, 21, 5, -0.33],\n [6, 58, 173, 15, 9, 0.67],\n [7, 210, 479, 27, 4, -0.02],\n [8, 202, 504, 14, 4, -0.05],\n [9, 198, 510, 30, 11, -0.84],\n [10, 158, 416, 16, 7, -0.69],\n [11, 165, 393, 14, 5, 0.30],\n [12, 201, 442, 25, 5, -0.46],\n [13, 157, 317, 52, 5, -0.03],\n [14, 131, 311, 16, 6, 0.50],\n [15, 166, 400, 34, 6, 0.73],\n [16, 160, 337, 31, 5, -0.52],\n [17, 186, 423, 42, 9, 0.90],\n [18, 125, 334, 26, 8, 0.40],\n [19, 218, 533, 16, 6, -0.78],\n [20, 146, 344, 22, 5, -0.56]]),\n columns=['id','x','y','sigma_y','sigma_x','rho_xy'])\n\n\n## for convenience zero-base the 'id' and use as index\ndfhogg['id'] = dfhogg['id'] - 1\ndfhogg.set_index('id', inplace=True)\n\n## standardize (mean center and divide by 1 sd)\ndfhoggs = (dfhogg[['x','y']] - dfhogg[['x','y']].mean(0)) / dfhogg[['x','y']].std(0)\ndfhoggs['sigma_y'] = dfhogg['sigma_y'] / dfhogg['y'].std(0)\ndfhoggs['sigma_x'] = dfhogg['sigma_x'] / dfhogg['x'].std(0)\n\n## create xlims ylims for plotting\nxlims = (dfhoggs['x'].min() - np.ptp(dfhoggs['x'])/5\n ,dfhoggs['x'].max() + np.ptp(dfhoggs['x'])/5)\nylims = (dfhoggs['y'].min() - np.ptp(dfhoggs['y'])/5\n ,dfhoggs['y'].max() + np.ptp(dfhoggs['y'])/5)\n\n## scatterplot the standardized data\ng = sns.FacetGrid(dfhoggs, size=8)\n_ = g.map(plt.errorbar, 'x', 'y', 'sigma_y', 'sigma_x', marker=\"o\", ls='')\n_ = g.axes[0][0].set_ylim(ylims)\n_ = g.axes[0][0].set_xlim(xlims)\n\nplt.subplots_adjust(top=0.92)\n_ = g.fig.suptitle('Scatterplot of Hogg 2010 dataset after standardization', fontsize=16)", "_____no_output_____" ] ], [ [ "**Observe**: \n\n+ Even judging just by eye, you can see these datapoints mostly fall on / around a straight line with positive gradient\n+ It looks like a few of the datapoints may be outliers from such a line", "_____no_output_____" ], [ "---\n\n---", "_____no_output_____" ], [ "# Create Conventional OLS Model", "_____no_output_____" ], [ "The *linear model* is really simple and conventional:\n\n$$\\bf{y} = \\beta^{T} \\bf{X} + \\bf{\\sigma}$$\n\nwhere: \n\n$\\beta$ = coefs = $\\{1, \\beta_{j \\in X_{j}}\\}$ \n$\\sigma$ = the measured error in $y$ in the dataset `sigma_y`", "_____no_output_____" ], [ "##### Define model\n\n**NOTE:**\n+ We're using a simple linear OLS model with Normally distributed priors so that it behaves like a ridge regression", "_____no_output_____" ] ], [ [ "with pm.Model() as mdl_ols:\n \n ## Define weakly informative Normal priors to give Ridge regression\n b0 = pm.Normal('b0_intercept', mu=0, sd=100)\n b1 = pm.Normal('b1_slope', mu=0, sd=100)\n \n ## Define linear model\n yest = b0 + b1 * dfhoggs['x']\n \n ## Use y error from dataset, convert into theano variable\n sigma_y = thno.shared(np.asarray(dfhoggs['sigma_y'],\n dtype=thno.config.floatX), name='sigma_y')\n\n ## Define Normal likelihood\n likelihood = pm.Normal('likelihood', mu=yest, sd=sigma_y, observed=dfhoggs['y'])\n", "_____no_output_____" ] ], [ [ "##### Sample", "_____no_output_____" ] ], [ [ "with mdl_ols:\n\n ## find MAP using Powell, seems to be more robust\n start_MAP = pm.find_MAP(fmin=optimize.fmin_powell, disp=True)\n\n ## take samples\n traces_ols = pm.sample(2000, start=start_MAP, step=pm.NUTS(), progressbar=True)", "Optimization terminated successfully.\n Current function value: 145.777745\n Iterations: 3\n Function evaluations: 118\n [-----------------100%-----------------] 2000 of 2000 complete in 0.9 sec" ] ], [ [ "##### View Traces\n\n**NOTE**: I'll 'burn' the traces to only retain the final 1000 samples", "_____no_output_____" ] ], [ [ "_ = pm.traceplot(traces_ols[-1000:], figsize=(12,len(traces_ols.varnames)*1.5),\n lines={k: v['mean'] for k, v in pm.df_summary(traces_ols[-1000:]).iterrows()})", "_____no_output_____" ] ], [ [ "**NOTE:** We'll illustrate this OLS fit and compare to the datapoints in the final plot", "_____no_output_____" ], [ "---\n\n---", "_____no_output_____" ], [ "# Create Robust Model: Student-T Method", "_____no_output_____" ], [ "I've added this brief section in order to directly compare the Student-T based method exampled in Thomas Wiecki's notebook in the [PyMC3 documentation](http://pymc-devs.github.io/pymc3/GLM-robust/)\n\nInstead of using a Normal distribution for the likelihood, we use a Student-T, which has fatter tails. In theory this allows outliers to have a smaller mean square error in the likelihood, and thus have less influence on the regression estimation. This method does not produce inlier / outlier flags but is simpler and faster to run than the Signal Vs Noise model below, so a comparison seems worthwhile.\n\n**Note:** we'll constrain the Student-T 'degrees of freedom' parameter `nu` to be an integer, but otherwise leave it as just another stochastic to be inferred: no need for prior knowledge.", "_____no_output_____" ], [ "##### Define Model", "_____no_output_____" ] ], [ [ "with pm.Model() as mdl_studentt:\n \n ## Define weakly informative Normal priors to give Ridge regression\n b0 = pm.Normal('b0_intercept', mu=0, sd=100)\n b1 = pm.Normal('b1_slope', mu=0, sd=100)\n \n ## Define linear model\n yest = b0 + b1 * dfhoggs['x']\n \n ## Use y error from dataset, convert into theano variable\n sigma_y = thno.shared(np.asarray(dfhoggs['sigma_y'],\n dtype=thno.config.floatX), name='sigma_y')\n \n ## define prior for Student T degrees of freedom\n nu = pm.DiscreteUniform('nu', lower=1, upper=100)\n\n ## Define Student T likelihood\n likelihood = pm.StudentT('likelihood', mu=yest, sd=sigma_y, nu=nu\n ,observed=dfhoggs['y'])\n", "_____no_output_____" ] ], [ [ "##### Sample", "_____no_output_____" ] ], [ [ "with mdl_studentt:\n\n ## find MAP using Powell, seems to be more robust\n start_MAP = pm.find_MAP(fmin=optimize.fmin_powell, disp=True)\n\n ## two-step sampling to allow Metropolis for nu (which is discrete)\n step1 = pm.NUTS([b0, b1])\n step2 = pm.Metropolis([nu])\n \n ## take samples\n traces_studentt = pm.sample(2000, start=start_MAP, step=[step1, step2], progressbar=True)", "Optimization terminated successfully.\n Current function value: 107.488021\n Iterations: 3\n Function evaluations: 77\n [-----------------100%-----------------] 2000 of 2000 complete in 1.0 sec" ] ], [ [ "##### View Traces", "_____no_output_____" ] ], [ [ "_ = pm.traceplot(traces_studentt[-1000:]\n ,figsize=(12,len(traces_studentt.varnames)*1.5)\n ,lines={k: v['mean'] for k, v in pm.df_summary(traces_studentt[-1000:]).iterrows()})", "_____no_output_____" ] ], [ [ "**Observe:**\n\n+ Both parameters `b0` and `b1` show quite a skew to the right, possibly this is the action of a few samples regressing closer to the OLS estimate which is towards the left\n+ The `nu` parameter seems very happy to stick at `nu = 1`, indicating that a fat-tailed Student-T likelihood has a better fit than a thin-tailed (Normal-like) Student-T likelihood.\n+ The inference sampling also ran very quickly, almost as quickly as the conventional OLS\n\n\n**NOTE:** We'll illustrate this Student-T fit and compare to the datapoints in the final plot", "_____no_output_____" ], [ "---\n\n---", "_____no_output_____" ], [ "# Create Robust Model with Outliers: Hogg Method", "_____no_output_____" ], [ "Please read the paper (Hogg 2010) and Jake Vanderplas' code for more complete information about the modelling technique.\n\nThe general idea is to create a 'mixture' model whereby datapoints can be described by either the linear model (inliers) or a modified linear model with different mean and larger variance (outliers).\n\n\nThe likelihood is evaluated over a mixture of two likelihoods, one for 'inliers', one for 'outliers'. A Bernouilli distribution is used to randomly assign datapoints in N to either the inlier or outlier groups, and we sample the model as usual to infer robust model parameters and inlier / outlier flags:\n\n$$\n\\mathcal{logL} = \\sum_{i}^{i=N} log \\left[ \\frac{(1 - B_{i})}{\\sqrt{2 \\pi \\sigma_{in}^{2}}} exp \\left( - \\frac{(x_{i} - \\mu_{in})^{2}}{2\\sigma_{in}^{2}} \\right) \\right] + \\sum_{i}^{i=N} log \\left[ \\frac{B_{i}}{\\sqrt{2 \\pi (\\sigma_{in}^{2} + \\sigma_{out}^{2})}} exp \\left( - \\frac{(x_{i}- \\mu_{out})^{2}}{2(\\sigma_{in}^{2} + \\sigma_{out}^{2})} \\right) \\right]\n$$\n\nwhere: \n$\\bf{B}$ is Bernoulli-distibuted $B_{i} \\in [0_{(inlier)},1_{(outlier)}]$\n\n", "_____no_output_____" ], [ "##### Define model", "_____no_output_____" ] ], [ [ "def logp_signoise(yobs, is_outlier, yest_in, sigma_y_in, yest_out, sigma_y_out):\n '''\n Define custom loglikelihood for inliers vs outliers. \n NOTE: in this particular case we don't need to use theano's @as_op \n decorator because (as stated by Twiecki in conversation) that's only \n required if the likelihood cannot be expressed as a theano expression.\n We also now get the gradient computation for free.\n ''' \n \n # likelihood for inliers\n pdfs_in = T.exp(-(yobs - yest_in + 1e-4)**2 / (2 * sigma_y_in**2)) \n pdfs_in /= T.sqrt(2 * np.pi * sigma_y_in**2)\n logL_in = T.sum(T.log(pdfs_in) * (1 - is_outlier))\n\n # likelihood for outliers\n pdfs_out = T.exp(-(yobs - yest_out + 1e-4)**2 / (2 * (sigma_y_in**2 + sigma_y_out**2))) \n pdfs_out /= T.sqrt(2 * np.pi * (sigma_y_in**2 + sigma_y_out**2))\n logL_out = T.sum(T.log(pdfs_out) * is_outlier)\n\n return logL_in + logL_out\n", "_____no_output_____" ], [ "with pm.Model() as mdl_signoise:\n \n ## Define weakly informative Normal priors to give Ridge regression\n b0 = pm.Normal('b0_intercept', mu=0, sd=100)\n b1 = pm.Normal('b1_slope', mu=0, sd=100)\n \n ## Define linear model\n yest_in = b0 + b1 * dfhoggs['x']\n\n ## Define weakly informative priors for the mean and variance of outliers\n yest_out = pm.Normal('yest_out', mu=0, sd=100)\n sigma_y_out = pm.HalfNormal('sigma_y_out', sd=100)\n\n ## Define Bernoulli inlier / outlier flags according to a hyperprior \n ## fraction of outliers, itself constrained to [0,.5] for symmetry\n frac_outliers = pm.Uniform('frac_outliers', lower=0., upper=.5)\n is_outlier = pm.Bernoulli('is_outlier', p=frac_outliers, shape=dfhoggs.shape[0]) \n \n ## Extract observed y and sigma_y from dataset, encode as theano objects\n yobs = thno.shared(np.asarray(dfhoggs['y'], dtype=thno.config.floatX), name='yobs')\n sigma_y_in = thno.shared(np.asarray(dfhoggs['sigma_y']\n , dtype=thno.config.floatX), name='sigma_y_in')\n \n ## Use custom likelihood using DensityDist\n likelihood = pm.DensityDist('likelihood', logp_signoise,\n observed={'yobs':yobs, 'is_outlier':is_outlier,\n 'yest_in':yest_in, 'sigma_y_in':sigma_y_in,\n 'yest_out':yest_out, 'sigma_y_out':sigma_y_out})\n", "_____no_output_____" ] ], [ [ "##### Sample", "_____no_output_____" ] ], [ [ "with mdl_signoise:\n\n ## two-step sampling to create Bernoulli inlier/outlier flags\n step1 = pm.NUTS([frac_outliers, yest_out, sigma_y_out, b0, b1])\n step2 = pm.BinaryMetropolis([is_outlier], tune_interval=100)\n\n ## find MAP using Powell, seems to be more robust\n start_MAP = pm.find_MAP(fmin=optimize.fmin_powell, disp=True)\n\n ## take samples\n traces_signoise = pm.sample(2000, start=start_MAP, step=[step1,step2], progressbar=True)", "Optimization terminated successfully.\n Current function value: 155.449990\n Iterations: 3\n Function evaluations: 213\n [-----------------100%-----------------] 2000 of 2000 complete in 169.1 sec" ] ], [ [ "##### View Traces", "_____no_output_____" ] ], [ [ "_ = pm.traceplot(traces_signoise[-1000:], figsize=(12,len(traces_signoise.varnames)*1.5),\n lines={k: v['mean'] for k, v in pm.df_summary(traces_signoise[-1000:]).iterrows()})", "_____no_output_____" ] ], [ [ "**NOTE:**\n\n+ During development I've found that 3 datapoints id=[1,2,3] are always indicated as outliers, but the remaining ordering of datapoints by decreasing outlier-hood is unstable between runs: the posterior surface appears to have a small number of solutions with very similar probability.\n+ The NUTS sampler seems to work okay, and indeed it's a nice opportunity to demonstrate a custom likelihood which is possible to express as a theano function (thus allowing a gradient-based sampler like NUTS). However, with a more complicated dataset, I would spend time understanding this instability and potentially prefer using more samples under Metropolis-Hastings.", "_____no_output_____" ], [ "---\n\n---", "_____no_output_____" ], [ "# Declare Outliers and Compare Plots", "_____no_output_____" ], [ "##### View ranges for inliers / outlier predictions", "_____no_output_____" ], [ "At each step of the traces, each datapoint may be either an inlier or outlier. We hope that the datapoints spend an unequal time being one state or the other, so let's take a look at the simple count of states for each of the 20 datapoints.", "_____no_output_____" ] ], [ [ "outlier_melt = pd.melt(pd.DataFrame(traces_signoise['is_outlier', -1000:],\n columns=['[{}]'.format(int(d)) for d in dfhoggs.index]),\n var_name='datapoint_id', value_name='is_outlier')\nax0 = sns.pointplot(y='datapoint_id', x='is_outlier', data=outlier_melt,\n kind='point', join=False, ci=None, size=4, aspect=2)\n\n_ = ax0.vlines([0,1], 0, 19, ['b','r'], '--')\n\n_ = ax0.set_xlim((-0.1,1.1))\n_ = ax0.set_xticks(np.arange(0, 1.1, 0.1))\n_ = ax0.set_xticklabels(['{:.0%}'.format(t) for t in np.arange(0,1.1,0.1)])\n\n_ = ax0.yaxis.grid(True, linestyle='-', which='major', color='w', alpha=0.4)\n_ = ax0.set_title('Prop. of the trace where datapoint is an outlier')\n_ = ax0.set_xlabel('Prop. of the trace where is_outlier == 1')", "_____no_output_____" ] ], [ [ "**Observe**:\n\n+ The plot above shows the number of samples in the traces in which each datapoint is marked as an outlier, expressed as a percentage.\n+ In particular, 3 points [1, 2, 3] spend >=95% of their time as outliers\n+ Contrastingly, points at the other end of the plot close to 0% are our strongest inliers.\n+ For comparison, the mean posterior value of `frac_outliers` is ~0.35, corresponding to roughly 7 of the 20 datapoints. You can see these 7 datapoints in the plot above, all those with a value >50% or thereabouts.\n+ However, only 3 of these points are outliers >=95% of the time. \n+ See note above regarding instability between runs.\n\nThe 95% cutoff we choose is subjective and arbitrary, but I prefer it for now, so let's declare these 3 to be outliers and see how it looks compared to Jake Vanderplas' outliers, which were declared in a slightly different way as points with means above 0.68.", "_____no_output_____" ], [ "##### Declare outliers\n\n**Note:**\n+ I will declare outliers to be datapoints that have value == 1 at the 5-percentile cutoff, i.e. in the percentiles from 5 up to 100, their values are 1. \n+ Try for yourself altering cutoff to larger values, which leads to an objective ranking of outlier-hood.", "_____no_output_____" ] ], [ [ "cutoff = 5\ndfhoggs['outlier'] = np.percentile(traces_signoise[-1000:]['is_outlier'],cutoff, axis=0)\ndfhoggs['outlier'].value_counts()", "_____no_output_____" ] ], [ [ "## Posterior Prediction Plots for OLS vs StudentT vs SignalNoise", "_____no_output_____" ] ], [ [ "g = sns.FacetGrid(dfhoggs, size=8, hue='outlier', hue_order=[True,False],\n palette='Set1', legend_out=False)\n\nlm = lambda x, samp: samp['b0_intercept'] + samp['b1_slope'] * x\n\npm.glm.plot_posterior_predictive(traces_ols[-1000:],\n eval=np.linspace(-3, 3, 10), lm=lm, samples=200, color='#22CC00', alpha=.2)\n\npm.glm.plot_posterior_predictive(traces_studentt[-1000:], lm=lm,\n eval=np.linspace(-3, 3, 10), samples=200, color='#FFA500', alpha=.5)\n\npm.glm.plot_posterior_predictive(traces_signoise[-1000:], lm=lm,\n eval=np.linspace(-3, 3, 10), samples=200, color='#357EC7', alpha=.3)\n\n_ = g.map(plt.errorbar, 'x', 'y', 'sigma_y', 'sigma_x', marker=\"o\", ls='').add_legend()\n\n_ = g.axes[0][0].annotate('OLS Fit: Green\\nStudent-T Fit: Orange\\nSignal Vs Noise Fit: Blue',\n size='x-large', xy=(1,0), xycoords='axes fraction',\n xytext=(-160,10), textcoords='offset points')\n_ = g.axes[0][0].set_ylim(ylims)\n_ = g.axes[0][0].set_xlim(xlims)", "_____no_output_____" ] ], [ [ "**Observe**:\n\n+ The posterior preditive fit for:\n + the **OLS model** is shown in **Green** and as expected, it doesn't appear to fit the majority of our datapoints very well, skewed by outliers\n + the **Robust Student-T model** is shown in **Orange** and does appear to fit the 'main axis' of datapoints quite well, ignoring outliers\n + the **Robust Signal vs Noise model** is shown in **Blue** and also appears to fit the 'main axis' of datapoints rather well, ignoring outliers.