Datasets:

---

bigcode

/

jupyter-parsed

like 4

Follow

BigCode 1.36k

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "# Datenvalidierung mit Voluptuous (Schemadefinitionen)\n\nIn diesem Notebook verwenden wir [Voluptuous](https://github.com/alecthomas/voluptuous), um Schemata für unsere Daten zu definieren. Wir können dann die Schemaprüfung an verschiedenen Stellen unserer Bereinigung verwenden, um sicherzustellen, dass wir die Kriterien erfüllen. Schließllich können wir Ausnahmen für die Schemaüberprüfung verwenden, um unreine oder ungültige Daten zu markieren, beiseite zu legen oder zu entfernen.", "_____no_output_____" ], [ "## 1. Importe", "_____no_output_____" ] ], [ [ "import logging\nimport pandas as pd\nfrom datetime import datetime\nfrom voluptuous import Schema, Required, Range, All, ALLOW_EXTRA\nfrom voluptuous.error import MultipleInvalid, Invalid", "_____no_output_____" ] ], [ [ "## 2. Logger", "_____no_output_____" ] ], [ [ "logger = logging.getLogger(0)\nlogger.setLevel(logging.WARNING)", "_____no_output_____" ] ], [ [ "## 3. Beispieldaten lesen", "_____no_output_____" ] ], [ [ "sales = pd.read_csv('https://raw.githubusercontent.com/kjam/data-cleaning-101/master/data/sales_data.csv')", "_____no_output_____" ] ], [ [ "## 4. Daten untersuchen", "_____no_output_____" ] ], [ [ "sales.head()", "_____no_output_____" ], [ "sales.dtypes", "_____no_output_____" ] ], [ [ "## 5. Schema definieren", "_____no_output_____" ] ], [ [ "schema = Schema({\n Required('sale_amount'): All(float, \n Range(min=2.50, max=1450.99)),\n}, extra=ALLOW_EXTRA)", "_____no_output_____" ], [ "error_count = 0\nfor s_id, sale in sales.T.to_dict().items():\n try:\n schema(sale)\n except MultipleInvalid as e:\n logging.warning('issue with sale: %s (%s) - %s', \n s_id, sale['sale_amount'], e)\n error_count += 1", "WARNING:root:issue with sale: 3 (-108.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 4 (-372.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 5 (-399.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 6 (-304.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 7 (-295.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 10 (-89.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 13 (-303.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 15 (-432.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 19 (-177.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 20 (-154.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 22 (-130.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 23 (1487.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 25 (-145.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 28 (1471.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 31 (-259.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 38 (-241.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 40 (-4.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 41 (1581.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 45 (1529.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 46 (-238.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 48 (-284.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 51 (-164.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 55 (-184.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 56 (-304.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 59 (1579.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 60 (-455.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 63 (1551.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 65 (-397.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 69 (-400.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 70 (1482.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 71 (-321.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 74 (-47.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 76 (-68.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 86 (1454.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 101 (-213.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 103 (-144.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 104 (-265.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 107 (-349.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 111 (-78.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 112 (-310.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 116 (1570.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 120 (1490.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 123 (-179.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 124 (-391.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 129 (1504.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 130 (-91.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 132 (-372.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 141 (1512.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 142 (-449.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 149 (1494.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 152 (-405.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 155 (1599.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 156 (1527.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 157 (-462.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 162 (-358.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 164 (-78.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 167 (-358.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 171 (-391.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 178 (-304.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 180 (-9.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 187 (1475.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 194 (-433.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 195 (-329.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 196 (-147.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 203 (-319.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 206 (-132.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 207 (-20.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 209 (1539.0) - value must be at most 1450.99 for dictionary value @ data['sale_amount']\nWARNING:root:issue with sale: 211 (-167.0) - value must be at least 2.5 for dictionary value @ data['sale_amount']\n" ], [ "error_count", "_____no_output_____" ], [ "sales.shape", "_____no_output_____" ] ], [ [ "Aktuell wissen wir jedoch noch nicht, ob\n\n* wir ein falsch definiertes Schema haben\n* möglicherweise negative Werte zurückgegeben oder falsch markiert werden\n* höhere Werte kombinierte Einkäufe oder Sonderverkäufe sind", "_____no_output_____" ], [ "## 6. Hinzufügen einer benutzerdefinierten Validierung", "_____no_output_____" ] ], [ [ "def ValidDate(fmt='%Y-%m-%d %H:%M:%S'):\n return lambda v: datetime.strptime(v, fmt)", "_____no_output_____" ], [ "schema = Schema({\n Required('timestamp'): All(ValidDate()),\n}, extra=ALLOW_EXTRA)", "_____no_output_____" ], [ "error_count = 0\nfor s_id, sale in sales.T.to_dict().items():\n try:\n schema(sale)\n except MultipleInvalid as e:\n logging.warning('issue with sale: %s (%s) - %s', \n s_id, sale['timestamp'], e)\n error_count += 1", "_____no_output_____" ], [ "error_count", "_____no_output_____" ] ], [ [ "## 7. Gültige Datumsstrukturen sind noch keine gültigen Daten", "_____no_output_____" ] ], [ [ "def ValidDate(fmt='%Y-%m-%d %H:%M:%S'):\n def validation_func(v):\n try:\n assert datetime.strptime(v, fmt) <= datetime.now()\n except AssertionError:\n raise Invalid('date is in the future! %s' % v)\n return validation_func", "_____no_output_____" ], [ "schema = Schema({\n Required('timestamp'): All(ValidDate()),\n}, extra=ALLOW_EXTRA)", "_____no_output_____" ], [ "error_count = 0\nfor s_id, sale in sales.T.to_dict().items():\n try:\n schema(sale)\n except MultipleInvalid as e:\n logging.warning('issue with sale: %s (%s) - %s', \n s_id, sale['timestamp'], e)\n error_count += 1", "_____no_output_____" ], [ "error_count", "_____no_output_____" ] ] ]

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

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "open boundary spin 1/2 1-D Heisenberg model\n\n$H = J \\sum_i [S_i^z S_{i+1}^z + \\frac{1}{2}(S_i^+ S_{i+1}^- + S_i^- S_{i+1}^+)]$\n\nexact result (Bethe Anstatz):\n\nL\t E/J\n\n16\t-6.9117371455749\n\n24\t-10.4537857604096\n\n32\t-13.9973156182243\n\n48\t-21.0859563143863\n\n64\t-28.1754248597421\n", "_____no_output_____" ] ], [ [ "from renormalizer.mps import Mps, Mpo, solver\nfrom renormalizer.model import MolList2, ModelTranslator\nfrom renormalizer.utils import basis as ba\nfrom renormalizer.utils import Op\nimport numpy as np\n\n# define the # of spins\nnspin = 16\n\n# define the model\n# sigma^+ = S^+\n# sigma^- = S^-\n# 1/2 sigma^x,y,z = S^x,y,z\n\nmodel = dict()\nfor ispin in range(nspin-1):\n model[(f\"e_{ispin}\", f\"e_{ispin+1}\")] = [(Op(\"sigma_z\",0),\n Op(\"sigma_z\",0), 1.0/4), (Op(\"sigma_+\",0), Op(\"sigma_-\",0), 1.0/2),\n (Op(\"sigma_-\",0), Op(\"sigma_+\", 0), 1.0/2)]\n\n# set the spin order and local basis\norder = {}\nbasis = []\nfor ispin in range(nspin):\n order[f\"e_{ispin}\"] = ispin\n basis.append(ba.BasisHalfSpin(sigmaqn=[0,0]))\n\n# construct MPO\nmol_list2 = MolList2(order, basis, model, ModelTranslator.general_model)\nmpo = Mpo(mol_list2)\nprint(f\"mpo_bond_dims:{mpo.bond_dims}\")\n\n# set the sweep paramter\nM=30\nprocedure = [[M, 0.2], [M, 0], [M, 0], [M,0], [M,0]]\n\n# initialize a random MPS\nqntot = 0\nmps = Mps.random(mol_list2, qntot, M)\n\nmps.optimize_config.procedure = procedure\nmps.optimize_config.method = \"2site\"\n\n# optimize MPS\nenergies = solver.optimize_mps_dmrg(mps.copy(), mpo)\nprint(\"gs energy:\", energies.min())\n", "2020-04-15 22:42:12,231[DEBUG] # of operator terms: 45\n2020-04-15 22:42:12,232[DEBUG] symbolic mpo algorithm: Hopcroft-Karp\n2020-04-15 22:42:12,285[DEBUG] mmax, percent: 30, 0.2\n2020-04-15 22:42:12,290[DEBUG] energy: -0.1085174478945915\n2020-04-15 22:42:12,290[DEBUG] current size: 151.2KiB, Matrix product bond dim:[1, 2, 4, 8, 16, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30, 1]\n" ], [ "# plot the microiterations vs energy error\nimport matplotlib.pyplot as plt\nimport logging \n\nmpl_logger = logging.getLogger('matplotlib') \nmpl_logger.setLevel(logging.WARNING) \n\nplt.rc('font', family='Times New Roman', size=16)\nplt.rc('axes', linewidth=1.5)\nplt.rcParams['lines.linewidth'] = 2\nstd = -6.9117371455749\nfig, ax = plt.subplots(figsize=(8,6))\nplt.plot(np.arange(len(energies)), np.array(energies)-std,\"o-\",ms=5)\n\nplt.yscale('log')\nplt.xlabel(\"microiterations\")\nplt.ylabel(\"$\\Delta$ E\")\nplt.ylim(1e-9, 10)\nplt.show()", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Striplog from CSV", "_____no_output_____" ] ], [ [ "%matplotlib inline\nimport matplotlib.pyplot as plt\nimport numpy as np\n\nimport striplog\nstriplog.__version__", "_____no_output_____" ], [ "from striplog import Lexicon, Decor, Component, Legend, Interval, Striplog", "_____no_output_____" ] ], [ [ "## Make legend\n\nMost of the stuff in the dicts you made were about **display** — so they are going to make `Decor` objects. A collection of `Decor`s makes a `Legend`. A `Legend` determines how a striplog is displayed.\n\nFirst I'll make the components, since those are easy. I'll move `'train'` into there too, since it is to do with the rocks, not the display. If it seems weird having `'train'` in the `Component` (which is really supposed to be about direct descriptions of the rock, but the idea is that it's always the same for all specimens of that rock so it does fit here) then you could put it in `data` instead. ", "_____no_output_____" ] ], [ [ "facies = {\n 's': Component({'lithology': 'sandstone', 'train':'y'}),\n 'i': Component({'lithology': 'interbedded', 'train':'y'}),\n 'sh': Component({'lithology': 'shale', 'train':'y'}),\n 'bs': Component({'lithology': 'sandstone', 'train': 'n'}),\n 't': Component({'lithology': 'turbidite', 'train':'y'}),\n 'nc': Component({'lithology': 'none', 'train':'n'}),\n}", "_____no_output_____" ], [ "sandstone = Decor({\n 'component': facies['s'],\n 'colour': 'yellow',\n 'hatch': '.',\n 'width': '3',\n})\n\ninterbedded = Decor({\n 'component': facies['i'],\n 'colour': 'darkseagreen',\n 'hatch': '--',\n 'width': '2',\n})\n\nshale = Decor({\n 'component': facies['sh'],\n 'colour': 'darkgray',\n 'hatch': '-',\n 'width': '1',\n})\n\nbadsand = Decor({\n 'component': facies['bs'],\n 'colour': 'orange',\n 'hatch': '.',\n 'width': '3',\n})\n\n\n# Not sure about the best way to do this, probably better\n# just to omit those intervals completely.\nnocore = Decor({\n 'component': facies['nc'],\n 'colour': 'white',\n 'hatch': '/',\n 'width': '5',\n})\n\nturbidite = Decor({\n 'component': facies['t'],\n 'colour': 'green',\n 'hatch': 'xxx',\n 'width': '3',\n})", "_____no_output_____" ], [ "legend = Legend([sandstone, badsand, interbedded, shale, turbidite, nocore])", "_____no_output_____" ], [ "legend", "_____no_output_____" ] ], [ [ "## Read CSV into striplog", "_____no_output_____" ] ], [ [ "strip = Striplog.from_csv('test.csv')", "_____no_output_____" ], [ "strip[0]", "_____no_output_____" ] ], [ [ "## Deal with lithology\n\nThe lithology has been turned into a component, but it's using the abbreviation... I can't figure out an elegant way to deal with this so, for now, we'll just loop over the striplog and fix it. We read the `data` item's lithology (`'s'` in the top layer), then look up the correct lithology name in our abbreviation dictionary, then add the new component in the proper place. Finally, we delete the `data` we had.", "_____no_output_____" ] ], [ [ "for s in strip:\n lith = s.data['lithology']\n s.components = [facies[lith]]\n s.data = {}", "_____no_output_____" ], [ "strip[0]", "_____no_output_____" ] ], [ [ "That's better!", "_____no_output_____" ] ], [ [ "strip.plot(legend)", "_____no_output_____" ] ], [ [ "## Remove non-training layers", "_____no_output_____" ] ], [ [ "strip", "_____no_output_____" ], [ "strip_train = Striplog([s for s in strip if s.primary['train'] == 'y'])", "_____no_output_____" ], [ "strip_train", "_____no_output_____" ], [ "strip_train.plot(legend)", "_____no_output_____" ] ] ]

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

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "This notebook aims to produce Figure S2 in \n\n Supplementary Material for the preprint:\n <Deriving information from missing data: implications for mood prediction>", "_____no_output_____" ] ], [ [ "import os\nimport random\nimport matplotlib.pyplot as plt\nimport numpy as np\nimport matplotlib.dates as mdates\nimport datetime\nimport time\nimport csv\nimport math\nimport scipy\nimport seaborn as sns\nfrom scipy.stats import iqr\nimport h5py\nimport pickle\nfrom tqdm import tqdm\nimport copy\nfrom sklearn.ensemble import RandomForestRegressor\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.metrics import accuracy_score\nfrom sklearn.preprocessing import LabelEncoder\nimport iisignature\nfrom datetime import date", "_____no_output_____" ], [ "from itertools import cycle\n\nfrom sklearn import svm, datasets\nfrom sklearn.metrics import roc_curve, auc\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.preprocessing import label_binarize\nfrom sklearn.multiclass import OneVsRestClassifier\nfrom scipy import interp", "_____no_output_____" ], [ "from classifiers import *\nfrom data_cleaning import *\nfrom data_transforms import *\nfrom ROC_functions import *", "_____no_output_____" ] ], [ [ "Load data", "_____no_output_____" ] ], [ [ "test_path='./all-true-colours-matlab-2017-02-27-18-38-12-nick/'\nparticipants_list, participants_data_list,\\\n participants_time_list=loadParticipants(test_path)\n \n\nParticipants=make_classes(participants_data_list,\\\n participants_time_list,\\\n participants_list)\ncohort=cleaning_sameweek_data(cleaning_same_data(Participants))", "14050\n" ] ], [ [ "**Length=20w**", "_____no_output_____" ], [ "* Missing-response-incorporated signature-based classification model (MRSCM, level2)", "_____no_output_____" ] ], [ [ "y_tests_mc, y_scores_mc=model_roc(cohort, minlen=20, training=0.7,order=2, sample_size=100) )\nplot_roc(y_tests_mc, y_scores_mc, \"signature_method_missingcount_2ndtrial\", n_classes=3,lw=2)", "The PostScript backend does not support transparency; partially transparent artists will be rendered opaque.\nThe PostScript backend does not support transparency; partially transparent artists will be rendered opaque.\n" ] ], [ [ "* Missing-response-incorporated signature-based classification model (MRSCM, level3)", "_____no_output_____" ] ], [ [ "y_tests_mc3, y_scores_mc3=model_roc(cohort, minlen=20, training=0.7,order=3, sample_size=100)\nplot_roc(y_tests_mc3, y_scores_mc3, \"signature_method_missingcount3\", n_classes=3,lw=2)", "The PostScript backend does not support transparency; partially transparent artists will be rendered opaque.\nThe PostScript backend does not support transparency; partially transparent artists will be rendered opaque.\n" ] ], [ [ "* Naive classification model", "_____no_output_____" ] ], [ [ "if __name__ == \"__main__\":\n\n\n \n y_tests_naive, y_scores_naive=model_roc(cohort, minlen=20,\\\n training=0.7,\\\n order=None,\\\n sample_size=100,\\\n standardise=False,\\\n count=False,\\\n feedforward=False,\\\n naive=True,\\\n time=False)\n \n plot_roc(y_tests_naive, y_scores_naive, \"naive_method\", n_classes=3,lw=2)", "The PostScript backend does not support transparency; partially transparent artists will be rendered opaque.\nThe PostScript backend does not support transparency; partially transparent artists will be rendered opaque.\n" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import importlib\nimport sys,os,math,time\n\ngostNetsFolder = os.path.dirname(os.getcwd())\nsys.path.insert(0, gostNetsFolder)\nfrom GOSTNets import GOSTnet as gn\nfrom GOSTNets import OSMParser\nimportlib.reload(gn)\n\nimport networkx as nx\nimport geopandas as gpd\nimport numpy as np\nimport osmnx as ox\nimport pandas as pd\nimport rasterio\n\nfrom shapely.geometry import Point\nfrom shapely.ops import nearest_points", "peartree version: 0.6.0 \nnetworkx version: 2.2 \nmatplotlib version: 2.2.2 \nosmnx version: 0.8.2 \n" ], [ "inputPBF = r\"Q:\\AFRICA\\MRT\\INFRA\\mauritania-latest_20190103_OSMLR12.osm.xml\"\ninputBorders = r\"Q:\\AFRICA\\MRT\\INFRA\\MRT_Fake_Border.shp\"\noriginFile = r\"Q:\\AFRICA\\MRT\\INFRA\\Origins.shp\"\ndestFile = r\"Q:\\AFRICA\\MRT\\INFRA\\Destinations.shp\"", "_____no_output_____" ], [ "#Read in the network object and convert to GDF\nimportlib.reload(OSMParser)\nimportlib.reload(gn)\n\nG = OSMParser.read_osm(inputPBF)\nG = gn.convert_network_to_time(G, distance_tag = 'length', speed_dict = gn.speed_dict)\nedgeDF = gn.edge_gdf_from_graph(G, xCol='lon', yCol='lat')\nnodeDF = gn.node_gdf_from_graph(G, xCol='lon', yCol='lat')\n\ninB = gpd.read_file(inputBorders)\ninO = gpd.read_file(originFile)\ninD = gpd.read_file(destFile)", "peartree version: 0.6.0 \nnetworkx version: 2.2 \nmatplotlib version: 2.2.2 \nosmnx version: 0.8.2 \n" ], [ "#Identify all network edges that intersect the borders file\nadjustments = {}\nfor idx, row in inB.iterrows():\n #Select the roads that intersect this border\n ### TODO - use a spatial index here, it is too slow\n intersections = edgeDF[edgeDF.intersects(row['geometry'])]\n for selIdx, selRow in intersections.iterrows():\n adjustments[selRow['id']] = row['id']", "_____no_output_____" ], [ "#Loop through edges in network and add time\nG_adj = G.copy()\nfor u,v,data in G_adj.edges(data=True):\n if data['id'] in adjustments.keys():\n data['time'] = data['time'] + adjustments[data['id']]", "_____no_output_____" ], [ "# identify intersecting nodes for input and output nodes\nimportlib.reload(gn)\n\ninNodes = []\ndestNodes = []\nfor idx, row in inO.iterrows():\n nPt = nodeDF.distance(row['geometry'])\n inNodes.append(nodeDF['node_ID'][nPt.idxmin()]) \n\ndestNodes = []\nfor idx, row in inD.iterrows():\n nPt = nodeDF.distance(row['geometry'])\n destNodes.append(nodeDF['node_ID'][nPt.idxmin()]) \n", "peartree version: 0.6.0 \nnetworkx version: 2.2 \nmatplotlib version: 2.2.2 \nosmnx version: 0.8.2 \n" ], [ "#Run OD matrix\nundistrubed = gn.calculate_OD(G, inNodes, destNodes, -1)\ndistrubed = gn.calculate_OD(G_adj, inNodes, destNodes, -1)", "_____no_output_____" ], [ "distrubed", "_____no_output_____" ], [ "undistrubed", "_____no_output_____" ] ] ]

[ "code" ]

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "##### Copyright 2018 The TensorFlow Authors.", "_____no_output_____" ] ], [ [ "#@title 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# https://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.", "_____no_output_____" ], [ "#@title MIT License\n#\n# Copyright (c) 2017 François Chollet\n#\n# Permission is hereby granted, free of charge, to any person obtaining a\n# copy of this software and associated documentation files (the \"Software\"),\n# to deal in the Software without restriction, including without limitation\n# the rights to use, copy, modify, merge, publish, distribute, sublicense,\n# and/or sell copies of the Software, and to permit persons to whom the\n# Software is furnished to do so, subject to the following conditions:\n#\n# The above copyright notice and this permission notice shall be included in\n# all copies or substantial portions of the Software.\n#\n# THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL\n# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING\n# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER\n# DEALINGS IN THE SOFTWARE.", "_____no_output_____" ] ], [ [ "# はじめてのニューラルネットワーク：分類問題の初歩", "_____no_output_____" ], [ "<table class=\"tfo-notebook-buttons\" align=\"left\">\n <td>\n <a target=\"_blank\" href=\"https://www.tensorflow.org/tutorials/keras/classification\"><img src=\"https://www.tensorflow.org/images/tf_logo_32px.png\" />View on TensorFlow.org</a>\n </td>\n <td>\n <a target=\"_blank\" href=\"https://colab.research.google.com/github/tensorflow/docs-l10n/blob/master/site/ja/tutorials/keras/classification.ipynb\"><img src=\"https://www.tensorflow.org/images/colab_logo_32px.png\" />Run in Google Colab</a>\n </td>\n <td>\n <a target=\"_blank\" href=\"https://github.com/tensorflow/docs-l10n/blob/master/site/ja/tutorials/keras/classification.ipynb\"><img src=\"https://www.tensorflow.org/images/GitHub-Mark-32px.png\" />View source on GitHub</a>\n </td>\n</table>", "_____no_output_____" ], [ "Note: これらのドキュメントは私たちTensorFlowコミュニティが翻訳したものです。コミュニティによる 翻訳は**ベストエフォート**であるため、この翻訳が正確であることや[英語の公式ドキュメント](https://www.tensorflow.org/?hl=en)の 最新の状態を反映したものであることを保証することはできません。 この翻訳の品質を向上させるためのご意見をお持ちの方は、GitHubリポジトリ[tensorflow/docs](https://github.com/tensorflow/docs)にプルリクエストをお送りください。 コミュニティによる翻訳やレビューに参加していただける方は、 [[email protected] メーリングリスト](https://groups.google.com/a/tensorflow.org/forum/#!forum/docs-ja)にご連絡ください。", "_____no_output_____" ], [ "このガイドでは、スニーカーやシャツなど、身に着けるものの写真を分類するニューラルネットワークのモデルを訓練します。すべての詳細を理解できなくても問題ありません。TensorFlowの全体を早足で掴むためのもので、詳細についてはあとから見ていくことになります。\n\nこのガイドでは、TensorFlowのモデルを構築し訓練するためのハイレベルのAPIである [tf.keras](https://www.tensorflow.org/guide/keras)を使用します。", "_____no_output_____" ] ], [ [ "try:\n # Colab only\n %tensorflow_version 2.x\nexcept Exception:\n pass\n", "_____no_output_____" ], [ "from __future__ import absolute_import, division, print_function, unicode_literals\n\n# TensorFlow と tf.keras のインポート\nimport tensorflow as tf\nfrom tensorflow import keras\n\n# ヘルパーライブラリのインポート\nimport numpy as np\nimport matplotlib.pyplot as plt\n\nprint(tf.__version__)", "_____no_output_____" ] ], [ [ "## ファッションMNISTデータセットのロード", "_____no_output_____" ], [ "このガイドでは、[Fashion MNIST](https://github.com/zalandoresearch/fashion-mnist)を使用します。Fashion MNISTには10カテゴリーの白黒画像70,000枚が含まれています。それぞれは下図のような1枚に付き1種類の衣料品が写っている低解像度（28×28ピクセル）の画像です。\n\n<table>\n <tr><td>\n <img src=\"https://tensorflow.org/images/fashion-mnist-sprite.png\"\n alt=\"Fashion MNIST sprite\" width=\"600\">\n </td></tr>\n <tr><td align=\"center\">\n <b>Figure 1.</b> <a href=\"https://github.com/zalandoresearch/fashion-mnist\">Fashion-MNIST samples</a> (by Zalando, MIT License).<br/>&nbsp;\n </td></tr>\n</table>\n\nFashion MNISTは、画像処理のための機械学習での\"Hello, World\"としてしばしば登場する[MNIST](http://yann.lecun.com/exdb/mnist/) データセットの代替として開発されたものです。MNISTデータセットは手書きの数字（0, 1, 2 など）から構成されており、そのフォーマットはこれから使うFashion MNISTと全く同じです。\n\nFashion MNISTを使うのは、目先を変える意味もありますが、普通のMNISTよりも少しだけ手応えがあるからでもあります。どちらのデータセットも比較的小さく、アルゴリズムが期待したとおりに機能するかどうかを確かめるために使われます。プログラムのテストやデバッグのためには、よい出発点になります。\n\nここでは、60,000枚の画像を訓練に、10,000枚の画像を、ネットワークが学習した画像分類の正確性を評価するのに使います。TensorFlowを使うと、下記のようにFashion MNISTのデータを簡単にインポートし、ロードすることが出来ます。", "_____no_output_____" ] ], [ [ "fashion_mnist = keras.datasets.fashion_mnist\n\n(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()", "_____no_output_____" ] ], [ [ "ロードしたデータセットは、NumPy配列になります。\n\n* `train_images` と `train_labels` の2つの配列は、モデルの訓練に使用される**訓練用データセット**です。\n* 訓練されたモデルは、 `test_images` と `test_labels` 配列からなる**テスト用データセット**を使ってテストします。\n\n画像は28×28のNumPy配列から構成されています。それぞれのピクセルの値は0から255の間の整数です。**ラベル**（label）は、0から9までの整数の配列です。それぞれの数字が下表のように、衣料品の**クラス**（class）に対応しています。\n\n<table>\n <tr>\n <th>Label</th>\n <th>Class</th> \n </tr>\n <tr>\n <td>0</td>\n <td>T-shirt/top</td> \n </tr>\n <tr>\n <td>1</td>\n <td>Trouser</td> \n </tr>\n <tr>\n <td>2</td>\n <td>Pullover</td> \n </tr>\n <tr>\n <td>3</td>\n <td>Dress</td> \n </tr>\n <tr>\n <td>4</td>\n <td>Coat</td> \n </tr>\n <tr>\n <td>5</td>\n <td>Sandal</td> \n </tr>\n <tr>\n <td>6</td>\n <td>Shirt</td> \n </tr>\n <tr>\n <td>7</td>\n <td>Sneaker</td> \n </tr>\n <tr>\n <td>8</td>\n <td>Bag</td> \n </tr>\n <tr>\n <td>9</td>\n <td>Ankle boot</td> \n </tr>\n</table>\n\n画像はそれぞれ単一のラベルに分類されます。データセットには上記の**クラス名**が含まれていないため、後ほど画像を出力するときのために、クラス名を保存しておきます。", "_____no_output_____" ] ], [ [ "class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat', \n 'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']", "_____no_output_____" ] ], [ [ "## データの観察\n\nモデルの訓練を行う前に、データセットのフォーマットを見てみましょう。下記のように、訓練用データセットには28×28ピクセルの画像が60,000枚含まれています。", "_____no_output_____" ] ], [ [ "train_images.shape", "_____no_output_____" ] ], [ [ "同様に、訓練用データセットには60,000個のラベルが含まれます。", "_____no_output_____" ] ], [ [ "len(train_labels)", "_____no_output_____" ] ], [ [ "ラベルはそれぞれ、0から9までの間の整数です。", "_____no_output_____" ] ], [ [ "train_labels", "_____no_output_____" ] ], [ [ "テスト用データセットには、10,000枚の画像が含まれます。画像は28×28ピクセルで構成されています。", "_____no_output_____" ] ], [ [ "test_images.shape", "_____no_output_____" ] ], [ [ "テスト用データセットには10,000個のラベルが含まれます。", "_____no_output_____" ] ], [ [ "len(test_labels)", "_____no_output_____" ] ], [ [ "## データの前処理\n\nネットワークを訓練する前に、データを前処理する必要があります。最初の画像を調べてみればわかるように、ピクセルの値は0から255の間の数値です。", "_____no_output_____" ] ], [ [ "plt.figure()\nplt.imshow(train_images[0])\nplt.colorbar()\nplt.grid(False)\nplt.show()", "_____no_output_____" ] ], [ [ "ニューラルネットワークにデータを投入する前に、これらの値を0から1までの範囲にスケールします。そのためには、画素の値を255で割ります。\n\n**訓練用データセット**と**テスト用データセット**は、同じように前処理することが重要です。", "_____no_output_____" ] ], [ [ "train_images = train_images / 255.0\n\ntest_images = test_images / 255.0", "_____no_output_____" ] ], [ [ "**訓練用データセット**の最初の25枚の画像を、クラス名付きで表示してみましょう。ネットワークを構築・訓練する前に、データが正しいフォーマットになっていることを確認します。", "_____no_output_____" ] ], [ [ "plt.figure(figsize=(10,10))\nfor i in range(25):\n plt.subplot(5,5,i+1)\n plt.xticks([])\n plt.yticks([])\n plt.grid(False)\n plt.imshow(train_images[i], cmap=plt.cm.binary)\n plt.xlabel(class_names[train_labels[i]])\nplt.show()", "_____no_output_____" ] ], [ [ "## モデルの構築\n\nニューラルネットワークを構築するには、まずモデルの階層を定義し、その後モデルをコンパイルします。", "_____no_output_____" ], [ "### 層の設定\n\nニューラルネットワークを形作る基本的な構成要素は**層**（layer）です。層は、入力されたデータから「表現」を抽出します。それらの「表現」は、今取り組もうとしている問題に対して、より「意味のある」ものであることが期待されます。\n\nディープラーニングモデルのほとんどは、単純な層の積み重ねで構成されています。`tf.keras.layers.Dense` のような層のほとんどには、訓練中に学習されるパラメータが存在します。", "_____no_output_____" ] ], [ [ "model = keras.Sequential([\n keras.layers.Flatten(input_shape=(28, 28)),\n keras.layers.Dense(128, activation='relu'),\n keras.layers.Dense(10, activation='softmax')\n])", "_____no_output_____" ] ], [ [ "このネットワークの最初の層は、`tf.keras.layers.Flatten` です。この層は、画像を（28×28ピクセルの）2次元配列から、28×28＝784ピクセルの、1次元配列に変換します。この層が、画像の中に積まれているピクセルの行を取り崩し、横に並べると考えてください。この層には学習すべきパラメータはなく、ただデータのフォーマット変換を行うだけです。\n\nピクセルが１次元化されたあと、ネットワークは2つの `tf.keras.layers.Dense` 層となります。これらの層は、密結合あるいは全結合されたニューロンの層となります。最初の `Dense` 層には、128個のノード（あるはニューロン）があります。最後の層でもある2番めの層は、10ノードの**softmax**層です。この層は、合計が1になる10個の確率の配列を返します。それぞれのノードは、今見ている画像が10個のクラスのひとつひとつに属する確率を出力します。\n\n### モデルのコンパイル\n\nモデルが訓練できるようになるには、いくつかの設定を追加する必要があります。それらの設定は、モデルの**コンパイル**(compile）時に追加されます。\n\n* **損失関数**（loss function） —訓練中にモデルがどれくらい正確かを測定します。この関数の値を最小化することにより、訓練中のモデルを正しい方向に向かわせようというわけです。\n* **オプティマイザ**（optimizer）—モデルが見ているデータと、損失関数の値から、どのようにモデルを更新するかを決定します。\n* **メトリクス**（metrics） —訓練とテストのステップを監視するのに使用します。下記の例では*accuracy* （正解率）、つまり、画像が正しく分類された比率を使用しています。", "_____no_output_____" ] ], [ [ "model.compile(optimizer='adam', \n loss='sparse_categorical_crossentropy',\n metrics=['accuracy'])", "_____no_output_____" ] ], [ [ "## モデルの訓練\n\nニューラルネットワークの訓練には次のようなステップが必要です。\n\n1. モデルに訓練用データを投入します—この例では `train_images` と `train_labels` の２つの配列です。\n2. モデルは、画像とラベルの対応関係を学習します。\n3. モデルにテスト用データセットの予測（分類）を行わせます—この例では `test_images` 配列です。その後、予測結果と `test_labels` 配列を照合します。 \n\n訓練を開始するには、`model.fit` メソッドを呼び出します。モデルを訓練用データに \"fit\"（適合）させるという意味です。", "_____no_output_____" ] ], [ [ "model.fit(train_images, train_labels, epochs=5)", "_____no_output_____" ] ], [ [ "モデルの訓練の進行とともに、損失値と正解率が表示されます。このモデルの場合、訓練用データでは0.88（すなわち88%）の正解率に達します。", "_____no_output_____" ], [ "## 正解率の評価\n\n次に、テスト用データセットに対するモデルの性能を比較します。", "_____no_output_____" ] ], [ [ "test_loss, test_acc = model.evaluate(test_images, test_labels, verbose=2)\n\nprint('\\nTest accuracy:', test_acc)", "_____no_output_____" ] ], [ [ "ご覧の通り、テスト用データセットでの正解率は、訓練用データセットでの正解率よりも少し低くなります。この訓練時の正解率とテスト時の正解率の差は、**過学習**（over fitting）の一例です。過学習とは、新しいデータに対する機械学習モデルの性能が、訓練時と比較して低下する現象です。", "_____no_output_____" ], [ "## 予測する\n\nモデルの訓練が終わったら、そのモデルを使って画像の分類予測を行うことが出来ます。", "_____no_output_____" ] ], [ [ "predictions = model.predict(test_images)", "_____no_output_____" ] ], [ [ "これは、モデルがテスト用データセットの画像のひとつひとつを分類予測した結果です。最初の予測を見てみましょう。", "_____no_output_____" ] ], [ [ "predictions[0]", "_____no_output_____" ] ], [ [ "予測結果は、10個の数字の配列です。これは、その画像が10の衣料品の種類のそれぞれに該当するかの「確信度」を表しています。どのラベルが一番確信度が高いかを見てみましょう。", "_____no_output_____" ] ], [ [ "np.argmax(predictions[0])", "_____no_output_____" ] ], [ [ "というわけで、このモデルは、この画像が、アンクルブーツ、`class_names[9]` である可能性が最も高いと判断したことになります。これが正しいかどうか、テスト用ラベルを見てみましょう。", "_____no_output_____" ] ], [ [ "test_labels[0]", "_____no_output_____" ] ], [ [ "10チャンネルすべてをグラフ化してみることができます。", "_____no_output_____" ] ], [ [ "def plot_image(i, predictions_array, true_label, img):\n predictions_array, true_label, img = predictions_array[i], true_label[i], img[i]\n plt.grid(False)\n plt.xticks([])\n plt.yticks([])\n\n plt.imshow(img, cmap=plt.cm.binary)\n\n predicted_label = np.argmax(predictions_array)\n if predicted_label == true_label:\n color = 'blue'\n else:\n color = 'red'\n\n plt.xlabel(\"{} {:2.0f}% ({})\".format(class_names[predicted_label],\n 100*np.max(predictions_array),\n class_names[true_label]),\n color=color)\n\ndef plot_value_array(i, predictions_array, true_label):\n predictions_array, true_label = predictions_array[i], true_label[i]\n plt.grid(False)\n plt.xticks([])\n plt.yticks([])\n thisplot = plt.bar(range(10), predictions_array, color=\"#777777\")\n plt.ylim([0, 1]) \n predicted_label = np.argmax(predictions_array)\n\n thisplot[predicted_label].set_color('red')\n thisplot[true_label].set_color('blue')", "_____no_output_____" ] ], [ [ "0番目の画像と、予測、予測配列を見てみましょう。", "_____no_output_____" ] ], [ [ "i = 0\nplt.figure(figsize=(6,3))\nplt.subplot(1,2,1)\nplot_image(i, predictions, test_labels, test_images)\nplt.subplot(1,2,2)\nplot_value_array(i, predictions, test_labels)\nplt.show()", "_____no_output_____" ], [ "i = 12\nplt.figure(figsize=(6,3))\nplt.subplot(1,2,1)\nplot_image(i, predictions, test_labels, test_images)\nplt.subplot(1,2,2)\nplot_value_array(i, predictions, test_labels)\nplt.show()", "_____no_output_____" ] ], [ [ "予測の中のいくつかの画像を、予測値とともに表示してみましょう。正しい予測は青で、誤っている予測は赤でラベルを表示します。数字は予測したラベルのパーセント（100分率）を示します。自信があるように見えても間違っていることがあることに注意してください。", "_____no_output_____" ] ], [ [ "# X個のテスト画像、予測されたラベル、正解ラベルを表示します。\n# 正しい予測は青で、間違った予測は赤で表示しています。\nnum_rows = 5\nnum_cols = 3\nnum_images = num_rows*num_cols\nplt.figure(figsize=(2*2*num_cols, 2*num_rows))\nfor i in range(num_images):\n plt.subplot(num_rows, 2*num_cols, 2*i+1)\n plot_image(i, predictions, test_labels, test_images)\n plt.subplot(num_rows, 2*num_cols, 2*i+2)\n plot_value_array(i, predictions, test_labels)\nplt.show()", "_____no_output_____" ] ], [ [ "最後に、訓練済みモデルを使って1枚の画像に対する予測を行います。", "_____no_output_____" ] ], [ [ "# テスト用データセットから画像を1枚取り出す\nimg = test_images[0]\n\nprint(img.shape)", "_____no_output_____" ] ], [ [ "`tf.keras` モデルは、サンプルの中の**バッチ**（batch）あるいは「集まり」について予測を行うように作られています。そのため、1枚の画像を使う場合でも、リスト化する必要があります。", "_____no_output_____" ] ], [ [ "# 画像を1枚だけのバッチのメンバーにする\nimg = (np.expand_dims(img,0))\n\nprint(img.shape)", "_____no_output_____" ] ], [ [ "そして、予測を行います。", "_____no_output_____" ] ], [ [ "predictions_single = model.predict(img)\n\nprint(predictions_single)", "_____no_output_____" ], [ "plot_value_array(0, predictions_single, test_labels)\n_ = plt.xticks(range(10), class_names, rotation=45)", "_____no_output_____" ] ], [ [ "`model.predict` メソッドの戻り値は、リストのリストです。リストの要素のそれぞれが、バッチの中の画像に対応します。バッチの中から、（といってもバッチの中身は１つだけですが）予測を取り出します。", "_____no_output_____" ] ], [ [ "np.argmax(predictions_single[0])", "_____no_output_____" ] ], [ [ "というわけで、モデルは9というラベルを予測しました。", "_____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" ]