\n \n \n+ We see that the **Robust Signal vs Noise model** also yields specific estimates of _which_ datapoints are outliers:\n + 17 'inlier' datapoints, in **Blue** and\n + 3 'outlier' datapoints shown in **Red**.\n + From a simple visual inspection, the classification seems fair, and agrees with Jake Vanderplas' findings.\n \n \n+ Overall, it seems that:\n + the **Signal vs Noise model** behaves as promised, yielding a robust regression estimate and explicit labelling of inliers / outliers, but\n + the **Signal vs Noise model** is quite complex and whilst the regression seems robust and stable, the actual inlier / outlier labelling seems slightly unstable\n + if you simply want a robust regression without inlier / outlier labelling, the **Student-T model** may be a good compromise, offering a simple model, quick sampling, and a very similar estimate.", "_____no_output_____" ], [ "---", "_____no_output_____" ], [ "Example originally contributed by Jonathan Sedar 2015-12-21 [github.com/jonsedar](https://github.com/jonsedar)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Sample plots", "_____no_output_____" ] ], [ [ "# load libraries\nimport os\nimport numpy as np\nimport pandas as pd\nimport seaborn as sns\nimport matplotlib.pyplot as plt\n\n%matplotlib inline", "_____no_output_____" ], [ "os.listdir()", "_____no_output_____" ] ], [ [ "### sample plot 1", "_____no_output_____" ] ], [ [ "# load csv\ntrain = pd.read_csv('restaurant-and-market-health-inspections.csv')", "_____no_output_____" ], [ "train.info()", "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 58872 entries, 0 to 58871\nData columns (total 20 columns):\nserial_number 58872 non-null object\nactivity_date 58872 non-null object\nfacility_name 58872 non-null object\nscore 58872 non-null int64\ngrade 58872 non-null object\nservice_code 58872 non-null int64\nservice_description 58872 non-null object\nemployee_id 58872 non-null object\nfacility_address 58872 non-null object\nfacility_city 58872 non-null object\nfacility_id 58872 non-null object\nfacility_state 58872 non-null object\nfacility_zip 58872 non-null object\nowner_id 58872 non-null object\nowner_name 58872 non-null object\npe_description 58872 non-null object\nprogram_element_pe 58872 non-null int64\nprogram_name 58740 non-null object\nprogram_status 58872 non-null object\nrecord_id 58872 non-null object\ndtypes: int64(3), object(17)\nmemory usage: 5.2+ MB\n" ], [ "train.head()", "_____no_output_____" ], [ "train['grade'].unique()", "_____no_output_____" ], [ "train.select_dtypes('object')", "_____no_output_____" ], [ "# plot the count of Unique Values in integer Columns\ntrain.select_dtypes(np.int64).nunique().value_counts().sort_index().plot.bar(color = 'red', \n figsize = (9, 6), edgecolor = 'c', linewidth = 2)\nplt.xlabel('Number of Unique Values')\nplt.ylabel('Count')\nplt.title('Count of Unique Values in Integer Columns')", "_____no_output_____" ] ], [ [ "# sample 2", "_____no_output_____" ] ], [ [ "# plotting different categories\nfrom collections import OrderedDict\n\nplt.figure(figsize = (20, 16))\nplt.style.use('fivethirtyeight')\n\n# Color mapping\ncolors = OrderedDict({'A': 'red', 'B': 'orange', 'C': 'blue', ' ': 'green'})\nmapping = OrderedDict({'A': 'extreme', 'B': 'moderate', 'C': 'vulnerable', ' ': 'non vulnerable'})\n\n# Iterate through the float columns\nfor i, col in enumerate(train.select_dtypes('int64')):\n ax = plt.subplot(4, 2, i + 1)\n # Iterate through the poverty levels\n for level, color in colors.items():\n # Plot each poverty level as a separate line\n sns.kdeplot(train.loc[train['grade'] == level, col].dropna(), \n ax = ax, color = color, label = mapping[level])\n \n plt.title(f'{col.capitalize()} Distribution'); plt.xlabel(f'{col}')\n plt.ylabel('Density')\n\nplt.subplots_adjust(top = 2)", "c:\\py37\\lib\\site-packages\\scipy\\stats\\stats.py:1713: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.\n return np.add.reduce(sorted[indexer] * weights, axis=axis) / sumval\nc:\\py37\\lib\\site-packages\\numpy\\core\\_methods.py:140: RuntimeWarning: Degrees of freedom <= 0 for slice\n keepdims=keepdims)\nc:\\py37\\lib\\site-packages\\numpy\\core\\_methods.py:132: RuntimeWarning: invalid value encountered in double_scalars\n ret = ret.dtype.type(ret / rcount)\nc:\\py37\\lib\\site-packages\\statsmodels\\nonparametric\\bandwidths.py:20: RuntimeWarning: invalid value encountered in minimum\n return np.minimum(np.std(X, axis=0, ddof=1), IQR)\nc:\\py37\\lib\\site-packages\\numpy\\core\\fromnumeric.py:83: RuntimeWarning: invalid value encountered in reduce\n return ufunc.reduce(obj, axis, dtype, out, **passkwargs)\n" ] ], [ [ "## Plot Categoricals", "_____no_output_____" ] ], [ [ "# plot two categorical variables\ndef plot_categoricals(x, y, data, annotate=False):\n \"\"\"Plot counts of two categoricals.\n Size is raw count for each grouping.\n Percentages are for a given value of y.\"\"\"\n \n # Raw counts \n raw_counts = pd.DataFrame(data.groupby(y)[x].value_counts(normalize = False))\n raw_counts = raw_counts.rename(columns = {x: 'raw_count'})\n \n # Calculate counts for each group of x and y\n counts = pd.DataFrame(data.groupby(y)[x].value_counts(normalize = True))\n \n # Rename the column and reset the index\n counts = counts.rename(columns = {x: 'normalized_count'}).reset_index()\n counts['percent'] = 100 * counts['normalized_count']\n \n # Add the raw count\n counts['raw_count'] = list(raw_counts['raw_count'])\n \n plt.figure(figsize = (14, 10))\n # Scatter plot sized by percent\n plt.scatter(counts[x], counts[y], edgecolor = 'k', color = 'lightgreen',\n s = 100 * np.sqrt(counts['raw_count']), marker = 'o',\n alpha = 0.6, linewidth = 1.5)\n \n if annotate:\n # Annotate the plot with text\n for i, row in counts.iterrows():\n # Put text with appropriate offsets\n plt.annotate(xy = (row[x] - (1 / counts[x].nunique()), \n row[y] - (0.15 / counts[y].nunique())),\n color = 'navy',\n s = f\"{round(row['percent'], 1)}%\")\n \n # Set tick marks\n plt.yticks(counts[y].unique())\n plt.xticks(counts[x].unique())\n \n # Transform min and max to evenly space in square root domain\n sqr_min = int(np.sqrt(raw_counts['raw_count'].min()))\n sqr_max = int(np.sqrt(raw_counts['raw_count'].max()))\n \n # 5 sizes for legend\n msizes = list(range(sqr_min, sqr_max,\n int(( sqr_max - sqr_min) / 5)))\n markers = []\n \n # Markers for legend\n for size in msizes:\n markers.append(plt.scatter([], [], s = 100 * size, \n label = f'{int(round(np.square(size) / 100) * 100)}', \n color = 'lightgreen',\n alpha = 0.6, edgecolor = 'k', linewidth = 1.5))\n \n # Legend and formatting\n plt.legend(handles = markers, title = 'Counts',\n labelspacing = 3, handletextpad = 2,\n fontsize = 16,\n loc = (1.10, 0.19))\n \n plt.annotate(f'* Size represents raw count while % is for a given y value.',\n xy = (0, 1), xycoords = 'figure points', size = 10)\n \n # Adjust axes limits\n #plt.xlim((counts[x].min() - (6 / counts[x].nunique()), \n # counts[x].max() + (6 / counts[x].nunique())))\n #plt.ylim((counts[y].min() - (4 / counts[y].nunique()), \n # counts[y].max() + (4 / counts[y].nunique())))\n plt.grid(None)\n plt.xlabel(f\"{x}\"); plt.ylabel(f\"{y}\"); plt.title(f\"{y} vs {x}\")\n\n\n# plots two categorical columns in the dataframe train\nplot_categoricals('grade', 'program_status', train);", "_____no_output_____" ] ], [ [ "## Value Count Plots", "_____no_output_____" ] ], [ [ "# plot value counts of a column\ndef plot_value_counts(df, col, condition=False):\n \"\"\"Plot value counts of a column, optionally with only the heads of a household\"\"\"\n # apply condition here if required\n if condition:\n # define condition below <>\n df = df.loc[df[col] == condition].copy()\n \n plt.figure(figsize = (8, 6))\n df[col].value_counts().sort_index().plot.bar(color = 'red',\n edgecolor = 'g',\n linewidth = 2)\n plt.xlabel(f'{col}')\n plt.title(f'{col} Value Counts')\n plt.ylabel('Count')\n plt.show()\n\n# plot\nplot_value_counts(train, 'grade')\nplot_value_counts(train, 'program_status')\n", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "a = ord('a')\nz = ord('z')\n\nfor c in range(a, z + 1):\n print(chr(c), end='')", "abcdefghijklmnopqrstuvwxyz" ], [ "def printAlphabet(size):\n a = ord('a')\n output = ''\n piramid = []\n\n if (0 >= size or size > 26 or size == 1):\n print('a')\n else:\n for c in range(a + size - 1, a - 1, -1):\n output = chr(c) + output\n piramid.append((output[len(output):0:-1] + output).center(size + size - 1, '-'))\n \n for step in piramid[len(piramid) - 2::-1]:\n piramid.append(step)\n \n for char in [*piramid]:\n print(*char, sep=\"-\")", "_____no_output_____" ], [ "printAlphabet(3)", "----c----\n--c-b-c--\nc-b-a-b-c\n--c-b-c--\n----c----\n" ], [ "printAlphabet(5)", "--------e--------\n------e-d-e------\n----e-d-c-d-e----\n--e-d-c-b-c-d-e--\ne-d-c-b-a-b-c-d-e\n--e-d-c-b-c-d-e--\n----e-d-c-d-e----\n------e-d-e------\n--------e--------\n" ], [ "printAlphabet(10)", "------------------j------------------\n----------------j-i-j----------------\n--------------j-i-h-i-j--------------\n------------j-i-h-g-h-i-j------------\n----------j-i-h-g-f-g-h-i-j----------\n--------j-i-h-g-f-e-f-g-h-i-j--------\n------j-i-h-g-f-e-d-e-f-g-h-i-j------\n----j-i-h-g-f-e-d-c-d-e-f-g-h-i-j----\n--j-i-h-g-f-e-d-c-b-c-d-e-f-g-h-i-j--\nj-i-h-g-f-e-d-c-b-a-b-c-d-e-f-g-h-i-j\n--j-i-h-g-f-e-d-c-b-c-d-e-f-g-h-i-j--\n----j-i-h-g-f-e-d-c-d-e-f-g-h-i-j----\n------j-i-h-g-f-e-d-e-f-g-h-i-j------\n--------j-i-h-g-f-e-f-g-h-i-j--------\n----------j-i-h-g-f-g-h-i-j----------\n------------j-i-h-g-h-i-j------------\n--------------j-i-h-i-j--------------\n----------------j-i-j----------------\n------------------j------------------\n" ], [ "printAlphabet(15)", "----------------------------o----------------------------\n--------------------------o-n-o--------------------------\n------------------------o-n-m-n-o------------------------\n----------------------o-n-m-l-m-n-o----------------------\n--------------------o-n-m-l-k-l-m-n-o--------------------\n------------------o-n-m-l-k-j-k-l-m-n-o------------------\n----------------o-n-m-l-k-j-i-j-k-l-m-n-o----------------\n--------------o-n-m-l-k-j-i-h-i-j-k-l-m-n-o--------------\n------------o-n-m-l-k-j-i-h-g-h-i-j-k-l-m-n-o------------\n----------o-n-m-l-k-j-i-h-g-f-g-h-i-j-k-l-m-n-o----------\n--------o-n-m-l-k-j-i-h-g-f-e-f-g-h-i-j-k-l-m-n-o--------\n------o-n-m-l-k-j-i-h-g-f-e-d-e-f-g-h-i-j-k-l-m-n-o------\n----o-n-m-l-k-j-i-h-g-f-e-d-c-d-e-f-g-h-i-j-k-l-m-n-o----\n--o-n-m-l-k-j-i-h-g-f-e-d-c-b-c-d-e-f-g-h-i-j-k-l-m-n-o--\no-n-m-l-k-j-i-h-g-f-e-d-c-b-a-b-c-d-e-f-g-h-i-j-k-l-m-n-o\n--o-n-m-l-k-j-i-h-g-f-e-d-c-b-c-d-e-f-g-h-i-j-k-l-m-n-o--\n----o-n-m-l-k-j-i-h-g-f-e-d-c-d-e-f-g-h-i-j-k-l-m-n-o----\n------o-n-m-l-k-j-i-h-g-f-e-d-e-f-g-h-i-j-k-l-m-n-o------\n--------o-n-m-l-k-j-i-h-g-f-e-f-g-h-i-j-k-l-m-n-o--------\n----------o-n-m-l-k-j-i-h-g-f-g-h-i-j-k-l-m-n-o----------\n------------o-n-m-l-k-j-i-h-g-h-i-j-k-l-m-n-o------------\n--------------o-n-m-l-k-j-i-h-i-j-k-l-m-n-o--------------\n----------------o-n-m-l-k-j-i-j-k-l-m-n-o----------------\n------------------o-n-m-l-k-j-k-l-m-n-o------------------\n--------------------o-n-m-l-k-l-m-n-o--------------------\n----------------------o-n-m-l-m-n-o----------------------\n------------------------o-n-m-n-o------------------------\n--------------------------o-n-o--------------------------\n----------------------------o----------------------------\n" ], [ "printAlphabet(20)", "--------------------------------------t--------------------------------------\n------------------------------------t-s-t------------------------------------\n----------------------------------t-s-r-s-t----------------------------------\n--------------------------------t-s-r-q-r-s-t--------------------------------\n------------------------------t-s-r-q-p-q-r-s-t------------------------------\n----------------------------t-s-r-q-p-o-p-q-r-s-t----------------------------\n--------------------------t-s-r-q-p-o-n-o-p-q-r-s-t--------------------------\n------------------------t-s-r-q-p-o-n-m-n-o-p-q-r-s-t------------------------\n----------------------t-s-r-q-p-o-n-m-l-m-n-o-p-q-r-s-t----------------------\n--------------------t-s-r-q-p-o-n-m-l-k-l-m-n-o-p-q-r-s-t--------------------\n------------------t-s-r-q-p-o-n-m-l-k-j-k-l-m-n-o-p-q-r-s-t------------------\n----------------t-s-r-q-p-o-n-m-l-k-j-i-j-k-l-m-n-o-p-q-r-s-t----------------\n--------------t-s-r-q-p-o-n-m-l-k-j-i-h-i-j-k-l-m-n-o-p-q-r-s-t--------------\n------------t-s-r-q-p-o-n-m-l-k-j-i-h-g-h-i-j-k-l-m-n-o-p-q-r-s-t------------\n----------t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t----------\n--------t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-e-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t--------\n------t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-e-d-e-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t------\n----t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-e-d-c-d-e-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t----\n--t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-e-d-c-b-c-d-e-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t--\nt-s-r-q-p-o-n-m-l-k-j-i-h-g-f-e-d-c-b-a-b-c-d-e-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t\n--t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-e-d-c-b-c-d-e-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t--\n----t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-e-d-c-d-e-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t----\n------t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-e-d-e-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t------\n--------t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-e-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t--------\n----------t-s-r-q-p-o-n-m-l-k-j-i-h-g-f-g-h-i-j-k-l-m-n-o-p-q-r-s-t----------\n------------t-s-r-q-p-o-n-m-l-k-j-i-h-g-h-i-j-k-l-m-n-o-p-q-r-s-t------------\n--------------t-s-r-q-p-o-n-m-l-k-j-i-h-i-j-k-l-m-n-o-p-q-r-s-t--------------\n----------------t-s-r-q-p-o-n-m-l-k-j-i-j-k-l-m-n-o-p-q-r-s-t----------------\n------------------t-s-r-q-p-o-n-m-l-k-j-k-l-m-n-o-p-q-r-s-t------------------\n--------------------t-s-r-q-p-o-n-m-l-k-l-m-n-o-p-q-r-s-t--------------------\n----------------------t-s-r-q-p-o-n-m-l-m-n-o-p-q-r-s-t----------------------\n------------------------t-s-r-q-p-o-n-m-n-o-p-q-r-s-t------------------------\n--------------------------t-s-r-q-p-o-n-o-p-q-r-s-t--------------------------\n----------------------------t-s-r-q-p-o-p-q-r-s-t----------------------------\n------------------------------t-s-r-q-p-q-r-s-t------------------------------\n--------------------------------t-s-r-q-r-s-t--------------------------------\n----------------------------------t-s-r-s-t----------------------------------\n------------------------------------t-s-t------------------------------------\n--------------------------------------t--------------------------------------\n" ], [ "printAlphabet(2)", "--b--\nb-a-b\n--b--\n" ], [ "printAlphabet(1)", "a\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Demos: Lecture 4", "_____no_output_____" ] ], [ [ "import pennylane as qml\nimport numpy as np", "/opt/conda/envs/pennylane/lib/python3.8/site-packages/_distutils_hack/__init__.py:30: UserWarning: Setuptools is replacing distutils.\n warnings.warn(\"Setuptools is replacing distutils.