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# PA005: High Value Customer Identification (Insiders)", "_____no_output_____" ], [ "# Ciclo 0 - Planejamento da solução (IOT)", "_____no_output_____" ], [ "# Ciclo 1 - Métricas de Validação de Clusters", "_____no_output_____" ], [ "1. Feature Engineering\n* Recency\n* Frequency\n* Monetary\n\n2. Métricas de Validação de clusters\n* WSS - (Within-Cluster Sum of Squares)\n* SS - (Silhouette Score)\n\n3. Cluster Analisys\n* Plot 3d\n* Cluster profile", "_____no_output_____" ], [ "# Ciclo 2 - Análise de Silhouette", "_____no_output_____" ], [ "1. Feature Engineering\n* Average Ticket\n\n\n2. Análise de Silhouette\n* Silhouette Analysis\n\n\n3. Clustering visualization\n* UMAP\n\n\n4. Cluster análise de perfil\n* Descrição dos centroides dos clusters", "_____no_output_____" ], [ "# Ciclo 3 - Statistical Descriptive", "_____no_output_____" ], [ "1. Análise Descritiva\n* Atributos Numéricos\n* Atributos Categóricos\n\n2. Data Filtering\n* Retornos e Compras\n\n3. Feature Engineering\n* Utilização dos dados de compra\n* Average Recency\n* Number of returns\n\n", "_____no_output_____" ], [ "## Input - Entrada", "_____no_output_____" ], [ "1. Problema de negócio\n\n --Selecionar os clientes mais valiosos para integrar um programa de fidelização\n \n2. Conjunto de dados\n\n --Vendas de um e-commerce, durante um período de um ano", "_____no_output_____" ], [ "## Output - Saída", "_____no_output_____" ], [ "1. A idicação de clientes que farão parte do programa de Insiders\n\n --Lista: client_id | is_insider\n 455534 | yes\n 433524 | no\n \n2. Relatório com as respostas das perguntas de negócios\n\n 2.0. Quem são as pessoas elegíveis para fazer parte do grupo de Insiders?\n\n 2.1. Quantos clientes farão parte do grupo?\n \n 2.2. Quais as principais características desses clientes?\n \n 2.3. Qual a porcentagem de contribuição do faturamento vindo dos Insiders?\n \n 2.4. Qual a expectativa de faturamento desse grupo para os próximos meses?\n \n 2.5. Quais as condições para umma pessoa ser elegível ao Insiders?\n \n 2.6. Quais as condições para umma pessoa ser removida do Insiders?\n \n 2.7. Qual a garantia que o programa Insiders é melhor que o restante da base?\n \n 2.8. Quais ações o time de marketing pode realizar para aumentar o faturamento?\n \n ", "_____no_output_____" ], [ "## Tasks - Tarefas", "_____no_output_____" ], [ " 0. Quem são as pessoas elegíveis para fazer parte do grupo de Insiders? \n - O que são clientes de maior \"valor\"?\n - Faturamento\n - Alto ticket médio\n - Alto LTV\n - Baixa recência\n - Baixa probabilidade de churn\n - Alto basket size\n - Alta previsão LTV\n - Alta propensão de compra\n \n - Custo\n - Baixa taxa de devolução\n \n - Experiência de compra\n - Média alta das avaliações\n \n 1. Quantos clientes farão parte do grupo?\n - Número total de clientes\n - % do grupo de Insiders\n \n 2. Quais as principais características desses clientes?\n -Características dos clientes:\n - Idade\n - Localização\n \n - Características de consumo:\n - Atributos da clusterização\n \n \n 3. Qual a porcentagem de contribuição do faturamento vindo dos Insiders?\n - Faturamento do ano\n - Faturamento dos Insiders\n \n 4. Qual a expectativa de faturamento desse grupo para os próximos meses?\n - LTV do grupo Insiders\n - Análise de Cohort\n \n 5. Quais as condições para umma pessoa ser elegível ao Insiders?\n - Definir periodicidade\n - A pessoa precisa ser similar ou parecido com uma pessoa do grupo\n \n 6. Quais as condições para umma pessoa ser removida do Insiders?\n - Definir periodicidade\n - A pessoa precisa ser dissimilar ou parecido com uma pessoa do grupo\n \n 7. Qual a garantia que o programa Insiders é melhor que o restante da base?\n - Teste A/B\n - Teste A/B Bayesiano\n - Teste de hipóteses\n \n 8. Quais ações o time de marketing pode realizar para aumentar o faturamento?\n - Desconto\n - Preferência de compra\n - Frente\n - Visita a empresa", "_____no_output_____" ], [ "## Benchmark de soluções", "_____no_output_____" ], [ "### 1. Desk research", "_____no_output_____" ], [ "Modelo RFM\n1. Recency\n a) Tempo desde a última compra\n b) Responsividade\n\n2. Frequency\n a) Tempo médio entre as transações\n b) Engajamento\n \n3. Monetary \n a) Total gasto, faturamento\n b) 'High-value purchases'", "_____no_output_____" ], [ "# 0.0. Imports", "_____no_output_____" ] ], [ [ "\nfrom yellowbrick.cluster import KElbowVisualizer\nfrom yellowbrick.cluster import SilhouetteVisualizer\n\nfrom matplotlib import pyplot as plt\nfrom sklearn import cluster as c\nfrom sklearn import metrics as m\nfrom sklearn import preprocessing as pp\nfrom plotly import express as px\n\nimport pandas as pd\nimport seaborn as sns\nimport numpy as np\n\nimport re", "_____no_output_____" ], [ "import umap.umap_ as umap\n", "/home/heitor/repos/insiders_clustering/venv/lib/python3.8/site-packages/tqdm/auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n from .autonotebook import tqdm as notebook_tqdm\n" ] ], [ [ "## 0.2. Helper Functions", "_____no_output_____" ] ], [ [ "pd.set_option('display.float_format', lambda x: '%.2f' % x)\n\ndef num_attributes(df1):\n \n num_attributes = df1.select_dtypes(['int64', 'float64'])\n\n #central tendency\n ct1 = pd.DataFrame(num_attributes.apply(np.mean)).T\n ct2 = pd.DataFrame(num_attributes.apply(np.median)).T\n\n #dispersion\n d1 = pd.DataFrame(num_attributes.apply(np.min)).T\n d2 = pd.DataFrame(num_attributes.apply(np.max)).T\n d3 = pd.DataFrame(num_attributes.apply(lambda x: x.max() - x.min())).T\n d4 = pd.DataFrame(num_attributes.apply(np.std)).T\n d5 = pd.DataFrame(num_attributes.apply(lambda x: x.skew())).T\n d6 = pd.DataFrame(num_attributes.apply(lambda x: x.kurtosis())).T\n\n m = pd.concat( [d1, d2, d3, ct1, ct2, d4, d5, d6] ).T.reset_index()\n m.columns = ['attributes', 'min', 'max', 'range', 'mean', 'median', 'std','skew', 'kurtosis']\n return m", "_____no_output_____" ] ], [ [ "## 0.3. Load Data", "_____no_output_____" ] ], [ [ "df_raw = pd.read_csv(r'../data/Ecommerce.csv')\n", "_____no_output_____" ] ], [ [ "# 1.0. Data Description", "_____no_output_____" ] ], [ [ "df1 = df_raw.copy()", "_____no_output_____" ] ], [ [ "## 1.1. Rename Columns", "_____no_output_____" ] ], [ [ "df1.columns", "_____no_output_____" ], [ "df1.columns = ['invoice_no', 'stock_code', 'description', 'quantity', 'invoice_date',\n 'unit_price', 'customer_id', 'country']", "_____no_output_____" ] ], [ [ "## 1.2. Data Shape", "_____no_output_____" ] ], [ [ "print(f'Number of rows: {df1.shape[0]}')\nprint(f'Number of columns: {df1.shape[1]}')", "Number of rows: 541909\nNumber of columns: 8\n" ] ], [ [ "## 1.3. Data Types", "_____no_output_____" ] ], [ [ "df1.dtypes", "_____no_output_____" ] ], [ [ "## 1.4. Check NAs\n", "_____no_output_____" ] ], [ [ "df1.isna().sum()", "_____no_output_____" ] ], [ [ "## 1.5. Fill NAs", "_____no_output_____" ] ], [ [ "#remove na\ndf1 = df1.dropna(axis=0)\n\nprint('Data removed: {:.0f}%'.format((1-(len(df1)/len(df_raw)))*100))", "Data removed: 25%\n" ] ], [ [ "## 1.6. Change dtypes", "_____no_output_____" ] ], [ [ "df1.dtypes", "_____no_output_____" ], [ "#invoice_no \n# df1['invoice_no'] = df1['invoice_no'].astype(int)\n\n\n#stock_code \n# df1['stock_code'] = df1['stock_code'].astype(int)\n\n\n#invoice_date --> Month --> b\ndf1['invoice_date'] = pd.to_datetime(df1['invoice_date'], format=('%d-%b-%y'))\n\n\n#customer_id\ndf1['customer_id'] = df1['customer_id'].astype(int)\ndf1.dtypes", "_____no_output_____" ] ], [ [ "## 1.7. Descriptive statistics", "_____no_output_____" ] ], [ [ "num_attributes = df1.select_dtypes(['int64', 'float64'])\ncat_attributes = df1.select_dtypes(exclude = ['int64', 'float64', 'datetime64[ns]'])", "_____no_output_____" ] ], [ [ "### 1.7.1. Numerical Attributes", "_____no_output_____" ] ], [ [ "m = num_attributes(df1)\nm", "_____no_output_____" ] ], [ [ "#### 1.7.1.1 Investigating", "_____no_output_____" ], [ "1. Negative quantity (devolution?)\n2. Price = 0 (Promo?)", "_____no_output_____" ], [ "## 1.7.2. Categorical Attributes", "_____no_output_____" ] ], [ [ "cat_attributes.head()", "_____no_output_____" ] ], [ [ "#### Invoice no", "_____no_output_____" ] ], [ [ "\n#invoice_no -- some of them has one char\ndf_invoice_char = df1.loc[df1['invoice_no'].apply(lambda x: bool(re.search('[^0-9]+', x))), :] \n\nlen(df_invoice_char[df_invoice_char['quantity']<0])\n\n\nprint('Total of invoices with letter: {}'.format(len(df_invoice_char)))\nprint('Total of negative quantaty: {}'.format(len(df1[df1['quantity']<0])))\nprint('Letter means negative quantity')", "_____no_output_____" ] ], [ [ "#### Stock Code", "_____no_output_____" ] ], [ [ "#all stock codes with char\ndf1.loc[df1['stock_code'].apply(lambda x: bool(re.search('^[a-zA-Z]+$', x))), 'stock_code'].unique()\n\n#remove stock code in ['POST', 'D', 'M', 'PADS', 'DOT', 'CRUK']\n# df1 = df1[-df1.isin(['POST', 'D', 'M', 'PADS', 'DOT', 'CRUK'])]", "_____no_output_____" ] ], [ [ "#### Description", "_____no_output_____" ] ], [ [ "#remove description\n# df1 = df1.drop('description', axis=1)", "_____no_output_____" ] ], [ [ "#### Country\n", "_____no_output_____" ] ], [ [ "df1['country'].value_counts(normalize='True').head()", "_____no_output_____" ], [ "df1[['country', 'customer_id']].drop_duplicates().groupby('country').count().reset_index().sort_values('customer_id', ascending=False).head()", "_____no_output_____" ] ], [ [ "# 2.0. Data Filtering", "_____no_output_____" ] ], [ [ "df2 = df1.copy()", "_____no_output_____" ], [ "df2['country'].unique()", "_____no_output_____" ], [ "# ====================== Numerical Attributes ====================== \n\n#unit price != 0\n# df2.sort_values('unit_price').head()\ndf2 = df2[df2['unit_price']>0.004]\n\n# ====================== Categorical Attributes ====================== \n\n#stock code\ndf2 = df2[~df2.isin(['POST', 'D', 'M', 'PADS', 'DOT', 'CRUK'])]\n\n#description\ndf2 = df2.drop('description', axis=1)\n\n\n#map\ndf2 = df2[~df2.isin(['European Community', 'Unspecified'])]\n\n# #quantity - negative numbers mean return\ndf_return = df2[df2['quantity']<0]\ndf_purchase = df2[df2['quantity']>0]\n\n\n", "_____no_output_____" ] ], [ [ "# 3.0. Feature Engineering", "_____no_output_____" ] ], [ [ "df3 = df2.copy()", "_____no_output_____" ] ], [ [ "## 3.1. Feature Creation", "_____no_output_____" ] ], [ [ "# df_purchase.loc[:, 'gross_revenue'] = df_purchase.loc[:, 'quantity'] * df_purchase.loc[:, 'unit_price'] \ndf_purchase.loc[:,'gross_revenue'] = df_purchase.loc[:, 'quantity'] * df_purchase.loc[:, 'unit_price'] \n\n#data reference\ndf_ref = df3.drop(['invoice_no', 'stock_code', 'quantity', 'invoice_date',\n 'unit_price', 'country'], axis=1).drop_duplicates(ignore_index=True)\n\n\ndf_monetary = df_purchase.loc[:,['customer_id', 'gross_revenue']].groupby('customer_id').sum().reset_index()\ndf_ref = pd.merge(df_ref, df_monetary, on='customer_id', how= 'left')\ndf_ref.isna().sum()", "/tmp/ipykernel_19796/959218557.py:6: SettingWithCopyWarning: \nA value is trying to be set on a copy of a slice from a DataFrame.\nTry using .loc[row_indexer,col_indexer] = value instead\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 df_purchase.loc[:,'gross_revenue'] = df_purchase.loc[:, 'quantity'] * df_purchase.loc[:, 'unit_price']\n" ], [ "#recency --> max date\n\ndf_recency = df_purchase.loc[:,['customer_id', 'invoice_date']].groupby('customer_id').max().reset_index()\ndf_recency['recency_days'] = (df3['invoice_date'].max() - df_recency['invoice_date']).dt.days\ndf_recency = df_recency[['customer_id', 'recency_days']].copy()\ndf_ref = pd.merge(df_ref, df_recency, on='customer_id', how='left')\n\n#frequency\ndf_freq = df_purchase.loc[:,['customer_id', 'invoice_no']].drop_duplicates().groupby('customer_id').count().reset_index()\ndf_ref = pd.merge(df_ref, df_freq, on='customer_id', how='left')\n\n\n#avg ticket\ndf_avgticket = df_purchase.loc[:,['customer_id', 'gross_revenue']].groupby('customer_id').mean().rename(columns={'gross_revenue':'avg_ticket'}).reset_index()\ndf_ref = pd.merge(df_ref, df_avgticket, on='customer_id', how='left')\n\n\n\n\n\n", "_____no_output_____" ], [ "df_ref.head()", "_____no_output_____" ] ], [ [ "# 4.0. Exploratory Data Analisys", "_____no_output_____" ] ], [ [ "df4 = df_ref.dropna().copy()", "_____no_output_____" ] ], [ [ "# 5.0. Data Preparation", "_____no_output_____" ] ], [ [ "df5 = df4.copy()\n\nss = pp.StandardScaler()\nmms = pp.MinMaxScaler()", "_____no_output_____" ], [ "df5['gross_revenue'] = ss.fit_transform(df5[['gross_revenue']])\ndf5['recency_days'] = ss.fit_transform(df5[['recency_days']])\ndf5['invoice_no'] = ss.fit_transform(df5[['invoice_no']])\ndf5['avg_ticket'] = ss.fit_transform(df5[['avg_ticket']])", "_____no_output_____" ], [ "df5.head()", "_____no_output_____" ] ], [ [ "# 6.0. Feature Selection", "_____no_output_____" ] ], [ [ "df6 = df5.copy()", "_____no_output_____" ] ], [ [ "# 7.0. Hyperparameter Fine-Tunning", "_____no_output_____" ] ], [ [ "X = df6.drop('customer_id', axis=1)\nclusters = [2,3,4,5,6, 7]", "_____no_output_____" ] ], [ [ "## 7.1. Within-Cluster Sum of Squares (WSS)", "_____no_output_____" ] ], [ [ "#Easy way\n\nkmeans = KElbowVisualizer(c.KMeans(), k=clusters, timings=False)\nkmeans.fit(X)\nkmeans.show();\n\n", "_____no_output_____" ] ], [ [ "## 7.2. Silhouette Score", "_____no_output_____" ] ], [ [ "#Easy way\nkmeans = KElbowVisualizer(c.KMeans(), k=clusters, metric = 'silhouette', timings = False)\nkmeans.fit(X)\nkmeans.show();", "_____no_output_____" ] ], [ [ "## 7.3. Silhouette Analysis", "_____no_output_____" ] ], [ [ "fig, ax = plt.subplots(3, 2, figsize=(25,18))\n\nfor k in clusters:\n km = c.KMeans(n_clusters=k, init='random', n_init=10, max_iter=100, random_state=3)\n q, mod = divmod(k, 2)\n \n visualizer = SilhouetteVisualizer(estimator = km ,colors='yellowbrick', ax=ax[q-1][mod])\n visualizer.fit(X)\n visualizer.finalize()", "_____no_output_____" ] ], [ [ "# 8.0. Model Training", "_____no_output_____" ], [ "## 8.1. KMeans", "_____no_output_____" ] ], [ [ "#model definition\nk = 3\nkmeans = c.KMeans(init='random', n_clusters=k, n_init=10, max_iter=300, random_state=3)\n\n#model training\nkmeans.fit(X)\n\n#clustering\nlabels = kmeans.labels_", "_____no_output_____" ] ], [ [ "# 9.0. Cluster Analisys", "_____no_output_____" ] ], [ [ "df9 = df6.copy()\ndf9['cluster'] = labels\n", "_____no_output_____" ] ], [ [ "## 9.1. Visualization Inspection", "_____no_output_____" ] ], [ [ "\n\nvisualizer = SilhouetteVisualizer(kmeans, colors='yellowbrick')\nvisualizer.fit(X)\nvisualizer.finalize()", "/home/heitor/repos/insiders_clustering/venv/lib/python3.8/site-packages/sklearn/base.py:450: UserWarning: X does not have valid feature names, but KMeans was fitted with feature names\n warnings.warn(\n" ] ], [ [ "## 9.2. 2d Plot", "_____no_output_____" ] ], [ [ "# df_vis = df9.drop('customer_id', axis=1)\n# sns.pairplot(df_vis, hue='cluster')\n\n", "_____no_output_____" ] ], [ [ "## 9.3. UMAP t-SNE", "_____no_output_____" ] ], [ [ "reducer = umap.UMAP(n_neighbors=30, random_state=3)\nembedding = reducer.fit_transform(X)\n\ndf_vis['embedding_x'] = embedding[:,0]\ndf_vis['embedding_y'] = embedding[:,1]\nsns.scatterplot(x='embedding_x', y='embedding_y', hue='cluster',\n palette = sns.color_palette('hls', n_colors = len(df_vis['cluster'].unique())),\n data = df_vis)", "_____no_output_____" ] ], [ [ "## 9.1. Visualization Inspection", "_____no_output_____" ] ], [ [ "#WSS\nprint('WSS: {}'.format(kmeans.inertia_))\n\n#SS\nprint('SS: {}'.format(m.silhouette_score(X,labels, metric='euclidean')))", "WSS: 9429.573769591378\nSS: 0.5879176205734721\n" ], [ "px.scatter_3d(df9, x='recency_days', y='invoice_no', z='gross_revenue', color='cluster')", "_____no_output_____" ] ], [ [ "## 9.2. Cluster Profile", "_____no_output_____" ] ], [ [ "aux1 = df9.groupby('cluster').mean().reset_index()\n\naux1 = aux1.drop('customer_id', axis=1)\naux2 = df9[['customer_id', 'cluster']].groupby('cluster').count().reset_index()\n\ndf_cluster = pd.merge(aux1, aux2, on='cluster', how='left')\n", "_____no_output_____" ], [ "df_cluster['perc'] = 100*df_cluster['customer_id']/df_cluster['customer_id'].sum()\n\ndf_cluster", "_____no_output_____" ] ], [ [ "## Cluster 0 (insiders)\n\n* Número de clientes: 6\n* Percentual de clientes: 0,01%\n* Faturamento médio: $182182\n* Recência: 7 dias\n* Frequência: 89 compras\n\n## Cluster 1 (Hibernating)\n\n* Número de clientes: 4335\n* Percentual de clientes: 99%\n* Faturamento: $1372\n* Recência: 92 dias\n* Frequência: 4 compras\n\n## Cluster 2 (potenciais clientes fieis)\n\n* Número de clientes: 31\n* Percentual de clientes: 0,07%\n* Faturamento: $40543\n* Recência: 13 dias\n* Frequência: 53 compras\n\n\n", "_____no_output_____" ], [ "# 10.0. Deploy to Production", "_____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" ]

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

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ [ [ "# Invariantin massan histogrammin piirtäminen", "_____no_output_____" ], [ "Tässä harjoituksessa opetellaan piirtämään invariantin massan histogrammi Pythonilla. Käytetään datana CMS-kokeen vuonna 2011 keräämää dataa kahden myonin törmäyksistä [1]. Tässä harjoituksessa käytettävään CSV-tiedostoon on karsittu edellä mainitusta datasta kiinnostavia tapahtumia, joissa myonille laskettu invariantti massa on välillä 8–12 GeV [2].\n\nTutustu alla oleviin koodisoluihin ja niissä #-merkillä erotettuihin kommenttiriveihin sekä aja koodia. Huomaa, että normaalisti koodia ei kommentoitaisi näin runsaasti, nyt kommenteissa kerrotaan lisätietoa käytetyistä komennoista.\n<br>\n<br>\n<br>\n[1] CMS collaboration (2016). DoubleMu primary dataset in AOD format from RunA of 2011 (/DoubleMu/Run2011A-12Oct2013-v1/AOD). CERN Open Data Portal. DOI: [10.7483/OPENDATA.CMS.RZ34.QR6N](http://doi.org/10.7483/OPENDATA.CMS.RZ34.QR6N).\n<br>\n[2] Thomas McCauley (2016). Ymumu. Jupyter Notebook file. https://github.com/tpmccauley/cmsopendata-jupyter/blob/hst-0.1/Ymumu.ipynb.", "_____no_output_____" ], [ "### 1) Alustus ", "_____no_output_____" ] ], [ [ "# Haetaan tarvittavat moduulit\nimport pandas\nimport matplotlib.pyplot as plt", "_____no_output_____" ] ], [ [ "### 2) Datan hakeminen", "_____no_output_____" ], [ "Alkuvalmisteluiden jälkeen siirrytään hakemaan CMS:n dataa käyttöömme notebookiin.", "_____no_output_____" ] ], [ [ "# Luodaan DataFrame-rakenne (periaatteessa taulukko), johon kirjataan kaikki CSV-tiedostossa oleva data.\n# Annetaan luomallemme DataFramelle nimi 'datasetti'.\ndatasetti = pandas.read_csv('https://raw.githubusercontent.com/cms-opendata-education/cms-jupyter-materials-finnish/master/Data/Ymumu_Run2011A.csv')\n\n# Luodaan muuttuja 'invariantti_massa', johon tallennetaan 'datasetin' sarakkeella 'M' olevat arvot, eli\n# kahden myonin invariantille massalle valmiiksi tiedostoon lasketut arvot.\ninvariantti_massa = datasetti['M']", "_____no_output_____" ] ], [ [ "### 3) Histogrammin piirtäminen", "_____no_output_____" ], [ "Nyt jäljellä on enää vaihe, jossa luomme histogrammin hakemistamme invariantin massan arvoista. Histogrammi on pylväskaavio, joka kuvaa kuinka monta törmäystapahtumaa on osunut kunkin invariantin massan arvon kohdalle. Huomaa, että alla käytämme yhteensä 500 pylvästä.", "_____no_output_____" ] ], [ [ "# Suoritetaan histogrammin piirtäminen pyplot-moduulin avulla:\n# (http://matplotlib.org/api/pyplot_api.html?highlight=matplotlib.pyplot.hist#matplotlib.pyplot.hist).\n# 'Bins' määrittelee histogrammin pylväiden lukumäärän.\nplt.hist(invariantti_massa, bins=500)\n\n# Näillä riveillä ainoastaan määritellään otsikko sekä akseleiden tekstit.\nplt.xlabel('Invariantti massa [GeV]')\nplt.ylabel('Tapahtumien lukumäärä')\nplt.title('Kahden myonin invariantin massan histogrammi \\n') # \\n luo uuden rivin otsikon muotoilua varten\n\n# Tehdään kuvaaja näkyväksi.\nplt.show()", "_____no_output_____" ] ], [ [ "### 4) Analyysi", "_____no_output_____" ], [ "- Mitä histogrammi kertoo?\n- Mitä tapahtuu noin 9,45 GeV:n kohdalla?", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## Burgers Equation - Forward Euler/0-estimation\n#### Parameter estimation for Burgers' Equation using Gaussian processes (Forward Euler scheme)\n\n\n#### Problem Setup\n\n$u_t + u u_{x} = \\nu u_{x}$\n\n$u(x,t) = \\frac{x}{1+t}$ => We'd expect $\\nu = 0$\n\n$u_0(x) := u(x,0) = x$\n\n$x \\in [0, 1], t \\in \\{0, \\tau \\}$\n\nUsing the forward Euler scheme, the equation can be re-written as:\n\n$\\frac{u_n - u_{n-1}}{\\tau} + u_{n-1} \\frac{d}{dx}u_{n-1} = \\nu \\frac{d^2}{dx}u_{n-1}$\n\nand setting the factor $u_{n-1}(x) = u_0(x) = x$ (no mean used! Should give a better result) to deal with the non-linearity:\n\n$\\tau \\nu \\frac{d^2}{dx}u_{n-1} - \\tau x \\frac{d}{dx}u_{n-1} + u_{n-1} = u_{n}$\n\n\nConsider $u_{n-1}$ to be a Gaussian process.\n\n$u_{n-1} \\sim \\mathcal{GP}(0, k_{uu}(x_i, x_j, \\theta))$\n\nAnd the linear operator:\n\n$\\mathcal{L}_x^\\nu = \\cdot + \\tau \\nu \\frac{d}{dx}\\cdot - \\tau x \\frac{d}{dx} \\cdot$\n\nso that\n\n$\\mathcal{L}_x^\\nu u_{n-1} = u_n$\n\nProblem at hand: estimate $\\nu$ (should be $\\nu = 0$ in the end).\n\nFor the sake of simplicity, take $u := u_{n-1}$ and $f := u_n$.\n\n\n#### step 1: Simulate data\n\nTake data points at $t = 0$ for $(u_{n-1})$ and $t = \\tau$ for $(u_n)$, where $\\tau$ is the time step.", "_____no_output_____" ] ], [ [ "import numpy as np\nimport sympy as sp\nfrom scipy.optimize import minimize\nimport matplotlib.pyplot as plt\nimport warnings\nimport time", "_____no_output_____" ], [ "tau = 0.001\ndef get_simulated_data(tau, n=20):\n x = np.random.rand(n)\n y_u = x\n y_f = x/(1+tau)\n return (x, y_u, y_f)\n\n(x, y_u, y_f) = get_simulated_data(tau)", "_____no_output_____" ], [ "plt.plot(x, y_u, 'ro')\nplt.show()", "_____no_output_____" ], [ "plt.plot(x, y_f, 'bo')\nplt.show()", "_____no_output_____" ] ], [ [ "#### Step 2:Evaluate kernels\n\n$k_{nn}(x_i, x_j; \\theta) = \\theta exp(-\\frac{1}{2l}(x_i-x_j)^2)$", "_____no_output_____" ] ], [ [ "x_i, x_j, theta, l, nu = sp.symbols('x_i x_j theta l nu')\nkuu_sym = theta*sp.exp(-1/(2*l)*((x_i - x_j)**2))\nkuu_fn = sp.lambdify((x_i, x_j, theta, l), kuu_sym, \"numpy\")\ndef kuu(x, theta, l):\n k = np.zeros((x.size, x.size))\n for i in range(x.size):\n for j in range(x.size):\n k[i,j] = kuu_fn(x[i], x[j], theta, l)\n return k", "_____no_output_____" ] ], [ [ "$k_{ff}(x_i,x_j;\\theta,\\phi) \\\\\n= \\mathcal{L}_{x_i}^\\nu \\mathcal{L}_{x_j}^\\nu k_{uu}(x_i, x_j; \\theta) \\\\\n= k_{uu} + \\tau \\nu \\frac{d}{dx_i}k_{uu} - \\tau x_i \\frac{d}{dx_i}k_{uu} + \\tau \\nu \\frac{d}{dx_j}k_{uu} + \\tau^2 \\nu^2 \\frac{d}{dx_i} \\frac{d}{dx_j}k_{uu} - \\tau^2 \\nu x_i\\frac{d^2}{dx_i dx_j} k_{uu} - \\tau x_j \\frac{d}{dx_j}k_{uu} - \\tau^2 \\nu x_j \\frac{d^2}{dx_i dx_j} k_{uu} + \\tau^2 x_i x_j \\frac{d^2}{dx_i dx_j}k_{uu}$", "_____no_output_____" ] ], [ [ "kff_sym = kuu_sym \\\n + tau*nu*sp.diff(kuu_sym, x_i) \\\n - tau*x_i*sp.diff(kuu_sym, x_i) \\\n + tau*nu*sp.diff(kuu_sym, x_j) \\\n + tau**2*nu**2*sp.diff(kuu_sym, x_j, x_i) \\\n - tau**2*nu*x_i*sp.diff(kuu_sym, x_j, x_i) \\\n - tau*x_j*sp.diff(kuu_sym, x_j) \\\n - tau**2*nu*x_j*sp.diff(kuu_sym, x_j, x_i) \\\n + tau**2*x_i*x_j*sp.diff(kuu_sym, x_j, x_i)\nkff_fn = sp.lambdify((x_i, x_j, theta, l, nu), kff_sym, \"numpy\")\ndef kff(x, theta, l, nu):\n k = np.zeros((x.size, x.size))\n for i in range(x.size):\n for j in range(x.size):\n k[i,j] = kff_fn(x[i], x[j], theta, l, nu)\n return k", "_____no_output_____" ] ], [ [ "$k_{fu}(x_i,x_j;\\theta,\\phi) \\\\\n= \\mathcal{L}_{x_i}^\\nu k_{uu}(x_i, x_j; \\theta) \\\\\n= k_{uu} + \\tau \\nu \\frac{d}{dx_i}k_{uu} - \\tau x_i\\frac{d}{dx_i}k_{uu}$", "_____no_output_____" ] ], [ [ "kfu_sym = kuu_sym + tau*nu*sp.diff(kuu_sym, x_i) - tau*x_i*sp.diff(kuu_sym, x_i)\nkfu_fn = sp.lambdify((x_i, x_j, theta, l, nu), kfu_sym, \"numpy\")\ndef kfu(x, theta, l, nu):\n k = np.zeros((x.size, x.size))\n for i in range(x.size):\n for j in range(x.size):\n k[i,j] = kfu_fn(x[i], x[j], theta, l, nu)\n return k", "_____no_output_____" ], [ "def kuf(x, theta, l, nu):\n return kfu(x,theta, l, nu).T", "_____no_output_____" ] ], [ [ "#### Step 3: Compute NLML", "_____no_output_____" ] ], [ [ "def nlml(params, x, y1, y2, s):\n theta_exp = np.exp(params[0]) \n l_exp = np.exp(params[1])\n K = np.block([\n [kuu(x, theta_exp, l_exp) + s*np.identity(x.size), kuf(x, theta_exp, l_exp, params[2])],\n [kfu(x, theta_exp, l_exp, params[2]), kff(x, theta_exp, l_exp, params[2]) + s*np.identity(x.size)]\n ])\n y = np.concatenate((y1, y2))\n val = 0.5*(np.log(abs(np.linalg.det(K))) + np.mat(y) * np.linalg.inv(K) * np.mat(y).T)\n return val.item(0)", "_____no_output_____" ] ], [ [ "#### Step 4: Optimise hyperparameters", "_____no_output_____" ] ], [ [ "m = minimize(nlml, np.random.rand(3), args=(x, y_u, y_f, 1e-7), method=\"Nelder-Mead\", options = {'maxiter' : 1000})", "_____no_output_____" ], [ "m.x[2]", "_____no_output_____" ], [ "m", "_____no_output_____" ] ], [ [ "#### Step 5: Analysis w.r.t. the number of data points (up to 25):", "_____no_output_____" ], [ "In this section we want to analyze the error of our algorithm using two different ways and its time complexity.", "_____no_output_____" ] ], [ [ "res = np.zeros((5,25))\ntiming = np.zeros((5,25))\nwarnings.filterwarnings(\"ignore\")\nfor k in range(5):\n for n in range(25):\n start_time = time.time()\n (x, y_u, y_f) = get_simulated_data(tau, n)\n m = minimize(nlml, np.random.rand(3), args=(x, y_u, y_f, 1e-7), method=\"Nelder-Mead\")\n res[k][n] = m.x[2]\n timing[k][n] = time.time() - start_time", "_____no_output_____" ] ], [ [ "###### Plotting the error in our estimate for $\\nu$ (Error = $| \\nu_{estimate} - \\nu_{true} |$):", "_____no_output_____" ] ], [ [ "lin = np.linspace(1, res.shape[1], res.shape[1])\nest = np.repeat(0.01, len(lin))\n\nf, (ax1, ax2) = plt.subplots(ncols=2, nrows=2, figsize=(13,7))\nax1[0].plot(lin, np.abs(res[0,:]), color = 'green')\nax1[0].plot(lin, est, color='blue', linestyle='dashed')\nax1[0].set(xlabel= r\"Number of data points\", ylabel= \"Error\")\nax1[1].plot(lin, np.abs(res[1,:]), color = 'green')\nax1[1].plot(lin, est, color='blue', linestyle='dashed')\nax1[1].set(xlabel= r\"Number of data points\", ylabel= \"Error\")\nax2[0].plot(lin, np.abs(res[2,:]), color = 'green')\nax2[0].plot(lin, est, color='blue', linestyle='dashed')\nax2[0].set(xlabel= r\"Number of data points\", ylabel= \"Error\")\nax2[1].plot(lin, np.abs(res[3,:]), color = 'green')\nax2[1].plot(lin, est, color='blue', linestyle='dashed')\nax2[1].set(xlabel= r\"Number of data points\", ylabel= \"Error\");", "_____no_output_____" ], [ "lin = np.linspace(1, res.shape[1], res.shape[1])\n\nfor i in range(res.shape[0]):\n plt.plot(lin, np.abs(res[i,:]))\n plt.ylabel('Error')\n plt.xlabel('Number of data points')\n plt.show()", "_____no_output_____" ] ], [ [ "All in one plot:", "_____no_output_____" ] ], [ [ "lin = np.linspace(1, res.shape[1], res.shape[1])\n\nfor i in range(res.shape[0]):\n plt.plot(lin, np.abs(res[i,:]))\n plt.ylabel('Error')\n plt.xlabel('Number of data points')\n\nest = np.repeat(0.01, len(lin))\nplt.plot(lin, est, color='blue', linestyle='dashed')\nplt.show()", "_____no_output_____" ] ], [ [ "We see that for n sufficiently large (in this case $n \\geq 3$), we can assume the error to be bounded by 0.01. <br>", "_____no_output_____" ], [ "###### Plotting the error between the solution and the approximative solution:", "_____no_output_____" ] ], [ [ "Another approach of plotting the error is by calculating the difference between the approximative solution and the true solution. <br>\nThat is: Let $\\tilde{\\nu}$ be the parameter, resulting from our algorithm. Set $\\Omega := ([0,1] \\times {0}) \\cup ([0,1] \\times {\\tau})$\nThen we can calculate the solution of the PDE \n\n\\begin{align}\n \\frac{d}{dt}\\tilde{u}(x,t) + \\tilde{\\nu}\\tilde{u}(x,t)\\frac{d}{dx}\\tilde{u}(x,t) = 0. \n\\end{align}\n\nand set the error to $\\lVert \\tilde{u}(x,t) - u(x,t) \\rVert_{\\Omega}$. The norm can be chosen freely. <br>\nIn our case, finding the solution to a given $\\tilde{\\nu}$ is very simple. It is given by $\\tilde{u}(x,t) = u(x,t) + \\tilde{\\nu} = \\frac{x}{1+t} + \\tilde{\\nu}$. <br>\nWe thus get:\n\n\\begin{align}\n\\lVert \\tilde{u}(x,t) - u(x,t) \\rVert_{\\Omega} = \\lVert u(x,t) + \\tilde{\\nu} - u(x,t) \\rVert_{\\Omega} \\propto \\vert \\tilde{\\nu} \\vert\n\\end{align}\n\nHere, the two error terms coincide.", "_____no_output_____" ] ], [ [ "###### Plotting the execution time:", "_____no_output_____" ] ], [ [ "lin = np.linspace(1, timing.shape[1], timing.shape[1])\n\nfor i in range(timing.shape[0]):\n plt.plot(lin, timing[i,:])\n plt.ylabel('Execution time in seconds')\n plt.xlabel('Number of data points')\n plt.show()", "_____no_output_____" ], [ "lin = np.linspace(1, timing.shape[1], timing.shape[1])\n\nfor i in range(timing.shape[0]):\n plt.plot(lin, timing[i,:])\n plt.ylabel('Execution time in seconds')\n plt.xlabel('Number of data points')\n\nest = lin**(1.25)\nplt.plot(lin, est, color='blue', linestyle='dashed')\nplt.show()", "_____no_output_____" ] ], [ [ "#Curiously, the time complexity seems to be around $\\mathcal{O}(n^{5/4})$ (blue-dashed line). <br>\n#Assuming an equal amount of function evaluations in the Nelder-Mead algorithm for different values of n,\n#we would expect a time complexity of $\\mathcal{O}(n^3)$, due to the computation of the inverse of an $n\\times n$-matrix in every evaluation of $\\textit{nlml}$.", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Collaboration and Competition\n\n---\n\nIn this notebook, you will learn how to use the Unity ML-Agents environment for the third project of the [Deep Reinforcement Learning Nanodegree](https://www.udacity.com/course/deep-reinforcement-learning-nanodegree--nd893) program.\n\n### 1. Start the Environment\n\nWe begin by importing the necessary packages. If the code cell below returns an error, please revisit the project instructions to double-check that you have installed [Unity ML-Agents](https://github.com/Unity-Technologies/ml-agents/blob/master/docs/Installation.md) and [NumPy](http://www.numpy.org/).", "_____no_output_____" ] ], [ [ "from unityagents import UnityEnvironment\nimport numpy as np", "_____no_output_____" ] ], [ [ "Next, we will start the environment! **_Before running the code cell below_**, change the `file_name` parameter to match the location of the Unity environment that you downloaded.\n\n- **Mac**: `\"path/to/Tennis.app\"`\n- **Windows** (x86): `\"path/to/Tennis_Windows_x86/Tennis.exe\"`\n- **Windows** (x86_64): `\"path/to/Tennis_Windows_x86_64/Tennis.exe\"`\n- **Linux** (x86): `\"path/to/Tennis_Linux/Tennis.x86\"`\n- **Linux** (x86_64): `\"path/to/Tennis_Linux/Tennis.x86_64\"`\n- **Linux** (x86, headless): `\"path/to/Tennis_Linux_NoVis/Tennis.x86\"`\n- **Linux** (x86_64, headless): `\"path/to/Tennis_Linux_NoVis/Tennis.x86_64\"`\n\nFor instance, if you are using a Mac, then you downloaded `Tennis.app`. If this file is in the same folder as the notebook, then the line below should appear as follows:\n```\nenv = UnityEnvironment(file_name=\"Tennis.app\")\n```", "_____no_output_____" ] ], [ [ "env = UnityEnvironment(file_name=\"Tennis.app\")", "INFO:unityagents:\n'Academy' started successfully!\nUnity Academy name: Academy\n Number of Brains: 1\n Number of External Brains : 1\n Lesson number : 0\n Reset Parameters :\n\t\t\nUnity brain name: TennisBrain\n Number of Visual Observations (per agent): 0\n Vector Observation space type: continuous\n Vector Observation space size (per agent): 8\n Number of stacked Vector Observation: 3\n Vector Action space type: continuous\n Vector Action space size (per agent): 2\n Vector Action descriptions: , \n" ] ], [ [ "Environments contain **_brains_** which are responsible for deciding the actions of their associated agents. Here we check for the first brain available, and set it as the default brain we will be controlling from Python.", "_____no_output_____" ] ], [ [ "# get the default brain\nbrain_name = env.brain_names[0]\nbrain = env.brains[brain_name]", "_____no_output_____" ] ], [ [ "### 2. Examine the State and Action Spaces\n\nIn this environment, two agents control rackets to bounce a ball over a net. If an agent hits the ball over the net, it receives a reward of +0.1. If an agent lets a ball hit the ground or hits the ball out of bounds, it receives a reward of -0.01. Thus, the goal of each agent is to keep the ball in play.\n\nThe observation space consists of 8 variables corresponding to the position and velocity of the ball and racket. Two continuous actions are available, corresponding to movement toward (or away from) the net, and jumping. \n\nRun the code cell below to print some information about the environment.", "_____no_output_____" ] ], [ [ "# reset the environment\nenv_info = env.reset(train_mode=True)[brain_name]\n\n# number of agents \nnum_agents = len(env_info.agents)\nprint('Number of agents:', num_agents)\n\n# size of each action\naction_size = brain.vector_action_space_size\nprint('Size of each action:', action_size)\n\n# examine the state space \nstates = env_info.vector_observations\nstate_size = states.shape[1]\nprint('There are {} agents. Each observes a state with length: {}'.format(states.shape[0], state_size))\nprint('The state for the first agent looks like:', states[0])", "Number of agents: 2\nSize of each action: 2\nThere are 2 agents. Each observes a state with length: 24\nThe state for the first agent looks like: [ 0. 0. 0. 0. 0. 0.\n 0. 0. 0. 0. 0. 0.\n 0. 0. 0. 0. -6.65278625 -1.5\n -0. 0. 6.83172083 6. -0. 0. ]\n" ] ], [ [ "### 3. Take Random Actions in the Environment\n\nIn the next code cell, you will learn how to use the Python API to control the agents and receive feedback from the environment.\n\nOnce this cell is executed, you will watch the agents' performance, if they select actions at random with each time step. A window should pop up that allows you to observe the agents.\n\nOf course, as part of the project, you'll have to change the code so that the agents are able to use their experiences to gradually choose better actions when interacting with the environment!", "_____no_output_____" ] ], [ [ "states[1,np.newaxis]", "_____no_output_____" ], [ "for i in range(1, 6): # play game for 5 episodes\n env_info = env.reset(train_mode=False)[brain_name] # reset the environment \n states = env_info.vector_observations # get the current state (for each agent)\n scores = np.zeros(num_agents) # initialize the score (for each agent)\n while True:\n actions = np.random.randn(num_agents, action_size) # select an action (for each agent)\n actions = np.clip(actions, -1, 1) # all actions between -1 and 1\n env_info = env.step(actions)[brain_name] # send all actions to tne environment\n next_states = env_info.vector_observations # get next state (for each agent)\n rewards = env_info.rewards # get reward (for each agent)\n dones = env_info.local_done # see if episode finished\n scores += env_info.rewards # update the score (for each agent)\n states = next_states # roll over states to next time step\n if np.any(dones): # exit loop if episode finished\n break\n print('Score (max over agents) from episode {}: {}'.format(i, np.max(scores)))", "Score (max over agents) from episode 1: 0.0\nScore (max over agents) from episode 2: 0.0\nScore (max over agents) from episode 3: 0.0\nScore (max over agents) from episode 4: 0.0\nScore (max over agents) from episode 5: 0.0\n" ] ], [ [ "When finished, you can close the environment.", "_____no_output_____" ], [ "### 4. It's Your Turn!\n\nNow it's your turn to train your own agent to solve the environment! When training the environment, set `train_mode=True`, so that the line for resetting the environment looks like the following:\n```python\nenv_info = env.reset(train_mode=True)[brain_name]\n```", "_____no_output_____" ] ], [ [ "import random\nimport torch\nimport numpy as np\nfrom collections import deque\nimport matplotlib.pyplot as plt\n%matplotlib inline\n\nfrom maddpg_agent import MAagent", "_____no_output_____" ], [ "def MA_ddpg(n_episodes=10000, max_t=1000, print_every=10,random_seed=2, noise_scalar_init=2.0, noise_reduction_factor=0.99, num_agents=num_agents,update_every=1, \\\n actor_fc1_units=400, actor_fc2_units=300,\\\n critic_fcs1_units=400, critic_fc2_units=300,\\\n gamma=0.99, tau=1e-3,lr_actor=1e-4, lr_critic=1e-3, weight_decay=0,\\\n mu=0., theta=0.15, sigma=0.2):\n maddpg_agents = MAagent(state_size=state_size, action_size=action_size, random_seed=random_seed,\\\n noise_scalar_init=noise_scalar_init,noise_reduction_factor=noise_reduction_factor, num_agents=num_agents,\\\n update_every=update_every, actor_fc1_units=actor_fc1_units, actor_fc2_units=actor_fc2_units,\\\n critic_fcs1_units=critic_fcs1_units, critic_fc2_units=critic_fc2_units,gamma=gamma, tau=tau, \\\n lr_actor=lr_actor, lr_critic=lr_critic,\\\n weight_decay=weight_decay,mu=mu, theta=theta, sigma=sigma)\n scores_deque = deque(maxlen=100)\n scores = []\n time_stamp = 0\n for i_episode in range(1, n_episodes+1):\n env_info = env.reset(train_mode=True)[brain_name]\n states = env_info.vector_observations\n for i in range(num_agents):\n maddpg_agents.agents[i].reset()\n score = np.zeros(num_agents)\n for t in range(max_t):\n actions = maddpg_agents.act(states)\n env_info = env.step(actions)[brain_name] # send all actions to tne environment\n next_states = env_info.vector_observations # get next state (for each agent)\n rewards = env_info.rewards # get reward (for each agent)\n dones = env_info.local_done # see if episode finished\n \n maddpg_agents.step(states, actions, rewards, next_states, dones)\n score += rewards # update the score (for each agent)\n states = next_states # roll over states to next time step\n time_stamp+=1\n if np.any(dones):\n break \n scores_deque.append(np.max(score))\n scores.append(np.max(score))\n print('\\rEpisode {}\\tAverage Score: {:.2f}'.format(i_episode, np.mean(scores_deque)), end=\"\")\n for i in range(num_agents):\n torch.save(maddpg_agents.agents[i].actor_local.state_dict(), str(i)+'checkpoint_actor.pth')\n torch.save(maddpg_agents.agents[i].critic_local.state_dict(),str(i)+'checkpoint_critic.pth')\n if i_episode % print_every == 0:\n print('\\rEpisode {}\\tAverage Score: {:.2f}'.format(i_episode, np.mean(scores_deque)))\n if np.mean(scores_deque)>=0.5:\n print('\\nEnvironment solved in {:d} episodes!\\tAverage100 Score: {:.2f}'.format(i_episode-100, np.mean(scores_deque)))\n for i in range(num_agents):\n torch.save(maddpg_agents.agents[i].actor_local.state_dict(), str(i)+'actor_checkpoint.pth')\n torch.save(maddpg_agents.agents[i].critic_local.state_dict(),str(i)+'critic_checkpoint.pth')\n break \n return scores", "_____no_output_____" ], [ "def plot_score(scores):\n fig = plt.figure()\n ax = fig.add_subplot(111)\n plt.plot(np.arange(1, len(scores)+1), scores)\n plt.ylabel('Score')\n plt.xlabel('Episode #')\n plt.show()", "_____no_output_____" ], [ "%%time\nprint(\"\\n\\nTest try #1: base line\")\nscores = MA_ddpg(n_episodes=10000, max_t=1000, print_every=100,random_seed=2, num_agents=num_agents,update_every=2,\\\n actor_fc1_units=200, actor_fc2_units=150,\\\n critic_fcs1_units=400, critic_fc2_units=300,\\\n gamma=0.99, tau=1e-2,lr_actor=1e-4, lr_critic=1e-3, weight_decay=0,\\\n mu=0., theta=0.15, sigma=0.2)\nplot_score(scores)", "\n\nTest try #1: base line\nEpisode 100\tAverage Score: 0.00\nEpisode 200\tAverage Score: 0.00\nEpisode 300\tAverage Score: 0.00\nEpisode 400\tAverage Score: 0.01\nEpisode 500\tAverage Score: 0.00\nEpisode 600\tAverage Score: 0.02\nEpisode 700\tAverage Score: 0.00\nEpisode 800\tAverage Score: 0.02\nEpisode 900\tAverage Score: 0.04\nEpisode 1000\tAverage Score: 0.05\nEpisode 1100\tAverage Score: 0.05\nEpisode 1200\tAverage Score: 0.06\nEpisode 1300\tAverage Score: 0.05\nEpisode 1400\tAverage Score: 0.06\nEpisode 1500\tAverage Score: 0.05\nEpisode 1600\tAverage Score: 0.04\nEpisode 1700\tAverage Score: 0.05\nEpisode 1800\tAverage Score: 0.05\nEpisode 1900\tAverage Score: 0.05\nEpisode 2000\tAverage Score: 0.06\nEpisode 2100\tAverage Score: 0.05\nEpisode 2200\tAverage Score: 0.06\nEpisode 2300\tAverage Score: 0.07\nEpisode 2400\tAverage Score: 0.07\nEpisode 2500\tAverage Score: 0.07\nEpisode 2600\tAverage Score: 0.09\nEpisode 2700\tAverage Score: 0.10\nEpisode 2800\tAverage Score: 0.12\nEpisode 2900\tAverage Score: 0.13\nEpisode 3000\tAverage Score: 0.26\nEpisode 3100\tAverage Score: 0.41\nEpisode 3126\tAverage Score: 0.51\nEnvironment solved in 3026 episodes!\tAverage100 Score: 0.51\n" ], [ "# load the weights from file\ndef let_agent_play():\n for i in range(num_agents):\n maddpg_agents.agents[i].actor_local.load_state_dict(torch.load(str(i)+'checkpoint_actor.pth'))\n maddpg_agents.agents[i].critic_local.load_state_dict(torch.load(str(i)+'checkpoint_critic.pth'))\n max_t=1000\n play_episodes=100\n play_scores = [] # list containing scores from each episode\n play_scores_window = deque(maxlen=100) # last 100 scores\n for i_episode in range(1, play_episodes+1):\n env_info = env.reset(train_mode=True)[brain_name]\n states = env_info.vector_observations \n for agent in maddpg_agents.agents:\n agent.reset()\n score = np.zeros(num_agents)\n for t in range(max_t):\n actions = maddpg_agents.act(states)\n env_info = env.step(actions)[brain_name] # send all actions to tne environment\n next_states = env_info.vector_observations # get next state (for each agent)\n rewards = env_info.rewards # get reward (for each agent)\n dones = env_info.local_done # see if episode finished\n score += rewards # update the score (for each agent)\n states = next_states # roll over states to next time step\n if np.any(dones):\n break \n play_scores_window.append(np.max(score)) # save most recent score\n play_scores.append(np.max(score)) # save most recent score\n print('\\rEpisode {}\\tAverage Score: {:.2f}'.format(i_episode, np.mean(play_scores_window)), end=\"\")\n if i_episode % 100 == 0:\n print('\\rEpisode {}\\tAverage Score: {:.2f}'.format(i_episode, np.mean(play_scores_window)))\n return play_scores", "_____no_output_____" ], [ "maddpg_agents = MAagent(state_size=state_size, action_size=action_size, \\\n noise_scalar_init=1.0, noise_reduction_factor=0.999,\\\n random_seed=2, num_agents=num_agents,update_every=2,\\\n actor_fc1_units=200, actor_fc2_units=150,\\\n critic_fcs1_units=400, critic_fc2_units=300,\\\n gamma=0.99, tau=1e-2,lr_actor=1e-4, lr_critic=1e-3, weight_decay=0,\\\n mu=0., theta=0.15, sigma=0.2)\nplay_scores=let_agent_play()\nplot_score(play_scores)", "Episode 100\tAverage Score: 0.58\n" ], [ "env.close()", "_____no_output_____" ], [ "# code for debugging. \n\n# maddpg_agents = MAagent(state_size=state_size, action_size=action_size, random_seed=2,num_agents=num_agents)\n\n# actions=[maddpg_agents.maddpg_agents[i].act(states[i, np.newaxis]).flatten() for i in range(maddpg_agents.num_agents)]\n# np.array(actions)\n\n# for i in range(1, 129): # play game for 5 episodes\n# env_info = env.reset(train_mode=True)[brain_name]\n# states = env_info.vector_observations \n# for i in range(num_agents):\n# maddpg_agents.agents[i].reset()\n# score = np.zeros(num_agents)\n# actions = maddpg_agents.act(states)\n# env_info = env.step(actions)[brain_name] # send all actions to tne environment\n# next_states = env_info.vector_observations # get next state (for each agent)\n# rewards = env_info.rewards # get reward (for each agent)\n# rewards\n# dones = env_info.local_done # see if episode finished\n# maddpg_agents.step(states, actions, rewards, next_states, dones)\n\n# experiences = random.sample(maddpg_agents.memory.memory, k=2)\n# print(experiences)\n\n# obs = [torch.from_numpy(np.vstack([e.state[i] for e in experiences if e is not None])).float() for i in range(num_agents)]\n# states = torch.from_numpy(np.vstack([e.state.flatten() for e in experiences if e is not None])).float() \n# actions = torch.from_numpy(np.vstack([e.action.flatten() for e in experiences if e is not None])).float()\n# rewards = [torch.from_numpy(np.vstack([e.reward[i] for e in experiences if e is not None])).float() \\\n# for i in range(num_agents)]\n# next_obs = [torch.from_numpy(np.vstack([e.next_state[i] for e in experiences if e is not None])).float() \\\n# for i in range(num_agents)]\n# next_states = torch.from_numpy(np.vstack([e.next_state.flatten() for e in experiences if e is not None])).float()\n# dones = [torch.from_numpy(np.vstack([e.done[i] for e in experiences if e is not None]).astype(np.uint8)).float() for i in range(num_agents)]\n# experiences[0].reward\n\n# (obs, states, actions, rewards, next_obs, next_states, dones)=maddpg_agents.memory.sample()", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "'''\nMONTY HALL PROBLEM\nCSE 2ND YEAR\nGROUP 1\nM401 MICRO PROJECT\n'''\nimport random\ndoors=[\"\",\"\",\"\"] #LIST OF DOORS\ngoatdoor=[] #LIST OF DOORS NOT CONTAINING PRIZE\nswap=0 #NO. OF SWAPPED ATTEMPTS\nXswap=0 #NO. OF RETAINED ATTEMPTS\nswapWins=0 #WINS AFTER SWAPPING\nXswapWins=0 #WINS AFTER RETAINING (NOT SWAPPING)\nexit=False\nwhile(not exit):\n x=random.randint(0,2) #DOOR CONTAINING PRIZE\n doors[x]=\"PRIZE\" #INDEX x OF doors HAS PRIZE\n for i in range(0,3): #FORMING THE LIST goatdoor\n if i!=x:\n doors[i]=\"GOAT\" #VALUES IN doors NOT EQUAL TO \"PRIZE\"\n goatdoor.append(i) #ARE MADE \"GOAT\"\n ch1=int(input(\"Enter choice of door (1/2/3): \")) #USER INPUT OF DOOR CHOICE\n ch1-=1 #SINCE WE ARE WORKING WITH START INDEX 0 (SO 1/2/3 => 0/1/2)\n opendoor=random.choice(goatdoor) #GIVES US RANDOM INDEX FROM goatdoor\n while opendoor==ch1:\n opendoor=random.choice(goatdoor) #opendoor NOT EQUAL TO CHOICE ch1 OF USER\n print(\"Door%2d has GOAT\"%(opendoor+1)) #USER IS SHOWN WHAT'S BEHIND ONE OF THE REMAINING DOORS\n ch2=input(\"Do you want to swap? (Y/N): \") #PROMPT TO SWAP\n W=\"PLAYER WINS\" #STRINGS TO STORE\n L=\"PLAYER LOSES\" #WIN/LOSS MESSAGES\n if str.upper(ch2)==\"Y\": #SWAPS\n swap+=1\n if doors[ch1]==\"GOAT\": #INITIAL CHOICE GOAT: USER WINS\n print(W)\n swapWins+=1 ##WIN BY SWAPPING\n else: #OR ELSE USER LOSES\n print(L)\n else: #RETAINS\n Xswap+=1\n if doors[ch1]==\"GOAT\": #INITIAL CHOICE GOAT: USER LOSES\n print(L)\n else: #OR ELSE USER WINS\n print(W)\n XswapWins+=1 ##WIN BY RETAINING\n if str.upper(input(\"Exit? (Y/N): \"))==\"Y\":\n exit=True #EXIT CONDITION\nprint(\"\\n\")\nprint(\"Swap attempts:\",swap)\nprint(\"Wins after swapping: \",swapWins)\nprint(\"\\n\")\nprint(\"Retain attempts:\",Xswap)\nprint(\"Wins after retaining: \",XswapWins)", "Enter choice of door (1/2/3): 2\nDoor 3 has GOAT\nDo you want to swap? (Y/N): n\nPLAYER LOSES\nExit? (Y/N): N\nEnter choice of door (1/2/3): 1\nDoor 3 has GOAT\nDo you want to swap? (Y/N): Y\nPLAYER WINS\nExit? (Y/N): n\nEnter choice of door (1/2/3): 2\nDoor 1 has GOAT\nDo you want to swap? (Y/N): n\nPLAYER LOSES\nExit? (Y/N): N\nEnter choice of door (1/2/3): 3\nDoor 2 has GOAT\nDo you want to swap? (Y/N): N\nPLAYER LOSES\nExit? (Y/N): N\nEnter choice of door (1/2/3): 1\nDoor 2 has GOAT\nDo you want to swap? (Y/N): Y\nPLAYER LOSES\nExit? (Y/N): N\nEnter choice of door (1/2/3): 2\nDoor 1 has GOAT\nDo you want to swap? (Y/N): N\nPLAYER LOSES\nExit? (Y/N): N\nEnter choice of door (1/2/3): 3\nDoor 2 has GOAT\nDo you want to swap? (Y/N): Y\nPLAYER WINS\nExit? (Y/N): N\nEnter choice of door (1/2/3): 1\nDoor 3 has GOAT\nDo you want to swap? (Y/N): N\nPLAYER WINS\nExit? (Y/N): N\nEnter choice of door (1/2/3): 1\nDoor 2 has GOAT\nDo you want to swap? (Y/N): Y\nPLAYER LOSES\nExit? (Y/N): N\nEnter choice of door (1/2/3): 2\nDoor 1 has GOAT\nDo you want to swap? (Y/N): Y\nPLAYER WINS\nExit? (Y/N): Y\n\n\nSwap attempts: 5\nWins after swapping: 3\n\n\nRetain attempts: 5\nWins after retaining: 1\n" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import requests\nimport pandas as pd\nimport folium", "_____no_output_____" ], [ "BASE_REQUEST = 'http://router.project-osrm.org/route/v1/foot/'", "_____no_output_____" ], [ "def getRoute(*args,points_to_include=[],points_to_avoid=[]):\n k = ';'.join(args)\n req = BASE_REQUEST + k + '?overview=full&steps=true&alternatives=3'\n _r = requests.get(req)\n _ = []\n for _route in range(len(_r.json()['routes'])):\n _tmp = pd.DataFrame()\n for _leg in range(len(_r.json()['routes'][_route]['legs'])):\n _tmp = _tmp.append(pd.json_normalize(_r.json()['routes'][_route]['legs'][_leg]['steps']))\n _.append(_tmp.reset_index(drop=True))\n \n return _\n \n\ndef getLocations(_df):\n loc = []\n for row,col in _df.iterrows():\n _ = pd.json_normalize(col['intersections'])\n for _row,_col in _.iterrows():\n loc.append((_col['location'][1],_col['location'][0]))\n \n return loc", "_____no_output_____" ], [ "\ncoord1 = '-0.076303,51.50815' # Tower of London\ncoord2 = '-0.108711,51.50457' # London Waterloo East Tube Station\ncoord3 = '-0.124613,51.50106' # Big Ben\ncoord4 = '-0.60604413,51.48402' # Windsor castle\n\nres = getRoute(coord1,coord4)\n\nprint(f'Amount of routes found: {len(res)}')\nfor i in range(len(res)):\n print(f'Length of route 1: {sum(res[i][\"distance\"])} meters')", "Amount of routes found: 2\nLength of route 1: 39587.2 meters\nLength of route 1: 41887.6 meters\n" ], [ "map = folium.Map(location=[51.5, -0.1], zoom_start=13)\ncolors=['#000000','#FF0000','#00FF00','#0000FF']\nfor routes in range(len(res)):\n l = getLocations(res[routes])\n for point in l:\n folium.CircleMarker(point, radius=5, color=colors[routes]).add_to(map)\nmap", "_____no_output_____" ], [ "df2 = pd.json_normalize(df.iloc[1]['intersections'])\ndf2", "_____no_output_____" ], [ "sum(df['distance'])", "_____no_output_____" ] ] ]

[ "code" ]

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import os, shutil\nimport numpy as np\nimport tensorflow as tf\nbase_dir = 'tiles'\n\ntrain_dir = os.path.join(base_dir, 'train')\nvalidation_dir = os.path.join(base_dir, 'validation')", "/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:516: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_qint8 = np.dtype([(\"qint8\", np.int8, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:517: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_quint8 = np.dtype([(\"quint8\", np.uint8, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:518: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_qint16 = np.dtype([(\"qint16\", np.int16, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:519: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_quint16 = np.dtype([(\"quint16\", np.uint16, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:520: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_qint32 = np.dtype([(\"qint32\", np.int32, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:525: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n np_resource = np.dtype([(\"resource\", np.ubyte, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:541: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_qint8 = np.dtype([(\"qint8\", np.int8, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:542: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_quint8 = np.dtype([(\"quint8\", np.uint8, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:543: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_qint16 = np.dtype([(\"qint16\", np.int16, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:544: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_quint16 = np.dtype([(\"quint16\", np.uint16, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:545: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n _np_qint32 = np.dtype([(\"qint32\", np.int32, 1)])\n/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:550: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n np_resource = np.dtype([(\"resource\", np.ubyte, 1)])\n" ], [ "config = tf.ConfigProto()\nconfig.gpu_options.allow_growth = True\nsess = tf.Session(config=config)", "_____no_output_____" ], [ "import keras\nfrom keras import layers\nfrom keras import models\nfrom keras.applications import VGG16\nfrom keras.preprocessing import image\nfrom keras.preprocessing.image import ImageDataGenerator\n\nimg_width, img_height = 256, 256\nconv_base= VGG16(weights=None, include_top=False, input_shape=(img_width,img_height,3))\n\nmodel = models.Sequential()\nmodel.add(conv_base)\nmodel.add(layers.MaxPooling2D(pool_size=(2,2),strides=(2,2)))\nmodel.add(layers.Flatten())\nmodel.add(layers.Dense(units=4096,activation=\"relu\")) #si no entrena bien: init = 'he_normal' #he_uniform\nmodel.add(layers.Dense(units=4096,activation=\"relu\"))\nmodel.add(layers.Dense(units=1, activation=\"sigmoid\"))\n\nmodel.summary()", "WARNING:tensorflow:From /home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n\nWARNING:tensorflow:From /home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n\nWARNING:tensorflow:From /home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n\nWARNING:tensorflow:From /home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:3976: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.\n\n" ], [ "from keras import optimizers\n\nmodel.compile(loss='binary_crossentropy', optimizer=optimizers.Adam(lr=1e-4), metrics=['acc']) #si es demasiado lento, Adam(lr=0.001)", "WARNING:tensorflow:From /home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/keras/optimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n\nWARNING:tensorflow:From /home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:3376: The name tf.log is deprecated. Please use tf.math.log instead.\n\nWARNING:tensorflow:From /home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/tensorflow/python/ops/nn_impl.py:180: add_dispatch_support.<locals>.wrapper (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\nInstructions for updating:\nUse tf.where in 2.0, which has the same broadcast rule as np.where\n" ], [ "from keras.preprocessing.image import ImageDataGenerator\ndatagen = ImageDataGenerator(rescale=1. /255)\ntrain_datagen = ImageDataGenerator(rescale=1. / 255, rotation_range=25, \n width_shift_range=0.1, height_shift_range=0.1, \n zoom_range=0.1, horizontal_flip=True, vertical_flip=True, \n fill_mode='nearest')\n\ntrain_generator = train_datagen.flow_from_directory(train_dir, target_size=(256, 256),\n batch_size=50, class_mode='binary')\nvalidation_datagen = ImageDataGenerator(rescale=1. / 255)\nvalidation_generator = validation_datagen.flow_from_directory(validation_dir, target_size=(256, 256), \n batch_size=50,class_mode='binary')", "Found 484048 images belonging to 2 classes.\nFound 26893 images belonging to 2 classes.\n" ], [ "train_steps_per_epoch = np.math.ceil(train_generator.samples / train_generator.batch_size)\nvalidation_steps_per_epoch = np.math.ceil(validation_generator.samples / validation_generator.batch_size)\n\n#teniendo en cuenta el desbalanceo de clases\nfrom sklearn.utils import class_weight\nclass_weights = class_weight.compute_class_weight('balanced',np.unique(train_generator.classes), train_generator.classes)\nclass_weights", "/home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/sklearn/utils/validation.py:70: FutureWarning: Pass classes=[0 1], y=[0 0 0 ... 1 1 1] as keyword args. From version 0.25 passing these as positional arguments will result in an error\n FutureWarning)\n" ], [ "history = model.fit_generator(train_generator, steps_per_epoch= train_steps_per_epoch, \n epochs=50, validation_data=validation_generator,\n validation_steps=validation_steps_per_epoch, class_weight=class_weights) #callbacks = [tensorboard]\n", "WARNING:tensorflow:From /home/miguelmmanso/anaconda3/envs/pruebas/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:986: The name tf.assign_add is deprecated. Please use tf.compat.v1.assign_add instead.\n\nEpoch 1/50\n9681/9681 [==============================] - 4988s 515ms/step - loss: 0.5138 - acc: 0.7290 - val_loss: 0.4430 - val_acc: 0.7851\nEpoch 2/50\n9681/9681 [==============================] - 4960s 512ms/step - loss: 0.4188 - acc: 0.8000 - val_loss: 0.3853 - val_acc: 0.8152\nEpoch 3/50\n9681/9681 [==============================] - 4958s 512ms/step - loss: 0.3933 - acc: 0.8117 - val_loss: 0.3899 - val_acc: 0.8208\nEpoch 4/50\n9681/9681 [==============================] - 4959s 512ms/step - loss: 0.3778 - acc: 0.8184 - val_loss: 0.3661 - val_acc: 0.8282\nEpoch 5/50\n9681/9681 [==============================] - 4959s 512ms/step - loss: 0.3674 - acc: 0.8223 - val_loss: 0.3650 - val_acc: 0.8269\nEpoch 6/50\n9681/9681 [==============================] - 4965s 513ms/step - loss: 0.3584 - acc: 0.8251 - val_loss: 0.3695 - val_acc: 0.8236\nEpoch 7/50\n9681/9681 [==============================] - 4968s 513ms/step - loss: 0.3522 - acc: 0.8267 - val_loss: 0.3729 - val_acc: 0.8257\nEpoch 8/50\n9681/9681 [==============================] - 4970s 513ms/step - loss: 0.3465 - acc: 0.8291 - val_loss: 0.3492 - val_acc: 0.8298\nEpoch 9/50\n9681/9681 [==============================] - 4968s 513ms/step - loss: 0.3427 - acc: 0.8308 - val_loss: 0.3519 - val_acc: 0.8301\nEpoch 10/50\n9681/9681 [==============================] - 4970s 513ms/step - loss: 0.3395 - acc: 0.8319 - val_loss: 0.3887 - val_acc: 0.8235\nEpoch 11/50\n9681/9681 [==============================] - 4969s 513ms/step - loss: 0.3364 - acc: 0.8332 - val_loss: 0.3598 - val_acc: 0.8321\nEpoch 12/50\n9681/9681 [==============================] - 4967s 513ms/step - loss: 0.3340 - acc: 0.8343 - val_loss: 0.3414 - val_acc: 0.8326\nEpoch 13/50\n9681/9681 [==============================] - 4960s 512ms/step - loss: 0.3314 - acc: 0.8353 - val_loss: 0.3396 - val_acc: 0.8351\nEpoch 14/50\n9681/9681 [==============================] - 5019s 518ms/step - loss: 0.3299 - acc: 0.8357 - val_loss: 0.3442 - val_acc: 0.8341\nEpoch 15/50\n9681/9681 [==============================] - 4969s 513ms/step - loss: 0.3272 - acc: 0.8374 - val_loss: 0.3436 - val_acc: 0.8340\nEpoch 16/50\n 624/9681 [>.............................] - ETA: 1:15:31 - loss: 0.3224 - acc: 0.8376" ], [ "import matplotlib.pyplot as plt\nacc = history.history['acc']\nval_acc = history.history['val_acc']\nplt.figure(figsize=(20,10))\nloss = history.history['loss']\nval_loss = history.history['val_loss']\nepochs = range(1, len(acc) +1) \nplt.rcParams.update({'font.size':18})\nplt.plot(epochs, acc, 'bo--', color='r', label='Training acc')\nplt.plot(epochs, val_acc, 'bo--', color='b', label='Validation acc')\nplt.legend()\nplt.plot(epochs, loss, color= 'r', label='Training loss')\nplt.plot(epochs, val_loss, color=\"b\", label='Validation loss')\nplt.legend()\nplt.show()", "_____no_output_____" ], [ "model.save(\"models1/vgg16_noweights-3.h5\")", "_____no_output_____" ], [ "#EARLY STOPPING\n#from keras.callbacks import ModelCheckpoint, EarlyStopping\n#checkpoint = ModelCheckpoint(\"vgg16_1.h5\", monitor='val_acc', verbose=1,\n# save_best_only=True, save_weights_only=False, mode='auto', \n# period=5)\n#early = EarlyStopping(monitor='val_acc', min_delta=0, patience=50, verbose=1, mode='auto')\n\n#hist = model.fit_generator(steps_per_epoch=100,generator=traindata, \n# validation_data= testdata, validation_steps=10,epochs=100,\n# callbacks=[checkpoint,early])", "_____no_output_____" ], [ "#from keras.preprocessing import imageimg = image.load_img(\"image.jpeg\",target_size=(224,224))\n#img = np.asarray(img)\n#plt.imshow(img)\n#img = np.expand_dims(img, axis=0)from keras.models import load_model\n#saved_model = load_model(\"vgg16_1.h5\")output = saved_model.predict(img)\n#if output[0][0] > output[0][1]:\n# print(\"Road\")\n#else:\n# print('No Road')", "_____no_output_____" ] ] ]

[ "code" ]