\")\n" ] ], [ [ "## Demo 1: `qml.ctrl`", "_____no_output_____" ] ], [ [ "def some_function():\n qml.PauliX(wires=1)\n qml.CNOT(wires=[1, 2])\n \n qml.Hadamard(wires=2)\n qml.CRX(0.3, wires=[2, 1])", "_____no_output_____" ], [ "dev = qml.device('default.qubit', wires=3)\n\[email protected](dev)\ndef control_the_thing():\n qml.Hadamard(wires=0)\n \n qml.ctrl(some_function, control=0)()\n \n return qml.state()", "_____no_output_____" ], [ "control_the_thing()", "_____no_output_____" ], [ "print(qml.draw(control_the_thing, expansion_strategy='device')())", " 0: ──H──╭C──╭C──╭ControlledPhaseShift(1.57)──╭C─────────╭ControlledPhaseShift(1.57)──╭C─────────╭C─────────╭C──╭C──────────╭C──╭C──────────╭┤ State \n 1: ─────╰X──├C──│────────────────────────────│──────────│────────────────────────────╰RZ(1.57)──╰RY(0.15)──├X──╰RY(-0.15)──├X──╰RZ(-1.57)──├┤ State \n 2: ─────────╰X──╰ControlledPhaseShift(1.57)──╰RX(1.57)──╰ControlledPhaseShift(1.57)────────────────────────╰C──────────────╰C──────────────╰┤ State \n\n" ] ], [ [ "## Demo 2: multi-qubit measurements", "_____no_output_____" ] ], [ [ "dev = qml.device('default.qubit', wires=3)#, shots=10)\n\[email protected](dev)\ndef something_parametrized(x, y):\n qml.Hadamard(wires=0)\n qml.CRX(x, wires=[0, 1])\n qml.CRY(y, wires=[1, 2])\n \n return qml.probs(wires=[0])", "_____no_output_____" ], [ "something_parametrized(0.1, 0.2)", "_____no_output_____" ] ], [ [ "## Demo 3: multi-qubit expectation values", "_____no_output_____" ] ], [ [ "dev = qml.device('default.qubit', wires=3)#, shots=10)\n\[email protected](dev)\ndef something_parametrized(x, y):\n qml.Hadamard(wires=0)\n qml.CRX(x, wires=[0, 1])\n qml.CRY(y, wires=[1, 2])\n \n return qml.expval(qml.PauliZ(0)), qml.expval(qml.PauliZ(1)), qml.expval(qml.PauliZ(2))", "_____no_output_____" ], [ "something_parametrized(0.3, 0.4)", "_____no_output_____" ], [ "dev = qml.device('default.qubit', wires=3)#, shots=10)\n\[email protected](dev)\ndef something_parametrized(x, y):\n qml.Hadamard(wires=0)\n qml.CRX(x, wires=[0, 1])\n qml.CRY(y, wires=[1, 2])\n \n return qml.expval(qml.PauliX(0) @ qml.PauliX(1) @ qml.PauliY(2))", "_____no_output_____" ], [ "something_parametrized(0.3, 0.4)", "_____no_output_____" ] ], [ [ "## Demo 4: superdense coding", "_____no_output_____" ], [ "<img src=\"fig/superdense.png\" width=\"600px\">", "_____no_output_____" ] ], [ [ "dev = qml.device('default.qubit', wires=2, shots=1)\n\ndef create_entangled_state(wires=None):\n qml.Hadamard(wires=wires[0])\n qml.CNOT(wires=[wires[0], wires[1]])\n \[email protected](dev)\ndef superdense_coding(b1=0, b2=0):\n create_entangled_state(wires=[0, 1])\n \n if b1 == 1:\n qml.PauliZ(wires=0)\n \n if b2 == 1:\n qml.PauliX(wires=0)\n \n qml.adjoint(create_entangled_state)(wires=[0, 1])\n \n return qml.sample()", "_____no_output_____" ], [ "superdense_coding(b1=1, b2=0)", "_____no_output_____" ] ], [ [ "## Demo 5: teleportation ", "_____no_output_____" ], [ "<img src=\"fig/teleportation_deferred.png\" width=\"800px\">", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# 1 - Getting Started", "_____no_output_____" ], [ "The main application for `scikit-gstat` is variogram analysis and [Kriging](https://en.wikipedia.org/wiki/Kriging). This Tutorial will guide you through the most basic functionality of `scikit-gstat`. There are other tutorials that will explain specific methods or attributes in `scikit-gstat` in more detail.\n\n#### What you will learn in this tutorial\n\n* How to instantiate `Variogram` and `OrdinaryKriging`\n* How to read a variogram\n* Perform an interpolation\n* Most basic plotting\n", "_____no_output_____" ] ], [ [ "import numpy as np \nimport pandas as pd\nimport matplotlib.pyplot as plt\nfrom pprint import pprint\nplt.style.use('ggplot')", "_____no_output_____" ] ], [ [ "The `Variogram` and `OrdinaryKriging` classes can be loaded directly from `skgstat`. This is the name of the Python module.", "_____no_output_____" ] ], [ [ "from skgstat import Variogram, OrdinaryKriging", "_____no_output_____" ] ], [ [ "At the current version, there are some deprecated attributes and method in the `Variogram` class. They do not raise `DeprecationWarning`s, but rather print a warning message to the screen. You can suppress this warning by adding an `SKG_SUPPRESS` environment variable", "_____no_output_____" ] ], [ [ "%set_env SKG_SUPPRESS=true", "env: SKG_SUPPRESS=true\n" ] ], [ [ "## 1.1 Load data", "_____no_output_____" ], [ "You can find a prepared example data set in the `./data` subdirectory. This example is extracted from a generated Gaussian random field. We can expect the field to be stationary and show a nice spatial dependence, because it was created that way.\nWe can load one of the examples and have a look at the data:", "_____no_output_____" ] ], [ [ "data = pd.read_csv('./data/sample_sr.csv')\nprint(\"Loaded %d rows and %d columns\" % data.shape)\ndata.head()", "Loaded 200 rows and 3 columns\n" ] ], [ [ "Get a first overview of your data by plotting the `x` and `y` coordinates and visually inspect how the `z` spread out.", "_____no_output_____" ] ], [ [ "fig, ax = plt.subplots(1, 1, figsize=(9, 9))\nart = ax.scatter(data.x,data.y, s=50, c=data.z, cmap='plasma')\nplt.colorbar(art);", "_____no_output_____" ] ], [ [ "We can already see a lot from here: \n\n* The small values seem to concentrate on the upper left and lower right corner\n* Larger values are arranged like a band from lower left to upper right corner\n* To me, each of these blobs seem to have a diameter of something like 30 or 40 units.\n* The distance between the minimum and maximum seems to be not more than 60 or 70 units.\n\nThese are already very important insights.", "_____no_output_____" ], [ "## 1.2 Build a Variogram", "_____no_output_____" ], [ "As a quick reminder, the variogram relates pair-wise separating distances of `coordinates` and relates them to the *semi-variance* of the corresponding `values` pairs. The default estimator used is the Matheron estimator:\n\n$$ \\gamma (h) = \\frac{1}{2N(h)} * \\sum_{i=1}^{N(h)}(Z(x_i) - Z(x_{i + h}))^2 $$\n\nFor more details, please refer to the [User Guide](https://mmaelicke.github.io/scikit-gstat/userguide/variogram.html#experimental-variograms)", "_____no_output_____" ], [ "The `Variogram` class takes at least two arguments. The `coordinates` and the `values` observed at these locations. \nYou should also at least set the `normalize` parameter to explicitly, as it changes it's default value in version `0.2.8` to `False`. This attribute affects only the plotting, not the variogram values.\nAdditionally, the number of bins is set to 15, because we have fairly many observations and the default value of 10 is unnecessarily small. The `maxlag` set the maximum distance for the last bin. We know from the plot above, that more than 60 units is not really meaningful", "_____no_output_____" ] ], [ [ "V = Variogram(data[['x', 'y']].values, data.z.values, normalize=False, maxlag=60, n_lags=15)\nfig = V.plot(show=False)", "_____no_output_____" ] ], [ [ "The upper subplot show the histogram for the count of point-pairs in each lag class. You can see various things here:\n\n* As expected, there is a clear spatial dependency, because semi-variance increases with distance (blue dots)\n* The default `spherical` variogram model is well fitted to the experimental data\n* The shape of the dependency is **not** captured quite well, but fair enough for this example\n\nThe sill of the variogram should correspond with the field variance. The field is unknown, but we can compare the sill to the *sample* variance:", "_____no_output_____" ] ], [ [ "print('Sample variance: %.2f Variogram sill: %.2f' % (data.z.var(), V.describe()['sill']))", "Sample variance: 1.10 Variogram sill: 1.26\n" ] ], [ [ "The `describe` method will return the most important parameters as a dictionary. And we can simply print the variogram ob,ect to the screen, to see all parameters.", "_____no_output_____" ] ], [ [ "pprint(V.describe())", "{'effective_range': 39.50027313170537,\n 'estimator': 'matheron',\n 'name': 'spherical',\n 'nugget': 0,\n 'sill': 1.2553698556802062}\n" ], [ "print(V)", "spherical Variogram\n-------------------\nEstimator: matheron\n \rEffective Range: 39.50\n \rSill: 1.26\n \rNugget: 0.00\n \n" ] ], [ [ "## 1.3 Kriging", "_____no_output_____" ], [ "The Kriging class will now use the Variogram from above to estimate the Kriging weights for each grid cell. This is done by solving a linear equation system. For an unobserved location $s_0$, we can use the distances to 5 observation points and build the system like:\n\n$$\n\\begin{pmatrix}\n\\gamma(s_1, s_1) & \\gamma(s_1, s_2) & \\gamma(s_1, s_3) & \\gamma(s_1, s_4) & \\gamma(s_1, s_5) & 1\\\\\n\\gamma(s_2, s_1) & \\gamma(s_2, s_2) & \\gamma(s_2, s_3) & \\gamma(s_2, s_4) & \\gamma(s_2, s_5) & 1\\\\\n\\gamma(s_3, s_1) & \\gamma(s_3, s_2) & \\gamma(s_3, s_3) & \\gamma(s_3, s_4) & \\gamma(s_3, s_5) & 1\\\\\n\\gamma(s_4, s_1) & \\gamma(s_4, s_2) & \\gamma(s_4, s_3) & \\gamma(s_4, s_4) & \\gamma(s_4, s_5) & 1\\\\\n\\gamma(s_5, s_1) & \\gamma(s_5, s_2) & \\gamma(s_5, s_3) & \\gamma(s_5, s_4) & \\gamma(s_5, s_5) & 1\\\\\n1 & 1 & 1 & 1 & 1 & 0 \\\\\n\\end{pmatrix} *\n\\begin{bmatrix}\n\\lambda_1 \\\\\n\\lambda_2 \\\\\n\\lambda_3 \\\\\n\\lambda_4 \\\\\n\\lambda_5 \\\\\n\\mu \\\\\n\\end{bmatrix} =\n\\begin{pmatrix}\n\\gamma(s_0, s_1) \\\\\n\\gamma(s_0, s_2) \\\\\n\\gamma(s_0, s_3) \\\\\n\\gamma(s_0, s_4) \\\\\n\\gamma(s_0, s_5) \\\\\n1 \\\\\n\\end{pmatrix}\n$$\n\nFor more information, please refer to the [User Guide](https://mmaelicke.github.io/scikit-gstat/userguide/kriging.html#kriging-equation-system)", "_____no_output_____" ], [ "Consequently, the `OrdinaryKriging` class needs a `Variogram` object as a mandatory attribute. Two very important optional attributes are `min_points` and `max_points`. They will limit the size of the Kriging equation system. As we have 200 observations, we can require at least 5 neighbors within the range. More than 15 will only unnecessarily slow down the computation. The `mode='exact'` attribute will advise the class to build and solve the system above for each location.", "_____no_output_____" ] ], [ [ "ok = OrdinaryKriging(V, min_points=5, max_points=15, mode='exact')", "_____no_output_____" ] ], [ [ "The `transform` method will apply the interpolation for passed arrays of coordinates. It requires each dimension as a single 1D array. We can easily build a meshgrid of 100x100 coordinates and pass them to the interpolator. To recieve a 2D result, we can simply reshape the result. The Kriging error will be available as the `sigma` attribute of the interpolator.", "_____no_output_____" ] ], [ [ "# build the target grid\nxx, yy = np.mgrid[0:99:100j, 0:99:100j]\nfield = ok.transform(xx.flatten(), yy.flatten()).reshape(xx.shape)\ns2 = ok.sigma.reshape(xx.shape)", "_____no_output_____" ] ], [ [ "And finally, we can plot the result.", "_____no_output_____" ] ], [ [ "fig, axes = plt.subplots(1, 2, figsize=(16, 8))\n\nart = axes[0].matshow(field.T, origin='lower', cmap='plasma')\naxes[0].set_title('Interpolation')\naxes[0].plot(data.x, data.y, '+k')\naxes[0].set_xlim((0,100))\naxes[0].set_ylim((0,100))\nplt.colorbar(art, ax=axes[0])\nart = axes[1].matshow(s2.T, origin='lower', cmap='YlGn_r')\naxes[1].set_title('Kriging Error')\nplt.colorbar(art, ax=axes[1])\naxes[1].plot(data.x, data.y, '+w')\naxes[1].set_xlim((0,100))\naxes[1].set_ylim((0,100));", "_____no_output_____" ] ], [ [ "From the Kriging error map, you can see how the interpolation is very certain close to the observation points, but rather high in areas with only little coverage (like the upper left corner).", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import pickle\nimport numpy as np\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom sklearn.model_selection import train_test_split, cross_val_score\nfrom sklearn.metrics import classification_report, confusion_matrix\nfrom sklearn.model_selection import GridSearchCV\nfrom sklearn.model_selection import StratifiedShuffleSplit\n\nfrom sklearn import metrics\nfrom sklearn.ensemble import RandomForestClassifier\n\n#試取資料\nfile = open(\"/Users/peggy/Documents/109-2(2-2)/machine_learning/MLGame/games/snake/log/snake1.pickle\", \"rb\")\ndata = pickle.load(file)\nfile.close()\ntype(data['ml'])", "_____no_output_____" ], [ "game_info = data[\"ml\"][\"scene_info\"]\ngame_command = data[\"ml\"][\"command\"]\n#print(game_info)\n#print(game_command)", "_____no_output_____" ], [ "\nfor i in range(2,3): \n path = \"/Users/peggy/Documents/109-2(2-2)/machine_learning/MLGame/games/snake/log/snake\" + str(i) + \".pickle\"\n file = open(path, \"rb\")\n data = pickle.load(file)\n game_info = game_info + data['ml']['scene_info']\n game_command = game_command + data['ml']['command']\n file.close() \n\nprint(len(game_info))\nprint(len(game_command))\n", "158034\n158034\n" ], [ "g = game_info[1]\n\nfeature = np.array([g[\"snake_head\"][0], g[\"snake_head\"][1]])\nprint(feature)\nprint(game_command[1])\nif game_command[1] == \"UP\": game_command[1] = 0\nelif game_command[1] == \"DOWN\": game_command[1] = 1\nelif game_command[1] == \"LEFT\": game_command[1] = 2\nelif game_command[1] == \"RIGHT\": game_command[1] = 3\nelse: game_command[1] = 4 ", "[40 50]\nLEFT\n" ], [ "for i in range(2, len(game_info) - 1):\n g = game_info[i]\n \n feature = np.vstack((feature, [g[\"snake_head\"][0], g[\"snake_head\"][1]]))\n if game_command[i] == \"UP\": game_command[i] = 0\n elif game_command[i] == \"DOWN\": game_command[i] = 1\n elif game_command[i] == \"LEFT\": game_command[i] = 2\n elif game_command[i] == \"RIGHT\": game_command[i] = 3\n else: game_command[i] = 4 \n \nanswer = np.array(game_command[1:-1])\nprint(feature)\nprint(feature.shape)\nprint(answer)\nprint(answer.shape)", "[[ 40 50]\n [ 30 50]\n [ 20 50]\n ...\n [270 0]\n [280 0]\n [290 0]]\n(158032, 2)\n[2 2 2 ... 