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## Stackexchange Dataset", "_____no_output_____" ] ], [ [ "! wget https://drive.switch.ch/index.php/s/I6hiqbHZRCFwZGj/download -O /data/dataset/stackexchange.zip", "_____no_output_____" ], [ "!ls -lisah /data/dataset/stackexchange.zip", "_____no_output_____" ], [ "!unzip /data/dataset/stackexchange.zip -d /data/dataset/", "_____no_output_____" ], [ "!ls -lisah /data/dataset/stackexchange.com/unix.stackexchange.com/json/", "_____no_output_____" ] ], [ [ "## Init Spark", "_____no_output_____" ] ], [ [ "import findspark\nfindspark.init()", "_____no_output_____" ], [ "from pyspark.sql import SparkSession\n\nspark = SparkSession \\\n .builder \\\n .appName(\"unix.stackexchange.com\") \\\n .getOrCreate()", "_____no_output_____" ] ], [ [ "## Badges", "_____no_output_____" ] ], [ [ "!head /data/dataset/stackexchange.com/unix.stackexchange.com/json/Badges.json", "_____no_output_____" ], [ "path = \"file:///data/dataset/stackexchange.com/unix.stackexchange.com/json/Badges.json\"", "_____no_output_____" ], [ "badges = spark.read.json(path)", "_____no_output_____" ], [ "badges.show(3, truncate=False)", "_____no_output_____" ], [ "badges.printSchema()", "_____no_output_____" ] ], [ [ "# Inspect all Files", "_____no_output_____" ] ], [ [ "!hdfs dfs -rm -r \"/dataset/unix.stackexchange.com\"", "_____no_output_____" ], [ "def get_info(name):\n print(f\"info for {name}\")\n print(\"------------------------------------\")\n path = f\"file:///data/dataset/stackexchange.com/unix.stackexchange.com/json/{name}.json\"\n df = spark.read.json(path)\n df.show(3, truncate=False)\n df.printSchema()\n return df", "_____no_output_____" ], [ "all_names = !ls /data/dataset/stackexchange.com/unix.stackexchange.com/json/", "_____no_output_____" ], [ "all_names = [name[:-5] for name in all_names]", "_____no_output_____" ], [ "all_names", "_____no_output_____" ] ], [ [ "# Save as Parquet", "_____no_output_____" ] ], [ [ "def save_as_parquet(name, df):\n print(f\"saving {name}\")\n print(\"------------------------------------\")\n \n df.show(3, truncate=False)\n df.printSchema()\n \n lower_name = name.lower()\n df.repartition(15).write.parquet(f\"/dataset/unix.stackexchange.com/{lower_name}.parquet\")", "_____no_output_____" ] ], [ [ "## Badges", "_____no_output_____" ] ], [ [ "df = get_info('Badges')", "_____no_output_____" ], [ "# https://spark.apache.org/docs/latest/api/python/_modules/pyspark/sql/functions.html\n# https://sparkbyexamples.com/spark/spark-sql-functions/\nfrom pyspark.sql import functions as f\n\ndf.select(f.min(\"Class\"), f.max(\"Class\")).collect()", "_____no_output_____" ], [ "# https://sparkbyexamples.com/pyspark/pyspark-cast-column-type/#:~:text=In%20PySpark%2C%20you%20can%20cast,Boolean%20e.t.c%20using%20PySpark%20examples.\n\n\n# Another way would be via Types\n# https://sparkbyexamples.com/pyspark/pyspark-sql-types-datatype-with-examples/\n# https://spark.apache.org/docs/latest/sql-ref-datatypes.html\n# from pyspark.sql.types import *\n\n\nsave_as_parquet(\"Badges\", df.selectExpr(\\\n \"cast(Id as int) id\", \\\n \"cast(UserId as int) user_id\", \\\n \"cast(Class as byte) class\", \\\n \"cast(Name as string) name\", \\\n \"cast(TagBased as boolean) tag_based\", \\\n \"cast(Date as timestamp) date\" \\\n ))\n\n", "_____no_output_____" ] ], [ [ "## Comments", "_____no_output_____" ] ], [ [ "df = get_info('Comments')", "_____no_output_____" ], [ "df.filter(\"UserDisplayName is not null\").show(2)", "_____no_output_____" ], [ "save_as_parquet(\"Comments\", df.selectExpr(\\\n \"cast(Id as int) id\", \\\n \"cast(PostId as int) post_id\", \\\n \"cast(UserId as int) user_id\", \\\n \"cast(Score as byte) score\", \\\n \"cast(ContentLicense as string) content_license\", \\\n \"cast(UserDisplayName as string) user_display_name\", \\\n \"cast(Text as String) text\", \\\n \"cast(CreationDate as timestamp) creation_date\" \\\n ))", "_____no_output_____" ] ], [ [ "## PostHistory", "_____no_output_____" ] ], [ [ "df = get_info('PostHistory')", "_____no_output_____" ], [ "save_as_parquet(\"PostHistory\", df.selectExpr(\\\n \"cast(Id as int) id\", \\\n \"cast(PostId as int) post_id\", \\\n \"cast(UserId as int) user_id\", \\\n \"cast(PostHistoryTypeId as byte) post_history_type_id\", \\\n \"cast(UserDisplayName as string) user_display_name\", \\\n \"cast(ContentLicense as string) content_license\", \\\n \"cast(RevisionGUID as string) revision_guid\", \\\n \"cast(Text as String) text\", \\\n \"cast(Comment as String) comment\", \\\n \"cast(CreationDate as timestamp) creation_date\" \\\n ))", "_____no_output_____" ] ], [ [ "## PostLinks", "_____no_output_____" ] ], [ [ "df = get_info('PostLinks')", "_____no_output_____" ], [ "save_as_parquet(\"PostLinks\", df.selectExpr(\\\n \"cast(Id as int) id\", \\\n \"cast(RelatedPostId as int) related_post_id\", \\\n \"cast(PostId as int) post_id\", \\\n \"cast(LinkTypeId as byte) link_type_id\", \\\n \"cast(CreationDate as timestamp) creation_date\" \\\n ))", "_____no_output_____" ] ], [ [ "## Posts", "_____no_output_____" ] ], [ [ "df = get_info('Posts')", "_____no_output_____" ], [ "save_as_parquet(\"Posts\", df.selectExpr(\\\n \"cast(Id as int) id\", \\\n \"cast(OwnerUserId as int) owner_user_id\", \\\n \"cast(LastEditorUserId as int) last_editor_user_id\", \\\n \"cast(PostTypeId as short) post_type_id\", \\\n \"cast(AcceptedAnswerId as int) accepted_answer_id\", \\\n \"cast(Score as int) score\", \\\n \"cast(ParentId as int) parent_id\", \\\n \"cast(ViewCount as int) view_count\", \\\n \"cast(AnswerCount as int) answer_count\", \\\n \"cast(CommentCount as int) comment_count\", \\\n \"cast(OwnerDisplayName as string) owner_display_name\", \\\n \"cast(LastEditorDisplayName as string) last_editor_display_name\", \\\n \"cast(Title as String) title\", \\\n \"cast(Tags as String) tags\", \\\n \"cast(ContentLicense as string) content_license\", \\\n \"cast(Body as string) body\", \\\n \"cast(FavoriteCount as int) favorite_count\", \\\n \"cast(CreationDate as timestamp) creation_date\", \\\n \"cast(CommunityOwnedDate as timestamp) community_owned_date\", \\\n \"cast(ClosedDate as timestamp) closed_date\", \\\n \"cast(LastEditDate as timestamp) last_edit_date\", \\\n \"cast(LastActivityDate as timestamp) last_activity_date\" \\\n ))", "_____no_output_____" ] ], [ [ "## Tags", "_____no_output_____" ] ], [ [ "df = get_info('Tags')", "_____no_output_____" ], [ "save_as_parquet(\"Tags\", df.selectExpr(\\\n \"cast(Id as int) id\", \\\n \"cast(ExcerptPostId as int) excerpt_post_id\", \\\n \"cast(WikiPostId as int) wiki_post_id\", \\\n \"cast(TagName as string) tag_name\", \\\n \"cast(Count as int) count\" \\\n ))", "_____no_output_____" ] ], [ [ "## Users", "_____no_output_____" ] ], [ [ "df = get_info('Users')", "_____no_output_____" ], [ "save_as_parquet(\"Users\", df.selectExpr(\\\n \"cast(Id as int) id\", \\\n \"cast(AccountId as int) account_id\", \\\n \"cast(Reputation as int) reputation\", \\\n \"cast(Views as int) views\", \\\n \"cast(DownVotes as int) down_votes\", \\\n \"cast(UpVotes as int) up_votes\", \\\n \"cast(DisplayName as string) display_name\", \\\n \"cast(Location as string) location\", \\\n \"cast(ProfileImageUrl as string) profile_image_url\", \\\n \"cast(WebsiteUrl as string) website_url\", \\\n \"cast(AboutMe as string) about_me\", \\\n \"cast(CreationDate as timestamp) creation_date\", \\\n \"cast(LastAccessDate as timestamp) last_access_date\" \\\n ))", "_____no_output_____" ] ], [ [ "## Votes", "_____no_output_____" ] ], [ [ "df = get_info('Votes')", "_____no_output_____" ], [ "save_as_parquet(\"Votes\", df.selectExpr(\\\n \"cast(Id as int) id\", \\\n \"cast(UserId as int) user_id\", \\\n \"cast(PostId as int) post_id\", \\\n \"cast(VoteTypeId as byte) vote_type_id\", \\\n \"cast(BountyAmount as byte) bounty_amount\", \\\n \"cast(CreationDate as timestamp) creation_date\" \\\n ))", "_____no_output_____" ] ], [ [ "# Analysing", "_____no_output_____" ] ], [ [ "def get_path(name):\n return f\"/dataset/unix.stackexchange.com/{name}.parquet\"", "_____no_output_____" ] ], [ [ "## Read all Parquets", "_____no_output_____" ] ], [ [ "badges = spark.read.parquet(get_path(\"badges\"))\ncomments = spark.read.parquet(get_path(\"comments\"))\nposthistory = spark.read.parquet(get_path(\"posthistory\"))\npostlinks = spark.read.parquet(get_path(\"postlinks\"))\nposts = spark.read.parquet(get_path(\"posts\"))\ntags = spark.read.parquet(get_path(\"tags\"))\nusers = spark.read.parquet(get_path(\"users\"))\nvotes = spark.read.parquet(get_path(\"votes\"))", "_____no_output_____" ], [ "print(\"badges\")\nbadges.show(3)\nprint(\"comments\")\ncomments.show(3)\nprint(\"posthistory\")\nposthistory.show(3)\nprint(\"postlinks\")\npostlinks.show(3)\nprint(\"posts\")\nposts.show(3)\nprint(\"tags\")\ntags.show(3)\nprint(\"users\")\nusers.show(3)\nprint(\"votes\")\nvotes.show(3)", "_____no_output_____" ] ], [ [ "## Tags", "_____no_output_____" ] ], [ [ "# tags = spark.read.parquet(get_path(\"tags\"))", "_____no_output_____" ], [ "tags.show()", "_____no_output_____" ], [ "from pyspark.sql import functions as f\n\ntags.filter(f.col(\"tag_name\") == \"async\").show()", "_____no_output_____" ], [ "tags.filter(\"tag_name = 'async'\").show()", "_____no_output_____" ], [ "tags.filter(\"tag_name like '%async%'\").show()", "_____no_output_____" ], [ "tags.filter(f.col(\"tag_name\").like('%async%')).show()", "_____no_output_____" ], [ "tags.select(\"tag_name\", \"count\").orderBy(f.col(\"count\").desc()).show(20)", "_____no_output_____" ] ], [ [ "### Wordcloud", "_____no_output_____" ], [ "Needs the `wordcloud` (and `matplotlib` which comes as a dependency) python package\n\n```\npip install wordcloud\n```\n\nsee [documentation](https://github.com/amueller/word_cloud)", "_____no_output_____" ] ], [ [ "filtered_tags = tags.select(\"tag_name\", \"count\").orderBy(f.col(\"count\").desc()).filter(\"count > 100\")", "_____no_output_____" ], [ "filtered_tags.show(2)\nfiltered_tags.count()", "_____no_output_____" ], [ "frequencies = filtered_tags.toPandas().set_index('tag_name').T.to_dict('records')[0]", "_____no_output_____" ], [ "frequencies['linux']", "_____no_output_____" ], [ "from wordcloud import WordCloud\nimport matplotlib.pyplot as plt\n\n\nwordcloud = WordCloud(width=2000, height=1000)\nwordcloud.generate_from_frequencies(frequencies)\n\n\nplt.figure(figsize=(20,30))\nplt.imshow(wordcloud, interpolation='bilinear')\nplt.axis('off')\n\n\nplt.savefig(\"./wordcloud.png\")\n", "_____no_output_____" ] ], [ [ "## Users ", "_____no_output_____" ] ], [ [ "users.printSchema()", "_____no_output_____" ], [ "print(users.count())\nprint(users.filter(\"id is not null\").count())\nprint(users.filter(\"id is not null\").distinct().count())", "_____no_output_____" ], [ "users. \\\n select(\"account_id\", \"display_name\", \"views\", \"down_votes\", \"up_votes\", \"reputation\"). \\\n show(2)\n", "_____no_output_____" ], [ "# most reputation\n# https://stackexchange.com/users/{account_id}/\nusers. \\\n select(\"account_id\", \"display_name\", \"views\", \"down_votes\", \"up_votes\", \"reputation\"). \\\n orderBy(f.col(\"reputation\").desc()). \\\n show(10, False)\n", "_____no_output_____" ], [ "# most viewed\nusers. \\\n select(\"account_id\", \"display_name\", \"views\", \"down_votes\", \"up_votes\", \"reputation\"). \\\n orderBy(f.col(\"views\").desc()). \\\n show(10, False)", "_____no_output_____" ], [ "# downvoters\nusers. \\\n select(\"account_id\", \"display_name\", \"views\", \"down_votes\", \"up_votes\", \"reputation\"). \\\n orderBy(f.col(\"down_votes\").desc()). \\\n show(10, False)", "_____no_output_____" ] ], [ [ "## Analysing a Question\n\n- [83577](https://unix.stackexchange.com/questions/83577/how-to-invoke-vim-with-line-numbers-shown)", "_____no_output_____" ] ], [ [ "posts.filter(\"id = 83577\").toPandas().T", "_____no_output_____" ], [ "posts.filter(\"id = 648583\").toPandas().T", "_____no_output_____" ], [ "posts.filter(\"id = 648608\").toPandas().T", "_____no_output_____" ], [ "posts.select(\"parent_id\").groupBy(\"parent_id\").count().sort(f.desc(\"count\")).show(20)", "_____no_output_____" ] ], [ [ "## Counts\n\n- inspired from [davidvrba](https://github.com/davidvrba/Stackoverflow-Data-Analysis)", "_____no_output_____" ] ], [ [ "posts.count()", "_____no_output_____" ], [ "# 1 = Question\n# 2 = Answer\n# 3 = Orphaned tag wiki\n# 4 = Tag wiki excerpt\n# 5 = Tag wiki\n# 6 = Moderator nomination\n# 7 = \"Wiki placeholder\" (seems to only be the election description)\n# 8 = Privilege wiki\n\nquestions = posts.filter(f.col('post_type_id') == 1)\nanswers = posts.filter(f.col('post_type_id') == 2)", "_____no_output_____" ], [ "print(questions.count())\nprint(answers.count())", "_____no_output_____" ], [ "# questions with accepted answer\n\nquestions.filter(f.col('accepted_answer_id').isNotNull()).count()", "_____no_output_____" ], [ "# count users\n\nprint(posts.filter(f.col('owner_user_id').isNotNull()).select('owner_user_id').distinct().count())\nprint(users.filter(\"id is not null\").select(\"id\").distinct().count())", "_____no_output_____" ] ], [ [ "## Response Time", "_____no_output_____" ] ], [ [ "response_time = (\n questions.alias('questions')\n .join(answers.alias('answers'), f.col('questions.accepted_answer_id') == f.col('answers.id'))\n .select(\n f.col('questions.id'),\n f.col('questions.creation_date').alias('question_time'),\n f.col('answers.creation_date').alias('answer_time')\n )\n .withColumn('response_time', f.unix_timestamp('answer_time') - f.unix_timestamp('question_time'))\n .filter('response_time > 0')\n .orderBy('response_time')\n)", "_____no_output_____" ], [ "response_time.show(2, False)", "_____no_output_____" ], [ "response_time = (\n questions.alias('questions')\n .join(answers.alias('answers'), f.col('questions.accepted_answer_id') == f.col('answers.id'))\n .filter(f.col(\"questions.owner_user_id\") != f.col(\"answers.owner_user_id\"))\n .select(\n f.col('questions.id'),\n f.col('questions.creation_date').alias('question_time'),\n f.col('answers.creation_date').alias('answer_time')\n )\n .withColumn('response_time', f.unix_timestamp('answer_time') - f.unix_timestamp('question_time'))\n .filter('response_time > 0')\n .orderBy('response_time')\n)\n \n", "_____no_output_____" ], [ "response_time.show(5, False)", "_____no_output_____" ] ], [ [ "## Hourly Data", "_____no_output_____" ] ], [ [ "hourly_data = (\n response_time\n .withColumn('hours', f.hour(\"answer_time\"))\n).show(2)", "_____no_output_____" ], [ "hourly_data = (\n response_time\n .withColumn('hours', f.hour(\"answer_time\"))\n .groupBy('hours')\n .count()\n .orderBy('hours')\n .limit(24)\n).toPandas()", "_____no_output_____" ], [ "hourly_data.plot(\n x='hours', y='count', figsize=(12, 6), \n title='Answer Hour',\n legend=False,\n kind='bar',\n xlabel='Hour',\n ylabel='Number of answered questions'\n)", "_____no_output_____" ], [ "year_data = (\n response_time\n .withColumn('years', f.year(\"answer_time\"))\n .groupBy('years')\n .count()\n .orderBy('years')\n).toPandas()", "_____no_output_____" ], [ "year_data.plot(\n x='years', y='count', figsize=(12, 6), \n title='Answer Year',\n legend=False,\n kind='bar',\n xlabel='Year',\n ylabel='Number of answered questions'\n)", "_____no_output_____" ], [ "response_hours = (\n response_time\n .withColumn('hours', f.ceil(f.col('response_time') / 3600))\n .groupBy('hours')\n .count()\n .orderBy('hours')\n .limit(48)\n).toPandas()", "_____no_output_____" ], [ "response_hours.plot(\n x='hours', y='count', figsize=(12, 6), \n title='Response time of questions',\n legend=False,\n kind='bar',\n xlabel='Hour',\n ylabel='Number of answered questions'\n)", "_____no_output_____" ] ], [ [ "## See the time evolution of the number of questions and answers", "_____no_output_____" ] ], [ [ "posts_grouped = (\n posts\n .filter('owner_user_id is not null')\n .groupBy(\n f.window('creation_date', '1 week')\n )\n .agg(\n f.sum(f.when(f.col('post_type_id') == 1, f.lit(1)).otherwise(f.lit(0))).alias('questions'),\n f.sum(f.when(f.col('post_type_id') == 2, f.lit(1)).otherwise(f.lit(0))).alias('answers')\n )\n .withColumn('date', f.col('window.start').cast('date'))\n .orderBy('date')\n).toPandas()", "_____no_output_____" ], [ "posts_grouped", "_____no_output_____" ], [ "posts_grouped.plot(\n x='date', \n figsize=(12, 6), \n title='Number of questions/answers per week',\n legend=True,\n xlabel='Date',\n ylabel='Number of answers',\n kind='line'\n)", "_____no_output_____" ], [ "posts_grouped_month = (\n posts\n .filter('owner_user_id is not null')\n .groupBy(\n f.window('creation_date', '4 weeks')\n )\n .agg(\n f.sum(f.when(f.col('post_type_id') == 1, f.lit(1)).otherwise(f.lit(0))).alias('questions'),\n f.sum(f.when(f.col('post_type_id') == 2, f.lit(1)).otherwise(f.lit(0))).alias('answers')\n )\n .withColumn('date', f.col('window.start').cast('date'))\n .orderBy('date')\n).toPandas()", "_____no_output_____" ], [ "posts_grouped_month.plot(\n x='date', \n figsize=(12, 6), \n title='Number of questions/answers per week',\n legend=True,\n xlabel='Date',\n ylabel='Number of answers',\n kind='line'\n)", "_____no_output_____" ] ], [ [ "# Tags", "_____no_output_____" ] ], [ [ "vi_sudo_tag = (\n questions\n .select('id', 'creation_date', 'tags')\n .groupBy(\n f.window('creation_date', \"4 weeks\")\n )\n .agg(\n f.sum(f.when(questions.tags.contains(\"nano\"), f.lit(1)).otherwise(f.lit(0))).alias('nano'),\n f.sum(f.when(questions.tags.contains(\"vim\"), f.lit(1)).otherwise(f.lit(0))).alias('vim')\n )\n .withColumn('date', f.col('window.start').cast('date'))\n .orderBy('date')\n).toPandas()", "_____no_output_____" ], [ "vi_sudo_tag", "_____no_output_____" ], [ "vi_sudo_tag.plot(\n x='date', \n figsize=(12, 6), \n legend=True,\n xlabel='Date',\n ylabel='Number of questions',\n kind='line'\n)", "_____no_output_____" ] ], [ [ "# Questions ", "_____no_output_____" ], [ "- Who asked the most questions?\n- How many people replied with an accepted answer by themselves?\n - what is the fraction of \"self-answerers\" against all users?\n- How many people never asked a question?\n- Which question took the longest to get an accepted answer?\n- Generate a plot where we can see at which month the most questions were raised.\n- Which post has the highest score? --> check it out https://unix.stackexchange.com/questions/{id}\n- Which question has the lowest score? --> https://unix.stackexchange.com/questions/{id}\n- Which post had the most comments?\n- Can you find an answer why \"When should I not kill -9 a process?\"", "_____no_output_____" ] ], [ [ "posts.select(\"id\", \"title\").filter(f.col(\"title\").contains(\"kill -9\")).show(20, False)", "+------+---------------------------------------------------------------------------------------------------------+\n|id |title |\n+------+---------------------------------------------------------------------------------------------------------+\n|62258 |What is the difference between exiting a process via Ctrl+C vs issuing a kill -9 command? |\n|674073|process not killed even we used kill -9 and process actually was belong to container that already removed|\n|612990|kill -9 $PPID kills the process? |\n|585409|Processes with states R and Rs not killable with kill -9 |\n|67166 |Why does Firefox refuse to die despite killing it with pkill -9? |\n|220175|Why kill -9 -1 doesn't work? |\n|212918|Still alive, still alive after kill -9 / SIGKILL |\n|456588|MacOS: su pkill -9 “process_name” = sorry |\n|558403|What if 'kill -9' still does not work? |\n|394298|Hanging process cannot be killed even via kill -9 -1 |\n|157133|How to get the pid of a process and invoke kill -9 on it in the shell script? |\n|8916 |When should I not kill -9 a process? |\n|38972 |Debian: Must pkill -9 twm and then login with twm |\n|281439|Why should I not use 'kill -9' / SIGKILL |\n|590984|How to alias `kj` to `kill -9 %?`(kill job) |\n|144918|Idiomatic way to kill -9 only if \"graceful\" way doesn't work? |\n|386380|wget can't be killed with 'kill -9' |\n|559910|How to terminate a process when kill -9 doesn't work? |\n|399438|Difference between `kill -9 <pid>` and `kill -INT <pid>`? |\n|660303|All processes using a device got hung and even `kill -9` does nothing |\n+------+---------------------------------------------------------------------------------------------------------+\nonly showing top 20 rows\n\n" ], [ "comments.filter(f.col(\"text\").contains(\"LOL\")).show(200, False)", "+-------+-------+-------+-----+---------------+-----------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----------------------+\n|id |post_id|user_id|score|content_license|user_display_name|text |creation_date |\n+-------+-------+-------+-----+---------------+-----------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----------------------+\n|357062 |210899 |47121 |0 |CC BY-SA 3.0 |null |LOL, true a stupid mistake:) |2015-06-19 20:03:08.107|\n|377732 |122731 |79008 |0 |CC BY-SA 3.0 |null |In my case I have the network manager along with lightdm. I wouldn't call those packages easily disposable expecially since I use them all the time. LOL |2015-08-10 18:46:40.583|\n|1302938|65564 |8965 |0 |CC BY-SA 4.0 |null |@Iguananaut LOL yeah about every two to three months I get a notification that somebody else stumbled upon this giant \"WTF?\" 9 years and counting... |2022-02-02 23:46:11.363|\n|133704 |89189 |7453 |0 |CC BY-SA 3.0 |null |@DravSloan - LOL, thanks for the laugh. It's something 8-) |2013-09-02 22:33:24.323|\n|141011 |93254 |39486 |0 |CC BY-SA 3.0 |null |try with `scp /Volumes/FX4\\ HDD/Users/matthewdavies/Downloads/NCIS.S11E01.HDTV.x264-LOL.mp4 [email protected]:\"/media/3TB/TV Shows/NCIS\"` |2013-10-02 05:59:04.82 |\n|270447 |164729 |38906 |0 |CC BY-SA 3.0 |null |@LOLKAT: Maybe `grep` saw your input as one line instead of many lines. Remember `-m` stop after the first line match, not the first match. |2014-10-29 02:49:02.747|\n|1184726|632729 |414777 |0 |CC BY-SA 4.0 |null |And no, not even the `kill $(jobs -p)` that you were probably meaning will not kill the background *jobs*, but only the *job leaders*. `sleep 1000 | sleep 1000 &`, then `kill -KILL $(jobs -p)`, `jobs -p` => LOL, still running. And you won't even be able to kill it again. |2021-02-05 19:18:24.403|\n|1206065|644007 |465598 |0 |CC BY-SA 4.0 |null |@KamilMaciorowski omg it was just that I don't get it LOL. I did command: ['sh', '-c', 'cp -rL /CustomTheme/custom_theme.tar /shared && cd /shared/ && tar -xvf custom_theme.tar &> file.txt']\\n and it works lol |2021-04-07 20:21:35.677|\n|1219537|499485 |411256 |0 |CC BY-SA 4.0 |null |OMG, people that have no direct experience with the problem described in the question should not be even commenting, WTF... The \"are you sure\" guys, or \"have you tried\" Andy LOL |2021-05-17 19:19:22.217|\n|1204814|643369 |200436 |0 |CC BY-SA 4.0 |null |LOL :-), thanks for the laugh! |2021-04-04 12:38:42.497|\n|984535 |505981 |39030 |0 |CC BY-SA 4.0 |null |@MyrddinEmrys: Ha! Yeah. FWIW, the package maintainer mentioned on one of the forums that they added a link to my answer on their wiki. I can't be the only one running Gentoo wanting more than one version of PHP, can I? Well, at least future me will benefit from this when I need to do it again in a few years! LOL |2019-07-24 18:47:35.947|\n|533158 |303595 |48650 |0 |CC BY-SA 3.0 |null |LOL. So, math is overrated, but first class functions are necessary for those complex object models frequently represented in scripts? :D I've written perl for around 21 years, and I led the conversion from ksh to perl at the company referenced. I still stand by David Korn when he says they have similar capabilities. https://www.usenix.org/legacy/publications/library/proceedings/vhll/full_papers/korn.ksh.a It really is worth learning more about; most people have no idea the power hiding right there under the covers in ksh93 (or even bash). KSH regex support is worth learning if nothing else.|2016-08-16 01:01:15.157|\n|309484 |185653 |103712 |0 |CC BY-SA 3.0 |null |Yep, it works. but looks like I reboot myself . LOL. how to do it in KRDC? |2015-02-19 09:18:20.323|\n|270448 |164729 |80886 |0 |CC BY-SA 3.0 |null |LOLKAT do you want the output to be exactly `ID: 4234235325` or just to print the line with the value of this id? My `grep (GNU grep) 2.16` prints only matched part of the first line with your expression, so believe it works as expected. |2014-10-29 02:50:13.4 |\n|35808 |26516 |977 |1 |CC BY-SA 3.0 |null |\"hopefully the SD cards aren't so stupid…\" <--- ROFLOL. Not likely! |2011-12-10 04:06:40.497|\n|25900 |18929 |9610 |1 |CC BY-SA 3.0 |null |When you were referring to /dev/disk/by-id I thought \"by-id\" was to be replaced by something and should be issued as a command. It didn't strike my mind that it could actually be a directory. Thanks for being patient with me. LOL, I was ridiculous :) |2011-08-17 22:22:32.187|\n|843169 |463276 |306119 |0 |CC BY-SA 4.0 |null |There is no option to register. It has been disabled and replaced with a message to send an email enclosing your required username & password. LOL security is obviously not high on agenda, but I have done so anyway and will see if I get any response. |2018-08-17 22:58:54.063|\n|576388 |327548 |138261 |3 |CC BY-SA 3.0 |null |OMG!!! Ponies!!! LOL!!! |2016-12-02 13:35:33.767|\n|1016322|547178 |8337 |0 |CC BY-SA 4.0 |null |Is it wifi? Is there a switch to turn wifi \"on/off\" on the keyboard (if it's a laptop LOL) |2019-10-19 01:32:41.77 |\n|1010783|544784 |29678 |0 |CC BY-SA 4.0 |null |En réalité: LOL. |2019-10-03 16:13:52.91 |\n|1061955|570795 |null |1 |CC BY-SA 4.0 |user314968 |Group share is a recurring problem with variations, as you point out. In this instance, software developers have access to the test release directory. It is theirs to muck with. They can work in that directory, push, pull, install in the virtual environment etc. The problem comes when they copy things in from another area they have been working in, and their umasks are not set with group permissions. It is confusing to them because file creation and copy by tar works fine. It is also funny that I have to give them a copy script that just does what cp does. 'acls' they are magic I say LOL |2020-03-03 10:28:13.783|\n|1063440|571293 |308316 |1 |CC BY-SA 4.0 |null |@bobsburner That's why the local client now checks the files sent by the remote against the glob (with an adhoc pattern matching implementation LOL) -- but that is very kludgy, because the shell on the remote machine may use a different glob syntax, you may want to quote or escape parts of the file name, etc. |2020-03-06 16:29:52.447|\n|293947 |177598 |37688 |16 |CC BY-SA 3.0 |null |Thanks for a link to the discussions. I guess the answer to this was NO. A funny quote from one of the links: \"modern users do not use terminals\". LOL. |2015-01-05 19:12:51.187|\n|799610 |439741 |null |0 |CC BY-SA 4.0 |user34720 |Yes, it is in Brazil. I can't convince my boss to spend 199R$(our currency) x 15 remote offices. LOL. |2018-05-04 18:44:55.117|\n|434347 |251322 |null |0 |CC BY-SA 3.0 |user86969 |LOL that is indeed the first time you mention it is a laptop :-D . |2015-12-28 16:55:14.203|\n|939420 |508458 |308316 |0 |CC BY-SA 4.0 |null |LOL what else is forbidden by the IT department's policy? looking askew at the monitor or typing with your thumbs? In `csh` the prompt is stored in the variable named ... `prompt` (`set prompt = 'foo% '`) not `PS1`. Your answer is completely bogus. |2019-03-25 12:20:35.26 |\n|838429 |460917 |209133 |0 |CC BY-SA 4.0 |null |LOL. Never thought of this. Nice. I’ll try it when I get home. |2018-08-06 23:09:14.807|\n|1295588|48555 |6622 |0 |CC BY-SA 4.0 |null |\"Finally, using ssh for a transport mechanism is both secure and carries very little overhead. \" LOL |2022-01-08 18:05:45.133|\n|248447 |152316 |7453 |0 |CC BY-SA 3.0 |null |@mikeserv- LOL, that's why I started testing all these A'ers, since they all appeared to be returning different results when I was reviewing them for you guys. We do have some Q's with like 10 A's and it's usually Q's like these, but they generally all return the same results 8-). |2014-08-27 01:57:26.147|\n|628396 |354645 |223413 |0 |CC BY-SA 3.0 |null |LOL. Noob indeed. I guess collation matters :) Edited. |2017-03-29 20:56:07.427|\n|1141065|611073 |4667 |0 |CC BY-SA 4.0 |null |\"John Jimmy Falloncamp\" LOL |2020-09-24 16:39:31.783|\n|131200 |87862 |7453 |0 |CC BY-SA 3.0 |null |@mattdm - LOL, thanks most certainly did not learn anything useful wrt FAA 8-). |2013-08-23 14:51:04.563|\n|532568 |303322 |181206 |0 |CC BY-SA 3.0 |null |Have you run LOL on a virtual machine? I was going to try that but someone said to try Play on Linux. |2016-08-14 12:49:51.647|\n|1029829|554422 |87673 |0 |CC BY-SA 4.0 |null |DOH! Thanks for pointing out the obvious that I missed LOL. Thanks again, was clear, concise and worked first time. |2019-11-27 20:13:55.87 |\n|1057818|568524 |null |0 |CC BY-SA 4.0 |user373503 |@UncleBilly 's link has this: \"This scheme rendered subdirectories sufficiently hard to use as to make them unused in practice.\" I think that explains Pike's story well (he is just not very precise). **sufficiently hard to use as to...** LOL, so there _is_ a limit to user-unfriendliness? |2020-02-21 07:23:38.697|\n|761189 |422759 |17208 |0 |CC BY-SA 3.0 |null |LOL, I mean Vim \"windows\", not MS Windows. Each window (file view) within a Vim session can have it's own value for each setting. |2018-02-08 16:15:47.543|\n|11555 |9120 |2343 |0 |CC BY-SA 2.5 |null |LOL :) .... My current status (still working on it): It seems that there is a 6 levels hierarchy of \"shell expansion\" (from `brace expansion` thru to `pathname expansion`). I can therefore understand that the example of `files=(*)` does not go through `word splitting`. However, it seems to me that the `pathname expansion` somehow internally(?) delimits the filenames by some other means than a space, because to my understanding the array asignment () would produce 4 items.. so is there some *special* interaction happening with these file-expansion words? This may be at the core of my puzzle. |2011-03-12 04:23:18.567|\n|47101 |34725 |16897 |0 |CC BY-SA 3.0 |null |LOL Unfortunately this is my current workplace where I have this problem. My ISP works fine! |2012-03-22 09:49:13.903|\n|363094 |212891 |116972 |0 |CC BY-SA 3.0 |null |@hjkefl LOL your input is outsized! Ok we need yet another approach for that large input, please be patient until next update. |2015-07-05 18:44:15.603|\n|853442 |468224 |308316 |0 |CC BY-SA 4.0 |null |LOL. I'm using gnu screen, which does NOT start login shells by default. Neither does the terminal emulator (mlterm). Maybe I should check my head ;-) |2018-09-11 14:01:15.323|\n|165831 |101824 |26026 |0 |CC BY-SA 3.0 |null |In fact I am just about to test it, belatedly. I plain forgot about it and just asked a similar question which caused this one to show as the top related question, LOL |2014-01-04 13:11:28.177|\n|298742 |179573 |99625 |0 |CC BY-SA 3.0 |null |LOL - Ubuntu 14.04 ***, fix it. |2015-01-17 18:33:09.067|\n|100361 |69226 |34963 |0 |CC BY-SA 3.0 |null |Mmm ok, that's worse than I imagined it. I hoped I could use some syntax like `\\d{1,2}` which I really find useful. Anyway, on OS X the command you wrote returns the whole output of `uptime`... LOL. Anyway, thank you for the answer, I'm gonna study `sed` in a deeper way. |2013-03-26 17:43:03.923|\n|579150 |329096 |113238 |0 |CC BY-SA 3.0 |null |LOL, my bad, I thought that /usr/bin/env was a directory, but of course env is program/executable, now I really don't why it doesn't work, but some progress is being made |2016-12-09 03:38:00.567|\n|11651 |9183 |2671 |0 |CC BY-SA 2.5 |null |LOL, I have no idea how that got there. |2011-03-13 12:04:04.74 |\n|42102 |31055 |15047 |1 |CC BY-SA 3.0 |null |@StéphaneGimenez done, bash is not the only thing to use. LOL |2012-02-07 13:30:23.157|\n|435241 |252380 |null |3 |CC BY-SA 3.0 |user34720 |Then, this should make you access the first and last array items at the same time, breaking all the logic of scanning an array ;) Could even create a blackhole. LOL |2015-12-30 18:13:19.72 |\n|881338 |298747 |257718 |1 |CC BY-SA 4.0 |null |LOL! This screenshot made me pretty happy. Thanks. |2018-11-13 19:17:29.78 |\n|209689 |130973 |4252 |4 |CC BY-SA 3.0 |null |LOL, \"cat abuse\" |2014-05-19 16:04:10.637|\n|639365 |360867 |204024 |0 |CC BY-SA 3.0 |null |Thanks again! LOL, the whole point of switching to a MAC was to not require a virtual machine anymore. |2017-04-24 03:49:33.507|\n|1143840|500438 |278933 |0 |CC BY-SA 4.0 |null |What happened here? I have a very similar situation where my specified DNS seems to be out of line with what's reported as current. I've also been experiencing random drops of my SSH on just the system I'm working on, which sounds a lot like an interface reset, where I am using a standard ens3. Thanks for the well-crafted question. An authoritative answer would be nice. (See what I did there? LOL) |2020-10-02 16:12:16.41 |\n|1127564|603624 |423323 |1 |CC BY-SA 4.0 |null |Hey i fix the problem, i run the recovery mode and after 40 min it got restored to default chrome. But it was the black screnn all the time... LOL ! |2020-08-13 09:02:03.5 |\n|404786 |236049 |11639 |0 |CC BY-SA 3.0 |null |There is a difference between \"nothing wrong\"and \"optimal setup.\" Personally the pool servers I admin all have local refclocks. I doubt you do because you assume pool clients must be stratum=3. Massively varying error? Thanks for the LOLz. I did'nt realize we were using a version of ntpd that does'nt groom the time sources. ntpd will smooth out the variance. Furthermore if you do add it to the pool list the pool daemon removes clocks with too much error. Adding capacity to the pool infrastructure is not negative. Run a server with `tos minsane 7 minclock 5` and tell me about your error |2015-10-15 23:43:05.21 |\n|1263871|670252 |373946 |0 |CC BY-SA 4.0 |null |LOL I am in full derp mode, it was not. is now but I had already started it just pointing it to the nfs mount. I think since I used the -Pv I am safe to ctrl-break it when I need to at any time? It seems to be working well a few files (that are on the remote not the host) give a strange error when the host gets back to them again: rsync: chown \"/mnt/archive_copy/archive/images/dcam/UT201006/.proc-d0355.fits.IGw8LT\" failed: Operation not permitted (1)\\n but other than that its all good. |2021-09-23 18:14:46.04 |\n|1308789|692058 |492642 |0 |CC BY-SA 4.0 |null |LOL! So UNIX has standards, but there's no enforcement. I just built svd2ada on MacOS and neither the build script nor the instructions advise on where and what files should be deployed to become part of the Ada toolchain. It's incredibly frustrating. MacOS outside UNIX has conventions that you dare not disobey, or you'll have users standing outside with pitchforks. |2022-02-25 00:45:37.72 |\n|120999 |81131 |19370 |0 |CC BY-SA 3.0 |null |@AlberteRomero LOL |2013-07-05 17:39:16.687|\n|142155 |10226 |19869 |1 |CC BY-SA 3.0 |null |don't ask like \"is it possible to use **sed** to do ....\" you can use sed to do anything within the area of text processing. LOL |2013-10-06 03:13:09.647|\n|609391 |344367 |113238 |0 |CC BY-SA 3.0 |null |LOL, ok I see, but it's not super clear, let me update your answer, thanks :) |2017-02-12 09:43:42.85 |\n|541013 |307757 |181999 |0 |CC BY-SA 3.0 |null |Dude you Data wont be on the disk if you boot linux. Your data is on a filesystem and linux got installed on a different filesystem that overwrote the current one. a filesystem is a type of agreement how a file should be stored what combination of ones and zeros indicate the beginning and the end of the file. bevor installing a so looking like \"hacker distro\" that gets previewed with he gnome desktop and 90% of the preview is the gnome desktop LOL. read something for yourself about how things work. yes I know reading sucks |2016-09-04 05:57:28.11 |\n|562011 |319749 |197700 |0 |CC BY-SA 3.0 |null |@don_crissti LOL. I like your comments buddy. ;) |2016-10-29 16:53:02.55 |\n|554456 |284033 |57439 |0 |CC BY-SA 3.0 |null |LOL. You were right: the `.sh` part disables the script from running. I don't know why this happens, but thanks you. |2016-10-10 14:13:25.58 |\n|255553 |155994 |84441 |4 |CC BY-SA 3.0 |null |well, while trying change from Frodo1 to Bilbo2 I got message that: \"You must wait longer to change your password\" LOL |2014-09-18 08:29:56.167|\n|1182523|631839 |325065 |0 |CC BY-SA 4.0 |null |Well, maybe not on Hurd. Thence \"unspecified\", LOL. |2021-01-31 00:41:44.283|\n|441329 |255483 |141443 |0 |CC BY-SA 3.0 |null |What's the exact difference between `[` and `[[`? (Searching on Google for `[ vs [[` gives me results for `vs` only, LOL) |2016-01-15 09:04:26.267|\n|496720 |284968 |83246 |0 |CC BY-SA 3.0 |null |Such evilness as files with spaces in their names!! How dare you bring that up LOL :) |2016-05-24 13:22:29.593|\n|665817 |373806 |113238 |0 |CC BY-SA 3.0 |null |The man pages for bash do not sound short LOL, combing for this information is not ideal. Now the next person can do a google search and hopefully find the information faster than reading the bash man pages, right? E.g.: https://tiswww.case.edu/php/chet/bash/bashref.html LOL |2017-06-29 23:17:30.24 |\n|1095911|588233 |413896 |0 |CC BY-SA 4.0 |null |LOL Also did you read the file in the git describing the assignment? One of the requirements is controlling ICMP traffic. Ping is an ICMP function. |2020-05-22 01:11:36.093|\n|1159964|619948 |435718 |0 |CC BY-SA 4.0 |null |@cutrightjm LOL |2020-11-18 07:33:54.437|\n|943226 |510315 |null |0 |CC BY-SA 4.0 |user34720 |@0xSheepdog - \"Question was moved to homework.stackexchange.com\"...LOL |2019-04-03 16:59:18.647|\n|1066433|573030 |308316 |0 |CC BY-SA 4.0 |null |@Slavistan: I think I already know your next __Q:__ Why does `export x=1; echo \"$0\" | entr -s 'x=2; echo set x to 2'; echo \"$x\"` NOT set `x` to `2`? __A:__ because Unix is not plan9, LOL ;-) |2020-03-15 17:06:35.173|\n|1000366|539210 |308316 |2 |CC BY-SA 4.0 |null |@markgraf LOL `chsh -s /bin/bash 7</dev/random` and Bob's your uncle (but only if the OP fixes the quoting in their awk command ;-)) |2019-09-05 15:20:17.907|\n|780172 |432100 |272806 |0 |CC BY-SA 3.0 |null |That variable index increment for a statically declared array, LOL. |2018-03-20 02:58:07.217|\n|841258 |462337 |305251 |0 |CC BY-SA 4.0 |null |Two downgrades! LOL |2018-08-13 17:15:16.68 |\n|1289303|681425 |491750 |0 |CC BY-SA 4.0 |null |hahaha LOL ... you just read my mind and the hidden agony, as that's exactly how i feel when i try to disconnect my laptop with my laptop behaving all erratic.\\n\\nI am hoping a powerful 'Linux Oracle' can intercede between my laptop and docking station and if things don't work out then 'exorcism' performed by a Linux priest of High Order may the last option. |2021-12-16 13:06:36.343|\n|1262110|669559 |90878 |0 |CC BY-SA 4.0 |null |LOL I actually installed FreeBSD a few days ago, but haven't even gotten a window manager running yet... :-D So can't help you there, unfortunately. I just did a quick test with a Win VM tho. I actually can kill the remote server, on the server itself; not by shutting down my client locally. And the local client still shows the server online. I can also close my local client's connection on the remote end, via my client. So... are you sure the users are actually disconnecting by closing their local client, not on the remote end? |2021-09-18 15:19:11.09 |\n|619621 |350046 |219835 |0 |CC BY-SA 3.0 |null |LOL, ok, I'm not a system guy, but I don't believe the whole server has only 200GB free space. I think that's free space of my afs, yet I can only use 50% of my allocated space? That's strange. |2017-03-08 18:54:04.467|\n|1133762|218839 |43781 |0 |CC BY-SA 4.0 |null |@MadsSkjern LOL.. I really did not expect that you will answer that :D but it's good, i think they all deserve the acceptance of their answers :)\\nBut better choose any answer instead of forgetting it 5 years - at least if it's correct and more than one sentence... Also your question was \"does dd copy everything\" -> this was already answered completely and sufficiently with the first post by Fiximan... Better answers will get their points anyway, while getting more upvotes than the accepted :D |2020-09-02 08:45:27.21 |\n|375255 |220633 |64157 |0 |CC BY-SA 3.0 |null |@Celada HA/LOL, but that isn't exactly the problem... |2015-08-06 20:53:09.58 |\n|371981 |15881 |32558 |0 |CC BY-SA 3.0 |null |LOL, this was migrated and the SO dupe that was not got more upvotes: http://stackoverflow.com/questions/11238457/disable-and-re-enable-address-space-layout-randomization-only-for-mysef |2015-07-28 13:27:57.107|\n|1261096|668138 |433561 |0 |CC BY-SA 4.0 |null |Yes, sorry, I was not clear at all it turns out! My system FS is EXT4. And I think that is where the problem lies. I don't really want to run my system on BTRFS, as I've read about instability issues compared to EXT4. I was thinking it could only do snaps if the STORAGE was BTRFS, but that the system didn't matter. I really want those snaps, so maybe I have to just bite the bullet. LOL\\nThank you for your response! I tried to mark your answer as the solution but even though it's my question, I don't have enough rep to vote. LOL So annoying. |2021-09-15 23:40:12.207|\n|115421 |78569 |40680 |0 |CC BY-SA 3.0 |null |LOL yeah I guess that's not my responsibilty, I just have to make the server secure. I think I'll hash the email address as well. |2013-06-07 11:30:08.517|\n|122730 |82831 |30048 |0 |CC BY-SA 3.0 |null |Ahhhh, it looks for a number! LOL. I tried words like \"high\". I'll give that a shot! |2013-07-12 21:24:19.37 |\n|532572 |303322 |184611 |0 |CC BY-SA 3.0 |null |LOL doesn't work on Wine, Probably It will doesn't on Play On Linux. You can try on Play On Linux, but I suggest you to use a virtual box with 2/4GB ram, It works good. |2016-08-14 12:56:16.397|\n|520746 |296553 |47085 |0 |CC BY-SA 3.0 |null |I must use them. I use my VM for gaming and LOL's launcher/client runs notoriously slow without hugepages. I enabled hugepages after I once almost accidentally dodged a ranked on promotion series because the launcher lagged during champion selection. |2016-07-18 08:50:10.223|\n|1050091|564029 |325065 |0 |CC BY-SA 4.0 |null |The entire premise of this question is wrong: `/dev/pts/10` is not \"connected to\" stdin, `/dev/pts/10` IS stdin. This question looks _purposely_ obtuse and a bit of text analysis \"connects it\" (LOL) to other similar questions on different matters. |2020-01-29 13:20:06.69 |\n|1016537|547651 |null |0 |CC BY-SA 4.0 |user373503 |@IliaGilmijarow LOL! btw I just answered, addressing that browser tab thing. Estimated time has three dots and is I think not so clear for all of us. Estimated time of arrival i only know. |2019-10-19 21:04:29.43 |\n|252877 |154486 |83488 |0 |CC BY-SA 3.0 |null |I realize I should've just used CLI PHP or Ruby, LOL, but I was too far into the project to turn back from shell scripting. Lesson learned. |2014-09-10 05:01:36.03 |\n|283575 |148551 |22858 |0 |CC BY-SA 3.0 |null |TY, cool stuff! Got them bookmarked. Gonna delve myself into those the next days for sure. BTW, please do no longer feel addressed when I say \"`sed` learners\". You definitely are no learner (LOL), more like a pro asking other uber-pros like Stéphane to squeeze out the very last quirks that remain. ;) *grin* |2014-12-05 17:15:22.163|\n|731255 |408364 |227199 |1 |CC BY-SA 3.0 |null |@RomanPerekhrest Greetings Ukraine PYTHON GENIUS LOL |2017-12-02 13:22:16.867|\n|444594 |257166 |79008 |0 |CC BY-SA 3.0 |null |Thanks for the tip @meuh however it seem that neither of the three options for freeing memory (global garbage collection, cyclic and sending \"heap-minimize\" notifications) results in something. Maybe these are just dummies? LOL Will have to investigate further on that. |2016-01-23 11:06:45.077|\n|504507 |288533 |121401 |1 |CC BY-SA 3.0 |null |Normally, I'd slap an IT person for doing something like this, but in this case, it's pretty humourous so I'd probably let it slide LOL |2016-06-08 20:51:23.93 |\n|954132 |492495 |202420 |0 |CC BY-SA 4.0 |null |@MartinOtto Indeed. I think unlocking the TDP would help a ton lot. We can cool it since the crazy fans. Also you can run \"Nitro Boost\" even on Linux with nbfc: https://forum.manjaro.org/t/control-fans-on-acer-nitro-5-an515-42-and-possibly-other-laptops-with-nbfc/80480 and that helps a lot in games.\\nAcer you should really unlock that TDP on Nitro 5. Like what ? Other OEMs limit it to 25W and they dont have as good cooling as we do and we are running 15W LOL.\\nI may post a bug on bugzilla. But do you have tearing on laptop display when playing some games on top of the screen ? |2019-04-30 19:20:31.92 |\n|1157219|618419 |243949 |0 |CC BY-SA 4.0 |null |512GB RAM. Have you tried adding more? LOL. Two more things to check: disk space and the logs. Has CentOS switched to systemd? |2020-11-08 19:32:59.007|\n|1130424|605378 |335415 |0 |CC BY-SA 4.0 |null |@AdminBee I thought all distribution is the same. LOL |2020-08-21 13:39:36.433|\n|121047 |81131 |10292 |10 |CC BY-SA 3.0 |null |@AlberteRomero That's it, more or less, at least most-ly,. LOL I really like the horizonital scroll in most. |2013-07-05 20:39:54.467|\n|82824 |59909 |27121 |0 |CC BY-SA 3.0 |null |It is new year, why you spending so much time on the server? LOL. HAPPY NEW YEAR!! |2012-12-31 04:02:31.72 |\n|598658 |338860 |null |1 |CC BY-SA 3.0 |user34720 |Mentally health people avoid newline character on filenames. LOL. |2017-01-20 12:35:01.67 |\n|993989 |536220 |252501 |0 |CC BY-SA 4.0 |null |@cas Oh man you saved my life LOL, you should put is as Answer instead of comment. Oh ya, another thing is use single `[` instead of double `[[` |2019-08-19 08:20:47.353|\n|48995 |36138 |null |0 |CC BY-SA 3.0 |user14517 |LOL. I re-read my question and I think I misled the audience, I am actually designing this as a productivity suite. |2012-04-10 13:28:14.18 |\n|906495 |493475 |117549 |0 |CC BY-SA 4.0 |null |LOL @StephenKitt; now the gears are turning |2019-01-09 14:47:18.36 |\n|822147 |451976 |null |0 |CC BY-SA 4.0 |user34720 |No problem at all. Do you have any hardware attached to PCI1 ? I'm searching for more ACPI+BIOS+Fedora+Your_hardware info on different sites and the best answers i could find is \"use Ubuntu 18.04 that is better supported by Dell\"... LOL |2018-06-29 14:14:57.853|\n|1265818|671167 |492628 |0 |CC BY-SA 4.0 |null |LOL, the contents of my ~/.profile said \"echo foo\". Nevertheless, the question is still a valid one I think. |2021-09-29 20:25:38.243|\n|1108544|594014 |11417 |0 |CC BY-SA 4.0 |null |@G-ManSays'ReinstateMonica' LOL. The original version just `rm -f`'ed the file which lead to a comment by another user about using the nuke option unneccessarily. But I agree that creating it before the loop is a bad idea. OTOH so is using `done > list-of-headers.csv` if you run the script in the same directory several times. |2020-06-21 07:03:31.22 |\n|1152828|226980 |39593 |1 |CC BY-SA 4.0 |null |Hey I've come here a SECOND time to relearn this LOL. A French accent (diacritic) in the text was causing grep to barf |2020-10-26 09:43:09.343|\n|238053 |146920 |78842 |0 |CC BY-SA 3.0 |null |That is a big 'LOL' :-) |2014-07-28 15:45:11.513|\n|1079886|579949 |286615 |0 |CC BY-SA 4.0 |null |ANSWER: in `bash` use `$ echo \"$RESPONSE\" | grep \"% packet loss\"`. Interesting logic... spend a quarter to save a dime LOL |2020-04-14 20:09:05.347|\n|1094481|4799 |333273 |2 |CC BY-SA 4.0 |null |Hey @frederix who let you in here? I see they will let just anyone in here these days! LOLOLOLOL |2020-05-18 20:53:03.27 |\n|1149630|614747 |431002 |1 |CC BY-SA 4.0 |null |I find it highly doubtful that that ever were the case. Maybe some \"Unix-like\" environment running on MSDOS (after all, `command.com` did support \"pipes\" using temporary files, LOL). Until you remember it, I'll consider that claim baseless. |2020-10-16 13:02:45.24 |\n|141055 |93254 |47009 |0 |CC BY-SA 3.0 |null |try narrowing the problem down; e.g. can you do `cd /Volumes/FX4\\ HDD/Users/matthewdavies/Downloads/; scp NCIS.S11E01.HDTV.x264-LOL.mp4 [email protected]:/tmp`? can you do `scp /Volumes/FX4\\ HDD/Users/matthewdavies/Downloads/NCIS.S11E01.HDTV.x264-LOL.mp4 [email protected]:/tmp`? |2013-10-02 10:40:06.21 |\n|122740 |82831 |30048 |0 |CC BY-SA 3.0 |null |Hah...you're right! LOL. That's funny. If it comes down to that being my one and only clue, I might just let this slide and abandon this part of the project as it's not vital for what I'm doing, just a \"would be nice\" kinda thing. |2013-07-12 21:37:18.3 |\n|979489 |529066 |124211 |0 |CC BY-SA 4.0 |null |LOL good point, I will edit with something more useful |2019-07-10 08:53:17.743|\n|1259780|668528 |90878 |0 |CC BY-SA 4.0 |null |LOL - @Lizardx ain't one to mince his words :-D Have to say I wholeheartedly agree. |2021-09-11 10:29:31.85 |\n|312263 |187142 |104700 |0 |CC BY-SA 3.0 |null |LOL - no, not dedication... it's a custom kernel/init where I simply haven't updated the repository in quite some time! :) HOWEVER, I just tinkered with block quotes and the likes and believe I have things down-pat manually (ie: 4 spaces after an empty line starts a code block, etc.) |2015-02-27 08:06:30.07 |\n|1193068|636791 |286615 |0 |CC BY-SA 4.0 |null |@FET: Good for you :) All rationality ever got anybody was boredom LOL. You don't want GRUB - there should be some instructions to that effect in [this procedure](https://askubuntu.com/a/1245535/831935). **Pay attention to the `ubiquity -b` step!!!** You can install `rEFInd` without installing Ubuntu. And note in my rev. answer you will need an external drive to install Linux. |2021-02-28 09:27:43.8 |\n|141013 |93254 |39486 |0 |CC BY-SA 3.0 |null |you don't have to escape the whitespace ` ` in double-quotes. actually, `I think it had to be scp /Volumes/FX4\\ HDD/Users/matthewdavies/Downloads/NCIS.S11E01.HDTV.x264-LOL.mp4 \"[email protected]:/media/3TB/TV Shows/NCIS\"` because the last part(destination) is counted as a whole argument. |2013-10-02 06:12:59.753|\n|80396 |57593 |19178 |2 |CC BY-SA 3.0 |null |That is a much simpler solution than the one I came up with, LOL! |2012-12-12 18:57:29.76 |\n|552243 |314191 |193279 |0 |CC BY-SA 3.0 |null |@maulinglawns - I think you've mostly covered the bases, here. But, I'm just saying that \"not all greps are built alike\" and \"OP hasn't given clear enough information\" for a definitive answer... that is all. There are some systems that are \"strange\" out there (eg. Solaris). (And honestly, I'm still not convinced this isn't a homework assignment... LOL) |2016-10-04 10:21:33.847|\n|10905 |8678 |3860 |7 |CC BY-SA 2.5 |null |LOL \"Experty friendly, user antagonistic\". It made my day. +1! |2011-03-06 09:21:17.33 |\n|500351 |287029 |172514 |0 |CC BY-SA 3.0 |null |well I'm glad you solved. I kinda figure that out when I finally saw that beautiful string that states that this question were asked **4 months ago**, I _really_ should pay more attention to the site UI, LOL!. and yup, let's leave this Q/A here in case you or anyone else trample into the same problem. |2016-06-01 23:14:10.953|\n|451434 |260694 |150530 |5 |CC BY-SA 3.0 |null |That sounds like that rick roll kernel module LOL |2016-02-08 05:11:54.437|\n|932117 |504937 |278319 |0 |CC BY-SA 4.0 |null |Thank you very much @JdeBP for useful links, I found the answer from your given links. LOL where were they when I was searching, I was not using proper keywords perhaps. |2019-03-07 15:33:31.75 |\n|419424 |244229 |11639 |0 |CC BY-SA 3.0 |null |Epic? LOL. I have no idea what the purpose of the bell is but it took me longer to read your perl script than it did to double check two man pages and type `$ nice -n 1 watch -b -p -n1 \"date +%s.%N;false\"` |2015-11-20 03:19:25.133|\n|358571 |211679 |22858 |0 |CC BY-SA 3.0 |null |@ikrabbe It definitely doesn't. :) So I now get it, you were the initiator of inverting the logic, and user179... just jumped on the bandwagon. LOL. |2015-06-23 20:57:22.007|\n|183480 |117550 |60618 |0 |CC BY-SA 3.0 |null |Thank you! I knew I was missing something. I even tried '$someday' earlier. LOL |2014-02-28 21:41:59.577|\n|682299 |383262 |105631 |0 |CC BY-SA 3.0 |null |LOL, I'm sorry. The code is anything but \"self evident\". There are three different layers and I have no idea which layer I should be looking for, nor in what kind of file to look for them. Is Java using some kind of reflection class to determine the properties? If so, no hope. If it's some kind of enumerated class, like you would do in C, I have been unable to find it.\\n\\nFurther, I cannot connect to our infastructure via the SDK because our admins have restricted it for some reason. I just need a list. |2017-08-03 18:37:30.887|\n|1117098|402489 |315425 |0 |CC BY-SA 4.0 |null |LOL just ran into this now ... must be way too tired, but thanks for the solution :) |2020-07-14 14:05:58.55 |\n|28019 |20464 |8965 |0 |CC BY-SA 3.0 |null |LOL my bad. I was trying to confirm the Cygwin part for ya (IIRC your original post had a possible command). It does specifically say \"ctime (time of last modification of file status information)\" if I had paid attention. |2011-09-13 02:02:12.223|\n|463430 |267357 |9503 |0 |CC BY-SA 3.0 |null |I managed to get a full list of files and they were around 6500. This is very less number, I thought there were millions LOL. I would assume this is nothing for `ext4` file system. [Any thoughts?](http://serverfault.com/questions/761381/why-would-a-folder-with-around-6500-in-a-ext4-partition-become-inaccesible?noredirect=1#comment957762_761381) |2016-03-04 07:27:00.52 |\n|962900 |520955 |281844 |0 |CC BY-SA 4.0 |null |`FNR==NR` Yuck!!! (btw, \"Yuck!\" is probably the proper way to pronounce `awk`.) LOL |2019-05-25 13:32:22.06 |\n|949204 |513271 |254900 |0 |CC BY-SA 4.0 |null |............. LOL |2019-04-18 19:15:28.28 |\n|139768 |92475 |22858 |0 |CC BY-SA 3.0 |null |LOL. YMMD with that great comparison.:) But for real, I think it might make a difference to a system that is *not yet* out of memory. That is, if you try adjusting swappiness *before* copying the huge-size stuff, you might have a slight chance of getting swap utilized *way earlier* (ideally preventing vital processes from being killed). Or couldn't you? |2013-09-26 19:13:38.927|\n|602160 |340593 |16792 |0 |CC BY-SA 3.0 |null |Is this another meaning of LOL? |2017-01-27 17:53:59.427|\n|275964 |167457 |91207 |0 |CC BY-SA 3.0 |null |Haha sorry about that. I fixed it! What happened was I changed the whole export PATH= TextEdit file. I opened by .bash_history and went through every detail. I found my original PATH and copied that into the TextEdit and it works. Basically, I deleted PATH..LOL. At least learned some new things about terminal! Thank you all for your help. Really appreciate it :) |2014-11-13 00:45:50 |\n|723744 |404680 |79008 |0 |CC BY-SA 3.0 |null |On Ubuntu it does...And Ubuntu is from the same Debian family... LOL Talking about diversity in the Linux world. -_- Also @Kusalananda is right - I cannot guarantee that SMP will be present in the output. |2017-11-15 09:26:44.047|\n|1180261|630636 |138197 |0 |CC BY-SA 4.0 |null |Of course I'll do as you suggest, but I can't believe it doesn't exist a proper way to make it work LOL |2021-01-24 09:04:33 |\n|398355 |16640 |32558 |0 |CC BY-SA 3.0 |null |http://stackoverflow.com/questions/5920333/how-to-check-size-of-a-file LOL for the migrate :-) |2015-10-01 08:06:07.693|\n|816943 |3506 |79008 |0 |CC BY-SA 4.0 |null |I no longer use it...from approx 1 year. Just saying. LOL |2018-06-15 13:06:00.147|\n|179095 |115266 |16369 |0 |CC BY-SA 3.0 |null |What DNS is for? LOL! Will DNS re-start your disrupted TCP streams? |2014-02-15 02:01:14.77 |\n|656964 |369330 |null |0 |CC BY-SA 3.0 |user34720 |This request made me remember this - https://xkcd.com/1172/ - LOL |2017-06-06 19:38:02.817|\n|33252 |24740 |9382 |0 |CC BY-SA 3.0 |null |I just read what I'd written... LOL. Actually, I'm not sure about *reinvoke* (*re-invoke* maybe?), but thanks for telling :) I always appreciate corrections. |2011-11-16 19:38:54.293|\n|20059 |14944 |1974 |0 |CC BY-SA 3.0 |null |@alex: Nevermind. I knew Cairo sounded familiar... it's familiar because that's what Inkscape uses to display SVG and convert them to PDFs. LOL. |2011-06-13 20:40:03.9 |\n|496786 |284968 |83246 |0 |CC BY-SA 3.0 |null |I will do no such horrible thing... :) ok I did and it does work.. Someone should be flogged for this!! LOL. |2016-05-24 15:37:42.207|\n|89816 |62886 |19617 |0 |CC BY-SA 3.0 |null |LOL- well, that was easy (read: obvious)! thx |2013-01-28 21:25:18.07 |\n|1018658|548771 |8337 |0 |CC BY-SA 4.0 |null |What if you set it to start \"last\" that help? https://superuser.com/a/573761/39364 This is like debugging via chat LOL |2019-10-26 00:31:17.227|\n|761653 |41647 |153769 |0 |CC BY-SA 3.0 |null |@mat - I'm a relative newcomer here. I see that you made an edit but left in the phrase \"You see, I'm a programmer ...\" which had me LOL-ing. Is there guidance on this kind of humor? Maybe it's a question for meta ... |2018-02-09 12:30:12.14 |\n|1201040|640662 |392770 |3 |CC BY-SA 4.0 |null |@GeoRie But this will also match some **unrelated** things. Check it with `echo \"AZAZA LOL 23 WTF\" | grep -oP '\\b(?<!\\.)\\d+(?!\\.)\\b'` - it returns `23`. So this is a bit **inaccurate solution**. |2021-03-23 11:52:18.47 |\n|821136 |452127 |null |0 |CC BY-SA 4.0 |user34720 |Old broadcom wifi user here so, this wiki was almost my homepage years ago. LOL. @EvanCarroll, added some probing with `dmesg` |2018-06-27 01:55:19.837|\n|1266364|671406 |90878 |0 |CC BY-SA 4.0 |null |LOL :-D Didn't check as AdminBee had just commented on missing code highlighting :-) Thx for letting me know, I can remove my redundant comment :-) |2021-10-01 11:16:51.17 |\n|1078175|579207 |308316 |0 |CC BY-SA 4.0 |null |LOL, does ssh supports reCAPTCHAs? That would certainly shut me out for good -- Google's artificial stupidity always treats me (a human) as a bot. Seriously, if you want an `expect(1)` solution for multiple ssh password auth, then say so explicitly. That would be much more useful for people coming here via searches. |2020-04-10 18:24:20.977|\n|683627 |382543 |245310 |0 |CC BY-SA 3.0 |null |If you knew someone who did this, you would be obliged to type in a LOLCAT message in the terminal. CAN HAS CHEEZBURGER? |2017-08-07 02:21:09.06 |\n|686815 |386095 |114317 |0 |CC BY-SA 3.0 |null |LOL, yes, you are right I suppose. ansible, docker and restore from backup every time something fails. I am getting too old for this stuff! :-) I did ask the question at Superuser, but that got me a Tumbleweed badge. Serverfault seems more about networking stuff. The question seems to be too arcane for other forums than this one. |2017-08-14 20:03:19.38 |\n|1146029|613618 |414777 |0 |CC BY-SA 4.0 |null |Weren't all plan9 fonts variable-width? I clearly remember that being a strong esthetical and technical point made by the designers of plan9. The interface to their command windows (\"xterms\" LOL) in the rio window manager was very deliberately rejecting any column alignment and any curses-like interface. |2020-10-09 02:34:55.38 |\n|142248 |93811 |48637 |0 |CC BY-SA 3.0 |null |May I add, that there are periods where it seems to post what I would say is acceptable in speed response, if I do a 'refresh' page (whether in Chrome or Firefox) it will take a perceived 200-300ms, then maybe some time later a mind boggling 1.5 mins. You can imagine trying to make multiple edits to a View setup, or Config change, I'm sitting here wondering if I should go make (another!) brew LOL |2013-10-06 15:54:54.14 |\n|81568 |58947 |27121 |0 |CC BY-SA 3.0 |null |LOL, Thx! Yes, I enjoy X'mas! |2012-12-21 14:56:42.067|\n|1040102|559688 |308316 |0 |CC BY-SA 4.0 |null |what \"console window\" is that? (`gnome-terminal`, `konsole`, `putty`, linux virtual terminal, etc). Try `tput smcup; tput cup \"$((LINES/2))\" \"$((COLUMNS/2-2))\"; echo -n LOL; read var; tput rmcup` |2019-12-31 13:02:01.49 |\n|1019071|548771 |8337 |0 |CC BY-SA 4.0 |null |I didn't think it would be useful, but in retrospect it might be: add an \"after=network-online.target\" though a handful of people have found even that isn't enough so you have to add a restart https://github.com/google/cloud-print-connector/issues/140 (or add a \"sleep 10\" into the bash side of things LOL). GL! |2019-10-28 04:04:20.697|\n|1057214|531173 |308316 |0 |CC BY-SA 4.0 |null |@ilkkachu thread ids and process ids live in the same namespace, At least in Linux. But, yes, it's probably a good idea to not create children in the first place, I will not argue against that, LOL. As to practical scenarios, ANY scenario where you want to do ANYTHING with the PID returned by waipid() other than check if it's != -1 is problematical. |2020-02-19 20:44:16.453|\n|1013360|230451 |8337 |1 |CC BY-SA 4.0 |null |Sadly, SO_KEEPALIVE seem to only be sent every couple of hours (at least by default) so a better name for them would be \"check if dead's\" LOL |2019-10-10 20:14:27.15 |\n|761556 |419361 |142735 |0 |CC BY-SA 3.0 |null |I'd have to do some research to find out how to enter the commands, so I'll just consider my itch scratched. LOL. I'll just have to hope that running a desktop session on an Internet-connected computer with user tom in the wheel group doesn't hose me. |2018-02-09 08:03:01.43 |\n|1221737|651007 |174665 |0 |CC BY-SA 4.0 |null |This `find` does lack `-print0`, unfortunately. That would've been pretty convenient. And the `xargs` lacks `-I`. It's as if this stuff was disabled with the specific intent of preventing what I'm trying to do. [sigh] \\n\\nThe `while` version DID work. I didn't try that approach before asking here, because people vent all the time, and even write big tutorials, about how allegedly terrible it is to use `while` loops to process files and directories in the shell. LOL.\\n\\nI didn't test the fancy script, not having names to deal with containing newlines. Thanks for the thorough answer! |2021-05-24 00:39:44.603|\n|56230 |40988 |15732 |0 |CC BY-SA 3.0 |null |So you trust more some perl script based portknocking than openssh which is developted by OpenBSD guys with security in mind? :D Is not great portknocking l33t tool running under root? LOL. |2012-06-19 09:02:47.813|\n|941650 |410050 |205857 |0 |CC BY-SA 4.0 |null |@dreua Yeah, the typo is quite funny. LOL |2019-03-29 20:02:26.697|\n|951871 |265646 |138261 |0 |CC BY-SA 4.0 |null |LOL 3 years after the answer after someone upvoted it, and I just noticed this answer is telling me about a typo in the question; corrected to /proc/interrupts |2019-04-24 10:53:59.417|\n|937506 |507379 |141494 |0 |CC BY-SA 4.0 |null |LOL. TBH it really didn't -- this question is three years old. But since it didn't have an answer yet I went ahead and tested your solution to confirm it. ¯\\_(ツ)_/¯ |2019-03-20 22:23:22.417|\n|900188 |490726 |325065 |0 |CC BY-SA 4.0 |null |So not being able to exchange accurate diff(1)s is a feature now? LOL. |2018-12-24 10:45:02.14 |\n|937342 |507361 |342648 |0 |CC BY-SA 4.0 |null |That means my -x \"..*\" strategy looks into lot of unnecessary places... I would need to update this everywhere, LOL....... |2019-03-20 17:20:14.933|\n|1262154|669603 |90878 |0 |CC BY-SA 4.0 |null |LOL :-D Once you get it working, please do write an answer to your own question with step-by-step instruction. I've been using KeePass for almost a decade and never found out how to use it with a terminal SSH client :-D |2021-09-18 18:31:10.653|\n|150146 |98443 |9491 |0 |CC BY-SA 3.0 |null |@JosephR. I'm not sure of how could I interleave the songs \"in such a way as to minimize adjacency of songs by the same artist\". Thanks for the info, though! Maybe because yesterday and today are holidays in my country, I'm a bit slow to think, LOL! |2013-11-02 17:36:45.137|\n|441864 |255744 |80389 |0 |CC BY-SA 3.0 |null |LOL was only an error syntax i use - instead of . because usually tar.gz are archived as name-version.tar.gz ,sendmail use name.version.tar.gz add your comment as answer so i can vote |2016-01-16 19:45:36.237|\n|440175 |254907 |83246 |1 |CC BY-SA 3.0 |null |You beat me to it :) LOL. |2016-01-12 16:18:19.323|\n|398863 |233466 |23692 |0 |CC BY-SA 3.0 |null |LOL. I see you solved it that way a couple of seconds before I suggested it. Nice. |2015-10-02 11:15:44.963|\n|225093 |139434 |60539 |0 |CC BY-SA 3.0 |null |I should do that, and will when I get around to it, may not be right this instant or today, but in the next few days at the latest, I will. LOL, \"Please rerun the make command\" for each \"module\" if you will, does seem rather ridiculous. The error was always, something along the lines of \"... make file is out of date with respect to Makefile.PL\" Maybe a timestamp issue with respect to file creation? I did notice that `date` always returned UTC instead of CDT, although the UTC time was correct +/- a minute. |2014-06-27 21:48:55.157|\n+-------+-------+-------+-----+---------------+-----------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----------------------+\n\n" ] ], [ [ "# Stopping Spark", "_____no_output_____" ] ], [ [ "spark.stop()", "_____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", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "from selenium import webdriver\nfrom selenium.webdriver.common.keys import Keys\nfrom PIL import Image\nimport time\nimport numpy as np\nimport pytesseract as ocr\nimport cv2", "_____no_output_____" ], [ "driver = webdriver.Chrome('/home/lucas.cardoso/fastfinger-tensorflow/chromedriver_linux64/chromedriver')\ndriver.get('https://10fastfingers.com/login')", "_____no_output_____" ], [ "login = input(\"Email: \")\npassword = input (\"Password: \")", "_____no_output_____" ], [ "driver.find_element('id', 'UserEmail').send_keys(login)\ndriver.find_element('id', 'UserPassword').send_keys(password)\ndriver.find_element('id', 'login-form-submit').click()", "_____no_output_____" ], [ "while True:\n try:\n highlightWord = driver.find_elements_by_class_name(\"highlight\")[0] \n except:\n break\n driver.find_element('id','inputfield').send_keys(highlightWord.text + \" \")\n time.sleep(0.075)", "_____no_output_____" ], [ "driver.get(\"https://10fastfingers.com/anticheat/view/1/1\")\ndriver.find_element('id', 'start-btn').click()\ndriver.set_window_size(1050, 1000)\nprint (driver.get_window_size())", "_____no_output_____" ], [ "driver.save_screenshot(\"screenshot.png\")", "_____no_output_____" ], [ "from PIL import Image\nimageObject = Image.open(\"screenshot.png\")\ncropped = imageObject.crop((300,190,880,320))\ncropped.save(\"screenshot.png\")", "_____no_output_____" ], [ "imagem = Image.open('screenshot.png').convert('RGB')\nnpimagem = np.asarray(imagem).astype(np.uint8) \nnpimagem[:, :, 0] = 0\nnpimagem[:, :, 2] = 0\nim = cv2.cvtColor(npimagem, cv2.COLOR_RGB2GRAY) \n\nret, thresh = cv2.threshold(im, 127, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU) \n\nbinimagem = Image.fromarray(thresh) \n\nphrase = ocr.image_to_string(binimagem, lang='por')\nprint(phrase) \n", "_____no_output_____" ], [ "textarea = driver.find_element('id', 'word-input')\ntextarea.send_keys (phrase)\ndriver.find_element('id','submit-anticheat').click()", "_____no_output_____" ] ] ]

[ "code" ]

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Final Project\n\n## Predict whether a mammogram mass is benign or malignant\n\nWe'll be using the \"mammographic masses\" public dataset from the UCI repository (source: https://archive.ics.uci.edu/ml/datasets/Mammographic+Mass)\n\nThis data contains 961 instances of masses detected in mammograms, and contains the following attributes:\n\n\n 1. BI-RADS assessment: 1 to 5 (ordinal) \n 2. Age: patient's age in years (integer)\n 3. Shape: mass shape: round=1 oval=2 lobular=3 irregular=4 (nominal)\n 4. Margin: mass margin: circumscribed=1 microlobulated=2 obscured=3 ill-defined=4 spiculated=5 (nominal)\n 5. Density: mass density high=1 iso=2 low=3 fat-containing=4 (ordinal)\n 6. Severity: benign=0 or malignant=1 (binominal)\n \nBI-RADS is an assesment of how confident the severity classification is; it is not a \"predictive\" attribute and so we will discard it. The age, shape, margin, and density attributes are the features that we will build our model with, and \"severity\" is the classification we will attempt to predict based on those attributes.\n\nAlthough \"shape\" and \"margin\" are nominal data types, which sklearn typically doesn't deal with well, they are close enough to ordinal that we shouldn't just discard them. The \"shape\" for example is ordered increasingly from round to irregular.\n\nA lot of unnecessary anguish and surgery arises from false positives arising from mammogram results. If we can build a better way to interpret them through supervised machine learning, it could improve a lot of lives.\n\n## Your assignment\n\nApply several different supervised machine learning techniques to this data set, and see which one yields the highest accuracy as measured with K-Fold cross validation (K=10). Apply:\n\n* Decision tree\n* Random forest\n* KNN\n* Naive Bayes\n* SVM\n* Logistic Regression\n* And, as a bonus challenge, a neural network using Keras.\n\nThe data needs to be cleaned; many rows contain missing data, and there may be erroneous data identifiable as outliers as well.\n\nRemember some techniques such as SVM also require the input data to be normalized first.\n\nMany techniques also have \"hyperparameters\" that need to be tuned. Once you identify a promising approach, see if you can make it even better by tuning its hyperparameters.\n\nI was able to achieve over 80% accuracy - can you beat that?\n\nBelow I've set up an outline of a notebook for this project, with some guidance and hints. If you're up for a real challenge, try doing this project from scratch in a new, clean notebook!\n", "_____no_output_____" ], [ "## Let's begin: prepare your data\n\nStart by importing the mammographic_masses.data.txt file into a Pandas dataframe (hint: use read_csv) and take a look at it.", "_____no_output_____" ], [ "Make sure you use the optional parmaters in read_csv to convert missing data (indicated by a ?) into NaN, and to add the appropriate column names (BI_RADS, age, shape, margin, density, and severity):", "_____no_output_____" ], [ "Evaluate whether the data needs cleaning; your model is only as good as the data it's given. Hint: use describe() on the dataframe.", "_____no_output_____" ], [ "There are quite a few missing values in the data set. Before we just drop every row that's missing data, let's make sure we don't bias our data in doing so. Does there appear to be any sort of correlation to what sort of data has missing fields? If there were, we'd have to try and go back and fill that data in.", "_____no_output_____" ], [ "If the missing data seems randomly distributed, go ahead and drop rows with missing data. Hint: use dropna().", "_____no_output_____" ], [ "Next you'll need to convert the Pandas dataframes into numpy arrays that can be used by scikit_learn. Create an array that extracts only the feature data we want to work with (age, shape, margin, and density) and another array that contains the classes (severity). You'll also need an array of the feature name labels.", "_____no_output_____" ], [ "Some of our models require the input data to be normalized, so go ahead and normalize the attribute data. Hint: use preprocessing.StandardScaler().", "_____no_output_____" ], [ "## Decision Trees\n\nBefore moving to K-Fold cross validation and random forests, start by creating a single train/test split of our data. Set aside 75% for training, and 25% for testing.", "_____no_output_____" ], [ "Now create a DecisionTreeClassifier and fit it to your training data.", "_____no_output_____" ], [ "Display the resulting decision tree.", "_____no_output_____" ], [ "Measure the accuracy of the resulting decision tree model using your test data.", "_____no_output_____" ], [ "Now instead of a single train/test split, use K-Fold cross validation to get a better measure of your model's accuracy (K=10). Hint: use model_selection.cross_val_score", "_____no_output_____" ], [ "Now try a RandomForestClassifier instead. Does it perform better?", "_____no_output_____" ], [ "## SVM\n\nNext try using svm.SVC with a linear kernel. How does it compare to the decision tree?", "_____no_output_____" ], [ "## KNN\nHow about K-Nearest-Neighbors? Hint: use neighbors.KNeighborsClassifier - it's a lot easier than implementing KNN from scratch like we did earlier in the course. Start with a K of 10. K is an example of a hyperparameter - a parameter on the model itself which may need to be tuned for best results on your particular data set.", "_____no_output_____" ], [ "Choosing K is tricky, so we can't discard KNN until we've tried different values of K. Write a for loop to run KNN with K values ranging from 1 to 50 and see if K makes a substantial difference. Make a note of the best performance you could get out of KNN.", "_____no_output_____" ], [ "## Naive Bayes\n\nNow try naive_bayes.MultinomialNB. How does its accuracy stack up? Hint: you'll need to use MinMaxScaler to get the features in the range MultinomialNB requires.", "_____no_output_____" ], [ "## Revisiting SVM\n\nsvm.SVC may perform differently with different kernels. The choice of kernel is an example of a \"hyperparamter.\" Try the rbf, sigmoid, and poly kernels and see what the best-performing kernel is. Do we have a new winner?", "_____no_output_____" ], [ "## Logistic Regression\n\nWe've tried all these fancy techniques, but fundamentally this is just a binary classification problem. Try Logisitic Regression, which is a simple way to tackling this sort of thing.", "_____no_output_____" ], [ "## Neural Networks\n\nAs a bonus challenge, let's see if an artificial neural network can do even better. You can use Keras to set up a neural network with 1 binary output neuron and see how it performs. Don't be afraid to run a large number of epochs to train the model if necessary.", "_____no_output_____" ], [ "## Do we have a winner?\n\nWhich model, and which choice of hyperparameters, performed the best? Feel free to share your results!", "_____no_output_____" ] ] ]

[ "markdown" ]

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

[ "CC0-1.0" ]

[ "CC0-1.0" ]

[ "CC0-1.0" ]

[ [ [ "#Installing packages and loading them into the enviroment\ninstall.packages(\"MASS\")\nlibrary(\"MASS\")\ninstall.packages(\"missMDA\")\nlibrary(\"missMDA\")\ninstall.packages(\"tidyverse\")\nlibrary(\"tidyverse\")\ninstall.packages(\"caret\")\nlibrary(\"caret\")\ninstall.packages(\"mice\")\nlibrary(\"mice\")", "Installing package into 'C:/Users/daeda/OneDrive/Documents/R/win-library/3.6'\n(as 'lib' is unspecified)\n\nWarning message:\n\"package 'MASS' is in use and will not be installed\"\nInstalling package into 'C:/Users/daeda/OneDrive/Documents/R/win-library/3.6'\n(as 'lib' is unspecified)\n\nWarning message:\n\"package 'missMDA' is in use and will not be installed\"\nInstalling package into 'C:/Users/daeda/OneDrive/Documents/R/win-library/3.6'\n(as 'lib' is unspecified)\n\nWarning message:\n\"package 'tidyverse' is in use and will not be installed\"\nInstalling package into 'C:/Users/daeda/OneDrive/Documents/R/win-library/3.6'\n(as 'lib' is unspecified)\n\nWarning message:\n\"package 'caret' is in use and will not be installed\"\nInstalling package into 'C:/Users/daeda/OneDrive/Documents/R/win-library/3.6'\n(as 'lib' is unspecified)\n\nWarning message:\n\"package 'mice' is in use and will not be installed\"\n" ], [ "#Loading all needed files,dropping first two coloumns, which are not needed for analysis (ID, species)\nionomics <- read.csv('spec_shoot_xyz_combined.csv', colClasses =c(\"NULL\",\"NULL\",NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,\n NA,NA,NA,NA,NA,NA,NA))\n#Imputation via missMDA\nimputeIonomics <- imputePCA(ionomics, method = \"Regularized\")", "_____no_output_____" ], [ "#Imputation via Mice\ntemp <- mice(ionomics, m = 5, maxit = 5, method = \"norm.boot\", seed = 123)\nionomics <- complete(temp,1)", "\n iter imp variable\n 1 1 spec_as_int S34 As75\n 1 2 spec_as_int S34 As75\n 1 3 spec_as_int S34 As75\n 1 4 spec_as_int S34 As75\n 1 5 spec_as_int S34 As75\n 2 1 spec_as_int S34 As75\n 2 2 spec_as_int S34 As75\n 2 3 spec_as_int S34 As75\n 2 4 spec_as_int S34 As75\n 2 5 spec_as_int S34 As75\n 3 1 spec_as_int S34 As75\n 3 2 spec_as_int S34 As75\n 3 3 spec_as_int S34 As75\n 3 4 spec_as_int S34 As75\n 3 5 spec_as_int S34 As75\n 4 1 spec_as_int S34 As75\n 4 2 spec_as_int S34 As75\n 4 3 spec_as_int S34 As75\n 4 4 spec_as_int S34 As75\n 4 5 spec_as_int S34 As75\n 5 1 spec_as_int S34 As75\n 5 2 spec_as_int S34 As75\n 5 3 spec_as_int S34 As75\n 5 4 spec_as_int S34 As75\n 5 5 spec_as_int S34 As75\n" ], [ "# Split the data into training (80%) and test set (20%)\nset.seed(123)\ntraining.samples <- ionomics[,2] %>%\n createDataPartition(p = 0.8, list = FALSE)\ntrain.data <- ionomics[training.samples, ]\ntest.data <- ionomics[-training.samples, ]\n\ntraining.samples1 <- imputeIonomics$completeObs[,2] %>%\n createDataPartition(p = 0.8, list = FALSE)\ntrain.data1 <- imputeIonomics$completeObs[training.samples1, ]\ntest.data1 <- imputeIonomics$completeObs[-training.samples1, ]\ntrain.data1 <- unlist(train.data1)\ntest.data1 <- unlist(test.data1)\ntrain.data1 <- as.data.frame(train.data1)\ntest.data1 <- as.data.frame(test.data1)", "_____no_output_____" ], [ "# Fit the model\nmodel <- lda(spec_as_int~., data = train.data)\n# Make predictions\npredictions <- model %>% predict(test.data)\n# Model accuracy\nmean(predictions$class==test.data$spec_as_int)\n#model\n\nmodel1 <- lda(spec_as_int~., data = train.data1)\n# Make predictions\npredictions1 <- model1 %>% predict(test.data1)\n# Model accuracy\nmean(predictions1$class == test.data1$spec_as_int)\n#model1", "_____no_output_____" ], [ "# Predicted classes\nhead(predictions$class, 6)\n# Predicted probabilities of class memebership.\nhead(predictions$posterior, 6) \n# Linear discriminants\nhead(predictions$x, 3) ", "_____no_output_____" ] ], [ [ "|species|spec_as_int|\n|---|---|\n|acerifolia_x|1|\n|aestivalis_x|2|\n|cinerea_x|3|\n|labrusca_x|4|\n|palmata_x|5|\n|riparia_x|6|\n|rupestris_x|7|\n|vulpina_x|8|\n|acerifolia_y|9|\n|aestivalis_y|10|\n|cinerea_y|11|\n|labrusca_y|12|\n|palmata_y|13|\n|riparia_y|14|\n|rupestris_y|15|\n|vulpina_y|16|\nacerifolia_z|17|\n|aestivalis_z|18|\n|cinerea_z|19|\n|labrusca_z|20|\n|palmata_z|21|\n|riparia_z|22|\n|rupestris_z|23|\n|vulpina_z|24|", "_____no_output_____" ] ], [ [ "table <- table(Predicted=predictions$class, Species=test.data$spec_as_int)\nprint(confusionMatrix(table))", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# TensorFlow 2 - Keras Activations\nActivations can either be an **Activation** layer, or used through the **activation** arguement supported by all forward layers.", "_____no_output_____" ], [ "### softmax\nsoftmax activation function", "_____no_output_____" ], [ "### elu\nExponential Linear unit", "_____no_output_____" ], [ "### selu\nScaled Eponential Linear unit", "_____no_output_____" ], [ "### softplus\n\nSoftplus activation function\n\n", "_____no_output_____" ], [ "### softsign\nSoftsign activation function", "_____no_output_____" ], [ "### relu\nRectified Linear unit", "_____no_output_____" ], [ "### tahn\nHyperbolic tangent activation function", "_____no_output_____" ], [ "### sigmoid\nSigmoid activation function", "_____no_output_____" ], [ "### hard_sigmoid\nHard Sigmoid activation function", "_____no_output_____" ] ] ]

[ "markdown" ]

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Densely Connected Networks (DenseNet)\n\nResNet significantly changed the view of how to parametrize the functions in deep networks. DenseNet is to some extent the logical extension of this. To understand how to arrive at it, let us take a small detour to theory. Recall the Taylor expansion for functions. For scalars it can be written as\n\n$$f(x) = f(0) + f'(x) x + \\frac{1}{2} f''(x) x^2 + \\frac{1}{6} f'''(x) x^3 + o(x^3).$$\n\n## Function Decomposition\n\nThe key point is that it decomposes the function into increasingly higher order terms. In a similar vein, ResNet decomposes functions into\n\n$$f(\\mathbf{x}) = \\mathbf{x} + g(\\mathbf{x}).$$\n\nThat is, ResNet decomposes $f$ into a simple linear term and a more complex\nnonlinear one. What if we want to go beyond two terms? A solution was proposed\nby :cite:`Huang.Liu.Van-Der-Maaten.ea.2017` in the form of\nDenseNet, an architecture that reported record performance on the ImageNet\ndataset.\n\n\n\n:label:`fig_densenet_block`\n\n\nAs shown in :numref:`fig_densenet_block`, the key difference between ResNet and DenseNet is that in the latter case outputs are *concatenated* rather than added. As a result we perform a mapping from $\\mathbf{x}$ to its values after applying an increasingly complex sequence of functions.\n\n$$\\mathbf{x} \\to \\left[\\mathbf{x}, f_1(\\mathbf{x}), f_2(\\mathbf{x}, f_1(\\mathbf{x})), f_3(\\mathbf{x}, f_1(\\mathbf{x}), f_2(\\mathbf{x}, f_1(\\mathbf{x})), \\ldots\\right].$$\n\nIn the end, all these functions are combined in an MLP to reduce the number of features again. In terms of implementation this is quite simple---rather than adding terms, we concatenate them. The name DenseNet arises from the fact that the dependency graph between variables becomes quite dense. The last layer of such a chain is densely connected to all previous layers. The main components that compose a DenseNet are dense blocks and transition layers. The former defines how the inputs and outputs are concatenated, while the latter controls the number of channels so that it is not too large. The dense connections are shown in :numref:`fig_densenet`.\n\n\n\n:label:`fig_densenet`\n\n\n\n## Dense Blocks\n\nDenseNet uses the modified \"batch normalization, activation, and convolution\"\narchitecture of ResNet (see the exercise in :numref:`sec_resnet`).\nFirst, we implement this architecture in the\n`conv_block` function.", "_____no_output_____" ] ], [ [ "%mavenRepo snapshots https://oss.sonatype.org/content/repositories/snapshots/\n\n%maven ai.djl:api:0.7.0-SNAPSHOT\n%maven ai.djl:model-zoo:0.7.0-SNAPSHOT\n%maven ai.djl:basicdataset:0.7.0-SNAPSHOT\n%maven org.slf4j:slf4j-api:1.7.26\n%maven org.slf4j:slf4j-simple:1.7.26\n \n%maven ai.djl.mxnet:mxnet-engine:0.7.0-SNAPSHOT\n%maven ai.djl.mxnet:mxnet-native-auto:1.7.0-a", "_____no_output_____" ], [ "%%loadFromPOM\n<dependency>\n <groupId>tech.tablesaw</groupId>\n <artifactId>tablesaw-jsplot</artifactId>\n <version>0.30.4</version>\n</dependency>", "_____no_output_____" ], [ "%load ../utils/plot-utils.ipynb\n%load ../utils/Training.java", "_____no_output_____" ], [ "import ai.djl.Model;\nimport ai.djl.basicdataset.FashionMnist;\nimport ai.djl.metric.Metrics;\nimport ai.djl.modality.cv.transform.Resize;\nimport ai.djl.modality.cv.transform.ToTensor;\nimport ai.djl.ndarray.NDArray;\nimport ai.djl.ndarray.NDArrays;\nimport ai.djl.ndarray.NDList;\nimport ai.djl.ndarray.NDManager;\nimport ai.djl.ndarray.types.DataType;\nimport ai.djl.ndarray.types.Shape;\nimport ai.djl.nn.*;\nimport ai.djl.nn.convolutional.Conv2d;\nimport ai.djl.nn.core.Linear;\nimport ai.djl.nn.norm.BatchNorm;\nimport ai.djl.nn.pooling.Pool;\nimport ai.djl.training.DefaultTrainingConfig;\nimport ai.djl.training.EasyTrain;\nimport ai.djl.training.ParameterStore;\nimport ai.djl.training.Trainer;\nimport ai.djl.training.dataset.ArrayDataset;\nimport ai.djl.training.dataset.Dataset;\nimport ai.djl.training.evaluator.Accuracy;\nimport ai.djl.training.initializer.XavierInitializer;\nimport ai.djl.training.listener.TrainingListener;\nimport ai.djl.training.loss.Loss;\nimport ai.djl.training.optimizer.Optimizer;\nimport ai.djl.training.optimizer.learningrate.LearningRateTracker;\nimport ai.djl.translate.Pipeline;\nimport ai.djl.util.PairList;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.HashMap;\nimport java.util.Map;\n\nimport tech.tablesaw.api.*;\nimport tech.tablesaw.plotly.api.*;\nimport tech.tablesaw.plotly.components.*;\nimport tech.tablesaw.plotly.Plot;\nimport tech.tablesaw.plotly.components.Figure;\nimport org.apache.commons.lang3.ArrayUtils;", "_____no_output_____" ], [ "public SequentialBlock convBlock(int numChannels) {\n\n SequentialBlock block = new SequentialBlock()\n .add(BatchNorm.builder().build())\n .add(Activation::relu)\n .add(Conv2d.builder()\n .setFilters(numChannels)\n .setKernelShape(new Shape(3, 3))\n .optPadding(new Shape(1, 1))\n .optStride(new Shape(1, 1))\n .build()\n );\n\n return block;\n}", "_____no_output_____" ] ], [ [ "A dense block consists of multiple `convBlock` units, each using the same number of output channels. In the forward computation, however, we concatenate the input and output of each block on the channel dimension.", "_____no_output_____" ] ], [ [ "class DenseBlock extends AbstractBlock {\n\n private static final byte VERSION = 2;\n public SequentialBlock net = new SequentialBlock();\n\n public DenseBlock(int numConvs, int numChannels) {\n super(VERSION);\n for (int i = 0; i < numConvs; i++) {\n this.net.add(\n addChildBlock(\"denseBlock\" + i, convBlock(numChannels))\n );\n }\n }\n\n @Override\n public String toString() {\n return \"DenseBlock()\";\n }\n\n @Override\n public NDList forward(ParameterStore parameterStore, NDList X, boolean training, PairList<String, Object> pairList) {\n\n NDArray Y;\n for (Block block : this.net.getChildren().values()) {\n Y = block.forward(parameterStore, X, training).singletonOrThrow();\n X = new NDList(NDArrays.concat(new NDList(X.singletonOrThrow(), Y), 1));\n }\n return X;\n }\n\n @Override\n public Shape[] getOutputShapes(NDManager ndManager, Shape[] inputs) {\n Shape[] shapesX = inputs;\n for (Block block : this.net.getChildren().values()) {\n Shape[] shapesY = block.getOutputShapes(ndManager, shapesX);\n shapesX[0] = new Shape(\n shapesX[0].get(0),\n shapesY[0].get(1) + shapesX[0].get(1),\n shapesX[0].get(2),\n shapesX[0].get(3)\n );\n }\n return shapesX;\n }\n\n @Override\n public void initializeChildBlocks(NDManager manager, DataType dataType, Shape... inputShapes) {\n Shape shapesX = inputShapes[0];\n\n for (Block block : this.net.getChildren().values()) {\n Shape[] shapesY = block.initialize(manager, DataType.FLOAT32, shapesX);\n shapesX = new Shape(\n shapesX.get(0),\n shapesY[0].get(1) + shapesX.get(1),\n shapesX.get(2),\n shapesX.get(3)\n );\n }\n }\n}", "_____no_output_____" ] ], [ [ "In the following example, we define a convolution block (`DenseBlock`) with two blocks of 10 output channels. When using an input with 3 channels, we will get an output with the $3+2\\times 10=23$ channels. The number of convolution block channels controls the increase in the number of output channels relative to the number of input channels. This is also referred to as the growth rate.", "_____no_output_____" ] ], [ [ "NDManager manager = NDManager.newBaseManager();\nSequentialBlock block = new SequentialBlock()\n .add(new DenseBlock(2, 10));\n\nNDArray X = manager.randomUniform(0f, 1.0f, new Shape(4, 3, 8, 8));\n\nblock.setInitializer(new XavierInitializer());\nblock.initialize(manager, DataType.FLOAT32, X.getShape());\n\nParameterStore parameterStore = new ParameterStore(manager, true);\n\nShape currentShape = X.getShape();\n\nfor (int i = 0; i < block.getChildren().size(); i++) {\n\n Shape[] newShape = block.getChildren().get(i).getValue().getOutputShapes(manager, new Shape[]{X.getShape()});\n currentShape = newShape[0]; \n}\n\ncurrentShape", "_____no_output_____" ] ], [ [ "## Transition Layers\n\nSince each dense block will increase the number of channels, adding too many of them will lead to an excessively complex model. A transition layer is used to control the complexity of the model. It reduces the number of channels by using the $1\\times 1$ convolutional layer and halves the height and width of the average pooling layer with a stride of 2, further reducing the complexity of the model.", "_____no_output_____" ] ], [ [ "public SequentialBlock transitionBlock(int numChannels) {\n SequentialBlock blk = new SequentialBlock()\n .add(BatchNorm.builder().build())\n .add(Activation::relu)\n .add(\n Conv2d.builder()\n .setFilters(numChannels)\n .setKernelShape(new Shape(1, 1))\n .optStride(new Shape(1, 1))\n .build()\n )\n .add(Pool.avgPool2dBlock(new Shape(2, 2), new Shape(2, 2)));\n\n return blk;\n}", "_____no_output_____" ] ], [ [ "Apply a transition layer with 10 channels to the output of the dense block in the previous example. This reduces the number of output channels to 10, and halves the height and width.", "_____no_output_____" ] ], [ [ "block = transitionBlock(10);\n\nblock.setInitializer(new XavierInitializer());\nblock.initialize(manager, DataType.FLOAT32, currentShape);\n\nfor (int i = 0; i < block.getChildren().size(); i++) {\n\n Shape[] newShape = block.getChildren().get(i).getValue().getOutputShapes(manager, new Shape[]{currentShape});\n currentShape = newShape[0];\n}\n\ncurrentShape", "_____no_output_____" ] ], [ [ "## DenseNet Model\n\nNext, we will construct a DenseNet model. DenseNet first uses the same single convolutional layer and maximum pooling layer as ResNet.", "_____no_output_____" ] ], [ [ "SequentialBlock net = new SequentialBlock()\n .add(Conv2d.builder()\n .setFilters(64)\n .setKernelShape(new Shape(7, 7))\n .optStride(new Shape(2, 2))\n .optPadding(new Shape(3, 3))\n .build())\n .add(BatchNorm.builder().build())\n .add(Activation::relu)\n .add(Pool.maxPool2dBlock(new Shape(3, 3), new Shape(2, 2), new Shape(1, 1)));", "_____no_output_____" ] ], [ [ "Then, similar to the four residual blocks that ResNet uses, DenseNet uses four dense blocks. Similar to ResNet, we can set the number of convolutional layers used in each dense block. Here, we set it to 4, consistent with the ResNet-18 in the previous section. Furthermore, we set the number of channels (i.e., growth rate) for the convolutional layers in the dense block to 32, so 128 channels will be added to each dense block.\n\nIn ResNet, the height and width are reduced between each module by a residual block with a stride of 2. Here, we use the transition layer to halve the height and width and halve the number of channels.", "_____no_output_____" ] ], [ [ "int numChannels = 64;\nint growthRate = 32;\n\nint[] numConvsInDenseBlocks = new int[]{4, 4, 4, 4};\n\nfor (int index = 0; index < numConvsInDenseBlocks.length; index++) {\n\n int numConvs = numConvsInDenseBlocks[index];\n net.add(new DenseBlock(numConvs, growthRate));\n\n numChannels += (numConvs * growthRate);\n\n if (index != (numConvsInDenseBlocks.length - 1)) {\n numChannels = (numChannels / 2);\n net.add(transitionBlock(numChannels));\n }\n}", "_____no_output_____" ] ], [ [ "Similar to ResNet, a global pooling layer and fully connected layer are connected at the end to produce the output.", "_____no_output_____" ] ], [ [ "net\n .add(BatchNorm.builder().build())\n .add(Activation::relu)\n .add(Pool.globalAvgPool2dBlock())\n .add(Linear.builder().setUnits(10).build());", "_____no_output_____" ] ], [ [ "## Data Acquisition and Training\n\nSince we are using a deeper network here, in this section, we will reduce the input height and width from 224 to 96 to simplify the computation.", "_____no_output_____" ] ], [ [ "int batchSize = 256;\nfloat lr = 0.1f;\nint numEpochs = 10;\n\ndouble[] trainLoss;\ndouble[] testAccuracy;\ndouble[] epochCount;\ndouble[] trainAccuracy;\n\nepochCount = new double[numEpochs];\n\nfor (int i = 0; i < epochCount.length; i++) {\n epochCount[i] = (i + 1);\n}\n\nFashionMnist trainIter =\n FashionMnist.builder()\n .optPipeline(new Pipeline().add(new Resize(96)).add(new ToTensor()))\n .optUsage(Dataset.Usage.TRAIN)\n .setSampling(batchSize, true)\n .build();\n\nFashionMnist testIter =\n FashionMnist.builder()\n .optPipeline(new Pipeline().add(new Resize(96)).add(new ToTensor()))\n .optUsage(Dataset.Usage.TEST)\n .setSampling(batchSize, true)\n .build();\n\ntrainIter.prepare();\ntestIter.prepare();\n\nModel model = Model.newInstance(\"cnn\");\nmodel.setBlock(net);\n\nLoss loss = Loss.softmaxCrossEntropyLoss();\n\nLearningRateTracker lrt = LearningRateTracker.fixedLearningRate(lr);\nOptimizer sgd = Optimizer.sgd().setLearningRateTracker(lrt).build();\n\nDefaultTrainingConfig config = new DefaultTrainingConfig(loss).optOptimizer(sgd) // Optimizer (loss function)\n .addEvaluator(new Accuracy()) // Model Accuracy\n .addTrainingListeners(TrainingListener.Defaults.logging()); // Logging\n\nTrainer trainer = model.newTrainer(config);\ntrainer.initialize(new Shape(1, 1, 96, 96));\n\nMap<String, double[]> evaluatorMetrics = new HashMap<>();\ndouble avgTrainTimePerEpoch = 0;", "_____no_output_____" ], [ "Training.trainingChapter6(trainIter, testIter, numEpochs, trainer, evaluatorMetrics, avgTrainTimePerEpoch);", "_____no_output_____" ], [ "trainLoss = evaluatorMetrics.get(\"train_epoch_SoftmaxCrossEntropyLoss\");\ntrainAccuracy = evaluatorMetrics.get(\"train_epoch_Accuracy\");\ntestAccuracy = evaluatorMetrics.get(\"validate_epoch_Accuracy\");\n\nSystem.out.printf(\"loss %.3f,\", trainLoss[numEpochs - 1]);\nSystem.out.printf(\" train acc %.3f,\", trainAccuracy[numEpochs - 1]);\nSystem.out.printf(\" test acc %.3f\\n\", testAccuracy[numEpochs - 1]);\nSystem.out.printf(\"%.1f examples/sec\", trainIter.size() / (avgTrainTimePerEpoch / Math.pow(10, 9)));\nSystem.out.println();", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ], [ [ "String[] lossLabel = new String[trainLoss.length + testAccuracy.length + trainAccuracy.length];\n\nArrays.fill(lossLabel, 0, trainLoss.length, \"train loss\");\nArrays.fill(lossLabel, trainAccuracy.length, trainLoss.length + trainAccuracy.length, \"train acc\");\nArrays.fill(lossLabel, trainLoss.length + trainAccuracy.length,\n trainLoss.length + testAccuracy.length + trainAccuracy.length, \"test acc\");\n\nTable data = Table.create(\"Data\").addColumns(\n DoubleColumn.create(\"epoch\", ArrayUtils.addAll(epochCount, ArrayUtils.addAll(epochCount, epochCount))),\n DoubleColumn.create(\"metrics\", ArrayUtils.addAll(trainLoss, ArrayUtils.addAll(trainAccuracy, testAccuracy))),\n StringColumn.create(\"lossLabel\", lossLabel)\n);\n\nrender(LinePlot.create(\"\", data, \"epoch\", \"metrics\", \"lossLabel\"),\"text/html\");", "_____no_output_____" ] ], [ [ "## Summary\n\n* In terms of cross-layer connections, unlike ResNet, where inputs and outputs are added together, DenseNet concatenates inputs and outputs on the channel dimension.\n* The main units that compose DenseNet are dense blocks and transition layers.\n* We need to keep the dimensionality under control when composing the network by adding transition layers that shrink the number of channels again.\n\n## Exercises\n\n1. Why do we use average pooling rather than max-pooling in the transition layer?\n1. One of the advantages mentioned in the DenseNet paper is that its model parameters are smaller than those of ResNet. Why is this the case?\n1. One problem for which DenseNet has been criticized is its high memory consumption.\n * Is this really the case? Try to change the input shape to $224\\times 224$ to see the actual (GPU) memory consumption.\n * Can you think of an alternative means of reducing the memory consumption? How would you need to change the framework?\n1. Implement the various DenseNet versions presented in Table 1 of :cite:`Huang.Liu.Van-Der-Maaten.ea.2017`.\n1. Why do we not need to concatenate terms if we are just interested in $\\mathbf{x}$ and $f(\\mathbf{x})$ for ResNet? Why do we need this for more than two layers in DenseNet?\n1. Design a DenseNet for fully connected networks and apply it to the Housing Price prediction task.", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "from owlready2 import *\nimport json\n\nonto = owlready2.get_ontology(\"ontology.owl\")\nonto.load()\nITO = onto.get_namespace(\"https://ai-strategies.org/ontology/\")", "_____no_output_____" ], [ "def class_to_dict(myclass):\n mydict = dict()\n if myclass.label[0]:\n mydict['name'] = myclass.label[0]\n if myclass.hasDefinition:\n mydict['description'] = myclass.hasDefinition\n if myclass.seeAlso:\n mydict['seeAlso'] = myclass.seeAlso\n if myclass.comment:\n mydict['comment'] = myclass.comment\n if myclass.has_input:\n mydict['has_input'] = myclass.has_input[0].label\n if myclass.has_output:\n mydict['has_output'] = myclass.has_output[0].label\n if myclass.hasExactSynonym:\n mydict['hasExactSynonym'] = myclass.hasExactSynonym\n if myclass.hasNarrowSynonym:\n mydict['hasNarrowSynonym'] = myclass.hasNarrowSynonym\n if myclass.hasBroadSynonym:\n mydict['hasBroadSynonym'] = myclass.hasBroadSynonym\n if myclass.refactor_comment:\n mydict['refactor_comment'] = myclass.refactor_comment\n if myclass.papers_with_code_id:\n mydict['papers_with_code_id'] = myclass.papers_with_code_id\n children = list()\n for subclass in myclass.subclasses():\n children.append(class_to_dict(subclass))\n if len(children) > 0:\n mydict['children'] = children\n return mydict", "_____no_output_____" ], [ "mydict = class_to_dict(ITO.ITO_00141)\nmydict", "_____no_output_____" ], [ "with open('NLP_class_hierarchy_21_12_2020.json', 'w', encoding='utf-8') as f:\n json.dump(mydict, f, indent=2)", "_____no_output_____" ] ] ]

[ "code" ]