3 3 3]\n(158032,)\n" ] ], [ [ "# KNN\n", "_____no_output_____" ] ], [ [ "import pickle\nimport numpy as np\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom sklearn.model_selection import train_test_split, cross_val_score\nfrom sklearn.metrics import classification_report, confusion_matrix\nfrom sklearn.model_selection import GridSearchCV\nfrom sklearn.model_selection import StratifiedShuffleSplit\n\n#資料劃分\nx_train, x_test, y_train, y_test = train_test_split(feature, answer, test_size=0.3, random_state=9)\n#參數區間\nparam_grid = {'n_neighbors':[1, 2, 3, 4, 5]}\n#交叉驗證 \ncv = StratifiedShuffleSplit(n_splits=2, test_size=0.3, random_state=12)\ngrid = GridSearchCV(KNeighborsClassifier(n_neighbors = 5), param_grid, cv=cv, verbose=10, n_jobs=-1) #n_jobs為平行運算的數量\ngrid.fit(x_train, y_train)\ngrid_predictions = grid.predict(x_test)\n\n#儲存\nfile = open('model_KNN.pickle', 'wb')\npickle.dump(grid, file)\nfile.close()\n", "Fitting 2 folds for each of 5 candidates, totalling 10 fits\n" ], [ "#最佳參數\nprint(grid.best_params_)\n#預測結果\n#print(grid_predictions)\n#混淆矩陣\nprint(confusion_matrix(y_test, grid_predictions))\n#分類結果\nprint(classification_report(y_test, grid_predictions))", "{'n_neighbors': 5}\n[[6164 75 373 345 0]\n [ 71 6402 356 352 0]\n [ 351 360 8975 105 0]\n [ 368 397 123 8683 0]\n [ 5 5 7 25 1]]\n precision recall f1-score support\n\n 0 0.89 0.89 0.89 6957\n 1 0.88 0.89 0.89 7181\n 2 0.91 0.92 0.91 9791\n 3 0.91 0.91 0.91 9571\n 4 1.00 0.02 0.05 43\n\n accuracy 0.90 33543\n macro avg 0.92 0.72 0.73 33543\nweighted avg 0.90 0.90 0.90 33543\n\n" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import requests \nfrom bs4 import BeautifulSoup \nimport re", "_____no_output_____" ], [ "# start_url = 'https://www.tripadvisor.com.au/Hotel_Review-g255100-d255744-Reviews-Atlantis_Hotel-Melbourne_Victoria.html'\n#start_url = 'https://www.tripadvisor.com.au/Restaurant_Review-g255100-d9554485-Reviews-or10-Hochi_Mama-Melbourne_Victoria.html#REVIEWS'\n# start_url = 'https://www.yelp.com/biz/the-meat-and-wine-co-melbourne-2'\nstart_url = 'https://www.yelp.com/biz/the-meat-and-wine-co-melbourne-2?start=30'", "_____no_output_____" ], [ "# split the url to different parts\nurl_parts = start_url.split('-Reviews-')\nurls = []\nreviewArr = []\nif \"Hotel_Review\" in start_url:\n if \"-or\" not in start_url:\n for page in range(0, 5):\n url = url_parts[0]+'-Reviews-'+'or{}-'.format(5*page)+url_parts[1]\n print(url)\n urls.append(url)\n else:\n # find all URLS of reviews (define how many pages are needed, here we set it for 5 page) for this hotel\n for page in range(int(int(url_parts[1][2:4])/5), int(int(url_parts[1][2:4])/5 + 5)):\n url = url_parts[0]+'-Reviews-'+'or{}-'.format(5*page)+url_parts[1][5:]\n print(url)\n urls.append(url)\n # extract all reviews from all urls and store in reviewArr(json objects)\n for url in urls:\n response = requests.get(url,timeout=10)\n try:\n status = response.status_code\n print(status)\n except Exception as e:\n print(e)\n content = BeautifulSoup(response.content,\"html.parser\")\n for review in content.findAll('div', attrs={\"class\": \"_2wrUUKlw _3hFEdNs8\"}):\n reviewObject = {\n \"review_title\": review.find('div', attrs={\"class\": \"glasR4aX\"}).text,\n \"review\": review.find('q', attrs={\"class\": \"IRsGHoPm\"}).get_text(separator='\\n'),\n \"review_rating\":str(review.find('div', attrs={\"class\": \"nf9vGX55\"}).find('span'))[-11:-10],\n \"date_of_stay\":review.find('span', attrs={\"class\": \"_34Xs-BQm\"}).text[14:],\n \"review_date\": review.find('div', attrs={\"class\": \"_2fxQ4TOx\"}).text}\n print(reviewObject)\n reviewArr.append(reviewObject)\nelif \"Restaurant_Review\" in start_url:\n if \"-or\" not in start_url:\n for page in range(0,5):\n url = url_parts[0]+'-Reviews-'+'or{}-'.format(10*page)+url_parts[1]\n print(url)\n urls.append(url)\n else: \n for page in range(int(int(url_parts[1][2:4])/10), int(int(url_parts[1][2:4])/10 + 5)):\n url = url_parts[0]+'-Reviews-'+'or{}-'.format(10*page)+url_parts[1][5:]\n print(url)\n urls.append(url)\n for url in urls:\n response = requests.get(url,timeout=10)\n try:\n status = response.status_code\n print(status)\n except Exception as e:\n print(e)\n content = BeautifulSoup(response.content,\"html.parser\")\n for review in content.findAll('div', attrs={\"class\": \"reviewSelector\"}):\n reviewObject = {\n \"review_title\": review.find('span', attrs={\"class\": \"noQuotes\"}).text,\n \"review\": review.find('p', attrs={\"class\": \"partial_entry\"}).text.replace(\"\\n\", \"\"),\n \"review_rating\":str(review.find('div',attrs={\"class\": \"ui_column is-9\"}).find('span'))[-11:-10],\n \"date_of_visit\":review.find('div', attrs={\"class\": \"prw_rup prw_reviews_stay_date_hsx\"}).text[15:],\n \"review_date\": review.find('span', attrs={\"class\": \"ratingDate\"}).text.strip()\n }\n print(reviewObject)\n reviewArr.append(reviewObject)\nelif \"www.yelp.com\" in start_url:\n if \"?start=\" not in start_url:\n for page in range(0, 5):\n url = start_url + '?start={}'.format(10*page)\n print(url)\n urls.append(url)\n else:\n for page in range(int(int(start_url[-2:])/10), int(int(start_url[-2:])/10 + 5)):\n url = start_url[:-9] + '?start={}'.format(10*page)\n print(url)\n urls.append(url)\n for url in urls:\n response = requests.get(url,timeout=10)\n try:\n status = response.status_code\n print(status)\n except Exception as e:\n print(e)\n content = BeautifulSoup(response.content,\"html.parser\")\n for review in content.findAll('div',{\"class\":\"review__373c0__13kpL border-color--default__373c0__2oFDT\"}):\n reviewObject = {\n \"review\": review.find('p', attrs={\"class\": \"comment__373c0__1M-px css-n6i4z7\"}).text.replace(\"\\xa0\", \"\"),\n \"review_rating\": review.select('[aria-label*=rating]')[0]['aria-label'][:1],\n \"review_date\": review.find('span', attrs={\"class\": \"css-e81eai\"}).text\n }\n print(reviewObject)\n reviewArr.append(reviewObject)\nelse: \n print(\"Please only paste Valid URL of hotel or restaurant review from Trip Advisor and Yelp!\")", "https://www.yelp.com/biz/the-meat-and-wine-co-melbourne-2?start=30\nhttps://www.yelp.com/biz/the-meat-and-wine-co-melbourne-2?start=40\nhttps://www.yelp.com/biz/the-meat-and-wine-co-melbourne-2?start=50\nhttps://www.yelp.com/biz/the-meat-and-wine-co-melbourne-2?start=60\nhttps://www.yelp.com/biz/the-meat-and-wine-co-melbourne-2?start=70\n200\n{'review': \"Came here for lunch on a Friday with a friend who works close by and boyyyyy was this place packed even at 1:45pm. The service was great! Daniel our wait staff was friendly and everyone was just uber nice, from the guy who took us to our tables, to the guy who brought out our food to the one who kept asking if we wanted another drink. Definitely 5 stars for service. Food was amazing... I got the pulled pork burger with chips and my friend got battered fish and chips - delightful! This is definitely my go to place whenever I'm in southgate.\", 'review_rating': '5', 'review_date': '3/6/2016'}\n{'review': 'Warm beer and tough steak cooked wrong... Epic fail... Steak ordered medium, served blue, sent back, same piece returned almost well done, shriveled up and not nice at all.. My boss was paying and said I should order the wagyu rib-eye as it would be the \"best\". It was tiny, (maybe half the 300g on the menu). It was also very sinewy. I cook better steak from the supermarket at home than this. I couldn\\'t wait to get back to my hotel room so I could go and get some food. Bad, bad service; we never got the veges we ordered. Other people in our party had ribs and said they were good...Really noisy and hard to hear the waiters too.', 'review_rating': '1', 'review_date': '2/16/2017'}\n{'review': 'Really liked this place.Had a nice ambiance and friendly staff.Had the rump which came with a salad.The meat was well presented and had an excellent taste at medium rare.The salad was impressively fresh with a nice blend of lettuce and vegetables, had a nice vinaigrette dressing and was bigger than the condiment-sized salads that seem to be a trademark in that area of town.All and all I thought it was a very good value considering its location.', 'review_rating': '4', 'review_date': '5/16/2016'}\n{'review': \"Fascinating decor and atmosphere. Just loved it! Came here with a large group of hungries (12). Although we didn't have reservation and they were BUSY we got a table in less than a half hour. And in the mean time we were well tended at the upstairs bar. Nice!I had the beef kabob with chips and a peri peri dip. Just outstanding. Cooked perfectly and A LOT of food. A number of folks had the Lamb shank and the feedback was that it was nicely executed as well. It looked gooood too!Great wine selection and oh if you are a vegetarian don't be scared away, they have an assortment of options that folks In our group found very good too. Service was good. Host staff took great care of us. Give this place a try and let me know what YOU think.\", 'review_rating': '4', 'review_date': '5/10/2014'}\n{'review': 'Visited this establishment on Thursday night and confirmed why this is my favourite restaurant. Service was excellent and the food was sensational. I have never had a bad experience here!', 'review_rating': '5', 'review_date': '11/14/2015'}\n{'review': \"Specialising in steaks i find that they do a decent job on cooking the steaks. I've had my luck with over cooked steak and some steaks cooked perfectly. The price is decent for a steak if you're having a date night with your partner. Interior design is quiet fancy and is very romantic. They also have private rooms for functions and parties. Love the wine list as they have so much to offer. Great service\", 'review_rating': '3', 'review_date': '10/14/2015'}\n{'review': 'Great Steak Restaurant on the Yarra Valley river front. Great food, and wine list. Large portions, totally delish. Highly recommend', 'review_rating': '5', 'review_date': '5/29/2018'}\n{'review': \"The kangaroo was fantastic! It comes with onion rings, baked potato and a delicious sweety sauce. The best dish I had ever! The environment is nice and The service is fast. The prices aren't such high. It's worth!\", 'review_rating': '5', 'review_date': '5/24/2016'}\n{'review': \"One of my favourite fine dining restaurants! Food is absolutely amazing every time I go, waiters are absolutely kind and understanding and the atmosphere is 'just right'! Can't complain!\", 'review_rating': '5', 'review_date': '10/30/2017'}\n{'review': \"Travel to Melbourne 3 or 4 times a year and come here most trips - have been pre and post renovation and really enjoyed the atmosphere and setup both before and after.What surprises me is that coming from Adelaide where we do have some excellent steak venues in the form of Argentinian (Sosta, Gaucho, Buenos Aires etc), is that the steaks are priced on par if not cheaper then the Adelaide equivalent. Considering the Southbank location and excellent quality of food I was expecting far higher prices. Anyway on to the food -Tasting platter is always a excellent entre. I always stick with either the rib eye, pork ribs or the dish which combines both. Always great and on par with the best ribs and steak's i've had anywhere. Only minor complaint last visit is that my side of mash was extremely rich / creamy - too much butter/ cheese I assume. Wasn't like this before so probably a one off. Normally when I visit Southbank its at the start of a night out and I'm with a group having a few drinks. This is where the extensive spirit and wine menu comes in handy - one of the most impressive lists I've come across in a resturant. Plenty of premium scotch to choose from!I'll be back next visit\", 'review_rating': '5', 'review_date': '8/17/2014'}\n200\n{'review': \"Second time I have been here, first time was in Sydney.We tried to book a table with very short notice and couldn't get one, however the young lady on the phone told us to just pop in and we could go up to the bar and wait for a table as people always cancel or we could just eat and order in the bar. We chose to just eat and order in the bar as we don't particularly like waiting around for any food for too long :)Me and my partner ordered ribs and just as I had suspected they were ridiculously good ribs, I mean they are ridiculous.Wish this place was in Adelaide so we can taste all the magnificent items on the menu!Do yourself a favour and stop in!!!!\", 'review_rating': '5', 'review_date': '10/1/2014'}\n{'review': \"I went there with my girlfriend on her birthday. For entree, I ordered the scallop and prawn. The prawn was a bit spicy. For main, I ordered a rib eye and lamb rib. The rib eye was a bit disappointed with considering the price, however, the lamb rib was stunning with excellent seasoning. Also, I ordered a creme bulee and semiferddo. The desserts are all fantastic. What's more, the waiting staff were friendly and nice.\", 'review_rating': '4', 'review_date': '9/11/2016'}\n{'review': \"Been here several times and it never disappoints! If you're after a great steak fix this is one of the best places to go! Staff were great, friendly and knowledgeable about the food and wine. Have used the private dinning room upstairs for birthday dinners which has a great view of Southbank and the service was excellent. All the food came out at around the same time for the whole group and everyone was happy!\", 'review_rating': '5', 'review_date': '5/25/2017'}\n{'review': 'Excellent as always, Staff are friendly, attentive and know the menu well and are helpful with suggestions we where served by (Shaz, she was great) . The food is excellent would happily return here again and again. Prices are surprisingly good for the quality of food.', 'review_rating': '4', 'review_date': '8/13/2017'}\n{'review': 'This places ambience was incredible, quite mellow and the meat basically melted in my mouth. The only downfall is that I almost tripped over their carpet upon entering and their hot mustard is way too intense.', 'review_rating': '4', 'review_date': '6/9/2017'}\n{'review': 'I feel like I missed the boat here, as I tried the kangaroo steak for novelty reasons and was unimpressed compared to the delicious beef steaks all of my friends got. Definitely a good meal, but I had steak envy.', 'review_rating': '4', 'review_date': '2/15/2014'}\n{'review': \"The Meat and Wine Co is a steakhouse of the highest calibre. If you want aged beef that has been basting in house and then cooked to perfection, there are few places I know better than the Meat and Wine Co (perhaps The Italian does a better steak).The Meat & Wine Co, Melbourne is located at Freshwater Place, Queensbridge Square in Southbank and has stunning views of the Yarra River and the city skyline. The steaks are mammoth and delicious. They also specialize in ribs - lamb, beef and pork. Although you could be mistaken for confusing them for Flintstonesesque dinosaur ribs (they are HUGE). For $49, you can give go all out and have a steak and ribs. You even get a massive bib for the massive mess you are about to make.The Meat and Wine Co isn't cheap. But you do get a quality feed that is well worth it. Enjoy pigging out!\", 'review_rating': '4', 'review_date': '10/15/2011'}\n{'review': \"Next to and with views of the Yarra River this hip looking steak house is great. Good wine selection although nothing from CA or OR. The meats are very good quality but they overdue the orders a little but not enough to complain at the restaurant. The meats are complimented with a variety of sauces, tried two of them, one was great and the other slightly bland. The portugués suace was amazing and the garlic sauce was bland and tasteless. Service was good and attentive although multiple servers would come and top up our glasses which I don't like much, I rather our main server attend to us more. It was good just good. Maybe I'll try it again for AO 2018.\", 'review_rating': '3', 'review_date': '1/27/2017'}\n{'review': 'Love this place. It is casual yet trendy and comfortable. Steaks amazing and hubby went for Wagu rib eye. He is in heaven. I had the sumptuous Salmon on a hanging skewer cooked medium. Drink list well composed and happy hour drinks upstairs.', 'review_rating': '4', 'review_date': '11/23/2015'}\n{'review': 'I went to this restaurant 3 days in a row, from Sydney to Melbourne you cud c how much I lv this restaurant. Unfortunately I had an embarrassing dinning experience tonite. I saw a cockroach climbing slowly on the wall while I was dinning happily w my brother. I almost shouted when I saw it. It was really disgusting to cover the good taste of half rack of the beef rib. I guess no more four days in a row.', 'review_rating': '3', 'review_date': '8/6/2014'}\n200\n{'review': 'Food was great, but the wait staff pulled the rating down. I stopped by on a Tuesday night, and as a singleton, they were available to seat me immediately. Unfortunately my seat was all the way at the back of the restaurant, but hey - \"Party of One\", I didn\\'t expect much better. I waited a longish time after being seated to see a waiter and I was seriously considering walking out. My waiter did finally appear and I ordered a smallish steak with a side order of onion strings and macaroni and cheese. The steak was very tasty as were the side dishes. Overall I was happy with my experience, but the service could have been better - at least speeding up the first contact with the client. I understand that this is a very busy place and a single diner may not be the highest priority but it did pull down my rating a bit.', 'review_rating': '4', 'review_date': '8/30/2013'}\n{'review': \"I have been to this brand over a few years but in Darling Harbour not the South Bank one and it was awesome every time. This one just doesn't hit the mark at all! It now makes perfect sense that it has McDonalds to the left and TGI to the right! It's in great company. Very disappointing.\", 'review_rating': '2', 'review_date': '3/9/2018'}\n{'review': 'Was here for a 40th dinner with friends. The service is amazing and the food was so lovely!! My suggestion is that if you like your steak medium ask for medium rare. My steak was a little over cooked but still yummy!!!', 'review_rating': '4', 'review_date': '12/12/2016'}\n{'review': 'fantastic experience and extremely enjoy the food and wine there. the service is great and the view is good too.', 'review_rating': '5', 'review_date': '10/2/2016'}\n{'review': 'So far, our favorite food spot on our trip to Melbourne. Our group of 11 from the USA, all raved about how well seasoned and flavorful their dishes were. I had the Portuguese Chicken and chick pea salad and both were awesome. Food was served quickly to our large group. Atmosphere was industrial modern and very clean.', 'review_rating': '5', 'review_date': '8/17/2015'}\n{'review': \"-Should have been 4 stars as steaks were really good - but the wait staff ... although polite, they messed up a few times- Few errors such as sauces on steaks when not requested, wrong entrees etc, but the big one; 2 waitresses did not know what steaks (as in literally not even the cut) they were serving us. We have a pretty good knowledge of cuts so we could sort it out, but for people who don't, that could mean the difference between eating a $34 Monte rib-eye when you've ordered a $59 wagyu rib-eye-Chewy steak sandwich; steak was well-done-TL;DR: Great steaks, staff need more training\", 'review_rating': '3', 'review_date': '12/1/2012'}\n{'review': \"It's a chain restaurant, I know. But the steak is bloody good as all the rest, wine selection really interesting and lots of different food that you can choose from. I loved it!\", 'review_rating': '4', 'review_date': '4/30/2015'}\n{'review': 'I would give it more of a 4.5, but it was excellent! Started off with the pork belly appetizer, which was just fantastic. We then got the kangaroo steak, medium rare. It was pretty perfect. My buddy said the kangaroo steak was \"AMAZING\". Great atmosphere and very relaxed. I\\'d just say to make sure to make reservations well in advance, especially on weekend nights.', 'review_rating': '5', 'review_date': '1/3/2015'}\n{'review': 'Had a fantastic experience here on Tuesday night. Took some clients for dinner and we were all super impressed, the service was fantastic and all of our steaks were cooked to perfection.....recommended!!!', 'review_rating': '5', 'review_date': '9/21/2016'}\n{'review': \"Well, I think this is quite a decent place to have a steak. I ordered Medium Rare Rib Eye, but the waiter recommended a Medium instead. So I accepted his recommendation. But when he came it looked Medium Rare. So was it an error. or was it a wrong order. or is that their Medium . Not sure. but it still tasted good, no complains. Loved the African Hot Sauce, suits you only if you can have spicy food. cos' i love spicy :) definitely will go once in awhile.\", 'review_rating': '4', 'review_date': '6/9/2014'}\n200\n{'review': \"Definitely still a fan of this place..but I guess it's just my meat loving instincts taking over.I'm not much of a salad eater but, from the salads I've had in my lifetimes, this place makes the best caesar salad around (minus the anchovies).\", 'review_rating': '4', 'review_date': '6/15/2012'}\n{'review': 'Finally had the chance to experience the hypeWell I am pleased to say Meat & Wine Co. lived up to it Although only being a in Melb a short time I have had the chance to try out a number of steak houses and usually stick with 2 staples .... Steak and ShirazThe Rib-eye selected had been wonderfully aged and seasoned, then perfectly cooked resulting in the best steak I have experienced in Melbourne thus farThe chips were crispy and hot (which is a plus compared to some places) and the peppercorn sauce was good also The wine list was of reasonable length and provided for a decent priced Shiraz to accompany the mealDespite the warning of prices going in I was pleasantly surprised to find them similar to comparable Steak HousesHighly recommend if your after a steak worth remembering P.S. No salad was harmed in the consumption of this meal', 'review_rating': '4', 'review_date': '9/12/2012'}\n{'review': 'Food www okay and service was mediocre. For the price had hoped for much better. Also, no a/c so definitely not a hot day spot.', 'review_rating': '2', 'review_date': '3/1/2017'}\n{'review': 'Great staff and service we had a red balloon voucher I asked if they could change one of the mains and they did. Also we did not like the table we were seated in so we asked to move and we did. One of my favourite restaurants to go to. Fantastic tasty food!', 'review_rating': '4', 'review_date': '2/8/2014'}\n{'review': 'Great steak and a romantic/bromantic atmosphere. Never a disappointment. One of the best steakhouses around!', 'review_rating': '5', 'review_date': '8/13/2013'}\n{'review': \"Not being the biggest meat eater of all time, I was a little worried about what to order at the Meat & Wine Co. I ordered the kangaroo fillet which was cooked how it should be, a little rare but not too bloody and it was marinated beautifully. I haven't sampled much else on the menu but I am eager to return. This place gets busy so its recommended to book. Also a great venue for groups and work functions.\", 'review_rating': '3', 'review_date': '11/7/2011'}\n{'review': 'Great food, pity about the crappy service, our waitress was no where to be seen the entire night; had to keep asking other servers when we wanted more bottles of wine or to top up water. Plus their inability to cater for even the most basic needs of a disabled customer we were entertaining was flabbergasting!!!', 'review_rating': '1', 'review_date': '4/14/2014'}\n{'review': \"The only place I've found in Melbourne that cooks full sized beef ribs. WHich is about the only thing that can tide me over between trips to Texas.Steak and ribs are superb here, and a reservation is definitely recommended.The burger, however, is a travesty. Stick with the basic cuts of meat, devour the chunky, crispy fries and enjoy the extra iron hit from the red meat.\", 'review_rating': '3', 'review_date': '11/5/2011'}\n{'review': 'Our first night in Melbourne we asked the concierge of our hotel if they knew of a place that served kangaroo meat. He called around and after about 15 minutes found this place. He even called and got us reservations for that night. We got there and the place was packed. So we knew that this place would be good. I had kangaroo steak and my wife had a regular cow steak. The food was really good. The atmosphere was a little loud and they tried to maximize the amount of people in the restaurant. But over all the experience was fantastic. Would love to go back here again.', 'review_rating': '4', 'review_date': '4/11/2012'}\n{'review': 'I have been here about a year ago and today visited again. I ordered garlic bread which was like rock with no flavour. I also ordered my usual favourite.... Pork ribs. Very much lacked flavour. The quality of food has certainly dropped, I would normal give a five star.', 'review_rating': '3', 'review_date': '9/20/2014'}\n200\n{'review': 'As a South African living in Melbourne I am very proud of this restaurant showcasing good quality South African wines to the rest of Aus and the world and also giving them a small taste of what a South African restaurant is all about. A little bit of home far from home. Doen so voort manne!', 'review_rating': '5', 'review_date': '6/7/2012'}\n{'review': 'Each time I have eaten here, trully enjoyed it. ABSOLUTELY love the food. RIBS were perfect! Pricing as you would expect , though worth it.', 'review_rating': '5', 'review_date': '7/18/2013'}\n" ], [ "response = requests.get(start_url,timeout=10)\ncontent = BeautifulSoup(response.content,\"html.parser\")", "_____no_output_____" ], [ "divs = content.findAll('div',attrs={\"class\":\"ui_pagination\"})", "_____no_output_____" ], [ "len(divs)", "_____no_output_____" ], [ "divs[0]", "_____no_output_____" ], [ "for c in divs[0].children:\n print(c)\n print()", "<span class=\"ui_button nav previous secondary disabled\">Previous</span>\n\n<a class=\"ui_button nav next primary\" href=\"/Hotel_Review-g255100-d255744-Reviews-or5-Atlantis_Hotel-Melbourne_Victoria.html\">Next</a>\n\n<div class=\"pageNumbers\"><span class=\"pageNum current disabled\">1</span><a class=\"pageNum\" href=\"/Hotel_Review-g255100-d255744-Reviews-or5-Atlantis_Hotel-Melbourne_Victoria.html\">2</a><a class=\"pageNum\" href=\"/Hotel_Review-g255100-d255744-Reviews-or10-Atlantis_Hotel-Melbourne_Victoria.html\">3</a><a class=\"pageNum\" href=\"/Hotel_Review-g255100-d255744-Reviews-or15-Atlantis_Hotel-Melbourne_Victoria.html\">4</a><a class=\"pageNum\" href=\"/Hotel_Review-g255100-d255744-Reviews-or20-Atlantis_Hotel-Melbourne_Victoria.html\">5</a><a class=\"pageNum\" href=\"/Hotel_Review-g255100-d255744-Reviews-or25-Atlantis_Hotel-Melbourne_Victoria.html\">6</a><span class=\"separator\">…</span><a class=\"pageNum\" href=\"/Hotel_Review-g255100-d255744-Reviews-or2120-Atlantis_Hotel-Melbourne_Victoria.html\">425</a></div>\n\n" ], [ "for c in divs[0].children:\n if 'pageNumbers' in c['class']:\n for cc in c.children:\n # print(cc)\n p = {}\n p['name'] = cc.string\n if 'disabled' in cc['class']:\n p['status'] = 'disable'\n else:\n p['status'] = None\n if 'href' in cc.attrs:\n p['url'] = cc['href']\n else:\n p['url'] = None\n print(p)\n else:\n p = {}\n p['name'] = c.string\n if 'disabled' in c['class']:\n p['status'] = 'disable'\n else:\n p['status'] = None\n if 'href' in c.attrs:\n p['url'] = c['href']\n else:\n p['url'] = None\n print(p)", "{'name': 'Previous', 'status': 'disable', 'url': None}\n{'name': 'Next', 'status': None, 'url': '/Hotel_Review-g255100-d255744-Reviews-or5-Atlantis_Hotel-Melbourne_Victoria.html'}\n{'name': '1', 'status': 'disable', 'url': None}\n{'name': '2', 'status': None, 'url': '/Hotel_Review-g255100-d255744-Reviews-or5-Atlantis_Hotel-Melbourne_Victoria.html'}\n{'name': '3', 'status': None, 'url': '/Hotel_Review-g255100-d255744-Reviews-or10-Atlantis_Hotel-Melbourne_Victoria.html'}\n{'name': '4', 'status': None, 'url': '/Hotel_Review-g255100-d255744-Reviews-or15-Atlantis_Hotel-Melbourne_Victoria.html'}\n{'name': '5', 'status': None, 'url': '/Hotel_Review-g255100-d255744-Reviews-or20-Atlantis_Hotel-Melbourne_Victoria.html'}\n{'name': '6', 'status': None, 'url': '/Hotel_Review-g255100-d255744-Reviews-or25-Atlantis_Hotel-Melbourne_Victoria.html'}\n{'name': '…', 'status': None, 'url': None}\n{'name': '425', 'status': None, 'url': '/Hotel_Review-g255100-d255744-Reviews-or2120-Atlantis_Hotel-Melbourne_Victoria.html'}\n" ], [ "divs[0].contents[0]['class']", "_____no_output_____" ], [ "'href' in divs[0].contents[1].attrs", "_____no_output_____" ], [ "len(divs[0].contents[1])", "_____no_output_____" ], [ "divs[0].contents[0].string", "_____no_output_____" ], [ "'next' in divs[0].contents[1]['class']", "_____no_output_____" ], [ "divs[0].contents[2].contents", "_____no_output_____" ], [ "divs = content.find_all(name=\"div\",attrs={\"class\":re.compile(r\"navigation-button-container(\\s\\w+)?\")})", "_____no_output_____" ], [ "divs[0]", "_____no_output_____" ], [ "divs[1]", "_____no_output_____" ], [ "len(divs)", "_____no_output_____" ], [ "divs[0].find(\"a\",attrs={\"class\":\"prev-link\"})", "_____no_output_____" ], [ "divs[1].find(\"a\",attrs={\"class\":\"next-link\"})", "_____no_output_____" ], [ "pagination-link-container", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Bayesian Regression Using NumPyro\n\nIn this tutorial, we will explore how to do bayesian regression in NumPyro, using a simple example adapted from Statistical Rethinking [[1](#References)]. In particular, we would like to explore the following:\n\n - Write a simple model using the `sample` NumPyro primitive.\n - Run inference using MCMC in NumPyro, in particular, using the No U-Turn Sampler (NUTS) to get a posterior distribution over our regression parameters of interest.\n - Learn about utilities such as `initialize_model` that are useful for running HMC.\n - Learn how we can use effect-handlers in NumPyro to generate execution traces, condition on sample sites, seed models with RNG seeds, etc., and use this to implement various utilities that will be useful for MCMC. e.g. computing model log likelihood, generating empirical distribution over the posterior predictive, etc.\n\n## Tutorial Outline:\n1. [Dataset](#Dataset)\n2. [Regression Model to Predict Divorce Rate](#Regression-Model-to-Predict-Divorce-Rate)\n - [Model-1: Predictor-Marriage Rate](#Model-1:-Predictor---Marriage-Rate)\n - [Posterior Distribution over the Regression Parameters](#Posterior-Distribution-over-the-Regression-Parameters)\n - [Posterior Predictive Distribution](#Posterior-Predictive-Distribution)\n - [Model Log Likelihood](#Model-Log-Likelihood)\n - [Model-2: Predictor-Median Age of Marriage](#Model-2:-Predictor---Median-Age-of-Marriage)\n - [Model-3: Predictor-Marriage Rate and Median Age of Marriage](Model-3:-Predictor---Marriage-Rate-and-Median-Age-of-Marriage)\n - [Divorce Rate Residuals by State](#Divorce-Rate-Residuals-by-State)\n3. [Regression Model with Measurement Error](#Regression-Model-with-Measurement-Error)\n - [Effect of Incorporating Measurement Noise on Residuals](#Effect-of-Incorporating-Measurement-Noise-on-Residuals)\n4. [References](#References)", "_____no_output_____" ] ], [ [ "%reset -s -f", "_____no_output_____" ], [ "import jax\nimport jax.numpy as np\nfrom jax import random, vmap\nfrom jax.config import config; config.update(\"jax_platform_name\", \"cpu\")\nfrom jax.scipy.special import logsumexp\nimport matplotlib\nimport matplotlib.pyplot as plt\nimport numpy as onp\nimport pandas as pd\nimport seaborn as sns\n\nfrom numpyro.diagnostics import hpdi\nimport numpyro.distributions as dist\nfrom numpyro.handlers import sample, seed, substitute, trace\nfrom numpyro.hmc_util import initialize_model\nfrom numpyro.mcmc import mcmc\n\n%matplotlib inline\nplt.style.use('bmh')\nplt.rcParams.update({'font.size': 16,\n 'xtick.labelsize': 14,\n 'ytick.labelsize': 14,\n 'axes.titlesize': 'large', \n 'axes.labelsize': 'medium'})", "_____no_output_____" ] ], [ [ "## Dataset\n\nFor this example, we will use the `WaffleDivorce` dataset from Chapter 05, Statistical Rethinking [[1](#References)]. The dataset contains divorce rates in each of the 50 states in the USA, along with predictors such as population, median age of marriage, whether it is a Southern state and, curiously, number of Waffle Houses.", "_____no_output_____" ] ], [ [ "DATASET_URL = 'https://raw.githubusercontent.com/rmcelreath/rethinking/master/data/WaffleDivorce.csv'\ndset = pd.read_csv(DATASET_URL, sep=';')\ndset", "_____no_output_____" ] ], [ [ "Let us plot the pair-wise relationship amongst the main variables in the dataset, using `seaborn.pairplot`. ", "_____no_output_____" ] ], [ [ "vars = ['Population', 'MedianAgeMarriage', 'Marriage', 'WaffleHouses', 'South', 'Divorce']\nsns.pairplot(dset, x_vars=vars, y_vars=vars, palette='husl');", "_____no_output_____" ] ], [ [ "From the plots above, we can clearly observe that there is a relationship between divorce rates and marriage rates in a state (as might be expected), and also between divorce rates and median age of marriage. \n\nThere is also a weak relationship between number of Waffle Houses and divorce rates, which is not obvious from the plot above, but will be clearer if we regress `Divorce` against `WaffleHouse` and plot the results. This is an example of a spurious association. We do not expect the number of Waffle Houses in a state to affect the divorce rate, but it is likely correlated with other factors that have an effect on the divorce rate. We will not delve into this spurious association in this tutorial, but the interested reader is encouraged to read Chapters 5 and 6 of [[1](#References)] which explores the problem of causal association in the presence of multiple predictors. \n\nFor simplicity, we will primarily focus on marriage rate and the median age of marriage as our predictors for divorce rate throughout the remaining tutorial.", "_____no_output_____" ] ], [ [ "sns.regplot('WaffleHouses', 'Divorce', dset);", "_____no_output_____" ] ], [ [ "## Regression Model to Predict Divorce Rate\n\nLet us now write a regressionn model in *NumPyro* to predict the divorce rate as a linear function of marriage rate and median age of marriage in each of the states. \n\nFirst, note that our predictor variables have somewhat different scales. It is a good practice to standardize our predictors and response variables to mean `0` and standard deviation `1`, which should result in faster inference. Refer to this [note](https://mc-stan.org/docs/2_19/stan-users-guide/standardizing-predictors-and-outputs.html) in the Stan manual for more details.", "_____no_output_____" ] ], [ [ "dset['AgeScaled'] = (dset.MedianAgeMarriage - onp.mean(dset.MedianAgeMarriage)) / onp.std(dset.MedianAgeMarriage)\ndset['MarriageScaled'] = (dset.Marriage - onp.mean(dset.Marriage)) / onp.std(dset.Marriage)\ndset['DivorceScaled'] = (dset.Divorce - onp.mean(dset.Divorce)) / onp.std(dset.Divorce)", "_____no_output_____" ] ], [ [ "We write the NumPyro model as follows. While the code should largely be self-explanatory, take note of the following:\n\n - In NumPyro, model code is any Python callable that can accept arguments and keywords. For HMC which we will be using for this tutorial, these arguments and keywords cannot change during model execution. This is convenient for passing in numpy arrays, or boolean arguments that might affect the execution path.