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## 1. Meet Professor William Sharpe\n<p>An investment may make sense if we expect it to return more money than it costs. But returns are only part of the story because they are risky - there may be a range of possible outcomes. How does one compare different investments that may deliver similar results on average, but exhibit different levels of risks?</p>\n<p><img style=\"float: left ; margin: 5px 20px 5px 1px;\" width=\"200\" src=\"https://assets.datacamp.com/production/project_66/img/sharpe.jpeg\"></p>\n<p>Enter William Sharpe. He introduced the <a href=\"https://web.stanford.edu/~wfsharpe/art/sr/sr.htm\"><em>reward-to-variability ratio</em></a> in 1966 that soon came to be called the Sharpe Ratio. It compares the expected returns for two investment opportunities and calculates the additional return per unit of risk an investor could obtain by choosing one over the other. In particular, it looks at the difference in returns for two investments and compares the average difference to the standard deviation (as a measure of risk) of this difference. A higher Sharpe ratio means that the reward will be higher for a given amount of risk. It is common to compare a specific opportunity against a benchmark that represents an entire category of investments.</p>\n<p>The Sharpe ratio has been one of the most popular risk/return measures in finance, not least because it's so simple to use. It also helped that Professor Sharpe won a Nobel Memorial Prize in Economics in 1990 for his work on the capital asset pricing model (CAPM).</p>\n<p>The Sharpe ratio is usually calculated for a portfolio and uses the risk-free interest rate as benchmark. We will simplify our example and use stocks instead of a portfolio. We will also use a stock index as benchmark rather than the risk-free interest rate because both are readily available at daily frequencies and we do not have to get into converting interest rates from annual to daily frequency. Just keep in mind that you would run the same calculation with portfolio returns and your risk-free rate of choice, e.g, the <a href=\"https://fred.stlouisfed.org/series/TB3MS\">3-month Treasury Bill Rate</a>. </p>\n<p>So let's learn about the Sharpe ratio by calculating it for the stocks of the two tech giants Facebook and Amazon. As benchmark we'll use the S&amp;P 500 that measures the performance of the 500 largest stocks in the US. When we use a stock index instead of the risk-free rate, the result is called the Information Ratio and is used to benchmark the return on active portfolio management because it tells you how much more return for a given unit of risk your portfolio manager earned relative to just putting your money into a low-cost index fund.</p>", "_____no_output_____" ] ], [ [ "# Importing required modules\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\n\n# Settings to produce nice plots in a Jupyter notebook\nplt.style.use('fivethirtyeight')\n%matplotlib inline\n\n# Reading in the data\nstock_data = pd.read_csv('datasets/stock_data.csv', parse_dates=['Date'], index_col='Date').dropna()\nbenchmark_data = pd.read_csv('datasets/benchmark_data.csv', parse_dates=['Date'], index_col='Date').dropna()", "_____no_output_____" ] ], [ [ "## 2. A first glance at the data\n<p>Let's take a look the data to find out how many observations and variables we have at our disposal.</p>", "_____no_output_____" ] ], [ [ "# Display summary for stock_data\nprint('Stocks\\n')\ndisplay(stock_data.info())\ndisplay(stock_data.head())\n\n# Display summary for benchmark_data\nprint('\\nBenchmarks\\n')\ndisplay(benchmark_data.info())\ndisplay(benchmark_data.head())", "Stocks\n\n<class 'pandas.core.frame.DataFrame'>\nDatetimeIndex: 252 entries, 2016-01-04 to 2016-12-30\nData columns (total 2 columns):\nAmazon 252 non-null float64\nFacebook 252 non-null float64\ndtypes: float64(2)\nmemory usage: 5.9 KB\n" ] ], [ [ "## 3. Plot & summarize daily prices for Amazon and Facebook\n<p>Before we compare an investment in either Facebook or Amazon with the index of the 500 largest companies in the US, let's visualize the data, so we better understand what we're dealing with.</p>", "_____no_output_____" ] ], [ [ "# visualize the stock_data\nstock_data.plot(subplots=True, title='Stock Data')\n\n# summarize the stock_data\nstock_data.describe()\n", "_____no_output_____" ] ], [ [ "## 4. Visualize & summarize daily values for the S&P 500\n<p>Let's also take a closer look at the value of the S&amp;P 500, our benchmark.</p>", "_____no_output_____" ] ], [ [ "# plot the benchmark_data\nbenchmark_data.plot(title='S&P 500')\n\n\n# summarize the benchmark_data\nbenchmark_data.describe()\n", "_____no_output_____" ] ], [ [ "## 5. The inputs for the Sharpe Ratio: Starting with Daily Stock Returns\n<p>The Sharpe Ratio uses the difference in returns between the two investment opportunities under consideration.</p>\n<p>However, our data show the historical value of each investment, not the return. To calculate the return, we need to calculate the percentage change in value from one day to the next. We'll also take a look at the summary statistics because these will become our inputs as we calculate the Sharpe Ratio. Can you already guess the result?</p>", "_____no_output_____" ] ], [ [ "# calculate daily stock_data returns\nstock_returns = stock_data.pct_change()\n\n# plot the daily returns\nstock_returns.plot()\n\n# summarize the daily returns\nstock_returns.describe()", "_____no_output_____" ] ], [ [ "## 6. Daily S&P 500 returns\n<p>For the S&amp;P 500, calculating daily returns works just the same way, we just need to make sure we select it as a <code>Series</code> using single brackets <code>[]</code> and not as a <code>DataFrame</code> to facilitate the calculations in the next step.</p>", "_____no_output_____" ] ], [ [ "# calculate daily benchmark_data returns\nsp_returns = benchmark_data['S&P 500'].pct_change()\n\n# plot the daily returns\nsp_returns.plot()\n\n# summarize the daily returns\nsp_returns.describe()", "_____no_output_____" ] ], [ [ "## 7. Calculating Excess Returns for Amazon and Facebook vs. S&P 500\n<p>Next, we need to calculate the relative performance of stocks vs. the S&amp;P 500 benchmark. This is calculated as the difference in returns between <code>stock_returns</code> and <code>sp_returns</code> for each day.</p>", "_____no_output_____" ] ], [ [ "# calculate the difference in daily returns\nexcess_returns = stock_returns.sub(sp_returns, axis=0)\n\n# plot the excess_returns\nexcess_returns.plot()\n\n# summarize the excess_returns\nexcess_returns.describe()", "_____no_output_____" ] ], [ [ "## 8. The Sharpe Ratio, Step 1: The Average Difference in Daily Returns Stocks vs S&P 500\n<p>Now we can finally start computing the Sharpe Ratio. First we need to calculate the average of the <code>excess_returns</code>. This tells us how much more or less the investment yields per day compared to the benchmark.</p>", "_____no_output_____" ] ], [ [ "# calculate the mean of excess_returns \n\navg_excess_return = excess_returns.mean()\n\n# plot avg_excess_returns\navg_excess_return.plot.bar()", "_____no_output_____" ] ], [ [ "## 9. The Sharpe Ratio, Step 2: Standard Deviation of the Return Difference\n<p>It looks like there was quite a bit of a difference between average daily returns for Amazon and Facebook.</p>\n<p>Next, we calculate the standard deviation of the <code>excess_returns</code>. This shows us the amount of risk an investment in the stocks implies as compared to an investment in the S&amp;P 500.</p>", "_____no_output_____" ] ], [ [ "# calculate the standard deviations\nsd_excess_return = excess_returns.std()\n\n# plot the standard deviations\nsd_excess_return.plot.bar()", "_____no_output_____" ] ], [ [ "## 10. Putting it all together\n<p>Now we just need to compute the ratio of <code>avg_excess_returns</code> and <code>sd_excess_returns</code>. The result is now finally the <em>Sharpe ratio</em> and indicates how much more (or less) return the investment opportunity under consideration yields per unit of risk.</p>\n<p>The Sharpe Ratio is often <em>annualized</em> by multiplying it by the square root of the number of periods. We have used daily data as input, so we'll use the square root of the number of trading days (5 days, 52 weeks, minus a few holidays): √252</p>", "_____no_output_____" ] ], [ [ "# calculate the daily sharpe ratio\ndaily_sharpe_ratio = avg_excess_return.div(sd_excess_return)\n\n# annualize the sharpe ratio\nannual_factor = np.sqrt(252)\nannual_sharpe_ratio = daily_sharpe_ratio.mul(annual_factor)\n\n# plot the annualized sharpe ratio\nannual_sharpe_ratio.plot.bar(title='Annualized Sharpe Ration: Stocks vs S&P 500')\n", "_____no_output_____" ] ], [ [ "## 11. Conclusion\n<p>Given the two Sharpe ratios, which investment should we go for? In 2016, Amazon had a Sharpe ratio twice as high as Facebook. This means that an investment in Amazon returned twice as much compared to the S&amp;P 500 for each unit of risk an investor would have assumed. In other words, in risk-adjusted terms, the investment in Amazon would have been more attractive.</p>\n<p>This difference was mostly driven by differences in return rather than risk between Amazon and Facebook. The risk of choosing Amazon over FB (as measured by the standard deviation) was only slightly higher so that the higher Sharpe ratio for Amazon ends up higher mainly due to the higher average daily returns for Amazon. </p>\n<p>When faced with investment alternatives that offer both different returns and risks, the Sharpe Ratio helps to make a decision by adjusting the returns by the differences in risk and allows an investor to compare investment opportunities on equal terms, that is, on an 'apples-to-apples' basis.</p>", "_____no_output_____" ] ], [ [ "# Uncomment your choice.\nbuy_amazon = True\n# buy_facebook = True", "_____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" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "### Global Validation ###\n\nThis notebook combines several validation notebooks: `global_validation_tasmax_v2.ipynb` and `global_validation_dtr_v2.ipynb` along with `check_aiqpd_downscaled_data.ipynb` to create a \"master\" global validation notebook. It also borrows validation code from the ERA-5 workflow, `validate_era5_hourlyORdaily_files.ipynb`. Parts of this will be incorporated into `dodola` such that minimal global validation is done automatically with every pipeline run. \n\n### Data Sources ###\n\nCoarse Resolution: \n- CMIP6 \n- Bias corrected data \n- ERA-5\n\nFine Resolution: \n- Bias corrected data \n- Downscaled data \n- ERA-5 (fine resolution)\n- ERA-5 (coarse resolution resampled to fine resolution) \n\n### Types of Validation ### \n\nBasic: \n- maxes, means, mins \n - CMIP6, bias corrected and downscaled (data)\n - historical, 2020-2040, 2040-2060, 2060-2080, 2080-2100 \n - 3 x 5 (in case we plot each data source separately in the pipeline) \n- NaN check, ranges, number of timesteps, file metadata attributes, variable names \n- spread between SSPs (this will not be a plot because it involves multiple runs)\n- differences between historical and future time periods for bias corrected and downscaled\n- differences between bias corrected and downscaled data \n\nVariable-specific: \n- GMST\n- days over 95 (TO-DO)\n- max values (precip) \n- negative values (DTR, precip) \n- number of days/gridcells that have wet day frequency correction applied (TO-DO)\n- max # of consecutive dry days, highest precip amt over 5-day rolling window (TO-DO) ", "_____no_output_____" ] ], [ [ "%matplotlib inline \nimport xarray as xr\nimport numpy as np\nimport matplotlib.pyplot as plt\nfrom cartopy import config\nimport cartopy.crs as ccrs\nimport cartopy.feature as cfeature\nimport os \nfrom matplotlib import cm\n\nfrom matplotlib.backends.backend_pdf import PdfPages\n\nfrom validation import *", "_____no_output_____" ] ], [ [ "### Azure (or GCS) authentication ### ", "_____no_output_____" ] ], [ [ "# ! pip install adlfs", "_____no_output_____" ], [ "from adlfs import AzureBlobFileSystem\nfs_az = AzureBlobFileSystem(\n account_name='dc6',\n account_key='', \n client_id=os.environ.get(\"AZURE_CLIENT_ID\", None),\n client_secret=os.environ.get(\"AZURE_CLIENT_SECRET\", None),\n tenant_id=os.environ.get(\"AZURE_TENANT_ID\", None))", "_____no_output_____" ] ], [ [ "### Filepaths ###\n\n#### Output data ####", "_____no_output_____" ] ], [ [ "data_dict = {'coarse': {'cmip6': {'historical': 'scratch/biascorrectdownscale-bk6n8/biascorrectdownscale-bk6n8-858077599/out.zarr', \n 'ssp370': 'scratch/biascorrectdownscale-bk6n8/biascorrectdownscale-bk6n8-269778292/out.zarr'}, \n 'bias_corrected': {'historical': 'az://biascorrected-stage/CMIP/NOAA-GFDL/GFDL-ESM4/historical/r1i1p1f1/day/tasmax/gr1/v20210920214427.zarr', \n 'ssp370': 'az://biascorrected-stage/ScenarioMIP/NOAA-GFDL/GFDL-ESM4/ssp370/r1i1p1f1/day/tasmax/gr1/v20210920214427.zarr'}, \n 'ERA-5':'az://scratch/biascorrectdownscale-bk6n8/biascorrectdownscale-bk6n8-131793962/out.zarr'}, \n 'fine': {'bias_corrected': {'historical': 'az://scratch/biascorrectdownscale-bk6n8/biascorrectdownscale-bk6n8-1362934973/regridded.zarr', \n 'ssp370': 'az://scratch/biascorrectdownscale-bk6n8/biascorrectdownscale-bk6n8-377595554/regridded.zarr'}, \n 'downscaled': {'historical': 'az://downscaled-stage/CMIP/NOAA-GFDL/GFDL-ESM4/historical/r1i1p1f1/day/tasmax/gr1/v20210920214427.zarr', \n 'ssp370': 'az://downscaled-stage/ScenarioMIP/NOAA-GFDL/GFDL-ESM4/ssp370/r1i1p1f1/day/tasmax/gr1/v20210920214427.zarr'}, \n 'ERA-5_fine': 'az://scratch/biascorrectdownscale-bk6n8/biascorrectdownscale-bk6n8-491178896/rechunked.zarr', \n 'ERA-5_coarse': 'az://scratch/biascorrectdownscale-bk6n8/biascorrectdownscale-bk6n8-1213790070/rechunked.zarr'}}", "_____no_output_____" ] ], [ [ "### Variables ###\n\nPossible variables: `tasmax`, `tasmin`, `pr`, `dtr`. Default is `tasmax`. ", "_____no_output_____" ] ], [ [ "variable = 'tasmax'", "_____no_output_____" ] ], [ [ "### other data inputs ### ", "_____no_output_____" ] ], [ [ "units = {'tasmax': 'K', 'tasmin': 'K', 'dtr': 'K', 'pr': 'mm'}\nyears = {'hist': {'start_yr': '1995', 'end_yr': '2014'}, \n '2020_2040': {'start_yr': '2020', 'end_yr': '2040'}, \n '2040_2060': {'start_yr': '2040', 'end_yr': '2060'}, \n '2060_2080': {'start_yr': '2060', 'end_yr': '2080'}, \n '2080_2100': {'start_yr': '2080', 'end_yr': '2100'}}\nyears_test = {'hist': {'start_yr': '1995', 'end_yr': '2014'}, \n '2020_2040': {'start_yr': '2020', 'end_yr': '2040'}, \n '2040_2060': {'start_yr': '2040', 'end_yr': '2060'}}\npdf_location = '/home/jovyan'", "_____no_output_____" ] ], [ [ "### Validation ### ", "_____no_output_____" ] ], [ [ "pdf_list = []", "_____no_output_____" ] ], [ [ "### basic diagnostic plots: means and maxes ### ", "_____no_output_____" ], [ "bias corrected", "_____no_output_____" ] ], [ [ "# plot bias corrected \nstore_hist = fs_az.get_mapper(data_dict['coarse']['bias_corrected']['historical'], check=False)\nds_hist = xr.open_zarr(store_hist)\nstore_future = fs_az.get_mapper(data_dict['coarse']['bias_corrected']['ssp370'], check=False)\nds_future = xr.open_zarr(store_future)\n\nnew_pdf = plot_diagnostic_climo_periods(ds_hist, \n ds_future, \n 'ssp370', \n years, 'tasmax', 'max', 'bias_corrected', units, pdf_location, vmin=280, vmax=320)\npdf_list.append(new_pdf)", "_____no_output_____" ], [ "new_pdf = plot_diagnostic_climo_periods(ds_hist, \n ds_future, \n 'ssp370', \n years, 'tasmax', 'mean', 'bias_corrected', units, pdf_location, vmin=280, vmax=320)\npdf_list.append(new_pdf)", "_____no_output_____" ] ], [ [ "cmip6", "_____no_output_____" ] ], [ [ "# plot bias corrected \nstore_hist = fs_az.get_mapper(data_dict['coarse']['cmip6']['historical'], check=False)\nds_hist = xr.open_zarr(store_hist)\nstore_future = fs_az.get_mapper(data_dict['coarse']['cmip6']['ssp370'], check=False)\nds_future = xr.open_zarr(store_future)\n\nnew_pdf = plot_diagnostic_climo_periods(ds_hist, \n ds_future, \n 'ssp370', \n years, 'tasmax', 'max', 'cmip6', units, pdf_location, vmin=280, vmax=320)\npdf_list.append(new_pdf)", "_____no_output_____" ], [ "new_pdf = plot_diagnostic_climo_periods(ds_hist, \n ds_future, \n 'ssp370', \n years, 'tasmax', 'mean', 'cmip6', units, pdf_location, vmin=280, vmax=320)\npdf_list.append(new_pdf)", "_____no_output_____" ] ], [ [ "downscaled", "_____no_output_____" ] ], [ [ "# plot bias corrected \nstore_hist = fs_az.get_mapper(data_dict['fine']['downscaled']['historical'], check=False)\nds_hist = xr.open_zarr(store_hist)\nstore_future = fs_az.get_mapper(data_dict['fine']['downscaled']['ssp370'], check=False)\nds_future = xr.open_zarr(store_future)\n\nnew_pdf = plot_diagnostic_climo_periods(ds_hist, \n ds_future, \n 'ssp370', \n years, 'tasmax', 'max', 'downscaled', units, pdf_location, vmin=280, vmax=320)\npdf_list.append(new_pdf)", "_____no_output_____" ], [ "new_pdf = plot_diagnostic_climo_periods(ds_hist, \n ds_future, \n 'ssp370', \n years, 'tasmax', 'mean', 'downscaled', units, pdf_location, vmin=280, vmax=320)\npdf_list.append(new_pdf)", "_____no_output_____" ] ], [ [ "GMST", "_____no_output_____" ] ], [ [ "store_hist_cmip6 = fs_az.get_mapper(data_dict['coarse']['cmip6']['historical'], check=False)\nds_hist_cmip6 = xr.open_zarr(store_hist_cmip6)\nstore_future_cmip6 = fs_az.get_mapper(data_dict['coarse']['cmip6']['ssp370'], check=False)\nds_future_cmip6 = xr.open_zarr(store_future_cmip6)\n\nstore_hist_bc = fs_az.get_mapper(data_dict['coarse']['bias_corrected']['historical'], check=False)\nds_hist_bc = xr.open_zarr(store_hist_bc)\nstore_future_bc = fs_az.get_mapper(data_dict['coarse']['bias_corrected']['ssp370'], check=False)\nds_future_bc = xr.open_zarr(store_future_bc)", "_____no_output_____" ], [ "gmst_pdf = plot_gmst_diagnostic(ds_hist_cmip6, \n ds_future_cmip6, \n ds_hist_bc, \n ds_future_bc, \n pdf_location, \n variable='tasmax', ssp='370', ds_hist_downscaled=None, ds_fut_downscaled=None)", "_____no_output_____" ] ], [ [ "Basic validation of zarr stores (general and variable-specific) ", "_____no_output_____" ] ], [ [ "test_dataset_allvars(ds_future_bc, 'tasmax', 'bias_corrected', time_period=\"future\")", "_____no_output_____" ], [ "test_dataset_allvars(ds_future_cmip6, 'tasmax', 'cmip6', time_period=\"future\")", "_____no_output_____" ], [ "test_dataset_allvars(ds_hist_cmip6, 'tasmax', 'cmip6', time_period=\"hist\")", "_____no_output_____" ], [ "test_temp_range(ds_hist_cmip6, 'tasmax')", "_____no_output_____" ] ], [ [ "create difference plots bw bias corrected and downscaled as well as historical/future bias corrected and downscaled", "_____no_output_____" ] ], [ [ "store_hist_bc = fs_az.get_mapper(data_dict['fine']['bias_corrected']['historical'], check=False)\nds_hist_bc = xr.open_zarr(store_hist_bc)\nstore_hist_ds = fs_az.get_mapper(data_dict['fine']['downscaled']['historical'], check=False)\nds_hist_ds = xr.open_zarr(store_hist_ds)\n\nstore_fut_bc = fs_az.get_mapper(data_dict['fine']['bias_corrected']['ssp370'], check=False)\nds_fut_bc = xr.open_zarr(store_fut_bc)\nstore_fut_ds = fs_az.get_mapper(data_dict['fine']['downscaled']['ssp370'], check=False)\nds_fut_ds = xr.open_zarr(store_fut_ds)", "_____no_output_____" ], [ "time_period = '2080_2100'\nda_hist_bc = ds_hist_bc[variable].sel(time=slice(years['hist']['start_yr'], years['hist']['end_yr'])).mean('time').load()\nda_hist_ds = ds_hist_ds[variable].sel(time=slice(years['hist']['start_yr'], years['hist']['end_yr'])).mean('time').load()\n\nda_future_bc = ds_fut_bc[variable].sel(time=slice(years[time_period]['start_yr'], years[time_period]['end_yr'])).mean('time').load()\nda_future_ds = ds_fut_ds[variable].sel(time=slice(years[time_period]['start_yr'], years[time_period]['end_yr'])).mean('time').load()", "_____no_output_____" ], [ "pdf_fname = plot_bias_correction_downscale_differences(da_hist_bc, da_hist_ds, da_future_bc, \n da_future_ds, 'downscaled_minus_biascorrected', 'downscaled', pdf_location,'tasmax',\n ssp='ssp370', time_period='2080_2100')", "_____no_output_____" ] ], [ [ "merge validation pdfs created so far ", "_____no_output_____" ] ], [ [ "# ! pip install PyPDF2", "_____no_output_____" ], [ "# ! ls -lh /home/jovyan/*.pdf", "_____no_output_____" ], [ "pdf_list = ['/home/jovyan/global_mean_tasmax_370.pdf', \n '/home/jovyan/tasmax_max_bias_corrected.pdf', \n '/home/jovyan/tasmax_max_cmip6.pdf', \n '/home/jovyan/tasmax_max_downscaled.pdf', \n '/home/jovyan/tasmax_mean_bias_corrected.pdf',\n '/home/jovyan/tasmax_mean_cmip6.pdf',\n '/home/jovyan/tasmax_mean_downscaled.pdf']", "_____no_output_____" ], [ "merge_validation_pdfs(pdf_list, '/home/jovyan/test_validation.pdf')", "_____no_output_____" ] ], [ [ "Days over 95 degrees F/extreme precip metrics will be added later. ", "_____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" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# CSX46 - Class 19 - MCODE\n\nIn this notebook, we will analyze a simple graph (`test.dot`) and then the Krogran network using the MCODE community detection algorithm.", "_____no_output_____" ] ], [ [ "import pygraphviz\nimport igraph\nimport numpy\nimport pandas\nimport sys\nfrom collections import defaultdict", "_____no_output_____" ], [ "test_graph = FILL IN HERE\nnodes = test_graph.nodes()\nedges = FILL IN HERE\ntest_igraph = FILL IN HERE\ntest_igraph.summary()", "_____no_output_____" ], [ "igraph.drawing.plot(FILL IN HERE)", "_____no_output_____" ] ], [ [ "Function `mcode` takes a graph adjacency list `adj_list` and a float parameter `vwp` (vertex weight probability), and returns a list of cluster assignments (of length equal to the number of clusters). Original code from True Price at UNC Chapel Hill [link to original code](https://github.com/trueprice/python-graph-clustering/blob/master/src/mcode.py).", "_____no_output_____" ] ], [ [ "def mcode(adj_list, vwp):\n \n # Stage 1: Vertex Weighting\n N = len(adj_list)\n edges = [[]]*N\n weights = dict((v, 1.) for v in range(0,N))\n\n edges=defaultdict(set)\n for i in range(0,N):\n edges[i] = # MAKE A SET FROM adj_list[i]\n \n res_clusters = []\n\n for i,v in enumerate(edges):\n neighborhood = # union of set((v,)) and edges[v]\n # if node has only one neighbor, we know everything we need to know\n if len(neighborhood) <= 2: continue\n\n k = 1 \n while neighborhood:\n k_core = # copy neighborhood object\n invalid_nodes = True\n while invalid_nodes and neighborhood:\n invalid_nodes = set(\n n for n in neighborhood if len(edges[n] & neighborhood) <= k)\n # remove invalid_nodes from neighborhood\n #increment k by one\n # vertex weight = k-core number * density of k-core\n weights[v] = (k-1) * (sum(len(edges[n] & k_core) for n in k_core) / \n (2. * len(k_core)**2))\n\n # Stage 2: Molecular Complex Prediction\n unvisited = set(edges)\n num_clusters = 0\n for seed in sorted(weights, key=weights.get, reverse=True):\n if seed not in unvisited: continue\n\n cluster, frontier = set((seed,)), set((seed,))\n w = weights[seed] * vwp\n while frontier:\n cluster.update(frontier)\n # remove frontier from unvisited\n frontier_plus_neighbors = set.union(*(edges[n] for n in frontier))\n frontier = set( \n n for n in frontier_plus_neighbors & unvisited if weights[n] > w)\n\n # haircut: only keep 2-core complexes\n invalid_nodes = True\n while invalid_nodes and cluster:\n invalid_nodes = set(n for n in cluster if len(edges[n] & cluster) < 2)\n # remove invalid_nodes from cluster\n\n if cluster:\n # make a list from `cluster` and add that list to `res_clusters`\n num_clusters += 1\n\n return(res_clusters)", "_____no_output_____" ] ], [ [ "Run mcode on the adjacency list for your toy graph, with vwp=0.8. How many clusters did it find? Do the cluster memberships make sense?", "_____no_output_____" ], [ "Load the Krogan et al. network edge-list data as a Pandas data frame", "_____no_output_____" ] ], [ [ "edge_list = pandas.read_csv(\"shared/krogan.sif\",\n sep=\"\\t\", \n names=[\"protein1\",\"protein2\"])", "_____no_output_____" ] ], [ [ "Make an igraph graph and print its summary", "_____no_output_____" ] ], [ [ "krogan_graph = FILL IN HERE\nkrogan_graph.summary()", "_____no_output_____" ] ], [ [ "Run mcode on your graph with vwp=0.1", "_____no_output_____" ] ], [ [ "res = FILL IN HERE", "_____no_output_____" ] ], [ [ "Get the cluster sizes", "_____no_output_____" ] ], [ [ "FILL IN HERE", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Test Hypothesis by Simulating Statistics\n## Mini-Lab 1: Hypothesis Testing", "_____no_output_____" ], [ "Welcome to your next mini-lab! Go ahead an run the following cell to get started. You can do that by clicking on the cell and then clickcing `Run` on the top bar. You can also just press `Shift` + `Enter` to run the cell.", "_____no_output_____" ] ], [ [ "from datascience import *\nimport numpy as np\nimport otter\n\nimport matplotlib\n%matplotlib inline\nimport matplotlib.pyplot as plots\nplots.style.use('fivethirtyeight')\n\ngrader = otter.Notebook(\"m7_l1_tests\")", "_____no_output_____" ] ], [ [ "In the previous two labs we've analyzed some data regarding COVID-19 test cases. Let's continue to analyze this data, specifically _claims_ about this data. Once again, we'll be be using ficitious statistics from Blockeley University.\n\nLet's say that Blockeley data science faculty are looking at the spread of COVID-19 across the realm of Minecraft. We have very specific data about Blockeley and the rest of Cubefornia but other realms' data isn't as clear cut or detailed. Let's say that a neighboring village has been reporting a COVID-19 infection rate of 26%. Should we trust these numbers?\n\nRegardless of whether or not you believe these claims, the job of a data scientist is to definitively substantiate or disprove such claims with data. You have access to the test results of similar sized village nearby and come up with the brilliant idea of running a hypothesis test with this data. Let's go ahead and load it! Run te cell below to import this data. If you want to explore this data further, go ahead and group by both columns! An empty cell is provided for you to do this.", "_____no_output_____" ] ], [ [ "test_results = Table.read_table(\"../datasets/covid19_village_tests.csv\")\ntest_results.show(5)", "_____no_output_____" ], [ "...", "_____no_output_____" ] ], [ [ "From here we can formulate our **Null Hypothesis** and **Alternate Hypothesis** Our *null hypothesis* is that this village truly has a 26% infection rate amongst the populations. Our *alternate hypothesis* is that this village does not in actuality have a 26% infection rate - it's way too low. Now we need our test statistic. Since we're looking at the infection rate inthe population, our test statistic should be:\n\n$$\\text{Test Statistic} = \\frac{\\text{Number of Positive Cases}}{\\text{Total Number of Cases}}$$\n\nWe've started the function declaration for you. Go ahead and complete `percent_positive` to calculate this test statistic.\n\n*Note*: Check out `np.count_nonzero` and built-in `len` function! These should be helpful for you.", "_____no_output_____" ] ], [ [ "def proportion_positive(test_results):\n numerator = ...\n denominator = ...\n return numerator / denominator", "_____no_output_____" ], [ "grader.check(\"q1\")", "_____no_output_____" ] ], [ [ "If you grouped by `Village Number` before, you would realize that there are roughly 3000 tests per village. Let's now create functions that will randomly take 3000 tests from the `test_results` table and to apply our test statistic. Complete the `sample_population` and `apply_statistic` functions below!\n\nThe `sample_population` function will take a `population_table` that is a table with all the data we want and will return a new table that has been sampled from this `population_table`. Please note that `with_replacement` should be `False`.\n\nThe `apply_statistic` function will take in a `sample_table` which is the table full of samples taken from a population table, a `column_name` which is the name of the column containing the data of interest, and a `statistic_function` which will be the test statistic that we will use. This function will return the result of using the `statistic_function` on the `sample_table`.", "_____no_output_____" ] ], [ [ "def sample_population(population_table):\n sampled_population = ...\n return sampled_population\n\n\ndef apply_statistic(sample_table, column_name, statistic_function):\n return statistic_function(...)", "_____no_output_____" ], [ "grader.check(\"q2\")", "_____no_output_____" ] ], [ [ "Now for the simulation portion! Complete the for loop below and fill in a reasonable number for the `iterations` variable. The `iterations` variable will determine just how many random samples that we will take in order to test our hypotheses. There is also code that will visualize your simulation and give you data regarding your simulation vs. the null hypothesis.", "_____no_output_____" ] ], [ [ "# Simulation code below. Fill out this portion!\n\niterations = ...\nsimulations = make_array()\n\nfor iteration in np.arange(iterations):\n sample_table = ...\n test_statistic = ...\n simulations = np.append(simulations, test_statistic)\n \n\n# This code is to tell you what percentage of our simulations are at or below the null hypothesis\n# There's no need to fill anything out but it is good to understand what's going on!\n\nnull_hypothesis = 0.26\nnum_below = np.count_nonzero(simulations <= null_hypothesis) / iterations\nprint(f\"Out of the {iterations} simulations, roughly {round(num_below * 100, 2)}% of test statistics \" +\n f\"are less than our null hypothesis of a {null_hypothesis * 100}% infection rate.\")\n\n\n# This code is to graph your simulation data and where our null hypothesis lies\n# There's no need to fill anything out but it is good to understand what's going on!\n\n\nsimulation_table = Table().with_column(\"Simulated Test Statistics\", simulations)\nsimulation_table.hist(bins=20)\nplots.scatter(null_hypothesis, 0, color='red', s=30);", "_____no_output_____" ], [ "grader.check(\"q3\")", "_____no_output_____" ] ], [ [ "Given our hypothesis test, what can you conclude about the village that reports having a 26% COVID-19 infection rate? Has your hypothesis changed before? Do you now trust or distrust these numbers? And if you do distrust these numbers, what do you think went wrong in the reporting?", "_____no_output_____" ], [ "Congratulations on finishing! Run the next cell to make sure that you passed all of the test cases.", "_____no_output_____" ] ], [ [ "grader.check_all()", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nfrom matplotlib import pyplot as plt\nfrom IPython import display\n\nimport torch\nfrom torch.utils.data import Dataset, DataLoader\n\nfrom data import H5Dataset\nfrom models import FCN\nfrom oc import OptControl\nfrom torch_pendulum import Pendulum", "_____no_output_____" ] ], [ [ "Given running cost $g(x_t,u_t)$ and terminal cost $h(x_T)$ the finite horizon $(t=0 \\ldots T)$ optimal control problem seeks to find the optimal control, \n$$u^*_{1:T} = \\text{argmin}_{u_{1:T}} L(x_{1:T},u_{1:T})$$ \n$$u^*_{1:T} = \\text{argmin}_{u_{1:T}} h(x_T) + \\sum_{t=0}^T g(x_t,u_t)$$\nsubject to the dynamics constraint: $x_{t+1} = f(x_t,u_t)$.\n\nThis notebook provides a dirty, brute forcing solution to problems of this form, using the inverted pendulum as an example, and assuming dynamics are not know a-priori. First, we gather state, actions, next state pairs, and use these to train a surrogate neural network dynamics model, $x_{t+1} \\sim \\hat{f}(x_t,u_t)$, approximating the true dynamics $f$.\n\nWe'll then set up a shooting-based trajectory optimisation problem, rolling out using the surrgoate dynamics $\\hat{f}$ for a sequence of controls $u^*_{1:T}$, evaluate the cost, then take gradient steps to minimise this, adjusting the values of the control. We'll use pytorch and Adam to accomplish this. We'll do this in a continuous control setting, but note that this is a practically infeasible control strategy, because solving this sort of optimisation online (and with any convergence guarantees) within the bandwidth of an inverted pendulum is a stretch. ", "_____no_output_____" ] ], [ [ "# NN parameters\nNsamples = 10000\nepochs = 500\n\nlatent_dim = 1024\nbatch_size = 8\nlr = 3e-4\n\n# Torch environment wrapping gym pendulum\ntorch_env = Pendulum()\n\n# Test parameters\nNsteps = 100", "_____no_output_____" ], [ "# Set up model (fully connected neural network)\n\nmodel = FCN(latent_dim=latent_dim,d=torch_env.d,ud=torch_env.ud)\noptimizer = torch.optim.Adam(model.parameters(), lr=lr)", "_____no_output_____" ], [ "# Load previously trained model\nmodel.load_state_dict(torch.load('./fcn.npy'))", "_____no_output_____" ], [ "# Or gather some training data\nstates_, actions, states = torch_env.get_data(Nsamples)\n\ndset = H5Dataset(np.array(states_),np.array(actions),np.array(states))\nsampler = DataLoader(dset, batch_size=batch_size, shuffle=True)", "_____no_output_____" ], [ "# and train model\n\nlosses = []\nfor epoch in range(epochs):\n \n batch_losses = []\n for states_,actions,states in sampler:\n \n recon_x = model(states_,actions)\n loss = model.loss_fn(recon_x,states)\n\n optimizer.zero_grad()\n loss.backward()\n optimizer.step()\n \n batch_losses.append(loss.item())\n \n losses.append(np.mean(batch_losses))\n plt.cla()\n plt.semilogy(losses)\n \n display.clear_output(wait=True)\n display.display(plt.gcf())", "_____no_output_____" ], [ "torch.save(model.state_dict(),'./fcn.npy')", "_____no_output_____" ], [ "# Test model rollouts - looks reasonable\n\nstates = []\n_states = []\n\ns = torch_env.env.reset()\nstates.append(s)\n_states.append(s.copy())\nfor i in range(30):\n a = torch_env.env.action_space.sample()\n s,r,_,_ = torch_env.env.step(a) # take a random action\n states.append(s)\n \n # roll-out with model\n _s = model(torch.from_numpy(_states[-1]).float().reshape(1,-1),torch.from_numpy(a).float().reshape(1,-1))\n _states.append(_s.detach().numpy())\n \n plt.cla()\n plt.plot(np.array(states),'--')\n plt.plot(np.vstack(_states))\n \n display.clear_output(wait=True)\n display.display(plt.gcf())", "_____no_output_____" ], [ "# Set up optimal controller\ncontroller = OptControl(model.dynamics, torch_env.running_cost, torch_env.term_cost, u_dim=torch_env.ud, umax=torch_env.umax, horizon=30,lr=1e-1)\n\n# Uncomment to use true dynamics\n# controller = OptControl(torch_env.dynamics, torch_env.running_cost, torch_env.term_cost, u_dim=torch_env.ud, umax=torch_env.umax, horizon=30,lr=1e-1)\n\n# Test controller\nplt.figure(figsize=(15,5))\ns = torch_env.env.reset()\nfor i in range(Nsteps):\n \n u,states,cost,costs = controller.minimize(torch.from_numpy(s).reshape(1,-1).float(),Nsteps=5) #OC\n \n s,r,_,_ = torch_env.env.step(u[:,0].detach().numpy()) # take a random action\n \n torch_env.env.render()\n \n plt.clf()\n plt.subplot(1,3,1)\n plt.plot(u.detach().numpy().T,'--')\n plt.ylim(-2,2)\n plt.ylabel('Controls')\n plt.subplot(1,3,2)\n plt.plot(np.squeeze(torch.stack(states).detach().numpy()))\n plt.legend({'Tip x','Tip y','Velocity'})\n plt.ylim(-8,8)\n plt.subplot(1,3,3)\n plt.plot(np.squeeze(torch.stack(costs).detach().numpy()))\n plt.ylabel('Cost')\n plt.ylim(0,15)\n \n display.clear_output(wait=True)\n display.display(plt.gcf())", "_____no_output_____" ], [ "torch_env.env.close()", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Генерация заголовков научных статей: слабый baseline", "_____no_output_____" ], [ "Источник: https://github.com/bentrevett/pytorch-seq2seq", "_____no_output_____" ] ], [ [ "# Если Вы запускаете ноутбук на colab,\n# выполните следующие строчки, чтобы подгрузить библиотеку dlnlputils:\n\n# !git clone https://github.com/Samsung-IT-Academy/stepik-dl-nlp.git\n# import sys; sys.path.append('/content/stepik-dl-nlp')", "_____no_output_____" ], [ "import torch\nimport torch.nn as nn\nimport torch.optim as optim\nimport torch.nn.functional as F\nfrom torchtext.data import Field, BucketIterator\n\nimport matplotlib.pyplot as plt\nimport matplotlib.ticker as ticker\n\nimport spacy\n\nimport random\nimport math\nimport time", "_____no_output_____" ], [ "SEED = 1234\n\nrandom.seed(SEED)\ntorch.manual_seed(SEED)\ntorch.backends.cudnn.deterministic = True", "_____no_output_____" ], [ "# возможно, Вам потребуется предварительно загрузить модели SpaCy для английского языка\n# !python -m spacy download en\n\nspacy_en = spacy.load('en')", "_____no_output_____" ], [ "def tokenize(text):\n \"\"\"\n Tokenizes English text from a string into a list of strings (tokens)\n \"\"\"\n return [tok.text for tok in spacy_en.tokenizer(text) if not tok.text.isspace()]", "_____no_output_____" ], [ "from torchtext import data, vocab\n\ntokenizer = data.get_tokenizer('spacy')\nTEXT = Field(tokenize=tokenize,\n init_token = '<sos>', \n eos_token = '<eos>', \n include_lengths = True,\n lower = True)\n\n", "_____no_output_____" ], [ "%%time\ntrn_data_fields = [(\"src\", TEXT),\n (\"trg\", TEXT)]\n\ndataset = data.TabularDataset(\n path='datasets/train.csv',\n format='csv',\n skip_header=True,\n fields=trn_data_fields\n)\n\ntrain_data, valid_data, test_data = dataset.split(split_ratio=[0.98, 0.01, 0.01])", "_____no_output_____" ], [ "TEXT.build_vocab(train_data, min_freq = 7)\nprint(f\"Unique tokens in vocabulary: {len(TEXT.vocab)}\")", "_____no_output_____" ], [ "BATCH_SIZE = 32\n\ndevice = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n\ntrain_iterator, valid_iterator, test_iterator = BucketIterator.splits(\n (train_data, valid_data, test_data), \n batch_size = BATCH_SIZE,\n sort_within_batch = True,\n sort_key = lambda x : len(x.src),\n device = device)", "_____no_output_____" ], [ "class Encoder(nn.Module):\n def __init__(self, input_dim, emb_dim, enc_hid_dim, dec_hid_dim, dropout):\n super().__init__()\n \n self.embedding = nn.Embedding(input_dim, emb_dim)\n \n self.rnn = nn.GRU(emb_dim, enc_hid_dim, bidirectional = True)\n \n self.fc = nn.Linear(enc_hid_dim * 2, dec_hid_dim)\n \n self.dropout = nn.Dropout(dropout)\n \n def forward(self, src, src_len):\n \n #src = [src sent len, batch size]\n #src_len = [src sent len]\n \n embedded = self.dropout(self.embedding(src))\n \n #embedded = [src sent len, batch size, emb dim]\n \n packed_embedded = nn.utils.rnn.pack_padded_sequence(embedded, src_len)\n \n packed_outputs, hidden = self.rnn(packed_embedded)\n \n #packed_outputs is a packed sequence containing all hidden states\n #hidden is now from the final non-padded element in the batch\n \n outputs, _ = nn.utils.rnn.pad_packed_sequence(packed_outputs) \n \n #outputs is now a non-packed sequence, all hidden states obtained\n # when the input is a pad token are all zeros\n \n #outputs = [sent len, batch size, hid dim * num directions]\n #hidden = [n layers * num directions, batch size, hid dim]\n \n #hidden is stacked [forward_1, backward_1, forward_2, backward_2, ...]\n #outputs are always from the last layer\n \n #hidden [-2, :, : ] is the last of the forwards RNN \n #hidden [-1, :, : ] is the last of the backwards RNN\n \n #initial decoder hidden is final hidden state of the forwards and backwards \n # encoder RNNs fed through a linear layer\n hidden = torch.tanh(self.fc(torch.cat((hidden[-2,:,:], hidden[-1,:,:]), dim = 1)))\n \n #outputs = [sent len, batch size, enc hid dim * 2]\n #hidden = [batch size, dec hid dim]\n \n return outputs, hidden", "_____no_output_____" ], [ "class Attention(nn.Module):\n def __init__(self, enc_hid_dim, dec_hid_dim):\n super().__init__()\n \n self.attn = nn.Linear((enc_hid_dim * 2) + dec_hid_dim, dec_hid_dim)\n self.v = nn.Parameter(torch.rand(dec_hid_dim))\n \n def forward(self, hidden, encoder_outputs, mask):\n \n #hidden = [batch size, dec hid dim]\n #encoder_outputs = [src sent len, batch size, enc hid dim * 2]\n #mask = [batch size, src sent len]\n \n batch_size = encoder_outputs.shape[1]\n src_len = encoder_outputs.shape[0]\n \n #repeat encoder hidden state src_len times\n hidden = hidden.unsqueeze(1).repeat(1, src_len, 1)\n \n encoder_outputs = encoder_outputs.permute(1, 0, 2)\n \n #hidden = [batch size, src sent len, dec hid dim]\n #encoder_outputs = [batch size, src sent len, enc hid dim * 2]\n \n energy = torch.tanh(self.attn(torch.cat((hidden, encoder_outputs), dim = 2))) \n \n #energy = [batch size, src sent len, dec hid dim]\n \n energy = energy.permute(0, 2, 1)\n \n #energy = [batch size, dec hid dim, src sent len]\n \n #v = [dec hid dim]\n \n v = self.v.repeat(batch_size, 1).unsqueeze(1)\n \n #v = [batch size, 1, dec hid dim]\n \n attention = torch.bmm(v, energy).squeeze(1)\n \n #attention = [batch size, src sent len]\n \n attention = attention.masked_fill(mask == 0, -1e10)\n \n return F.softmax(attention, dim = 1)", "_____no_output_____" ], [ "class Decoder(nn.Module):\n def __init__(self, output_dim, emb_dim, enc_hid_dim, dec_hid_dim, dropout, attention):\n super().__init__()\n\n self.output_dim = output_dim\n self.attention = attention\n \n self.embedding = nn.Embedding(output_dim, emb_dim)\n \n self.rnn = nn.GRU((enc_hid_dim * 2) + emb_dim, dec_hid_dim)\n \n self.out = nn.Linear((enc_hid_dim * 2) + dec_hid_dim + emb_dim, output_dim)\n \n self.dropout = nn.Dropout(dropout)\n \n def forward(self, input, hidden, encoder_outputs, mask):\n \n #input = [batch size]\n #hidden = [batch size, dec hid dim]\n #encoder_outputs = [src sent len, batch size, enc hid dim * 2]\n #mask = [batch size, src sent len]\n \n input = input.unsqueeze(0)\n \n #input = [1, batch size]\n \n embedded = self.dropout(self.embedding(input))\n \n #embedded = [1, batch size, emb dim]\n \n a = self.attention(hidden, encoder_outputs, mask)\n \n #a = [batch size, src sent len]\n \n a = a.unsqueeze(1)\n \n #a = [batch size, 1, src sent len]\n \n encoder_outputs = encoder_outputs.permute(1, 0, 2)\n \n #encoder_outputs = [batch size, src sent len, enc hid dim * 2]\n \n weighted = torch.bmm(a, encoder_outputs)\n \n #weighted = [batch size, 1, enc hid dim * 2]\n \n weighted = weighted.permute(1, 0, 2)\n \n #weighted = [1, batch size, enc hid dim * 2]\n \n rnn_input = torch.cat((embedded, weighted), dim = 2)\n \n #rnn_input = [1, batch size, (enc hid dim * 2) + emb dim]\n \n output, hidden = self.rnn(rnn_input, hidden.unsqueeze(0))\n \n #output = [sent len, batch size, dec hid dim * n directions]\n #hidden = [n layers * n directions, batch size, dec hid dim]\n \n #sent len, n layers and n directions will always be 1 in this decoder, therefore:\n #output = [1, batch size, dec hid dim]\n #hidden = [1, batch size, dec hid dim]\n #this also means that output == hidden\n assert (output == hidden).all()\n \n embedded = embedded.squeeze(0)\n output = output.squeeze(0)\n weighted = weighted.squeeze(0)\n \n output = self.out(torch.cat((output, weighted, embedded), dim = 1))\n \n #output = [bsz, output dim]\n \n return output, hidden.squeeze(0), a.squeeze(1)", "_____no_output_____" ], [ "class Seq2Seq(nn.Module):\n def __init__(self, encoder, decoder, pad_idx, sos_idx, eos_idx, device):\n super().__init__()\n \n self.encoder = encoder\n self.decoder = decoder\n self.pad_idx = pad_idx\n self.sos_idx = sos_idx\n self.eos_idx = eos_idx\n self.device = device\n \n def create_mask(self, src):\n mask = (src != self.pad_idx).permute(1, 0)\n return mask\n \n def forward(self, src, src_len, trg, teacher_forcing_ratio = 0.5):\n \n #src = [src sent len, batch size]\n #src_len = [batch size]\n #trg = [trg sent len, batch size]\n #teacher_forcing_ratio is probability to use teacher forcing\n #e.g. if teacher_forcing_ratio is 0.75 we use teacher forcing 75% of the time\n \n if trg is None:\n assert teacher_forcing_ratio == 0, \"Must be zero during inference\"\n inference = True\n trg = torch.zeros((100, src.shape[1])).long().fill_(self.sos_idx).to(src.device)\n else:\n inference = False\n \n batch_size = src.shape[1]\n max_len = trg.shape[0]\n trg_vocab_size = self.decoder.output_dim\n \n #tensor to store decoder outputs\n outputs = torch.zeros(max_len, batch_size, trg_vocab_size).to(self.device)\n \n #tensor to store attention\n attentions = torch.zeros(max_len, batch_size, src.shape[0]).to(self.device)\n \n #encoder_outputs is all hidden states of the input sequence, back and forwards\n #hidden is the final forward and backward hidden states, passed through a linear layer\n encoder_outputs, hidden = self.encoder(src, src_len)\n \n #first input to the decoder is the <sos> tokens\n input = trg[0,:]\n \n mask = self.create_mask(src)\n \n #mask = [batch size, src sent len]\n \n for t in range(1, max_len):\n \n #insert input token embedding, previous hidden state, all encoder hidden states \n # and mask\n #receive output tensor (predictions), new hidden state and attention tensor\n output, hidden, attention = self.decoder(input, hidden, encoder_outputs, mask)\n \n #place predictions in a tensor holding predictions for each token\n outputs[t] = output\n \n #place attentions in a tensor holding attention value for each input token\n attentions[t] = attention\n \n #decide if we are going to use teacher forcing or not\n teacher_force = random.random() < teacher_forcing_ratio\n \n #get the highest predicted token from our predictions\n top1 = output.argmax(1) \n \n #if teacher forcing, use actual next token as next input\n #if not, use predicted token\n input = trg[t] if teacher_force else top1\n \n #if doing inference and next token/prediction is an eos token then stop\n if inference and input.item() == self.eos_idx:\n return outputs[:t], attentions[:t]\n \n return outputs, attentions", "_____no_output_____" ], [ "INPUT_DIM = len(TEXT.vocab)\nOUTPUT_DIM = len(TEXT.vocab)\nENC_EMB_DIM = 128\nDEC_EMB_DIM = 128\nENC_HID_DIM = 64\nDEC_HID_DIM = 64\nENC_DROPOUT = 0.8\nDEC_DROPOUT = 0.8\nPAD_IDX = TEXT.vocab.stoi['<pad>']\nSOS_IDX = TEXT.vocab.stoi['<sos>']\nEOS_IDX = TEXT.vocab.stoi['<eos>']\n\nattn = Attention(ENC_HID_DIM, DEC_HID_DIM)\nenc = Encoder(INPUT_DIM, ENC_EMB_DIM, ENC_HID_DIM, DEC_HID_DIM, ENC_DROPOUT)\ndec = Decoder(OUTPUT_DIM, DEC_EMB_DIM, ENC_HID_DIM, DEC_HID_DIM, DEC_DROPOUT, attn)\n\nmodel = Seq2Seq(enc, dec, PAD_IDX, SOS_IDX, EOS_IDX, device).to(device)", "_____no_output_____" ], [ "def init_weights(m):\n for name, param in m.named_parameters():\n if 'weight' in name:\n nn.init.normal_(param.data, mean=0, std=0.01)\n else:\n nn.init.constant_(param.data, 0)\n \nmodel.apply(init_weights)", "_____no_output_____" ], [ "def count_parameters(model):\n return sum(p.numel() for p in model.parameters() if p.requires_grad)\n\nprint(f'The model has {count_parameters(model):,} trainable parameters')", "_____no_output_____" ], [ "optimizer = optim.Adam(model.parameters())", "_____no_output_____" ], [ "criterion = nn.CrossEntropyLoss(ignore_index = PAD_IDX)", "_____no_output_____" ] ], [ [ "### Обучение модели", "_____no_output_____" ] ], [ [ "import matplotlib\nmatplotlib.rcParams.update({'figure.figsize': (16, 12), 'font.size': 14})\nimport matplotlib.pyplot as plt\n%matplotlib inline\nfrom IPython.display import clear_output\n\n\ndef train(model, iterator, optimizer, criterion, clip, train_history=None, valid_history=None):\n \n model.train()\n \n epoch_loss = 0\n history = []\n for i, batch in enumerate(iterator):\n \n src, src_len = batch.src\n trg, trg_len = batch.trg\n \n optimizer.zero_grad()\n \n output, attetion = model(src, src_len, trg)\n \n #trg = [trg sent len, batch size]\n #output = [trg sent len, batch size, output dim]\n \n output = output[1:].view(-1, output.shape[-1])\n trg = trg[1:].view(-1)\n \n #trg = [(trg sent len - 1) * batch size]\n #output = [(trg sent len - 1) * batch size, output dim]\n \n loss = criterion(output, trg)\n \n loss.backward()\n \n torch.nn.utils.clip_grad_norm_(model.parameters(), clip)\n \n optimizer.step()\n \n epoch_loss += loss.item()\n \n history.append(loss.cpu().data.numpy())\n if (i+1)%10==0:\n fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(12, 8))\n\n clear_output(True)\n ax[0].plot(history, label='train loss')\n ax[0].set_xlabel('Batch')\n ax[0].set_title('Train loss')\n if train_history is not None:\n ax[1].plot(train_history, label='general train history')\n ax[1].set_xlabel('Epoch')\n if valid_history is not None:\n ax[1].plot(valid_history, label='general valid history')\n plt.legend()\n \n plt.show()\n \n return epoch_loss / len(iterator)", "_____no_output_____" ], [ "def evaluate(model, iterator, criterion):\n \n model.eval()\n \n epoch_loss = 0\n \n with torch.no_grad():\n \n for i, batch in enumerate(iterator):\n\n src, src_len = batch.src\n trg, trg_len = batch.trg\n\n output, attention = model(src, src_len, trg, 0) #turn off teacher forcing\n\n #trg = [trg sent len, batch size]\n #output = [trg sent len, batch size, output dim]\n\n output = output[1:].view(-1, output.shape[-1])\n trg = trg[1:].view(-1)\n\n #trg = [(trg sent len - 1) * batch size]\n #output = [(trg sent len - 1) * batch size, output dim]\n\n loss = criterion(output, trg)\n\n epoch_loss += loss.item()\n \n return epoch_loss / len(iterator)", "_____no_output_____" ], [ "def epoch_time(start_time, end_time):\n elapsed_time = end_time - start_time\n elapsed_mins = int(elapsed_time / 60)\n elapsed_secs = int(elapsed_time - (elapsed_mins * 60))\n return elapsed_mins, elapsed_secs", "_____no_output_____" ], [ "MODEL_NAME = 'models/lstm_baseline.pt'\nN_EPOCHS = 5\nCLIP = 1\n\ntrain_history = []\nvalid_history = []\n\nbest_valid_loss = float('inf')\n\nfor epoch in range(N_EPOCHS):\n \n start_time = time.time()\n \n train_loss = train(model, train_iterator, optimizer, criterion, CLIP, train_history, valid_history)\n valid_loss = evaluate(model, valid_iterator, criterion)\n \n end_time = time.time()\n \n epoch_mins, epoch_secs = epoch_time(start_time, end_time)\n \n if valid_loss < best_valid_loss:\n best_valid_loss = valid_loss\n torch.save(model.state_dict(), MODEL_NAME)\n \n \n train_history.append(train_loss)\n valid_history.append(valid_loss)\n \n print(f'Epoch: {epoch+1:02} | Time: {epoch_mins}m {epoch_secs}s')\n print(f'\\tTrain Loss: {train_loss:.3f} | Train PPL: {math.exp(train_loss):7.3f}')\n print(f'\\t Val. Loss: {valid_loss:.3f} | Val. PPL: {math.exp(valid_loss):7.3f}')", "_____no_output_____" ] ], [ [ "Finally, we load the parameters from our best validation loss and get our results on the test set.", "_____no_output_____" ] ], [ [ "# for cpu usage\nmodel.load_state_dict(torch.load(MODEL_NAME, map_location=torch.device('cpu')))\n\n# for gpu usage\n# model.load_state_dict(torch.load(MODEL_NAME), map_location=torch.device('cpu'))\n\n\ntest_loss = evaluate(model, test_iterator, criterion)\n\nprint(f'| Test Loss: {test_loss:.3f} | Test PPL: {math.exp(test_loss):7.3f} |')", "_____no_output_____" ] ], [ [ "### Генерация заголовков", "_____no_output_____" ] ], [ [ "def translate_sentence(model, tokenized_sentence):\n model.eval()\n tokenized_sentence = ['<sos>'] + [t.lower() for t in tokenized_sentence] + ['<eos>']\n numericalized = [TEXT.vocab.stoi[t] for t in tokenized_sentence] \n sentence_length = torch.LongTensor([len(numericalized)]).to(device) \n tensor = torch.LongTensor(numericalized).unsqueeze(1).to(device) \n translation_tensor_logits, attention = model(tensor, sentence_length, None, 0) \n translation_tensor = torch.argmax(translation_tensor_logits.squeeze(1), 1)\n translation = [TEXT.vocab.itos[t] for t in translation_tensor]\n translation, attention = translation[1:], attention[1:]\n return translation, attention", "_____no_output_____" ], [ "def display_attention(sentence, translation, attention):\n \n fig = plt.figure(figsize=(30,50))\n ax = fig.add_subplot(111)\n \n attention = attention.squeeze(1).cpu().detach().numpy().T\n \n cax = ax.matshow(attention, cmap='bone')\n \n ax.tick_params(labelsize=12)\n ax.set_yticklabels(['']+['<sos>']+[t.lower() for t in sentence]+['<eos>'])\n ax.set_xticklabels(['']+translation, rotation=80)\n\n ax.xaxis.set_major_locator(ticker.MultipleLocator(1))\n ax.yaxis.set_major_locator(ticker.MultipleLocator(1))\n\n plt.show()\n plt.close()", "_____no_output_____" ], [ "example_idx = 100\n\nsrc = vars(train_data.examples[example_idx])['src']\ntrg = vars(train_data.examples[example_idx])['trg']\n\nprint(f'src = {src}')\nprint(f'trg = {trg}')", "_____no_output_____" ], [ "translation, attention = translate_sentence(model, src)\n\nprint(f'predicted trg = {translation}')", "_____no_output_____" ], [ "display_attention(src, translation, attention)", "_____no_output_____" ], [ "for example_idx in range(100):\n src = vars(test_data.examples[example_idx])['src']\n trg = vars(test_data.examples[example_idx])['trg']\n translation, attention = translate_sentence(model, src)\n\n print('Оригинальный заголовок: ', ' '.join(trg))\n print('Предсказанный заголовок: ', ' '.join(translation))\n print('-----------------------------------')", "_____no_output_____" ], [ "example_idx = 0\n\nsrc = vars(valid_data.examples[example_idx])['src']\ntrg = vars(valid_data.examples[example_idx])['trg']\n\nprint(f'src = {src}')\nprint(f'trg = {trg}')", "_____no_output_____" ], [ "translation, attention = translate_sentence(model, src)\n\nprint(f'predicted trg = {translation}')\n\ndisplay_attention(src, translation, attention)", "_____no_output_____" ], [ "example_idx = 510\n\nsrc = vars(test_data.examples[example_idx])['src']\ntrg = vars(test_data.examples[example_idx])['trg']\n\nprint(f'src = {src}')\nprint(f'trg = {trg}')", "_____no_output_____" ], [ "translation, attention = translate_sentence(model, src)\n\nprint(f'predicted trg = {translation}')\n\ndisplay_attention(src, translation, attention)", "_____no_output_____" ] ], [ [ "### Считаем BLEU на train.csv", "_____no_output_____" ] ], [ [ "import nltk\n\nn_gram_weights = [0.3334, 0.3333, 0.3333]", "_____no_output_____" ], [ "test_len = len(test_data)", "_____no_output_____" ], [ "original_texts = []\ngenerated_texts = []\nmacro_bleu = 0\n\nfor example_idx in range(test_len):\n src = vars(test_data.examples[example_idx])['src']\n trg = vars(test_data.examples[example_idx])['trg']\n translation, _ = translate_sentence(model, src)\n\n original_texts.append(trg)\n generated_texts.append(translation)\n\n bleu_score = nltk.translate.bleu_score.sentence_bleu(\n [trg],\n translation,\n weights = n_gram_weights\n ) \n macro_bleu += bleu_score\n\nmacro_bleu /= test_len", "_____no_output_____" ], [ "# averaging sentence-level BLEU (i.e. macro-average precision)\nprint('Macro-average BLEU (LSTM): {0:.5f}'.format(macro_bleu))", "_____no_output_____" ] ], [ [ "### Делаем submission в Kaggle", "_____no_output_____" ] ], [ [ "import pandas as pd\n\nsubmission_data = pd.read_csv('datasets/test.csv')\nabstracts = submission_data['abstract'].values", "_____no_output_____" ] ], [ [ "Генерация заголовков для тестовых данных:", "_____no_output_____" ] ], [ [ "titles = []\nfor abstract in abstracts:\n title, _ = translate_sentence(model, abstract.split())\n titles.append(' '.join(title).replace('<unk>', ''))", "_____no_output_____" ] ], [ [ "Записываем полученные заголовки в файл формата `<abstract>,<title>`:", "_____no_output_____" ] ], [ [ "submission_df = pd.DataFrame({'abstract': abstracts, 'title': titles})\nsubmission_df.to_csv('datasets/predicted_titles.csv', index=False)", "_____no_output_____" ] ], [ [ "С помощью скрипта `generate_csv` приводим файл `submission_prediction.csv` в формат, необходимый для посылки в соревнование на Kaggle:", "_____no_output_____" ] ], [ [ "from create_submission import generate_csv\n\ngenerate_csv('datasets/predicted_titles.csv', 'datasets/kaggle_pred.csv', 'datasets/vocs.pkl')", "_____no_output_____" ], [ "!wc -l datasets/kaggle_pred.csv", "_____no_output_____" ], [ "!head datasets/kaggle_pred.csv", "_____no_output_____" ] ] ]