\n - In addition to regular Python statements, the model code also contains primitives like `sample`. These primitives can be interpreted with various side-effects by effect handlers used by inference algorithms in NumPyro. For more on effect handlers, refer to [[3](#References)], [[4](#References)]. For now, just remember that a `sample` statement makes this a stochastic function by sampling from some distribution of interest.\n - The reason why we have kept our predictors as optional keyword arguments is to be able to reuse the same model as we vary the set of predictors. Likewise, the reason why the response variable is optional is that we would like to reuse this model to sample from the posterior predictive distribution. See the [section](#Posterior-Predictive-Distribution) on plotting the posterior predictive distribution, as an example.", "_____no_output_____" ] ], [ [ "def model(marriage=None, age=None, divorce=None):\n a = sample('a', dist.Normal(0., 0.2))\n M, A = 0., 0.\n if marriage is not None:\n bM = sample('bM', dist.Normal(0., 0.5))\n M = bM * marriage\n if age is not None:\n bA = sample('bA', dist.Normal(0., 0.5))\n A = bA * age\n sigma = sample('sigma', dist.Exponential(1.))\n mu = a + M + A\n sample('obs', dist.Normal(mu, sigma), obs=divorce)", "_____no_output_____" ] ], [ [ "### Model 1: Predictor - Marriage Rate\n\nWe first try to model the divorce rate as depending on a single variable, marriage rate. As mentioned above, we can use the same `model` code as earlier, but only pass values for `marriage` and `divorce` keyword arguments. We will use the No U-Turn Sampler (see [[5](#References)] for more details on the NUTS algorithm) to run inference on this simple model.\n\n\nNote the following requirements for running HMC and NUTS in NumPyro:\n\n - The Hamiltonian Monte Carlo (or, the NUTS) implementation in Pyro takes in a potential energy function. This is the negative log joint density for the model. \n - The verlet integrator in HMC (or, NUTS) returns sample values simulated using Hamiltonian dynamics in the unconstrained space. As such, continuous variables with bounded support need to be transformed into unconstrained space using bijective transforms. We also need to transform these samples back to their constrained support before returning these values to the user.\n \nThankfully, all of this is handled on the backend for us. Let us go through the steps one by one.\n\n - JAX uses functional PRNGs. Unlike other languages / frameworks which maintain a global random state, in JAX, every call to a sampler requires an [explicit PRNGKey](https://github.com/google/jax#random-numbers-are-different). We will split our initial random seed for subsequent operations, so that we do not accidentally reuse the same seed.\n - The function [initialize_model](https://numpyro.readthedocs.io/en/latest/mcmc.html#numpyro.hmc_util.initialize_model) takes a model along with model arguments (and keyword arguments), and returns a tuple of initial parameters, potential energy function, and constrain function. The initial parameters are used to initiate the MCMC chain, the potential energy function is a callable that when given unconstrained sample values returns the potential energy at these sample values. This is used by the verlet integrator in HMC. Lastly, `constrain_fn` is a callable that transforms the unconstrained samples returned by HMC/NUTS to sample values that lie within the constrained support.\n - Finally, we use the [mcmc](https://numpyro.readthedocs.io/en/latest/mcmc.html#numpyro.mcmc.mcmc) function to run inference using the default `NUTS` sampler. Note that to run vanilla HMC, all you need to do is to pass `algo='HMC'` as argument to `mcmc` instead. This is a convenience utility that does all of the following:\n\n - Runs warmup - adapts steps size and mass matrix.\n - Uses the sample from the warmup phase to start MCMC.\n - Return samples from the posterior distribution and print diagnostic information.", "_____no_output_____" ] ], [ [ "# Start from this source of randomness. We will split keys for subsequent operations.\nrng = random.PRNGKey(0)\nrng_, rng = random.split(rng)\n\n# Initialize the model.\ninit_params, potential_fn, constrain_fn = initialize_model(rng_, model, \n marriage=dset.MarriageScaled.values, \n divorce=dset.DivorceScaled.values)\nnum_warmup, num_samples = 1000, 2000\n\n# Run NUTS.\nsamples_1 = mcmc(num_warmup, num_samples, init_params,\n potential_fn=potential_fn, \n trajectory_length=10, \n constrain_fn=constrain_fn)", "warmup: 100%|██████████| 1000/1000 [00:12<00:00, 78.24it/s, 1 steps of size 6.99e-01. acc. prob=0.79] \nsample: 100%|██████████| 2000/2000 [00:03<00:00, 515.37it/s, 3 steps of size 6.99e-01. acc. prob=0.88]\n" ] ], [ [ "#### Posterior Distribution over the Regression Parameters\n\nWe notice that the progress bar gives us online statistics on the acceptance probability, step size and number of steps taken per sample while running NUTS. In particular, during warmup, we adapt the step size and mass matrix to achieve a certain target acceptance probability (0.8, by default). We were able to successfully adapt our step size to achieve this target in the warmup phase.\n\nDuring warmup, the aim is to adapt or learn values for hyper-parameters such as step size and mass matrix (the HMC algorithm is very sensitive to these hyper-parameters), and to reach the typical set (see [[6](#References)] for more details). If there are any issues in the model specification, it might be reflected in low acceptance probabilities or very high number of steps. We use the sample from the end of the warmup phase to seed the MCMC chain (denoted by the second `sample` progress bar) from which we generate the desired number of samples from our target distribution.\n\nAt the end of inference, NumPyro prints the mean, std and 90% CI values for each of the latent parameters. Note that since we standardized our predictors and response variable, we would expect the intercept to have mean 0, as can be seen here. It also prints other convergence diagnostics on the latent parameters in the model, including [effective sample size](https://numpyro.readthedocs.io/en/latest/diagnostics.html#numpyro.diagnostics.effective_sample_size) and the [gelman rubin diagnostic](https://numpyro.readthedocs.io/en/latest/diagnostics.html#numpyro.diagnostics.gelman_rubin) ($\\hat{R}$). The value for these diagnostics indicates that the chain has converged to the target distribution. In our case, the \"target distribution\" is the posterior distribution over the latent parameters that we are interested in. Note that this is often worth verifying with multiple chains on more complicated models. In the end, `samples_1` is a collection (in our case, a `dict` since `init_samples` was a `dict`) containing samples from the posterior distribution for each of the latent parameters in the model.\n\nTo look at our regression fit, let us plot the regression line using our posterior estimates for the regression parameters, along with the 90% Credibility Interval (CI). Note that the [hpdi](https://numpyro.readthedocs.io/en/latest/diagnostics.html#numpyro.diagnostics.hpdi) function in NumPyro's diagnostics module can be used to compute CI. In the functions below, note that the collected samples from the posterior are all along the leading axis.\n\nWe can see from the plot, that the CI broadens towards the tails where values of the predictor variables are sparse, as can be expected.", "_____no_output_____" ] ], [ [ "def plot_regression(x, y_mean, y_hpdi):\n # Sort values for plotting by x axis\n idx = np.argsort(x)\n marriage = x[idx]\n mean = y_mean[idx]\n hpdi = y_hpdi[:, idx]\n divorce = dset.DivorceScaled.values[idx]\n\n # Plot\n fig, ax = plt.subplots(nrows=1, ncols=1, figsize=(6, 6))\n ax.plot(marriage, mean)\n ax.plot(marriage, divorce, 'o')\n ax.fill_between(marriage, hpdi[0], hpdi[1], alpha=0.3, interpolate=True)\n return ax\n\n# Compute empirical posterior distribution over mu\nposterior_mu = np.expand_dims(samples_1['a'], -1) + \\\n np.expand_dims(samples_1['bM'], -1) * dset.MarriageScaled.values\n\nmean_mu = np.mean(posterior_mu, axis=0)\nhpdi_mu = hpdi(posterior_mu, 0.9)\nax = plot_regression(dset.MarriageScaled.values, mean_mu, hpdi_mu)\nax.set(xlabel='Marriage rate', ylabel='Divorce rate', title='Regression line with 90% CI');", "_____no_output_____" ] ], [ [ "#### Posterior Predictive Distribution\n\nLet us now look at the posterior predictive distribution to see how our predictive distribution looks with respect to the observed divorce rates. To get samples from the posterior predictive distribution, we need to run the model by substituting the latent parameters with samples from the posterior. This sounds complicated, but this can be easily achieved by using effect handlers from the [handlers module](https://numpyro.readthedocs.io/en/latest/handlers.html).\n\nIn particular, note the use of the `substitute`, `seed` and `trace` effect handlers in the `predict` function.\n\n - The `seed` effect-handler is used to wrap a stochastic function with an initial `PRNGKey` seed. When a sample statement inside the model is called, it uses the existing seed to sample from a distribution but this effect-handler also splits the existing key to ensure that future `sample` calls in the model use the newly split key instead. This is to prevent us from having to explicitly pass in a `PRNGKey` to each `sample` statement.\n - The `substitute` effect handler simply substitutes the value for the site name present in the `post_samples` dict instead of sampling from the distribution, which can be useful for conditioning sample sites to certain values.\n - The `trace` effect handler runs the model and records the execution trace within an `OrderedDict`. This trace object contains execution metadata that is useful for computing quantities such as the log joint density.\n \nIt should be clear now that the `predict` function simply runs the model by substituting the latent parameters with samples from the posterior (generated by the `mcmc` function) to generate predictions, which are samples from the posterior predictive distribution. Note the use of JAX's auto-vectorization transform called [vmap](https://github.com/google/jax#auto-vectorization-with-vmap) to vectorize predictions. If we didn't use `vmap`, we would have to use a native for loop which for each sample which is much slower. Each draw from the posterior can be used to get predictions over all the 50 states. When we vectorize this over all the samples from the posterior using `vmap`, we will get a `predictions_1` array of shape `(num_samples, 50)`. We can then compute the mean and 90% CI of these samples to plot the posterior predictive distribution.", "_____no_output_____" ] ], [ [ "def predict(rng, post_samples, model, *args, **kwargs):\n model = substitute(seed(model, rng), post_samples)\n model_trace = trace(model).get_trace(*args, **kwargs)\n return model_trace['obs']['value']", "_____no_output_____" ], [ "# vectorize predictions via vmap\npredict_fn = vmap(lambda rng, samples: predict(rng, samples, model, marriage=dset.MarriageScaled.values))\nrng, rng_ = random.split(rng)\npredictions_1 = predict_fn(random.split(rng_, num_samples), samples_1)\n\nmean_pred = np.mean(predictions_1, axis=0)\nhpdi_pred = hpdi(predictions_1, 0.9)\n\nax = plot_regression(dset.MarriageScaled.values, mean_pred, hpdi_pred)\nax.set(xlabel='Marriage rate', ylabel='Divorce rate', title='Predictions with 90% CI');", "_____no_output_____" ] ], [ [ "We will use the same `plot_regression` function as earlier. We notice that our CI for the predictive distribution is much broader as compared to the last plot due to the additional noise introduced by the `sigma` parameter. Note that most data points lie well within the 90% CI, which indicates a good fit.\n\n#### Model Log Likelihood\n\nLikewise, making use of effect-handlers and `vmap`, we can also compute the log likelihood for this model given the dataset.", "_____no_output_____" ] ], [ [ "def log_lk(rng, params, model, *args, **kwargs):\n model = substitute(seed(model, rng), params)\n model_trace = trace(model).get_trace(*args, **kwargs)\n obs_node = model_trace['obs']\n return np.sum(obs_node['fn'].log_prob(obs_node['value']))\n \ndef expected_log_likelihood(rng, params, model, *args, **kwargs):\n n = list(params.values())[0].shape[0]\n log_lk_fn = vmap(lambda rng, params: log_lk(rng, params, model, *args, **kwargs))\n log_lk_vals = log_lk_fn(random.split(rng, n), params)\n return logsumexp(log_lk_vals) - np.log(n)", "_____no_output_____" ], [ "rng, rng_ = random.split(rng)\nprint('Log likelihood: {}'.format(expected_log_likelihood(rng_,\n samples_1, \n model,\n marriage=dset.MarriageScaled.values,\n divorce=dset.DivorceScaled.values)))", "Log likelihood: -68.14618682861328\n" ] ], [ [ "### Model 2: Predictor - Median Age of Marriage\n\nWe will now model the divorce rate as a function of the median age of marriage. The computations are mostly a reproduction of what we did for Model 1. Notice the following:\n\n - Divorce rate is inversely related to the age of marriage. Hence states where the median age of marriage is low will likely have a higher divorce rate.\n - We get a higher log likelihood of -60.92 as compared to -68.15 with Model 2, indicating that median age of marriage is likely a much better predictor of divorce rate. ", "_____no_output_____" ] ], [ [ "rng, rng_ = random.split(rng)\ninit_params, potential_fn, constrain_fn = initialize_model(rng_, model, \n age=dset.AgeScaled.values, \n divorce=dset.DivorceScaled.values)\n\nsamples_2 = mcmc(num_warmup, num_samples, init_params,\n potential_fn=potential_fn, \n trajectory_length=10, \n constrain_fn=constrain_fn)", "warmup: 100%|██████████| 1000/1000 [00:12<00:00, 79.17it/s, 3 steps of size 6.96e-01. acc. prob=0.79] \nsample: 100%|██████████| 2000/2000 [00:04<00:00, 470.51it/s, 3 steps of size 6.96e-01. acc. prob=0.88]\n" ], [ "posterior_mu = np.expand_dims(samples_2['a'], -1) + \\\n np.expand_dims(samples_2['bA'], -1) * dset.AgeScaled.values\nmean_mu = np.mean(posterior_mu, axis=0)\nhpdi_mu = hpdi(posterior_mu, 0.9)\nax = plot_regression(dset.AgeScaled.values, mean_mu, hpdi_mu)\nax.set(xlabel='Median marriage age', ylabel='Divorce rate', title='Regression line with 90% CI');", "_____no_output_____" ], [ "rng, rng_ = random.split(rng)\npredict_fn = vmap(lambda rng, samples: predict(rng, samples, model, age=dset.AgeScaled.values))\npredictions_2 = predict_fn(random.split(rng_, num_samples), samples_2)\n\nmean_pred = np.mean(predictions_2, axis=0)\nhpdi_pred = hpdi(predictions_2, 0.9)\n\nax = plot_regression(dset.AgeScaled.values, mean_pred, hpdi_pred)\nax.set(xlabel='Median Age', ylabel='Divorce rate', title='Predictions with 90% CI');", "_____no_output_____" ], [ "rng, rng_ = random.split(rng)\nprint('Log likelihood: {}'.format(expected_log_likelihood(rng_,\n samples_2, \n model,\n age=dset.AgeScaled.values,\n divorce=dset.DivorceScaled.values)))", "Log likelihood: -60.926387786865234\n" ] ], [ [ "### Model 3: Predictor - Marriage Rate and Median Age of Marriage\n\nFinally, we will also model divorce rate as depending on both marriage rate as well as the median age of marriage. Note that there is no increase in the model's log likelihood over Model 2 which likely indicates that the marginal information from marriage rate in predicting divorce rate is low when the median age of marriage is already known.", "_____no_output_____" ] ], [ [ "rng, rng_ = random.split(rng)\ninit_params, potential_fn, constrain_fn = initialize_model(rng_, model, \n marriage=dset.MarriageScaled.values,\n age=dset.AgeScaled.values,\n divorce=dset.DivorceScaled.values)\n\nsamples_3 = mcmc(num_warmup, num_samples, init_params,\n potential_fn=potential_fn, \n trajectory_length=10, \n constrain_fn=constrain_fn)", "warmup: 100%|██████████| 1000/1000 [00:10<00:00, 93.74it/s, 3 steps of size 6.48e-01. acc. prob=0.79] \nsample: 100%|██████████| 2000/2000 [00:04<00:00, 474.30it/s, 3 steps of size 6.48e-01. acc. prob=0.86]\n" ], [ "rng, rng_ = random.split(rng)\nprint('Log likelihood: {}'.format(expected_log_likelihood(rng_,\n samples_3,\n model,\n marriage=dset.MarriageScaled.values,\n age=dset.AgeScaled.values,\n divorce=dset.DivorceScaled.values)))", "Log likelihood: -61.04328918457031\n" ] ], [ [ "### Divorce Rate Residuals by State\n\nThe regression plots above shows that the observed divorce rates for many states differs considerably from the mean regression line. To dig deeper into how the last model (Model 3) under-predicts or over-predicts for each of the states, we will plot the posterior predictive and residuals (`Observed divorce rate - Predicted divorce rate`) for each of the states.", "_____no_output_____" ] ], [ [ "# Predictions for Model 3.\nrng, rng_ = random.split(rng)\npredict_fn = vmap(lambda rng, samples: predict(rng, samples, model,\n marriage=dset.MarriageScaled.values,\n age=dset.AgeScaled.values))\npredictions_3 = predict_fn(random.split(rng_, num_samples), samples_3)\ny = np.arange(50)\n\n\nfig, ax = plt.subplots(nrows=1, ncols=2, figsize=(12, 16))\npred_mean = np.mean(predictions_3, axis=0)\npred_hpdi = hpdi(predictions_3, 0.9)\nresiduals_3 = dset.DivorceScaled.values - predictions_3\nresiduals_mean = np.mean(residuals_3, axis=0)\nresiduals_hpdi = hpdi(residuals_3, 0.9)\nidx = np.argsort(residuals_mean)\n\n# Plot posterior predictive\nax[0].plot(np.zeros(50), y, '--')\nax[0].errorbar(pred_mean[idx], y, xerr=pred_hpdi[1, idx] - pred_mean[idx], \n marker='o', ms=5, mew=4, ls='none', alpha=0.8)\nax[0].plot(dset.DivorceScaled.values[idx], y, marker='o', \n ls='none', color='gray', alpha=0.5)\nax[0].set(xlabel='Posterior Predictive', ylabel='State', title='Posterior Predictive with 90% CI')\nax[0].set_yticks(y)\nax[0].set_yticklabels(dset.Loc.values[idx], fontsize=10);\n\n# Plot residuals\nresiduals_3 = dset.DivorceScaled.values - predictions_3\nresiduals_mean = np.mean(residuals_3, axis=0)\nresiduals_hpdi = hpdi(residuals_3, 0.9)\nerr = residuals_hpdi[1] - residuals_mean\n\nax[1].plot(np.zeros(50), y, '--')\nax[1].errorbar(residuals_mean[idx], y, xerr=err[idx], \n marker='o', ms=5, mew=4, ls='none', alpha=0.8)\nax[1].set(xlabel='Residuals', ylabel='State', title='Residuals with 90% CI')\nax[1].set_yticks(y)\nax[1].set_yticklabels(dset.Loc.values[idx], fontsize=10);", "_____no_output_____" ] ], [ [ "The plot on the left shows the mean predictions with 90% CI for each of the states using Model 3. The gray markers indicate the actual observed divorce rates. The right plot shows the residuals for each of the states, and both these plots are sorted by the residuals, i.e. at the bottom, we are looking at states where the model predictions are higher than the observed rates, whereas at the top, the reverse is true.\n\nOverall, the model fit seems good because most observed data points like within a 90% CI around the mean predictions. However, notice how the model over-predicts by a large margin for states like Idaho (bottom left), and on the other end under-predicts for states like Maine (top right). This is likely indicative of other factors that we are missing out in our model that affect divorce rate across different states. Even ignoring other socio-political variables, one such factor that we have not yet modeled is the measurement noise given by `Divorce SE` in the dataset. We will explore this in the next section.", "_____no_output_____" ], [ "## Regression Model with Measurement Error\n\nNote that in our previous models, each data point influences the regression line equally. Is this well justified? We will build on the previous model to incorporate measurement error given by `Divorce SE` variable in the dataset. Incorporating measurement noise will be useful in ensuring that observations that have higher confidence (i.e. lower measurement noise) have a greater impact on the regression line. On the other hand, this will also help us better model outliers with high measurement errors. For more details on modeling errors due to measurement noise, refer to Chapter 15 of [[1](#References)].\n\nTo do this, we will reuse Model 3, with the only change that the final observed value has a measurement error given by `divorce_sd` (notice that this has to be standardized since the `divorce` variable itself has been standardized to mean 0 and std 1).", "_____no_output_____" ] ], [ [ "def model_se(marriage, age, divorce_sd, divorce=None):\n a = sample('a', dist.Normal(0., 0.2))\n bM = sample('bM', dist.Normal(0., 0.5))\n M = bM * marriage\n bA = sample('bA', dist.Normal(0., 0.5))\n A = bA * age\n sigma = sample('sigma', dist.Exponential(1.))\n mu = a + M + A\n divorce_rate = sample('divorce_rate', dist.Normal(mu, sigma))\n sample('obs', dist.Normal(divorce_rate, divorce_sd), obs=divorce)", "_____no_output_____" ], [ "rng, rng_ = random.split(rng)\n# Standardize\ndset['DivorceScaledSD'] = dset['Divorce SE'] / np.std(dset.Divorce.values)\ninit_params, potential_fn, constrain_fn = initialize_model(rng_, model_se, \n marriage=dset.MarriageScaled.values,\n age=dset.AgeScaled.values,\n divorce_sd=dset.DivorceScaledSD.values,\n divorce=dset.DivorceScaled.values)", "_____no_output_____" ], [ "samples_4 = mcmc(num_warmup=1000, \n num_samples=3000, \n init_params=init_params,\n potential_fn=potential_fn,\n trajectory_length=10,\n target_accept_prob=0.9,\n constrain_fn=constrain_fn)", "warmup: 100%|██████████| 1000/1000 [00:19<00:00, 50.19it/s, 15 steps of size 2.16e-01. acc. prob=0.89] \nsample: 100%|██████████| 3000/3000 [00:06<00:00, 442.19it/s, 15 steps of size 2.16e-01. acc. prob=0.94]\n" ] ], [ [ "### Effect of Incorporating Measurement Noise on Residuals\n\nNotice that our values for the regression coefficients is very similar to Model 3. However, introducing measurement noise allows us to more closely match our predictive distributions to the observed values. We can see this if we plot the residuals as earlier. ", "_____no_output_____" ] ], [ [ "rng, rng_ = random.split(rng)\npredict_fn = vmap(lambda rng, samples: predict(rng, samples, model_se, \n marriage=dset.MarriageScaled.values,\n age=dset.AgeScaled.values,\n divorce_sd=dset.DivorceScaledSD.values))\npredictions_4 = predict_fn(random.split(rng_, 3000), samples_4)", "_____no_output_____" ], [ "sd = dset.DivorceScaledSD.values\nresiduals_4 = dset.DivorceScaled.values - predictions_4\nresiduals_mean = np.mean(residuals_4, axis=0)\nresiduals_hpdi = hpdi(residuals_4, 0.9)\nerr = residuals_hpdi[1] - residuals_mean\nidx = np.argsort(residuals_mean)\ny = np.arange(50)\nfig, ax = plt.subplots(nrows=1, ncols=1, figsize=(6, 16))\n\n\n# Plot Residuals\nax.plot(np.zeros(50), y, '--')\nax.errorbar(residuals_mean[idx], y, xerr=err[idx], \n marker='o', ms=5, mew=4, ls='none', alpha=0.4)\n\n# Plot SD \nax.errorbar(residuals_mean[idx], y, xerr=sd[idx], \n ls='none')\n\n# Plot earlier mean residual\nax.plot(np.mean(dset.DivorceScaled.values - predictions_3, 0)[idx], y,\n ls='none', marker='o', ms=5, color='gray', alpha=0.8)\n\nax.set(xlabel='Residuals', ylabel='State', title='Residuals with 90% CI')\nax.set_yticks(y)\nax.set_yticklabels(dset.Loc.values[idx], fontsize=10);", "_____no_output_____" ] ], [ [ "The plot above shows the residuals for each of the states, along with the measurement noise given by inner error bar. The gray dots are the mean residuals from our earlier Model 3. Notice how having an additional degree of freedom to model the measurement noise has shrunk the residuals. In particular, for Idaho and Maine, our predictions are now much closer to the observed values after incorporating measurement noise in the model.\n\nTo better see how measurement noise affects the movement of the regression line, let us plot the residuals with respect to the measurement noise.", "_____no_output_____" ] ], [ [ "fig, ax = plt.subplots(nrows=1, ncols=1, figsize=(10, 6))\nx = dset.DivorceScaledSD.values\ny1 = np.mean(residuals_3, 0)\ny2 = np.mean(residuals_4, 0)\nax.plot(x, y1, ls='none', marker='o')\nax.plot(x, y2, ls='none', marker='o')\nfor i, (j, k) in enumerate(zip(y1, y2)):\n ax.plot([x[i], x[i]], [j, k], '--', color='gray');\n\nax.set(xlabel='Measurement Noise', ylabel='Residual', title='Mean residuals (Model 4: red, Model 3: blue)');", "_____no_output_____" ] ], [ [ "The plot above shows what has happend in more detail - the regression line itself has moved to ensure a better fit for observations with low measurement noise (left of the plot) where the residuals have shrunk very close to 0. That is to say that data points with low measurement error have a concomitantly higher contribution in determining the regression line. On the other hand, for states with high measurement error (right of the plot), incorporating measurement noise allows us to move our posterior distribution mass closer to the observations resulting in a shrinkage of residuals as well.", "_____no_output_____" ], [ "## References\n\n1. McElreath, R. (2016). Statistical Rethinking: A Bayesian Course with Examples in R and Stan CRC Press.\n2. Stan Development Team. [Stan User's Guide](https://mc-stan.org/docs/2_19/stan-users-guide/index.html)\n3. Goodman, N.D., and StuhlMueller, A. (2014). [The Design and Implementation of Probabilistic Programming Languages](http://dippl.org/)\n4. Pyro Development Team. [Poutine: A Guide to Programming with Effect Handlers in Pyro](http://pyro.ai/examples/effect_handlers.html)\n5. Hoffman, M.D., Gelman, A. (2011). The No-U-Turn Sampler: Adaptively Setting Path Lengths in Hamiltonian Monte Carlo.\n6. Betancourt, M. (2017). A Conceptual Introduction to Hamiltonian Monte Carlo.\n7. JAX Development Team (2018). [Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more](https://github.com/google/jax)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/chuducthang77/coronavirus/blob/main/Mutation_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "# Mutation analysis\n1. Calculate the mutation rate (number of mutations/number of unique sequences) \n2. Consider only the sequences between 2000 and 2020\n3. Write the different files (Year : Mutation rate)", "_____no_output_____" ] ], [ [ "from google.colab import drive\ndrive.mount('/content/gdrive')", "Mounted at /content/gdrive\n" ], [ "%cd 'gdrive/MyDrive/Machine Learning/coronavirus/analysis'\n!ls", "/content/gdrive/MyDrive/Machine Learning/coronavirus/analysis\nH1N_H9N Mutation_analysis.ipynb\n" ], [ "import os\nimport pandas as pd\nimport numpy as np", "_____no_output_____" ], [ "#List all the directory names\ndir_name = os.listdir('./H1N_H9N')\n\nfor name in dir_name:\n df = pd.read_csv('./H1N_H9N/' + name)\n\n #Select only sequence between 2000 and 2020\n start_date = '2000-01-01'\n end_date = '2020-12-31'\n df = df[(df['Date'] >= start_date)]\n df = df[(df['Date'] < end_date)]\n\n #Create another column called year to select the number of unique sequence and number of mutations\n df['year'] = pd.to_datetime(df['Date']).dt.year\n\n start = 2000\n length = 20\n i = 0\n\n with open('output_{}.txt'.format(name[:-4]), 'a') as output:\n output.write('Year Mutation Rate\\n')\n\n while i <= length:\n specific_year = df[(df['year'] == start + i)]\n\n #Calculate the number of mutations\n num_mutations = np.float64(specific_year['year'].count())\n\n #Calculate the number of unique sequence\n num_unique = np.float64(specific_year['Accession ID'].nunique())\n\n #Calculate the mutation rate\n if num_unique != 0:\n mutation_rate = num_mutations/num_unique\n else:\n mutation_rate = 0\n \n output.write('{} {}\\n'.format(start + i, mutation_rate))\n\n i += 1\n", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Table of Contents\n <p><div class=\"lev3 toc-item\"><a href=\"#Cover-Slide-1\" data-toc-modified-id=\"Cover-Slide-1-001\"><span class=\"toc-item-num\">0.0.1&nbsp;&nbsp;</span>Cover Slide 1</a></div><div class=\"lev3 toc-item\"><a href=\"#Cover-Slide-2\" data-toc-modified-id=\"Cover-Slide-2-002\"><span class=\"toc-item-num\">0.0.2&nbsp;&nbsp;</span>Cover Slide 2</a></div><div class=\"lev1 toc-item\"><a href=\"#Headline-Slide\" data-toc-modified-id=\"Headline-Slide-1\"><span class=\"toc-item-num\">1&nbsp;&nbsp;</span>Headline Slide</a></div><div class=\"lev1 toc-item\"><a href=\"#Preprocessing\" data-toc-modified-id=\"Preprocessing-2\"><span class=\"toc-item-num\">2&nbsp;&nbsp;</span>Preprocessing</a></div><div class=\"lev1 toc-item\"><a href=\"#Headline-Subslide\" data-toc-modified-id=\"Headline-Subslide-3\"><span class=\"toc-item-num\">3&nbsp;&nbsp;</span>Headline Subslide</a></div><div class=\"lev1 toc-item\"><a href=\"#Fragment\" data-toc-modified-id=\"Fragment-4\"><span class=\"toc-item-num\">4&nbsp;&nbsp;</span>Fragment</a></div><div class=\"lev3 toc-item\"><a href=\"#Divider\" data-toc-modified-id=\"Divider-401\"><span class=\"toc-item-num\">4.0.1&nbsp;&nbsp;</span>Divider</a></div><div class=\"lev1 toc-item\"><a href=\"#Markdown-Examples\" data-toc-modified-id=\"Markdown-Examples-5\"><span class=\"toc-item-num\">5&nbsp;&nbsp;</span>Markdown Examples</a></div><div class=\"lev4 toc-item\"><a href=\"#Text\" data-toc-modified-id=\"Text-5001\"><span class=\"toc-item-num\">5.0.0.1&nbsp;&nbsp;</span>Text</a></div><div class=\"lev1 toc-item\"><a href=\"#Headline-Subslide\" data-toc-modified-id=\"Headline-Subslide-6\"><span class=\"toc-item-num\">6&nbsp;&nbsp;</span>Headline Subslide</a></div><div class=\"lev4 toc-item\"><a href=\"#Code\" data-toc-modified-id=\"Code-6001\"><span class=\"toc-item-num\">6.0.0.1&nbsp;&nbsp;</span>Code</a></div><div class=\"lev1 toc-item\"><a href=\"#Python-example\" data-toc-modified-id=\"Python-example-7\"><span class=\"toc-item-num\">7&nbsp;&nbsp;</span>Python example</a></div><div class=\"lev4 toc-item\"><a href=\"#Code\" data-toc-modified-id=\"Code-7001\"><span class=\"toc-item-num\">7.0.0.1&nbsp;&nbsp;</span>Code</a></div><div class=\"lev1 toc-item\"><a href=\"#Headline-Subslide\" data-toc-modified-id=\"Headline-Subslide-8\"><span class=\"toc-item-num\">8&nbsp;&nbsp;</span>Headline Subslide</a></div><div class=\"lev4 toc-item\"><a href=\"#Lists\" data-toc-modified-id=\"Lists-8001\"><span