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

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "import pandas as pd\nimport numpy as np\ntrain = pd.read_csv('train.csv')\ndev = pd.read_csv('dev.csv')\n# print(train+dev)\ntrain = train.iloc[:,:].values\ndev = dev.iloc[:,:].values\n# print(train)\n# print(dev)\ntrain = np.concatenate((train, dev),axis=0)\nprint(len(train))\ntrain = pd.DataFrame(train)\nprint(train)\ntrain.to_csv('totol_data.csv', index=False,header=True)", "10749\n 0 1 2 3 4\n0 0 咳血 剧烈运动后咯血,是怎么了? 剧烈运动后咯血是什么原因？ 1\n1 1 咳血 剧烈运动后咯血,是怎么了? 剧烈运动后为什么会咯血？ 1\n2 2 咳血 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，应该怎么处理？ 0\n3 3 咳血 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，需要就医吗？ 0\n4 4 咳血 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，是否很严重？ 0\n... ... .. ... ... ..\n10744 1997 哮喘 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘与过敏性哮喘一样吗？ 1\n10745 1998 哮喘 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘是否就是过敏性哮喘？ 1\n10746 1999 哮喘 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘的饮食禁忌有哪些？ 0\n10747 2000 哮喘 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘怎么治疗？ 0\n10748 2001 哮喘 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘能跑步吗？ 0\n\n[10749 rows x 5 columns]\n" ], [ "data = pd.read_csv('totol_data.csv')\nm,n=0,0\nfor i in data.index:\n if data.iloc[i,-1]==1:\n m+=1\n else:\n n+=1\nprint(m/n)", "0.6670285359801489\n" ], [ "data = pd.read_csv('total_data3.csv')\nimport os\nprint(data)\nfrom sklearn.model_selection import KFold\nkfold = KFold(n_splits=5, shuffle=True)\ni = 0\nos.system('cd fold5')\nfor train, test in kfold.split(data):\n print(\"%s %s\" % (train, test))\n print(len(train),len(test))\n train_data = data.loc[train]\n dev_data = data.loc[test]\n# print(train_data)\n try:\n os.system('mkdir data3_%d'%i)\n except:\n pass\n train_data.to_csv('data3_%d/train.csv'%i, index=False, header=True)\n dev_data.to_csv('data3_%d/dev.csv'%i, index=False, header=True)\n i += 1", " id query1 query2 label\n0 7114 干咳无痰吃什么水果恢复的快 干咳无痰并伴有哮鸣音怎么回事 0\n1 1375 肺气肿病人怎么治疗的？ 肺气肿病人最明显的表现有哪些呢? 0\n2 1946 肺炎会传染人吗 大叶性肺炎传染吗？ 0\n3 2012 感冒肺炎，住院三天医生说出院需要注意饮食 感冒肺炎，住院三天医生说出院后不要去人群密集的地方 0\n4 2211 严重的慢性肺炎传染严重吗？ 严重的慢性肺炎传染人吗？ 0\n... ... ... ... ...\n16134 3005 哮喘能否吃洋参？ 哮喘可不可以吃洋参？ 1\n16135 3006 过敏性哮喘患者的注意事项是什么？ 过敏性哮喘患者要注意的是什么？ 1\n16136 3007 过敏性哮喘平时应注意什么？ 患了过敏性哮喘平时需要注意哪些事？ 1\n16137 3008 如何治疗过敏性哮喘？ 过敏性哮喘要怎么治疗？ 1\n16138 3009 变应性哮喘与过敏性哮喘一样吗？ 变应性哮喘是否就是过敏性哮喘？ 1\n\n[16139 rows x 4 columns]\n[ 2 4 7 ... 16136 16137 16138] [ 0 1 3 ... 16127 16129 16130]\n12911 3228\n[ 0 1 3 ... 16135 16136 16138] [ 2 7 8 ... 16122 16126 16137]\n12911 3228\n[ 0 1 2 ... 16134 16136 16137] [ 10 11 20 ... 16123 16135 16138]\n12911 3228\n[ 0 1 2 ... 16135 16137 16138] [ 9 37 55 ... 16120 16124 16136]\n12911 3228\n[ 0 1 2 ... 16136 16137 16138] [ 4 12 17 ... 16132 16133 16134]\n12912 3227\n" ], [ "import pandas as pd\ndata = pd.read_csv('totol_data.csv')\ndata = data.drop(['category'],axis=1)\nsimilary = data[data['label']==1]\nunsimilary = data[data['label']==0]\nprint(similary)\nprint(unsimilary)\na = []\nfor i, inde in enumerate(similary.index[:-1]):\n# print(i,inde)\n# tmp = []\n if similary.loc[inde]['query1']==similary.loc[similary.index[i+1]]['query1']:\n tmp = [similary.loc[inde]['query2'], similary.loc[similary.index[i+1]]['query2']]\n a.append(tmp)\n try:\n if similary.loc[inde]['query1']==similary.loc[similary.index[i+2]]['query1']:\n tmp = [similary.loc[inde]['query2'], similary.loc[similary.index[i+2]]['query2']]\n a.append(tmp)\n except:\n print('index error')\nprint(len(a))\n\nb=[]\nfor i, inde in enumerate(unsimilary.index[:-1]):\n# print(i,inde)\n# tmp = []\n if unsimilary.loc[inde]['query1']==unsimilary.loc[unsimilary.index[i+1]]['query1']:\n tmp = [unsimilary.loc[inde]['query2'], unsimilary.loc[unsimilary.index[i+1]]['query2']]\n b.append(tmp)\n try:\n if unsimilary.loc[inde]['query1']==unsimilary.loc[unsimilary.index[i+2]]['query1']:\n tmp = [unsimilary.loc[inde]['query2'], unsimilary.loc[unsimilary.index[i+2]]['query2']]\n b.append(tmp)\n except:\n print('index error')\nprint(len(b))", " id query1 query2 label\n0 0 剧烈运动后咯血,是怎么了? 剧烈运动后咯血是什么原因？ 1\n1 1 剧烈运动后咯血,是怎么了? 剧烈运动后为什么会咯血？ 1\n5 5 百令胶囊需要注意什么？ 百令胶囊有什么注意事项？ 1\n6 6 百令胶囊需要注意什么？ 服用百令胶囊有什么需要特别注意的吗？ 1\n10 10 肝癌兼肺癌晚期能活多久？ 肝癌兼肺癌晚期还有多少寿命？ 1\n... ... ... ... ...\n10735 1988 过敏性哮喘平时应注意哪些问题？ 患了过敏性哮喘平时需要注意哪些事？ 1\n10739 1992 过敏性哮喘究竟怎样治疗? 如何治疗过敏性哮喘？ 1\n10740 1993 过敏性哮喘究竟怎样治疗? 过敏性哮喘要怎么治疗？ 1\n10744 1997 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘与过敏性哮喘一样吗？ 1\n10745 1998 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘是否就是过敏性哮喘？ 1\n\n[4301 rows x 4 columns]\n id query1 query2 label\n2 2 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，应该怎么处理？ 0\n3 3 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，需要就医吗？ 0\n4 4 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，是否很严重？ 0\n7 7 百令胶囊需要注意什么？ 百令胶囊如何服用？ 0\n8 8 百令胶囊需要注意什么？ 百令胶囊效果好吗？ 0\n... ... ... ... ...\n10742 1995 过敏性哮喘究竟怎样治疗? 过敏性哮喘有哪些症状？ 0\n10743 1996 过敏性哮喘究竟怎样治疗? 过敏性哮喘治好会不会复发？ 0\n10746 1999 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘的饮食禁忌有哪些？ 0\n10747 2000 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘怎么治疗？ 0\n10748 2001 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘能跑步吗？ 0\n\n[6448 rows x 4 columns]\nindex error\n3010\nindex error\n7295\n" ], [ "df = pd.DataFrame(data=a, columns=['query1', 'query2'])\ndf['label']=1\ndf['id']=df.index\nprint(df)\ndf[['id','query1','query2','label']].to_csv('similary_generate.csv',header=True,index=False)", " query1 query2 label id\n0 剧烈运动后咯血是什么原因？ 剧烈运动后为什么会咯血？ 1 0\n1 百令胶囊有什么注意事项？ 服用百令胶囊有什么需要特别注意的吗？ 1 1\n2 肝癌兼肺癌晚期还有多少寿命？ 肝癌兼肺癌晚期还有多少时间？ 1 2\n3 咳嗽咯血半年是怎么回事？ 咳嗽咯血半年是什么情况？ 1 3\n4 百令胶囊是否可以长时间服用? 百令胶囊长时间服用会有问题吗? 1 4\n... ... ... ... ...\n3005 哮喘能否吃洋参？ 哮喘可不可以吃洋参？ 1 3005\n3006 过敏性哮喘患者的注意事项是什么？ 过敏性哮喘患者要注意的是什么？ 1 3006\n3007 过敏性哮喘平时应注意什么？ 患了过敏性哮喘平时需要注意哪些事？ 1 3007\n3008 如何治疗过敏性哮喘？ 过敏性哮喘要怎么治疗？ 1 3008\n3009 变应性哮喘与过敏性哮喘一样吗？ 变应性哮喘是否就是过敏性哮喘？ 1 3009\n\n[3010 rows x 4 columns]\n" ], [ "df = pd.DataFrame(data=b, columns=['query1', 'query2'])\n# df['lable']=1\ndf['id']=df.index\nprint(df)\ndf[['id','query1','query2']].to_csv('unknow_generate.csv',header=True,index=False)", " query1 query2 id\n0 剧烈运动后咯血，应该怎么处理？ 剧烈运动后咯血，需要就医吗？ 0\n1 剧烈运动后咯血，应该怎么处理？ 剧烈运动后咯血，是否很严重？ 1\n2 剧烈运动后咯血，需要就医吗？ 剧烈运动后咯血，是否很严重？ 2\n3 百令胶囊如何服用？ 百令胶囊效果好吗？ 3\n4 百令胶囊如何服用？ 百令胶囊需要如何服用？ 4\n... ... ... ...\n7290 过敏性哮喘病因是什么？ 过敏性哮喘治好会不会复发？ 7290\n7291 过敏性哮喘有哪些症状？ 过敏性哮喘治好会不会复发？ 7291\n7292 变应性哮喘的饮食禁忌有哪些？ 变应性哮喘怎么治疗？ 7292\n7293 变应性哮喘的饮食禁忌有哪些？ 变应性哮喘能跑步吗？ 7293\n7294 变应性哮喘怎么治疗？ 变应性哮喘能跑步吗？ 7294\n\n[7295 rows x 3 columns]\n" ], [ "df = pd.read_csv('../result.csv')\n# print(df)\nprint(df[df['label']==1])\ndf1 = pd.read_csv('unknow_generate.csv')\ndf1['label'] = df.iloc[:,-1].values\nprint(df1)\ndf2 = df1[df1['label']==1]\ndf2_ = df1[df1['label']==0]\nprint(len(df2_))\ndf3 = pd.read_csv('total_data1.csv')\nprint(df3)\ndf4 = pd.concat([df2,df3],axis=0)\nprint(df4)\ndf5 = df2_.sample(n=2000, random_state=1)\nprint(df5)\ndf6=pd.concat([df5,df4],axis=0)\nprint(df6)\ndf6.to_csv('total_data3.csv',header=True,index=False)", " id label\n4 4 1\n7 7 1\n14 14 1\n15 15 1\n16 16 1\n... ... ...\n7014 7014 1\n7116 7116 1\n7170 7170 1\n7202 7202 1\n7257 7257 1\n\n[380 rows x 2 columns]\n id query1 query2 label\n0 0 剧烈运动后咯血，应该怎么处理？ 剧烈运动后咯血，需要就医吗？ 0\n1 1 剧烈运动后咯血，应该怎么处理？ 剧烈运动后咯血，是否很严重？ 0\n2 2 剧烈运动后咯血，需要就医吗？ 剧烈运动后咯血，是否很严重？ 0\n3 3 百令胶囊如何服用？ 百令胶囊效果好吗？ 0\n4 4 百令胶囊如何服用？ 百令胶囊需要如何服用？ 1\n... ... ... ... ...\n7290 7290 过敏性哮喘病因是什么？ 过敏性哮喘治好会不会复发？ 0\n7291 7291 过敏性哮喘有哪些症状？ 过敏性哮喘治好会不会复发？ 0\n7292 7292 变应性哮喘的饮食禁忌有哪些？ 变应性哮喘怎么治疗？ 0\n7293 7293 变应性哮喘的饮食禁忌有哪些？ 变应性哮喘能跑步吗？ 0\n7294 7294 变应性哮喘怎么治疗？ 变应性哮喘能跑步吗？ 0\n\n[7295 rows x 4 columns]\n6915\n id query1 query2 label\n0 0 剧烈运动后咯血,是怎么了? 剧烈运动后咯血是什么原因？ 1\n1 1 剧烈运动后咯血,是怎么了? 剧烈运动后为什么会咯血？ 1\n2 2 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，应该怎么处理？ 0\n3 3 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，需要就医吗？ 0\n4 4 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，是否很严重？ 0\n... ... ... ... ...\n13754 3005 哮喘能否吃洋参？ 哮喘可不可以吃洋参？ 1\n13755 3006 过敏性哮喘患者的注意事项是什么？ 过敏性哮喘患者要注意的是什么？ 1\n13756 3007 过敏性哮喘平时应注意什么？ 患了过敏性哮喘平时需要注意哪些事？ 1\n13757 3008 如何治疗过敏性哮喘？ 过敏性哮喘要怎么治疗？ 1\n13758 3009 变应性哮喘与过敏性哮喘一样吗？ 变应性哮喘是否就是过敏性哮喘？ 1\n\n[13759 rows x 4 columns]\n id query1 query2 label\n4 4 百令胶囊如何服用？ 百令胶囊需要如何服用？ 1\n7 7 肝癌兼肺癌晚期能治疗吗？ 肝癌兼肺癌晚期能治好吗？ 1\n14 14 百令胶囊怎么服用? 百令胶囊的服用方法是什么? 1\n15 15 我国最常见的咯血有哪些? 我国最常见的咯血的种类有多少? 1\n16 16 我国最常见的咯血有哪些? 我国最常见的咯血是什么? 1\n... ... ... ... ...\n13754 3005 哮喘能否吃洋参？ 哮喘可不可以吃洋参？ 1\n13755 3006 过敏性哮喘患者的注意事项是什么？ 过敏性哮喘患者要注意的是什么？ 1\n13756 3007 过敏性哮喘平时应注意什么？ 患了过敏性哮喘平时需要注意哪些事？ 1\n13757 3008 如何治疗过敏性哮喘？ 过敏性哮喘要怎么治疗？ 1\n13758 3009 变应性哮喘与过敏性哮喘一样吗？ 变应性哮喘是否就是过敏性哮喘？ 1\n\n[14139 rows x 4 columns]\n id query1 query2 label\n7114 7114 干咳无痰吃什么水果恢复的快 干咳无痰并伴有哮鸣音怎么回事 0\n1375 1375 肺气肿病人怎么治疗的？ 肺气肿病人最明显的表现有哪些呢? 0\n1946 1946 肺炎会传染人吗 大叶性肺炎传染吗？ 0\n2012 2012 感冒肺炎，住院三天医生说出院需要注意饮食 感冒肺炎，住院三天医生说出院后不要去人群密集的地方 0\n2211 2211 严重的慢性肺炎传染严重吗？ 严重的慢性肺炎传染人吗？ 0\n... ... ... ... ...\n3129 3129 宝宝吃过东西就吐是扁桃体发炎了吗？ 宝宝吃过东西就吐可以吃健胃消食片吗？ 0\n64 64 康尔佳益肺止咳胶囊有什么作用？ 康尔佳益肺止咳胶囊怎么服用？ 0\n365 365 怎么诊断小儿支原体肺炎？ 小儿支原体肺炎有什么症状？ 0\n106 106 达肺草能治疗肺纤维化吗 达肺草属于中草药吗 0\n7235 7235 怎样判断是过敏性哮喘 过敏跟季节有关系吗 0\n\n[2000 rows x 4 columns]\n id query1 query2 label\n7114 7114 干咳无痰吃什么水果恢复的快 干咳无痰并伴有哮鸣音怎么回事 0\n1375 1375 肺气肿病人怎么治疗的？ 肺气肿病人最明显的表现有哪些呢? 0\n1946 1946 肺炎会传染人吗 大叶性肺炎传染吗？ 0\n2012 2012 感冒肺炎，住院三天医生说出院需要注意饮食 感冒肺炎，住院三天医生说出院后不要去人群密集的地方 0\n2211 2211 严重的慢性肺炎传染严重吗？ 严重的慢性肺炎传染人吗？ 0\n... ... ... ... ...\n13754 3005 哮喘能否吃洋参？ 哮喘可不可以吃洋参？ 1\n13755 3006 过敏性哮喘患者的注意事项是什么？ 过敏性哮喘患者要注意的是什么？ 1\n13756 3007 过敏性哮喘平时应注意什么？ 患了过敏性哮喘平时需要注意哪些事？ 1\n13757 3008 如何治疗过敏性哮喘？ 过敏性哮喘要怎么治疗？ 1\n13758 3009 变应性哮喘与过敏性哮喘一样吗？ 变应性哮喘是否就是过敏性哮喘？ 1\n\n[16139 rows x 4 columns]\n" ], [ "df = pd.read_csv('total_data3.csv')\nprint(df[df['label']==0])", " id query1 query2 label\n0 7114 干咳无痰吃什么水果恢复的快 干咳无痰并伴有哮鸣音怎么回事 0\n1 1375 肺气肿病人怎么治疗的？ 肺气肿病人最明显的表现有哪些呢? 0\n2 1946 肺炎会传染人吗 大叶性肺炎传染吗？ 0\n3 2012 感冒肺炎，住院三天医生说出院需要注意饮食 感冒肺炎，住院三天医生说出院后不要去人群密集的地方 0\n4 2211 严重的慢性肺炎传染严重吗？ 严重的慢性肺炎传染人吗？ 0\n... ... ... ... ...\n13122 1995 过敏性哮喘究竟怎样治疗? 过敏性哮喘有哪些症状？ 0\n13123 1996 过敏性哮喘究竟怎样治疗? 过敏性哮喘治好会不会复发？ 0\n13126 1999 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘的饮食禁忌有哪些？ 0\n13127 2000 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘怎么治疗？ 0\n13128 2001 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘能跑步吗？ 0\n\n[8448 rows x 4 columns]\n" ], [ "df1 = pd.read_csv('totol_data.csv')\ndf1 = df1.drop(['category'],axis=1)\nprint(df1)\ndf2 = pd.read_csv('similary_generate.csv')\nprint(df2)\ndf = pd.concat([df1,df2],axis=0)\nprint(df)\ndf.to_csv('total_data1.csv',header=True,index=False)", " id query1 query2 label\n0 0 剧烈运动后咯血,是怎么了? 剧烈运动后咯血是什么原因？ 1\n1 1 剧烈运动后咯血,是怎么了? 剧烈运动后为什么会咯血？ 1\n2 2 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，应该怎么处理？ 0\n3 3 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，需要就医吗？ 0\n4 4 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，是否很严重？ 0\n... ... ... ... ...\n10744 1997 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘与过敏性哮喘一样吗？ 1\n10745 1998 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘是否就是过敏性哮喘？ 1\n10746 1999 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘的饮食禁忌有哪些？ 0\n10747 2000 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘怎么治疗？ 0\n10748 2001 变应性哮喘就是过敏性哮喘吗？ 变应性哮喘能跑步吗？ 0\n\n[10749 rows x 4 columns]\n id query1 query2 label\n0 0 剧烈运动后咯血是什么原因？ 剧烈运动后为什么会咯血？ 1\n1 1 百令胶囊有什么注意事项？ 服用百令胶囊有什么需要特别注意的吗？ 1\n2 2 肝癌兼肺癌晚期还有多少寿命？ 肝癌兼肺癌晚期还有多少时间？ 1\n3 3 咳嗽咯血半年是怎么回事？ 咳嗽咯血半年是什么情况？ 1\n4 4 百令胶囊是否可以长时间服用? 百令胶囊长时间服用会有问题吗? 1\n... ... ... ... ...\n3005 3005 哮喘能否吃洋参？ 哮喘可不可以吃洋参？ 1\n3006 3006 过敏性哮喘患者的注意事项是什么？ 过敏性哮喘患者要注意的是什么？ 1\n3007 3007 过敏性哮喘平时应注意什么？ 患了过敏性哮喘平时需要注意哪些事？ 1\n3008 3008 如何治疗过敏性哮喘？ 过敏性哮喘要怎么治疗？ 1\n3009 3009 变应性哮喘与过敏性哮喘一样吗？ 变应性哮喘是否就是过敏性哮喘？ 1\n\n[3010 rows x 4 columns]\n id query1 query2 label\n0 0 剧烈运动后咯血,是怎么了? 剧烈运动后咯血是什么原因？ 1\n1 1 剧烈运动后咯血,是怎么了? 剧烈运动后为什么会咯血？ 1\n2 2 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，应该怎么处理？ 0\n3 3 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，需要就医吗？ 0\n4 4 剧烈运动后咯血,是怎么了? 剧烈运动后咯血，是否很严重？ 0\n... ... ... ... ...\n3005 3005 哮喘能否吃洋参？ 哮喘可不可以吃洋参？ 1\n3006 3006 过敏性哮喘患者的注意事项是什么？ 过敏性哮喘患者要注意的是什么？ 1\n3007 3007 过敏性哮喘平时应注意什么？ 患了过敏性哮喘平时需要注意哪些事？ 1\n3008 3008 如何治疗过敏性哮喘？ 过敏性哮喘要怎么治疗？ 1\n3009 3009 变应性哮喘与过敏性哮喘一样吗？ 变应性哮喘是否就是过敏性哮喘？ 1\n\n[13759 rows x 4 columns]\n" ] ] ]

[ "code" ]

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import neilpy\nimport pandas as pd\nimport glob\nimport os\n\n# Load image names into a list\nimages_dir = 'POAS/*.jpg'\nfns = glob.glob(images_dir)\n\n# Read the geotags into a dataframe\nphotos_df = neilpy.read_geotags_into_df(fns,return_datetimes=False)\n\n# Fix names, as we don't want to include the path to the image, just the basename\nphotos_df['fn'] = photos_df['fn'].apply(os.path.basename)\n\n# Calculate the azimuths based on the tracks\nphotos_df['azimuth'] = neilpy.track2azimuth(photos_df['lat'].values,photos_df['lon'].values)\n\n# Based on a specified pitch estimated during the mounting, calculate Omega, Phi, and Kappa angles\npitch = -70\nphotos_df['omega'],photos_df['phi'],photos_df['kappa'] = neilpy.ypr2opk(photos_df['azimuth'].values,pitch)\n\n# Correct for GEOID\nphotos_df['alt'] = photos_df['alt'] + 35.356 \n\n# Define accuracy of measurements:\nphotos_df['xy_acc'] = 2.\nphotos_df['z_acc'] = 2.\n\n# Write out the values\noutfile = 'sept_poas_opk.csv'\ncols = ['fn','lat','lon','alt','omega','phi','kappa','xy_acc','z_acc']\nphotos_df.to_csv(outfile,index=False,header=False,columns=cols)", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "from scipy import signal\nimport numpy as np\n\n%matplotlib notebook\nimport matplotlib.pyplot as plt", "_____no_output_____" ], [ "sig_len_sec = 2\nFs = 100\nT = 1/Fs\nN = Fs * sig_len_sec\ntt = np.arange(0, N) * T - (N/2 * T)", "_____no_output_____" ], [ "ff = np.linspace(-1.0/(2.0*T), 1.0/(2.0*T), N, endpoint=False)", "_____no_output_____" ], [ "# Create a Frequency Mask, extending into negative frequency domain\nmask = np.ones(len(tt))\nmask[130::] = 0\nmask[0:70] = 0\nmask = 1.0 - mask", "_____no_output_____" ], [ "f = plt.figure(figsize=(5, 4))\nax = f.add_subplot(111)\n\nax.plot(ff, mask)\n\nax.set_xlim([0, Fs/2]) # zoom in on positive frequencies only\nax.set_xlabel('Frequency [Hz]')", "_____no_output_____" ], [ "y_mask = np.real(np.fft.fftshift(np.fft.ifft(np.fft.ifftshift(mask))))", "_____no_output_____" ], [ "f = plt.figure(figsize=(5, 4))\nax = f.add_subplot(111)\n\nax.plot(tt, y_mask, label='y_mask')\n\n# ax.legend()\nax.set_xlabel('Time [s]')", "_____no_output_____" ], [ "f1 = 2\ny1 = np.sin(2 * np.pi * f1 * tt)", "_____no_output_____" ], [ "f2 = 20\ny2 = 0.25 * np.sin(2 * np.pi * f2 * tt)", "_____no_output_____" ], [ "y3 = y1 + y2", "_____no_output_____" ], [ "y3_padded = np.hstack([y3, np.zeros(len(y3))])", "_____no_output_____" ], [ "f = plt.figure(figsize=(9, 4))\nax = f.add_subplot(111)\n\nax.plot(y3_padded, label='y3_padded')\n\nax.legend()\nax.set_xlabel('Time [s]')", "_____no_output_____" ], [ "def fir_filter(b, xx):\n yy = np.zeros_like(xx) # create buffer for output values\n delay = np.zeros_like(b) # create delay line\n \n for ii, x in enumerate(xx):\n delay[1:] = delay[:-1] # right-shift values in 'delay'\n delay[:1] = x # place new value into the delay line\n yy[ii] = np.sum(delay * b)\n \n return yy", "_____no_output_____" ], [ "y3_filt = fir_filter(y_mask, y3_padded)", "_____no_output_____" ], [ "f = plt.figure(figsize=(9, 4))\nax = f.add_subplot(111)\n\nax.plot(T * np.arange(len(y3_filt)), y3_filt, label='y_mask')\n\nax.legend()\nax.set_xlabel('Time [s]')", "_____no_output_____" ], [ "from scipy.signal import freqz\n\n# Calculate the frequency response 'h' at the complex frequencies 'w'\n# Note that 'w' is returned in the same units as 'Fs'\nw, h = freqz(y_mask, [1], worN=8192)\n\nw_hz = w * (Fs/(2*np.pi)) # 'convert 'w' from radians to Hz\nh_db = 20 * np.log10(np.abs(h)) # convert 'h' from complex magitude to dB\nangles = np.unwrap(np.angle(h)) * (180/np.pi)", "_____no_output_____" ], [ "f = plt.figure(figsize=(9, 4))\nax1 = f.add_subplot(111)\n\nax1.plot(w_hz, 20*np.log(np.abs(h)), color='xkcd:blue')\n\n# ax1.set_xscale('log')\nax1.set_xlim([1, Fs/2])\nax1.grid(which='both', axis='both')\nax1.set_ylabel('Amplitude [dB]', color='xkcd:blue')\nax1.set_title('Filer Frequency and Phase Response')\n\nax2 = ax1.twinx()\nax2.plot(w_hz, angles, color='xkcd:green')\nax2.set_ylabel('angle [deg]', color='xkcd:green')", "_____no_output_____" ] ] ]

[ "code" ]

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

[ "Unlicense" ]

[ "Unlicense" ]

[ "Unlicense" ]

[ [ [ "%cd C:\\Users\\Bruno Ferrari\\Documents\\Bruno\\2019\\2s\\MC\\artigos revisão\\Artigos Mes\\GD\\bdp\\dbdp_instances", "C:\\Users\\Bruno Ferrari\\Documents\\Bruno\\2019\\2s\\MC\\artigos revisão\\Artigos Mes\\GD\\bdp\\dbdp_instances\n" ], [ "import pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\n\n#from lazypredict.Supervised import LazyClassifier\n\nfrom sklearn.preprocessing import scale\n\nfrom sklearn.model_selection import train_test_split", "_____no_output_____" ], [ "dados_org = pd.read_excel('final_results.xlsx')", "_____no_output_____" ], [ "dados_org.head()", "_____no_output_____" ], [ "dados_org.tail()", "_____no_output_____" ], [ "dados_org.loc[:, dados_org.columns.str.contains('Crossing', na=False)].plot(logy=True, figsize=(16,12))\ndados_org.loc[:, dados_org.columns.str.contains('Crossing', na=False)]", "_____no_output_____" ], [ "dados_org.loc[:, dados_org.columns.str.contains('Time', na=False)].plot(logy=True, figsize=(16,12))\ndados_org.loc[:, dados_org.columns.str.contains('Time', na=False)]", "_____no_output_____" ], [ "dados_org.loc[:, dados_org.columns.str.contains('Crossing', na=False)].drop('Crossing_gs', axis=1).fillna(np.inf).idxmin(axis=1).value_counts(normalize=0)", "_____no_output_____" ], [ "np.unique(np.argmin(dados_org.loc[:, dados_org.columns.str.contains('Crossing', na=False)].drop('Crossing_gs', axis=1).values,axis=1), return_counts=True)", "_____no_output_____" ], [ "aux_dm = pd.concat(\n [\n #pd.Series(np.argmin(dados_org.loc[:, dados_org.columns.str.contains('Crossing', na=False)].drop('Crossing_gs', axis=1).fillna(np.inf).values,axis=1)),\n dados_org.loc[:, dados_org.columns.str.contains('Crossing', na=False)].drop('Crossing_gs', axis=1).fillna(np.inf).idxmin(axis=1) \n ],\n axis=1\n)", "_____no_output_____" ], [ "display(aux_dm)", "_____no_output_____" ], [ "from sklearn.preprocessing import LabelBinarizer", "_____no_output_____" ], [ "Y_labeled = (aux_dm.astype('category')[0].cat.codes)", "_____no_output_____" ], [ "Y_gs=pd.get_dummies(aux_dm, prefix='Best')['Best_Crossing_gs_vns']\nY_ts=pd.get_dummies(aux_dm, prefix='Best')['Best_Crossing_ts']\nY_vns=pd.get_dummies(aux_dm, prefix='Best')['Best_Crossing_vns']", "_____no_output_____" ], [ "display(Y_ts)\nprint(Y_ts.sum())\ndisplay(Y_vns)\nprint(Y_vns.sum())\ndisplay(Y_gs)\nprint(Y_gs.sum())", "_____no_output_____" ], [ "#Y_gs[Y_gs==1]\nY_gs[205:211] = 1\n\nY_labeled = Y_labeled.copy()\nY_labeled[205:211] = 0\nY_labeled.value_counts()", "_____no_output_____" ], [ "X = dados_org.iloc[:, 1:11]\nX.drop(['V1', 'V2'],axis=1)", "_____no_output_____" ], [ "from sklearn.model_selection import cross_val_score, cross_val_predict\nfrom sklearn.svm import SVC\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.tree import DecisionTreeClassifier\nfrom sklearn.neighbors import KNeighborsClassifier", "_____no_output_____" ], [ "svc = SVC(decision_function_shape='ovo', kernel=\"linear\")\nrf = RandomForestClassifier(random_state=42, n)\ndt = DecisionTreeClassifier(random_state=42, min_samples_leaf=10)\nknn = KNeighborsClassifier() ", "_____no_output_____" ], [ "#score_svc = cross_val_score(svc, X.drop(['V1', 'V2'],axis=1), Y_labeled, scoring='accuracy')\nscore_rf = cross_val_score(rf, X.drop(['V1', 'V2'],axis=1), Y_labeled, scoring='accuracy')\n#score_dt = cross_val_score(dt, X.drop(['V1', 'V2'],axis=1), Y_labeled, scoring='accuracy')\n#score_knn = cross_val_score(knn, X.drop(['V1', 'V2'],axis=1), Y_labeled, scoring='accuracy')", "_____no_output_____" ], [ "#score_svc.mean()\nscore_rf.mean()\n#score_dt.mean()\n#score_knn.mean()", "_____no_output_____" ], [ "score = cross_val_score(rf, X.drop(['V1', 'V2'],axis=1), Y_vns, scoring='accuracy')\nscore.mean()", "_____no_output_____" ], [ "score = cross_val_score(rf, X.drop(['V1', 'V2'],axis=1), Y_ts, scoring='accuracy')\nscore.mean()", "_____no_output_____" ], [ "score = cross_val_score(rf, X.drop(['V1', 'V2'],axis=1), Y_gs, scoring='accuracy')\nscore.mean()", "_____no_output_____" ], [ "test_vns = cross_val_predict(svc, X.drop(['V1', 'V2'],axis=1), Y_vns, method=\"decision_function\")\ntest_gs = cross_val_predict(rf, X.drop(['V1', 'V2'],axis=1), Y_gs)\ntest_ts = cross_val_predict(svc, X.drop(['V1', 'V2'],axis=1), Y_ts, method=\"decision_function\")\ntest_labeled = cross_val_predict(svc, X.drop(['V1', 'V2'],axis=1), Y_labeled, method=\"decision_function\")", "_____no_output_____" ], [ "pd.DataFrame(test_labeled).plot()", "_____no_output_____" ], [ "rf = RandomForestClassifier(n_estimators=10000, random_state=42, n_jobs=-1)\nprint(cross_val_score(rf, X, Y_labeled, scoring='accuracy', n_jobs=-1))\ny_hat=cross_val_predict(rf, X, Y_labeled, verbose=1, n_jobs=-1)\ny_hat", "[0.35164835 0.55555556 0.52222222 0.61111111 0.67777778]\n" ], [ "sum(Y_labeled.values==y_hat)/len(y_hat)", "_____no_output_____" ], [ "from sklearn.metrics import roc_curve", "_____no_output_____" ], [ "fpr,tpr,thresholds=roc_curve(y_train_5, y_scores)", "_____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" ]