class=\"toc-item-num\">8.0.0.1&nbsp;&nbsp;</span>Lists</a></div><div class=\"lev1 toc-item\"><a href=\"#Headline-Subslide\" data-toc-modified-id=\"Headline-Subslide-9\"><span class=\"toc-item-num\">9&nbsp;&nbsp;</span>Headline Subslide</a></div><div class=\"lev4 toc-item\"><a href=\"#Lists\" data-toc-modified-id=\"Lists-9001\"><span class=\"toc-item-num\">9.0.0.1&nbsp;&nbsp;</span>Lists</a></div><div class=\"lev1 toc-item\"><a href=\"#Headline-Subslide\" data-toc-modified-id=\"Headline-Subslide-10\"><span class=\"toc-item-num\">10&nbsp;&nbsp;</span>Headline Subslide</a></div><div class=\"lev4 toc-item\"><a href=\"#Blockquotes\" data-toc-modified-id=\"Blockquotes-10001\"><span class=\"toc-item-num\">10.0.0.1&nbsp;&nbsp;</span>Blockquotes</a></div><div class=\"lev4 toc-item\"><a href=\"#inline-code\" data-toc-modified-id=\"inline-code-10002\"><span class=\"toc-item-num\">10.0.0.2&nbsp;&nbsp;</span>inline code</a></div><div class=\"lev1 toc-item\"><a href=\"#Table\" data-toc-modified-id=\"Table-11\"><span class=\"toc-item-num\">11&nbsp;&nbsp;</span>Table</a></div><div class=\"lev1 toc-item\"><a href=\"#Images\" data-toc-modified-id=\"Images-12\"><span class=\"toc-item-num\">12&nbsp;&nbsp;</span>Images</a></div><div class=\"lev1 toc-item\"><a href=\"#Links\" data-toc-modified-id=\"Links-13\"><span class=\"toc-item-num\">13&nbsp;&nbsp;</span>Links</a></div><div class=\"lev3 toc-item\"><a href=\"#Q&amp;A-Slide\" data-toc-modified-id=\"Q&amp;A-Slide-1301\"><span class=\"toc-item-num\">13.0.1&nbsp;&nbsp;</span>Q&amp;A Slide</a></div>", "_____no_output_____" ] ], [ [ "# Add all necessary imports here\nimport matplotlib.pyplot as plt\n%matplotlib inline\nplt.style.reload_library()\nplt.style.use(\"ggplot\")", "_____no_output_____" ] ], [ [ "### Cover Slide 1", "_____no_output_____" ] ], [ [ "<image>\n<section data-background=\"img/cover.jpg\" data-state=\"img-transparent no-title-footer\">\n<div class=\"intro-body\">\n<div class=\"intro_h1\"><h1>Title</h1></div>\n<h3>Subtitle of the Presentation</h3>\n<p><strong><span class=\"a\">Speaker 1</span></strong> <span class=\"b\"></span> <span>Job Title</span></p>\n<p><strong><span class=\"a\">Speaker 2</span></strong> <span class=\"b\"></span> <span>Job Title</span></p>\n<p>&nbsp;</p>\n<p>&nbsp;</p>\n</div>\n</section>\n</image>", "_____no_output_____" ] ], [ [ "### Cover Slide 2", "_____no_output_____" ] ], [ [ "<image>\n<section data-state=\"no-title-footer\">\n<div class=\"intro_h1\"><h1>Title</h1></div>\n<h3>Subtitle of the Presentation</h3>\n<p><strong><span class=\"a\">Speaker 1</span></strong> <span class=\"b\"></span> <span>Job Title</span></p>\n<p><strong><span class=\"a\">Speaker 2</span></strong> <span class=\"b\"></span> <span>Job Title</span></p>\n<p>&nbsp;</p>\n<p>&nbsp;</p>\n</section>\n</image>", "_____no_output_____" ] ], [ [ "# Headline Slide", "_____no_output_____" ], [ "# Preprocessing", "_____no_output_____" ], [ "", "_____no_output_____" ], [ "", "_____no_output_____" ], [ "", "_____no_output_____" ], [ "", "_____no_output_____" ], [ "", "_____no_output_____" ], [ "", "_____no_output_____" ] ], [ [ "def f(x):\n \"\"\"a docstring\"\"\"\n return x**2", "_____no_output_____" ] ], [ [ "# Headline Subslide", "_____no_output_____" ] ], [ [ "plt.plot([1,2,3,4])\nplt.ylabel('some numbers')\nplt.show()", "_____no_output_____" ] ], [ [ "# Fragment\n\nPress the right arrow.", "_____no_output_____" ], [ "- I am a Fragement", "_____no_output_____" ], [ "- I am another one", "_____no_output_____" ], [ "### Divider", "_____no_output_____" ] ], [ [ "<image>\n</section>\n<section data-background=\"#F27C3A\" data-state=\"no-title-footer\">\n <div class=\"divider_h1\">\n <h1>Divider</h1>\n </div>\n</section>\n</image>", "_____no_output_____" ] ], [ [ "# Markdown Examples\n\n#### Text\n\nIt's very easy to make some words **bold** and other words *italic* with Markdown. You can even [link to Google!](http://google.com)", "_____no_output_____" ], [ "# Headline Subslide\n#### Code\n\n```javascript\nvar s = \"JavaScript syntax highlighting\";\nalert(s);\n```\n \n```python\ns = \"Python syntax highlighting\"\nprint s\n```\n \n```\nNo language indicated, so no syntax highlighting. \nBut let's throw in a <b>tag</b>.\n```", "_____no_output_____" ], [ "# Python example\n\n#### Code\n```python\n# This program adds up integers in the command line\nimport sys\ntry:\n total = sum(int(arg) for arg in sys.argv[1:])\n print 'sum =', total\nexcept ValueError:\n print 'Please supply integer arguments'\n```", "_____no_output_____" ], [ "# Headline Subslide\n\n#### Lists\n\nSometimes you want numbered lists:\n\n1. Item 1\n2. Item 2\n3. Item 3\n * Item 3a\n * Item 3b\n", "_____no_output_____" ], [ "# Headline Subslide\n\n#### Lists\n\nSometimes you want bullet points:\n\n* Item 1\n* Item 2\n * Item 2a\n * Item 2b\n* This is a long long long long long long long long long long long long long long long long long long long long long long long long long list", "_____no_output_____" ], [ "# Headline Subslide\n#### Blockquotes\n\nAs Kanye West said:\n\n> We're living the future so\n> the present is our past.\n\n#### inline code\nI think you should use an\n`<addr>` element here instead.", "_____no_output_____" ], [ "# Table\n\n| Tables | Are | Cool |\n| ------------- |:-------------:| -----:|\n| col 3 is | right-aligned | $1600 |\n| col 2 is | centered | $12 |\n| zebra stripes | are neat | $1 |", "_____no_output_____" ], [ "# Images\nIf you want to embed images, this is how you do it:\n\n", "_____no_output_____" ], [ "# Links\n\n- https://guides.github.com/features/mastering-markdown/\n\n- https://github.com/adam-p/markdown-here/wiki/Markdown-Cheatsheet", "_____no_output_____" ], [ "### Q&A Slide", "_____no_output_____" ] ], [ [ "<image>\n</section>\n<section data-background=\"#0093C9\" data-state=\"no-title-footer\">\n <div class=\"divider_h1\">\n <h1>Questions???</h1>\n </div>\n</section>\n</image>", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "raw", "markdown", "raw", "markdown", "code", "markdown", "code", "markdown", "raw", "markdown", "raw" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "raw" ], [ "markdown" ], [ "raw" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "raw" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "raw" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "test: 30% \ntrain-val 70%\n\nTo simulate noisy label acquisition, assume a labeler quality p. then use a random \n\nwe first hide the labels of all examples for each dataset. At the point in an experiment when a label is acquired, we generate a label according to the labeler quality p: we assign the example's original label with the probability p and the opposite value with the probability 1−p. \n\nFor each worker, the original true label was assigned to each instance with the probability p, and the opposite value was assigned with the probability 1−𝑝. In our experiments, the labeling quality p of each worker was generated randomly from a uniform distribution on the interval (0.3, 0.9)\n\n\nAfter obtaining the labels, we construct the training set to induce a classifier. The classifier is evaluated on the test set (with the true labels). Each experiment is repeated 10 times with a different random data partition, and average results are reported. \n\n## procedure\n1. generate a random p for each labeler\n2. generate a random number from a random variable for each instance and if the value is bigger than p, then assign the true label, but if its smaller than p then assign the opposite of true label for that worker \n3. train a network for these noisy labels for each labeler \n4. repeate this 10 times to be able to measure ethe uncertainty and average accuracy.\n\n\n## When it is necessary, we convert the target to binary\n\n- for thyroid we keep the negative class and integrate the other three classes into positive; \n- for splice, we integrate classes IE and EI; \n- for waveform, we integrate classes 1 and 2.)", "_____no_output_____" ] ], [ [ "%reload_ext autoreload\n%autoreload 2\n\nimport sys, os, wget\nsys.path.append('../../')\n\nimport pandas as pd\nimport numpy as np\nimport load_data \nimport ipywidgets \nfrom IPython.display import display", "_____no_output_____" ], [ "args = {'options': ['read', 'download'], \n 'label': 'read', \n 'description': 'mode:' , \n 'orientation': 'vertical', \n 'disabled': False, \n 'button_style': '', \n 'layout': {'width': 'max-content'},\n 'tooltips': ['reading the dataset', 'downloading the dataset'], \n }\n\ndataset_mode = ipywidgets.ToggleButtons( **args )\n\ndisplay(dataset_mode)", "_____no_output_____" ], [ "dataset = ipywidgets.Dropdown( options = [ ('1. kr-vs-kp' ,'kr-vs-kp'), \n ('2. mushroom' ,'mushroom'),\n ('3. sick' ,'sick'),\n ('4. spambase' ,'spambase'),\n ('5. tic-tac-toe' ,'tic-tac-toe'),\n ('6. splice' ,'splice'),\n ('7. thyroid' ,'thyroid'),\n ('8. waveform' ,'waveform'),\n ('9. biodeg' ,'biodeg'),\n ('10. horse-colic','horse-colic'),\n ('11. ionosphere' ,'ionosphere'),\n ('12. vote' ,'vote')], \n value = 'horse-colic')\n\n\[email protected](WHICH_DATASET = dataset)\ndef read_data(WHICH_DATASET):\n\n if WHICH_DATASET in ['splice','vote','thyroid']:\n\n print('dataset does not exist')\n\n else:\n\n global data, feature_columns\n data, feature_columns = load_data.aim1_3_read_download_UCI_database( WHICH_DATASET = WHICH_DATASET, \n mode = dataset_mode.value)\n print(data['train'].head(3) , '\\ntrain shape:',data['train'].shape , '\\ntest shape:',data['test'].shape)", "_____no_output_____" ], [ "data, feature_columns = load_data.aim1_3_read_download_UCI_database(WHICH_DATASET='biodeg', mode='read')\n", "_____no_output_____" ], [ "data", "_____no_output_____" ], [ "data, feature_columns = load_data.aim1_3_read_download_UCI_database(WHICH_DATASET='horse-colic', mode='read_raw')", "_____no_output_____" ], [ "data['train']", "_____no_output_____" ], [ "pd.read_csv('/groups/jjrodrig/projects/datasets/uci_multilabeler_aim1_3/UCI_horse-colic/horse-colic.data',delimiter=' ', index_col=None)", "_____no_output_____" ] ], [ [ "## guide for ipywidgets\nsource: <https://coderzcolumn.com/tutorials/python/interactive-widgets-in-jupyter-notebook-using-ipywidgets>", "_____no_output_____" ] ], [ [ "widgets.FloatRangeSlider(\n value=[5, 7.5],\n min=0,\n max=10.0,\n step=0.1,\n description='Test:',\n disabled=False,\n continuous_update=False,\n orientation='horizontal',\n readout=True,\n readout_format='.1f',\n)", "_____no_output_____" ], [ "b = widgets.BoundedFloatText(\n value=7.5,\n min=0,\n max=10.0,\n step=0.1,\n description='Text:',\n disabled=False\n)\nb", "_____no_output_____" ], [ "b.value ", "_____no_output_____" ], [ "a = widgets.ToggleButton(\n value=False,\n description='Run',\n button_style='', # 'success', 'info', 'warning', 'danger' or ''\n tooltip='Description',\n icon='check')\na", "_____no_output_____" ], [ "a.value", "_____no_output_____" ], [ "c = widgets.Checkbox(\n value=False,\n description='this',\n disabled=False,\n indent=False\n)\n\nc", "_____no_output_____" ], [ "c.value ", "_____no_output_____" ], [ "widgets.Valid(\n value=True,\n description='Status:',\n)", "_____no_output_____" ], [ "d = widgets.Dropdown(\n options=[('One', 1), ('Two', 2), ('Three', 3)],\n value=2,\n description='Number:')", "_____no_output_____" ], [ "d.value ", "_____no_output_____" ], [ "widgets.RadioButtons(\n options=['pepperoni', 'pineapple', 'anchovies'],\n# value='pineapple', # Defaults to 'pineapple'\n# layout={'width': 'max-content'}, # If the items' names are long\n description='Pizza topping:',\n disabled=False\n)", "_____no_output_____" ], [ "widgets.Box(\n [\n widgets.Label(value='Pizza topping with a very long label:'),\n widgets.RadioButtons(\n options=[\n 'pepperoni',\n 'pineapple',\n 'anchovies',\n 'and the long name tha'\n ],\n layout={'width': 'max-content'}\n )\n ]\n)", "_____no_output_____" ], [ "widgets.Select(\n options=['Linux', 'Windows', 'OSX'],\n value='OSX',\n # rows=10,\n description='OS:',\n disabled=False\n)", "_____no_output_____" ], [ "widgets.ToggleButtons(\n options=['Slow', 'Regular', 'Fast'],\n description='Speed:',\n disabled=False,\n button_style='', # 'success', 'info', 'warning', 'danger' or ''\n tooltips=['Description of slow', 'Description of regular', 'Description of fast'],\n icons=['check'] * 3\n)", "_____no_output_____" ], [ "a = widgets.ColorPicker(\n concise=False,\n description='Pick a color',\n value='blue',\n disabled=False\n)\na", "_____no_output_____" ], [ "widgets.FileUpload(\n accept='', # Accepted file extension e.g. '.txt', '.pdf', 'image/*', 'image/*,.pdf'\n multiple=False # True to accept multiple files upload else False\n)", "_____no_output_____" ], [ "from IPython.display import display\n\ndef func3(a,b,c):\n display((a+b)^c)\n\nw = interactive(func3, a=ipywidgets.IntSlider(min=10, max=50, value=25, step=2),\n b=ipywidgets.IntSlider(min=10, max=50, value=25, step=2),\n c=ipywidgets.IntSlider(min=10, max=50, value=25, step=2) ) \ndisplay(w)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import sys\nimport numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nimport math\nimport os\n\nimport boto3\nimport re\nfrom sagemaker import get_execution_role\nimport sagemaker\n\n# SDK 2 serializers and deserializers\nfrom sagemaker.serializers import CSVSerializer\nfrom sagemaker.deserializers import JSONDeserializer", "_____no_output_____" ] ], [ [ "<h1>XGBoost Cloud Prediction - Iris Classification</h1>\n<h4>Invoke SageMaker Prediction Service</h4>", "_____no_output_____" ] ], [ [ "# Acquire a realtime endpoint\nendpoint_name = 'xgboost-iris-v1'\npredictor = sagemaker.predictor.Predictor (endpoint_name=endpoint_name)", "_____no_output_____" ], [ "predictor.serializer = CSVSerializer()", "_____no_output_____" ], [ "# Test predictive quality against data in validation file\ndf_all = pd.read_csv('iris_validation.csv',\n names=['encoded_class','sepal_length','sepal_width','petal_length','petal_width'])", "_____no_output_____" ], [ "df_all.head()", "_____no_output_____" ], [ "df_all.columns", "_____no_output_____" ], [ "# Need to pass an array to the prediction\n# can pass a numpy array or a list of values [[19,1],[20,1]]\n# arr_test = df_all.as_matrix(['sepal_length', 'sepal_width', 'petal_length','petal_width'])\narr_test = df_all[['sepal_length', 'sepal_width', 'petal_length','petal_width']].values", "_____no_output_____" ], [ "type(arr_test)", "_____no_output_____" ], [ "arr_test.shape", "_____no_output_____" ], [ "arr_test[:5]", "_____no_output_____" ], [ "result = predictor.predict(arr_test[:2])", "_____no_output_____" ], [ "arr_test.shape", "_____no_output_____" ], [ "result", "_____no_output_____" ], [ "# For large number of predictions, we can split the input data and\n# Query the prediction service.\n# array_split is convenient to specify how many splits are needed\npredictions = []\nfor arr in np.array_split(arr_test,10):\n result = predictor.predict(arr)\n result = result.decode(\"utf-8\")\n result = result.split(',')\n print (arr.shape)\n predictions += [int(float(r)) for r in result]", "_____no_output_____" ], [ "len(predictions)", "_____no_output_____" ], [ "predictions[:5]", "_____no_output_____" ], [ "from sklearn import preprocessing\nle = preprocessing.LabelEncoder()\nle.fit(['Iris-setosa', 'Iris-versicolor', 'Iris-virginica'])", "_____no_output_____" ], [ "df_all['class'] = le.inverse_transform(df_all.encoded_class)", "_____no_output_____" ], [ "df_all['predicted_class']=le.inverse_transform(predictions)", "_____no_output_____" ], [ "df_all.head()", "_____no_output_____" ], [ "print('Confusion matrix - Actual versus Predicted')\npd.crosstab(df_all['class'], df_all['predicted_class'])", "_____no_output_____" ], [ "import sklearn.metrics as metrics\nprint(metrics.classification_report(df_all['class'], df_all['predicted_class']))", "_____no_output_____" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]