[ [ [ "# Basic Apach Spark Analysis\r\n\r\n- Ref: https://timw.info/ply\r\n- Notebook tutorial: https://timw.info/ekt\r\n\r\n\r\n\r\n\r\n\r\n", "_____no_output_____" ] ], [ [ "# Load NYC Taxi data\r\ndf = spark.read.load('abfss://[email protected]/NYCTripSmall.parquet', format='parquet')\r\ndisplay(df.limit(10))", "_____no_output_____" ], [ "# View the dataframe schema\r\ndf.printSchema()", "_____no_output_____" ], [ " # Load the NYC Taxi data into the Spark nyctaxi database\r\nspark.sql(\"CREATE DATABASE IF NOT EXISTS nyctaxi\")\r\ndf.write.mode(\"overwrite\").saveAsTable(\"nyctaxi.trip\")", "_____no_output_____" ], [ "# Display the taxi data\r\ndf = spark.sql(\"SELECT * FROM nyctaxi.trip\") \r\ndisplay(df)", "_____no_output_____" ], [ "# Analyze the data and save results to nyctaxi.passengercountstats table (select CHART)\r\ndf = spark.sql(\"\"\"\r\n SELECT PassengerCount,\r\n SUM(TripDistanceMiles) as SumTripDistance,\r\n AVG(TripDistanceMiles) as AvgTripDistance\r\n FROM nyctaxi.trip\r\n WHERE TripDistanceMiles > 0 AND PassengerCount > 0\r\n GROUP BY PassengerCount\r\n ORDER BY PassengerCount\r\n\"\"\") \r\ndisplay(df)\r\ndf.write.saveAsTable(\"nyctaxi.passengercountstats\")", "_____no_output_____" ], [ "", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

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

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ [ [ "# Germany: LK Kleve (Nordrhein-Westfalen)\n\n* Homepage of project: https://oscovida.github.io\n* [Execute this Jupyter Notebook using myBinder](https://mybinder.org/v2/gh/oscovida/binder/master?filepath=ipynb/Germany-Nordrhein-Westfalen-LK-Kleve.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(country=\"Germany\", subregion=\"LK Kleve\");", "_____no_output_____" ], [ "# load the data\ncases, deaths, region_label = germany_get_region(landkreis=\"LK Kleve\")\n\n# compose into one table\ntable = compose_dataframe_summary(cases, deaths)\n\n# show tables with up to 500 rows\npd.set_option(\"max_rows\", 500)\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/Germany-Nordrhein-Westfalen-LK-Kleve.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- 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" ], [ "markdown", "markdown" ], [ "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/nk555/AI-Projects/blob/master/GAN/MNIST_Style_GAN_v2.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "# MNIST learns your handwriting\n\nThis is a small project on using a GAN to generate numbers that look as someone else's handwriting when not trained on all numbers written by this person. For example say we had someone write the number 273 and we now want to write 481 in their own handwriting.\n\nThe main inspiration for this project is a paper I read recently called STAR GAN v2. In this paper they try to recognize diferent styles and features in images and transfer those into a different image. For example they are able to use image of different animals like dogs or tigers and making them look like a cat. Furthermore at the time of writing this it is currently a state-of-the-art method for this style translation tasks.\n\nSome of the results can be seen at the end of this notebook. Unfortunately it seems not that many features were captured and mostly it was only the thickness of the numbers that was preserved. A reason this happens might be that the size of the images is small being 28x28. However, some ways to allow for more variation might be by exteding the number of layers being used, by having higher dimensional spaces for the latent and style spaces, or by giving a higher weight to the style diversification loss (look at section loss functions to see more about this).\n\nThe main purpose of this notebook is to make a small showcase of the architecture used in a simple design so that the ideas are simple to follow. This notebook will also contain some explanations and comments on the architecture of the neural network so that it might be easier to follow.\n\nNote: another small thing I did in this project is to 'translate' STAR GAN code from pytorch to tensorflow. Redoing all of the work was useful to understand everything done on their code and having an option in tensorflow might be useful for some people.\n\nFor a small tutorial on how to write a simple GAN architecture: https://machinelearningmastery.com/how-to-develop-a-generative-adversarial-network-for-an-mnist-handwritten-digits-from-scratch-in-keras/\n\nLink to STAR GAN v2: https://app.wandb.ai/stacey/stargan/reports/Cute-Animals-and-Post-Modern-Style-Transfer%3A-StarGAN-v2-for-Multi-Domain-Image-Synthesis---VmlldzoxNzcwODQ\n\nFurther Reading on style domain techniques for image generation:\n\nLink to STAR GAN paper: https://arxiv.org/pdf/1912.01865.pdf\n\nLink to Multimodal Unsupervised Image-to-Image Translation: https://arxiv.org/pdf/1804.04732.pdf\n\nLink to Improving Style-Content Disentanglement Paper: https://arxiv.org/pdf/2007.04964.pdf", "_____no_output_____" ], [ "# Intitializing", "_____no_output_____" ] ], [ [ "import tensorflow as tf\nfrom tensorflow_addons.layers import InstanceNormalization\nimport numpy as np\nimport tensorflow.keras.layers as layers\nimport time\nfrom tensorflow.keras.datasets.mnist import load_data\nimport sys\nimport os\nimport datetime", "_____no_output_____" ] ], [ [ "# Layers\n\nThere are a few layers that were custom made. More importantly it is udeful to make this custom layers for the layers that try to incorporate style. This is as the inputs themselves are custom as you are inputing an image and a vector representing the style.\n\nResBlk is short for Residual Block, where it is predicting the residual (the difference between the original and the prediction).", "_____no_output_____" ] ], [ [ "class ResBlk(tf.keras.Model):\n def __init__(self, dim_in, dim_out, actv=layers.LeakyReLU(),\n normalize=False, downsample=False):\n super(ResBlk, self).__init__()\n self.actv = actv\n self.normalize = normalize\n self.downsample = downsample\n self.learned_sc = dim_in != dim_out\n self._build_weights(dim_in, dim_out)\n\n def _build_weights(self, dim_in, dim_out):\n self.conv1 = layers.Conv2D(dim_in, 3, padding='same')\n self.conv2 = layers.Conv2D(dim_out, 3, padding='same')\n if self.normalize:\n self.norm1 = InstanceNormalization()\n self.norm2 = InstanceNormalization()\n if self.learned_sc:\n self.conv1x1 = layers.Conv2D(dim_out, 1)\n\n def _shortcut(self, x):\n if self.learned_sc:\n x = self.conv1x1(x)\n if self.downsample:\n x = layers.AveragePooling2D(pool_size=(2,2), padding='same')(x)\n return x\n \n def _residual(self, x):\n if len(tf.shape(x))>4:\n x=tf.reshape(x,tf.shape(x)[1:])\n if self.normalize:\n x = self.norm1(x)\n x = self.actv(x)\n x = self.conv1(x)\n if self.downsample:\n x = layers.AveragePooling2D(pool_size=(2,2), padding='same')(x)\n if self.normalize:\n x = self.norm2(x)\n x = self.actv(x)\n x = self.conv2(x)\n return x\n\n def call(self, x):\n x = self._shortcut(x) + self._residual(x)\n return x / 2**(1/2) # unit variance", "_____no_output_____" ] ], [ [ "AdaIN stands for Adaptive Instance Normalization. It is a type of normalization that allows to 'mix' two inputs. In this case we use the style vector to mix with our input x which is the image or part of the process of constructing this image.", "_____no_output_____" ] ], [ [ "class AdaIn(tf.keras.Model):\n def __init__(self, style_dim, num_features):\n super(AdaIn,self).__init__()\n self.norm = InstanceNormalization()\n self.lin = layers.Dense(num_features*2)\n\n def call(self, x, s):\n h=self.lin(s)\n h=tf.reshape(h, [1, tf.shape(h)[0], 1, tf.shape(h)[1]])\n gamma,beta=tf.split(h, 2, axis=3)\n return (1+gamma)*self.norm(x)+beta", "_____no_output_____" ], [ "class AdainResBlk(tf.keras.Model):\n def __init__(self, dim_in, dim_out, style_dim=16,\n actv=layers.LeakyReLU(), upsample=False):\n super(AdainResBlk, self).__init__()\n self.actv = actv\n self.upsample = upsample\n self.learned_sc = dim_in != dim_out\n self._build_weights(dim_in, dim_out, style_dim)\n\n def _build_weights(self, dim_in, dim_out, style_dim=16):\n self.conv1 = layers.Conv2D(dim_out, 3, padding='same')\n self.conv2 = layers.Conv2D(dim_out, 3, padding='same')\n self.norm1 = AdaIn(style_dim, dim_in)\n self.norm2 = AdaIn(style_dim, dim_out)\n if self.learned_sc:\n self.conv1x1 = layers.Conv2D(dim_out, 1)\n\n def _shortcut(self, x):\n if self.upsample:\n x = layers.UpSampling2D(size=(2,2), interpolation='nearest')(x)\n if self.learned_sc:\n x = self.conv1x1(x)\n return x\n\n def _residual(self, x, s):\n x = self.norm1(x, s)\n x = self.actv(x)\n if self.upsample:\n x = layers.UpSampling2D(size=(2,2), interpolation='nearest')(x)\n x = self.conv1(x)\n x = self.norm2(x, s)\n x = self.actv(x)\n x = self.conv2(x)\n return x\n\n def call(self, x, s):\n x = self._shortcut(x) + self._residual(x,s)\n return x / 2**(1/2) # unit variance", "_____no_output_____" ] ], [ [ "# Generator Class\n\nIn the generator we have two steps one for encoding the image into lower level information and one to decode back to the image. In this particular architecture the decoding uses the style to build back the image as it is an important part of the process. The decoding does not do this as we have the style encoder as an architecture that deals with this issue of generating a style vector for a particular image.", "_____no_output_____" ] ], [ [ "class Generator(tf.keras.Model):\n def __init__(self, img_size=28, style_dim=24, dim_in=8, max_conv_dim=128, repeat_num=2):\n super(Generator, self).__init__()\n self.img_size=img_size\n self.from_bw=layers.Conv2D(dim_in, 3, padding='same', input_shape=(1,img_size,img_size,1))\n self.encode=[]\n self.decode=[]\n self.to_bw=tf.keras.Sequential([InstanceNormalization(), layers.LeakyReLU(), layers.Conv2D(1, 1, padding='same')])\n\n for _ in range(repeat_num):\n dim_out = min(dim_in*2, max_conv_dim)\n self.encode.append(ResBlk(dim_in, dim_out, normalize=True, downsample=True))\n self.decode.insert(0, AdainResBlk(dim_out, dim_in, style_dim, upsample=True))\n dim_in = dim_out\n\n # bottleneck blocks\n for _ in range(2):\n self.encode.append(ResBlk(dim_out, dim_out, normalize=True))\n self.decode.insert(0, AdainResBlk(dim_out, dim_out, style_dim))\n\n def call(self, x, s):\n x = self.from_bw(x)\n cache = {}\n for block in self.encode:\n x = block(x)\n for block in self.decode:\n x = block(x, s)\n return self.to_bw(x)", "_____no_output_____" ] ], [ [ "# Mapping Network\n\nThe Mapping Network and the Style encoder are the parts of this architecture that make a difference in allowing style to be analyzed and put into our images. The mapping network will take as an input a latent code (represents images as a vector in a high dimensional space) and the domain in this case the domain is the number we are representing. And the style encoder will take as inputs an image and a domain.", "_____no_output_____" ] ], [ [ "class MappingNetwork(tf.keras.Model):\n def __init__(self, latent_dim=16, style_dim=24, num_domains=10):\n super(MappingNetwork,self).__init__()\n map_layers = [layers.Dense(128)]\n map_layers += [layers.ReLU()]\n for _ in range(2):\n map_layers += [layers.Dense(128)]\n map_layers += [layers.ReLU()]\n self.shared = tf.keras.Sequential(layers=map_layers)\n\n self.unshared = []\n for _ in range(num_domains):\n self.unshared += [tf.keras.Sequential(layers=[layers.Dense(128),\n layers.ReLU(),\n layers.Dense(128),\n layers.ReLU(),\n layers.Dense(128),\n layers.ReLU(),\n layers.Dense(style_dim)])]\n\n def call(self, z, y):\n h = self.shared(z)\n out = []\n for layer in self.unshared:\n out += [layer(h)]\n out = tf.stack(out, axis=1) # (batch, num_domains, style_dim)\n s = tf.gather(out, y, axis=1) # (batch, style_dim)\n return s", "_____no_output_____" ] ], [ [ "# Style Encoder\n\nAn important thing to notice from the style encoder is that it takes as an input an image and outputs a style vector. Looking at the dimensions of these we notice we need to flatten out the image through the layers. This can usually be done in two ways. By flattening a 2 dimensional input to a 1 dimensional output a flatten layer, or as it was done hear by using enough pooling layers so that we downsample the size of our 2 dimensional input until it is one dimensional.", "_____no_output_____" ] ], [ [ "class StyleEncoder(tf.keras.Model):\n def __init__(self, img_size=28, style_dim=24, dim_in=16, num_domains=10, max_conv_dim=128, repeat_num=5):\n super(StyleEncoder,self).__init__()\n blocks = [layers.Conv2D(dim_in, 3, padding='same')]\n\n for _ in range(repeat_num): #repetition 1 sends to (b,14,14,d) 2 to (b,7,7,d) 3 to (b,4,4,d) 4 to (b,2,2,d) 5 to (b,1,1,d)\n dim_out = min(dim_in*2, max_conv_dim)\n blocks += [ResBlk(dim_in, dim_out, downsample=True)]\n dim_in = dim_out\n\n blocks += [layers.LeakyReLU()]\n blocks += [layers.Conv2D(dim_out, 4, padding='same')]\n blocks += [layers.LeakyReLU()]\n self.shared = tf.keras.Sequential(layers=blocks)\n\n self.unshared = []\n for _ in range(num_domains):\n self.unshared += [layers.Dense(style_dim)]\n\n def call(self, x, y):\n h = self.shared(x)\n h = tf.reshape(h,[tf.shape(h)[0], tf.shape(h)[3]])\n out = []\n for layer in self.unshared:\n out += [layer(h)]\n out = tf.stack(out, axis=1) # (batch, num_domains, style_dim)\n s = tf.gather(out, y, axis=1) # (batch, style_dim)\n return s", "_____no_output_____" ] ], [ [ "# Discriminator Class\n\nSimilarly to the Style encoder the input of the discriminator is an image and we need to downsample it until it is one dimensional.", "_____no_output_____" ] ], [ [ "class Discriminator(tf.keras.Model):\n def __init__(self, img_size=28, dim_in=16, num_domains=10, max_conv_dim=128, repeat_num=5):\n super(Discriminator, self).__init__()\n blocks = [layers.Conv2D(dim_in, 3, padding='same')]\n\n for _ in range(repeat_num): #repetition 1 sends to (b,14,14,d) 2 to (b,7,7,d) 3 to (b,4,4,d) 4 to (b,2,2,d) 5 to (b,1,1,d)\n dim_out = min(dim_in*2, max_conv_dim)\n blocks += [ResBlk(dim_in, dim_out, downsample=True)]\n dim_in = dim_out\n\n blocks += [layers.LeakyReLU()]\n blocks += [layers.Conv2D(dim_out, 4, padding='same')]\n blocks += [layers.LeakyReLU()]\n blocks += [layers.Conv2D(num_domains, 1, padding='same')]\n self.main = tf.keras.Sequential(layers=blocks)\n\n\n def call(self, x, y):\n out = self.main(x)\n out = tf.reshape(out, (tf.shape(out)[0], tf.shape(out)[3])) # (batch, num_domains)\n out = tf.gather(out, y, axis=1) # (batch)\n out = tf.reshape(out, [1])\n return out", "_____no_output_____" ] ], [ [ "# Loss Functions\n\nThe loss functions used are an important part of this model as it describes our goal when training and how to perform gradient descent. The discriminator loss function is the regular adversarial loss L_adv used in a GAN architecture. But furthermore we have three loss functions added.\n\nFor this loss functions if you want to see the mathematical formula I recommend looking at STAR GAN 2's paper. However I will explain what the loss tries to measure and a quick description of how it does so.\n\nL_sty is a style reconstruction loss. This tries to capture how well the style was captured on our output. It is computed as an expected value of the distance between the target style vector and the style vector that our style encoder predicts for the generated image.\n\nL_ds is a style diversification loss. It tries to capture that the images produced are different to promote a variety of images produced. It is computed as the expected value of the distance between the images (l_1 norm) generated when using two different styles and the same sources. \n\nL_cyc is a characteristic preserving loss. The cyc comes from cyclic as we measusre the distance between the original image and the image generated by using an image generated by this image and the style our style encoder provides as an input. (Notice we use the image generated by the image generated, so that we use the generator two times.)\n\nIn the end the total loss function is expressed as\n\nL_adv + lambda_sty * L_sty + lambda_ds * L_ds + lambda_cyc * L_cyc", "_____no_output_____" ] ], [ [ "def moving_average(model, model_test, beta=0.999):\n for i in range(len(model.weights)):\n model_test.weights[i] = (1-beta)*model.weights[i] + beta*model_test.weights[i]", "_____no_output_____" ], [ "def adv_loss(logits, target):\n assert target in [1, 0]\n targets = tf.fill(tf.shape(logits), target)\n loss = tf.keras.losses.BinaryCrossentropy(from_logits=True)(targets, logits)\n return loss", "_____no_output_____" ], [ "def r1_reg(d_out, x_in, g):\n # zero-centered gradient penalty for real images\n batch_size = tf.shape(x_in)[0]\n grad_dout=g.gradient(d_out, x_in)\n #grad_dout = tf.gradients(ys=d_out, xs=x_in)\n grad_dout2 = tf.square(grad_dout)\n grad_dout2 = tf.reshape(grad_dout2,[batch_size, tf.shape(grad_dout2)[1]*tf.shape(grad_dout2)[2]])\n reg = 0.5 * tf.math.reduce_mean(tf.math.reduce_sum(grad_dout2, axis=1))\n return reg", "_____no_output_____" ], [ "def compute_d_loss(nets, args, x_real, y_org, y_trg, z_trg=None, x_ref=None):\n assert (z_trg is None) != (x_ref is None)\n # with real images\n with tf.GradientTape() as g:\n g.watch(x_real)\n out = nets['discriminator'](x_real, y_org)\n loss_real = adv_loss(out, 1)\n loss_reg = r1_reg(out, x_real, g)\n\n # with fake images\n if z_trg is not None:\n s_trg = nets['mapping_network'](z_trg, y_trg)\n else: # x_ref is not None\n s_trg = nets['style_encoder'](x_ref, y_trg)\n\n x_fake = nets['generator'](x_real, s_trg)\n out = nets['discriminator'](x_fake, y_trg)\n loss_fake = adv_loss(out, 0)\n\n loss = loss_real + loss_fake + args['lambda_reg'] * loss_reg\n return loss, {'real': loss_real, 'fake':loss_fake, 'reg':loss_reg}", "_____no_output_____" ], [ "def compute_g_loss(nets, args, x_real, y_org, y_trg, z_trgs=None, x_refs=None):\n assert (z_trgs is None) != (x_refs is None)\n if z_trgs is not None:\n z_trg, z_trg2 = z_trgs\n if x_refs is not None:\n x_ref, x_ref2 = x_refs\n\n # adversarial loss\n if z_trgs is not None:\n s_trg = nets['mapping_network'](z_trg, y_trg)\n else:\n s_trg = nets['style_encoder'](x_ref, y_trg)\n\n x_fake = nets['generator'](x_real, s_trg)\n out = nets['discriminator'](x_fake, y_trg)\n loss_adv = adv_loss(out, 1)\n\n # style reconstruction loss\n s_pred = nets['style_encoder'](x_fake, y_trg)\n loss_sty = tf.math.reduce_mean(tf.abs(s_pred - s_trg))\n\n # diversity sensitive loss\n if z_trgs is not None:\n s_trg2 = nets['mapping_network'](z_trg2, y_trg)\n else:\n s_trg2 = nets['style_encoder'](x_ref2, y_trg)\n x_fake2 = nets['generator'](x_real, s_trg2)\n loss_ds = tf.math.reduce_mean(tf.abs(x_fake - x_fake2))\n\n # cycle-consistency loss\n s_org = nets['style_encoder'](x_real, y_org)\n x_rec = nets['generator'](x_fake, s_org)\n loss_cyc = tf.math.reduce_mean(tf.abs(x_rec - x_real))\n\n loss = loss_adv + args['lambda_sty'] * loss_sty \\\n - args['lambda_ds'] * loss_ds + args['lambda_cyc'] * loss_cyc\n return loss, {'adv':loss_adv, 'sty':loss_sty, 'ds':loss_ds, 'cyc':loss_cyc}", "_____no_output_____" ] ], [ [ "# The Model\n\nHere we introduce the class Solver which is the most important class as this will represent our whole model. It will initiate all of our neural networks as well as train our network.", "_____no_output_____" ] ], [ [ "class Solver(tf.keras.Model):\n def __init__(self, args):\n super(Solver, self).__init__()\n self.args = args\n self.step=0\n\n self.nets, self.nets_ema = self.build_model(self.args)\n # below setattrs are to make networks be children of Solver, e.g., for self.to(self.device)\n for name in self.nets.keys():\n setattr(self, name, self.nets[name])\n for name in self.nets_ema.keys():\n setattr(self, name + '_ema', self.nets_ema[name])\n\n if args['mode'] == 'train':\n self.optims = {}\n for net in self.nets.keys():\n self.optims[net] = tf.keras.optimizers.Adam(learning_rate= args['f_lr'] if net == 'mapping_network' else args['lr'], \n beta_1=args['beta1'], beta_2=args['beta2'], \n epsilon=args['weight_decay'])\n\n self.ckptios = [tf.train.Checkpoint(model=net) for net in self.nets.values()]\n self.ckptios += [tf.train.Checkpoint(model=net_ema) for net_ema in self.nets_ema.values()]\n self.ckptios += [tf.train.Checkpoint(optimizer=optim) for optim in self.optims.values()]\n else:\n self.ckptios = [tf.train.Checkpoint(model=net_ema) for net_ema in self.nets_ema.values()]\n\n #for name in self.nets.keys():\n # Do not initialize the FAN parameters\n # print('Initializing %s...' % name)\n #self.nets[name].apply(initializer=tf.keras.initializers.HeNormal)\n\n def build_model(self, args):\n generator = Generator(args['img_size'], args['style_dim'])\n mapping_network = MappingNetwork(args['latent_dim'], args['style_dim'], args['num_domains'])\n style_encoder = StyleEncoder(args['img_size'], args['style_dim'], args['num_domains'])\n discriminator = Discriminator(args['img_size'], args['num_domains'])\n generator_ema = Generator(args['img_size'], args['style_dim'])\n mapping_network_ema = MappingNetwork(args['latent_dim'], args['style_dim'], args['num_domains'])\n style_encoder_ema = StyleEncoder(args['img_size'], args['style_dim'], args['num_domains'])\n\n nets = {'generator':generator, 'mapping_network':mapping_network,\n 'style_encoder':style_encoder, 'discriminator':discriminator}\n nets_ema = {'generator':generator_ema, 'mapping_network':mapping_network_ema,\n 'style_encoder':style_encoder_ema}\n\n nets['discriminator'](inputs[0]['x_src'],inputs[0]['y_src'])\n s_trg = nets['mapping_network'](inputs[0]['z_trg'],inputs[0]['y_src'])\n nets['generator'](inputs[0]['x_src'],s_trg)\n nets['style_encoder'](inputs[0]['x_src'], inputs[0]['y_src'])\n s_trg = nets_ema['mapping_network'](inputs[0]['z_trg'],inputs[0]['y_src'])\n nets_ema['generator'](inputs[0]['x_src'],s_trg)\n nets_ema['style_encoder'](inputs[0]['x_src'], inputs[0]['y_src'])\n\n return nets, nets_ema\n\n def save(self):\n for net in solv.nets.keys():\n solv.nets[net].save_weights('MNIST_GAN_2/saved_model/'+net+'step'+str(self.step)+'.h5')\n for net in solv.nets_ema.keys():\n solv.nets[net].save_weights('MNIST_GAN_2/saved_model/'+net+'step'+str(self.step)+'_ema.h5')\n \n \n #for ckptio in self.ckptios:\n # ckptio.save(step)\n\n def load(self, step):\n self.step= step\n for net in solv.nets.keys():\n solv.nets[net].load_weights('MNIST_GAN_2/saved_model/'+net+'step'+str(step)+'.h5')\n for net in solv.nets_ema.keys():\n solv.nets[net].load_weights('MNIST_GAN_2/saved_model/'+net+'step'+str(step)+'_ema.h5')\n \n #for ckptio in self.ckptios:\n # ckptio.load(step)\n\n# def _reset_grad(self):\n# for optim in self.optims.values():\n# optim.zero_grad()\n\n def train(self, inputs, validations):\n \"\"\"\n inputs is a list of dictionaries that contains a source image, a reference image, domain and latent code information used to train the network\n validation is a list that contains validation images\n \"\"\"\n\n args = self.args\n nets = self.nets\n nets_ema = self.nets_ema\n optims = self.optims\n\n inputs_val=validations[0]\n\n # resume training if necessary\n if args['resume_iter'] > 0:\n self.load(args['resume_iter'])\n\n # remember the initial value of ds weight\n initial_lambda_ds = args['lambda_ds']\n\n print('Start training...')\n start_time = time.time()\n for i in range(args['resume_iter'], args['total_iters']):\n self.step+=1\n # fetch images and labels\n input= inputs[i-args['resume_iter']]\n\n x_real, y_org = input['x_src'], input['y_src']\n x_ref, x_ref2, y_trg = input['x_ref'], input['x_ref2'], input['y_ref']\n z_trg, z_trg2 = input['z_trg'], input['z_trg2']\n\n #print(1.5)\n\n # train the discriminator\n with tf.GradientTape() as g:\n g.watch(nets['discriminator'].weights)\n\n d_loss, d_losses_latent = compute_d_loss(\n nets, args, x_real, y_org, y_trg, z_trg=z_trg)\n #self._reset_grad()\n #d_loss.backward()\n grad=g.gradient(d_loss, nets['discriminator'].weights)\n #optims['discriminator'].get_gradients(d_loss, nets['discriminator'].weights)\n optims['discriminator'].apply_gradients(zip(grad, nets['discriminator'].weights))\n\n #print(2)\n\n with tf.GradientTape() as g:\n g.watch(nets['discriminator'].weights)\n d_loss, d_losses_ref = compute_d_loss(\n nets, args, x_real, y_org, y_trg, x_ref=x_ref)\n #self._reset_grad()\n #d_loss.backward()\n grad=g.gradient(d_loss, nets['discriminator'].weights)\n optims['discriminator'].apply_gradients(zip(grad, nets['discriminator'].weights))\n\n #print(3)\n\n # train the generator\n with tf.GradientTape(persistent=True) as g:\n g.watch(nets['generator'].weights)\n g.watch(nets['mapping_network'].weights)\n g.watch(nets['style_encoder'].weights)\n g_loss, g_losses_latent = compute_g_loss(\n nets, args, x_real, y_org, y_trg, z_trgs=[z_trg, z_trg2])\n #self._reset_grad()\n #g_loss.backward()\n grad=g.gradient(g_loss, nets['generator'].weights)\n optims['generator'].apply_gradients(zip(grad, nets['generator'].weights))\n grad=g.gradient(g_loss, nets['mapping_network'].weights)\n optims['mapping_network'].apply_gradients(zip(grad, nets['mapping_network'].weights))\n grad=g.gradient(g_loss, nets['style_encoder'].weights)\n optims['style_encoder'].apply_gradients(zip(grad, nets['style_encoder'].weights))\n del g\n\n #print(4)\n with tf.GradientTape(persistent=True) as g:\n g.watch(nets['generator'].weights)\n g_loss, g_losses_ref = compute_g_loss(\n nets, args, x_real, y_org, y_trg, x_refs=[x_ref, x_ref2])\n #self._reset_grad()\n #g_loss.backward()\n grad=g.gradient(g_loss, nets['generator'].weights)\n optims['generator'].apply_gradients(zip(grad, nets['generator'].weights))\n\n #print(5)\n\n # compute moving average of network parameters\n moving_average(nets['generator'], nets_ema['generator'], beta=0.999)\n moving_average(nets['mapping_network'], nets_ema['mapping_network'], beta=0.999)\n moving_average(nets['style_encoder'], nets_ema['style_encoder'], beta=0.999)\n\n #print(6)\n\n # decay weight for diversity sensitive loss\n if args['lambda_ds'] > 0:\n args['lambda_ds'] -= (initial_lambda_ds / args['ds_iter'])\n\n # print out log info\n if (i+1) % args['print_every'] == 0:\n elapsed = time.time() - start_time\n elapsed = str(datetime.timedelta(seconds=elapsed))[:-7]\n log = \"Elapsed time [%s], Iteration [%i/%i], \" % (elapsed, i+1, args['total_iters'])\n all_losses = {}\n for loss, prefix in [(d_losses_latent,'D/latent_'), (d_losses_ref,'D/ref_'), \n (g_losses_latent,'G/latent_'), (g_losses_ref,'G/ref_')]:\n for key, value in loss.items():\n all_losses[prefix + key] = value\n all_losses['G/lambda_ds'] = args['lambda_ds']\n for key, value in all_losses.items():\n if key!= 'G/lambda_ds':\n print(log+key, value.numpy())\n else:\n print(log+key, value)\n\n # generate images for debugging\n #if (i+1) % args['sample_every'] == 0:\n # os.makedirs(args['sample_dir'], exist_ok=True)\n # debug_image(nets_ema, args, inputs=inputs_val, step=i+1)\n\n # save model checkpoints\n if (i+1) % args['save_every'] == 0:\n for net in solv.nets.keys():\n solv.nets[net].save_weights('MNIST_GAN_2/saved_model/'+net+'step'+str(self.step)+'.h5')\n for net in solv.nets_ema.keys():\n solv.nets[net].save_weights('MNIST_GAN_2/saved_model/'+net+'step'+str(self.step)+'_ema.h5')\n \n \n # self._save_checkpoint(step=i+1)\n\n def sample(self, src, ref):\n \"\"\"\n src source image that we want to modify\n ref pair of reference image and domain\n\n generates an image that changes source image into the style of the reference image \n \"\"\"\n args = self.args\n nets_ema = self.nets_ema\n os.makedirs(args['result_dir'], exist_ok=True)\n self._load_checkpoint(args['resume_iter'])\n\n fname = ospj(args['result_dir'], 'reference.jpg')\n print('Working on {}...'.format(fname))\n translate_using_reference(nets_ema, args, src, ref[0], ref[1], fname)", "_____no_output_____" ] ], [ [ "# Data Loading and Preprocessing", "_____no_output_____" ] ], [ [ "(trainX, trainy), (valX, valy) = load_data()\n\ntrainX=tf.reshape(trainX, (60000,1,28,28,1))\nvalX=tf.reshape(valX, (10000,1,28,28,1))", "_____no_output_____" ], [ "\ninputs=[]\nlatent_dim=8\nfor i in range(6000):\n i=i+36000\n if i % 2000==1999:\n print(i+1)\n input={}\n input['x_src']=tf.cast(trainX[i],tf.float32)\n input['y_src']=int(trainy[i])\n n=np.random.randint(0,60000)\n input['x_ref']=tf.cast(trainX[n],tf.float32)\n input['x_ref2']=tf.cast(trainX[np.random.randint(0,60000)],tf.float32)\n input['y_ref']=int(trainy[n])\n input['z_trg']=tf.random.normal((1,latent_dim))\n input['z_trg2']=tf.random.normal((1,latent_dim))\n inputs.append(input)", "38000\n40000\n42000\n" ] ], [ [ "# Parameters\n\nThis dictionary contains the different parameters we use to run the model.", "_____no_output_____" ] ], [ [ "args={'img_size':28,\n 'style_dim':24,\n 'latent_dim':16,\n 'num_domains':10,\n 'lambda_reg':1, \n 'lambda_ds':1,\n 'lambda_sty':10,\n 'lambda_cyc':10,\n 'hidden_dim':128,\n 'resume_iter':0,\n 'ds_iter':6000, \n 'total_iters':6000,\n 'batch_size':8,\n 'val_batch_size':32, \n 'lr':1e-4,\n 'f_lr':1e-6,\n 'beta1':0,\n 'beta2':0.99,\n 'weight_decay':1e-4,\n 'num_outs_per_domain':4,\n 'mode': 'train', #train,sample,eval\n 'seed':0,\n 'train_img_dir':'GAN/data/train',\n 'val_img_dir': 'GAN/data/val',\n 'sample_dir':'GAN/res/samples',\n 'checkpoint_dir':'GAN/res/checkpoints',\n 'eval_dir':'GAN/res/eval',\n 'result_dir':'GAN/res/results',\n 'src_dir':'GAN/data/src', \n 'ref_dir':'GAN/data/ref',\n 'print_every': 500,\n 'sample_every':200,\n 'save_every':1000,\n 'eval_every':1000 }", "_____no_output_____" ] ], [ [ "# Load Model", "_____no_output_____" ] ], [ [ "solv=Solver(args)\nsolv.build_model(args)\nsolv.load(96000)", "_____no_output_____" ] ], [ [ "# Training", "_____no_output_____" ] ], [ [ "with tf.device('/device:GPU:0'):\n solv.train(inputs, inputs)", "Start training...\nWARNING:tensorflow:Calling GradientTape.gradient on a persistent tape inside its context is significantly less efficient than calling it outside the context (it causes the gradient ops to be recorded on the tape, leading to increased CPU and memory usage). Only call GradientTape.gradient inside the context if you actually want to trace the gradient in order to compute higher order derivatives.\nWARNING:tensorflow:Calling GradientTape.gradient on a persistent tape inside its context is significantly less efficient than calling it outside the context (it causes the gradient ops to be recorded on the tape, leading to increased CPU and memory usage). Only call GradientTape.gradient inside the context if you actually want to trace the gradient in order to compute higher order derivatives.\nWARNING:tensorflow:Calling GradientTape.gradient on a persistent tape inside its context is significantly less efficient than calling it outside the context (it causes the gradient ops to be recorded on the tape, leading to increased CPU and memory usage). Only call GradientTape.gradient inside the context if you actually want to trace the gradient in order to compute higher order derivatives.\nWARNING:tensorflow:Calling GradientTape.gradient on a persistent tape inside its context is significantly less efficient than calling it outside the context (it causes the gradient ops to be recorded on the tape, leading to increased CPU and memory usage). Only call GradientTape.gradient inside the context if you actually want to trace the gradient in order to compute higher order derivatives.\nElapsed time [0:18:08], Iteration [500/6000], D/latent_real 5.933643e-06\nElapsed time [0:18:08], Iteration [500/6000], D/latent_fake 1.3552595e-08\nElapsed time [0:18:08], Iteration [500/6000], D/latent_reg 5.7575995e-05\nElapsed time [0:18:08], Iteration [500/6000], D/ref_real 5.965053e-06\nElapsed time [0:18:08], Iteration [500/6000], D/ref_fake 1.6118185e-16\nElapsed time [0:18:08], Iteration [500/6000], D/ref_reg 5.745954e-05\nElapsed time [0:18:08], Iteration [500/6000], G/latent_adv 18.11265\nElapsed time [0:18:08], Iteration [500/6000], G/latent_sty 0.03147651\nElapsed time [0:18:08], Iteration [500/6000], G/latent_ds 35.073742\nElapsed time [0:18:08], Iteration [500/6000], G/latent_cyc 35.35375\nElapsed time [0:18:08], Iteration [500/6000], G/ref_adv 37.213753\nElapsed time [0:18:08], Iteration [500/6000], G/ref_sty 941.0832\nElapsed time [0:18:08], Iteration [500/6000], G/ref_ds 0.6424262\nElapsed time [0:18:08], Iteration [500/6000], G/ref_cyc 35.773685\nElapsed time [0:18:08], Iteration [500/6000], G/lambda_ds 0.9166666666666758\nElapsed time [0:34:51], Iteration [1000/6000], D/latent_real 3.2877884e-05\nElapsed time [0:34:51], Iteration [1000/6000], D/latent_fake 5.9242324e-05\nElapsed time [0:34:51], Iteration [1000/6000], D/latent_reg 1.7900673e-05\nElapsed time [0:34:51], Iteration [1000/6000], D/ref_real 3.2785334e-05\nElapsed time [0:34:51], Iteration [1000/6000], D/ref_fake 3.3252056e-07\nElapsed time [0:34:51], Iteration [1000/6000], D/ref_reg 1.7910741e-05\nElapsed time [0:34:51], Iteration [1000/6000], G/latent_adv 9.766323\nElapsed time [0:34:51], Iteration [1000/6000], G/latent_sty 0.04155442\nElapsed time [0:34:51], Iteration [1000/6000], G/latent_ds 33.543137\nElapsed time [0:34:51], Iteration [1000/6000], G/latent_cyc 30.618359\nElapsed time [0:34:51], Iteration [1000/6000], G/ref_adv 14.91778\nElapsed time [0:34:51], Iteration [1000/6000], G/ref_sty 4931.98\nElapsed time [0:34:51], Iteration [1000/6000], G/ref_ds 3.2844734\nElapsed time [0:34:51], Iteration [1000/6000], G/ref_cyc 30.256683\nElapsed time [0:34:51], Iteration [1000/6000], G/lambda_ds 0.8333333333333517\nElapsed time [0:51:21], Iteration [1500/6000], D/latent_real 3.895065e-06\nElapsed time [0:51:21], Iteration [1500/6000], D/latent_fake 1.8259102e-09\nElapsed time [0:51:21], Iteration [1500/6000], D/latent_reg 2.7209067e-05\nElapsed time [0:51:21], Iteration [1500/6000], D/ref_real 3.889516e-06\nElapsed time [0:51:21], Iteration [1500/6000], D/ref_fake 3.044195e-13\nElapsed time [0:51:21], Iteration [1500/6000], D/ref_reg 2.7181919e-05\nElapsed time [0:51:21], Iteration [1500/6000], G/latent_adv 20.124987\nElapsed time [0:51:21], Iteration [1500/6000], G/latent_sty 0.042978738\nElapsed time [0:51:21], Iteration [1500/6000], G/latent_ds 6.781559\nElapsed time [0:51:21], Iteration [1500/6000], G/latent_cyc 24.195248\nElapsed time [0:51:21], Iteration [1500/6000], G/ref_adv 28.627546\nElapsed time [0:51:21], Iteration [1500/6000], G/ref_sty 2035.4679\nElapsed time [0:51:21], Iteration [1500/6000], G/ref_ds 1.2784641\nElapsed time [0:51:21], Iteration [1500/6000], G/ref_cyc 21.998203\nElapsed time [0:51:21], Iteration [1500/6000], G/lambda_ds 0.7500000000000275\nElapsed time [1:07:58], Iteration [2000/6000], D/latent_real 2.5237994e-06\nElapsed time [1:07:58], Iteration [2000/6000], D/latent_fake 1.6464643e-12\nElapsed time [1:07:58], Iteration [2000/6000], D/latent_reg 2.9553012e-05\nElapsed time [1:07:58], Iteration [2000/6000], D/ref_real 2.5142215e-06\nElapsed time [1:07:58], Iteration [2000/6000], D/ref_fake 1.6905084e-09\nElapsed time [1:07:58], Iteration [2000/6000], D/ref_reg 2.9505836e-05\nElapsed time [1:07:58], Iteration [2000/6000], G/latent_adv 27.145195\nElapsed time [1:07:58], Iteration [2000/6000], G/latent_sty 0.083152466\nElapsed time [1:07:58], Iteration [2000/6000], G/latent_ds 32.264793\nElapsed time [1:07:58], Iteration [2000/6000], G/latent_cyc 35.44804\nElapsed time [1:07:58], Iteration [2000/6000], G/ref_adv 20.16832\nElapsed time [1:07:58], Iteration [2000/6000], G/ref_sty 2833.9543\nElapsed time [1:07:58], Iteration [2000/6000], G/ref_ds 5.3448505\nElapsed time [1:07:58], Iteration [2000/6000], G/ref_cyc 35.21791\nElapsed time [1:07:58], Iteration [2000/6000], G/lambda_ds 0.6666666666667034\nElapsed time [1:24:39], Iteration [2500/6000], D/latent_real 0.00011105127\nElapsed time [1:24:39], Iteration [2500/6000], D/latent_fake 1.5896394e-14\nElapsed time [1:24:39], Iteration [2500/6000], D/latent_reg 0.002541282\nElapsed time [1:24:39], Iteration [2500/6000], D/ref_real 0.000100846126\nElapsed time [1:24:39], Iteration [2500/6000], D/ref_fake 7.2359145e-16\nElapsed time [1:24:39], Iteration [2500/6000], D/ref_reg 0.0024638632\nElapsed time [1:24:39], Iteration [2500/6000], G/latent_adv 31.634012\nElapsed time [1:24:39], Iteration [2500/6000], G/latent_sty 0.03502376\nElapsed time [1:24:39], Iteration [2500/6000], G/latent_ds 17.42342\nElapsed time [1:24:39], Iteration [2500/6000], G/latent_cyc 18.593584\nElapsed time [1:24:39], Iteration [2500/6000], G/ref_adv 34.839787\nElapsed time [1:24:39], Iteration [2500/6000], G/ref_sty 3970.8281\nElapsed time [1:24:39], Iteration [2500/6000], G/ref_ds 0.8345002\nElapsed time [1:24:39], Iteration [2500/6000], G/ref_cyc 18.072935\nElapsed time [1:24:39], Iteration [2500/6000], G/lambda_ds 0.5833333333333792\nElapsed time [1:41:22], Iteration [3000/6000], D/latent_real 5.3445833e-06\nElapsed time [1:41:22], Iteration [3000/6000], D/latent_fake 8.820166e-10\nElapsed time [1:41:22], Iteration [3000/6000], D/latent_reg 6.4297914e-05\nElapsed time [1:41:22], Iteration [3000/6000], D/ref_real 5.36975e-06\nElapsed time [1:41:22], Iteration [3000/6000], D/ref_fake 3.0929835e-13\nElapsed time [1:41:22], Iteration [3000/6000], D/ref_reg 6.230037e-05\nElapsed time [1:41:22], Iteration [3000/6000], G/latent_adv 20.843857\nElapsed time [1:41:22], Iteration [3000/6000], G/latent_sty 0.03444063\nElapsed time [1:41:22], Iteration [3000/6000], G/latent_ds 9.778823\nElapsed time [1:41:22], Iteration [3000/6000], G/latent_cyc 28.097446\nElapsed time [1:41:22], Iteration [3000/6000], G/ref_adv 28.774145\nElapsed time [1:41:22], Iteration [3000/6000], G/ref_sty 2043.1276\nElapsed time [1:41:22], Iteration [3000/6000], G/ref_ds 0.5735893\nElapsed time [1:41:22], Iteration [3000/6000], G/ref_cyc 28.21649\nElapsed time [1:41:22], Iteration [3000/6000], G/lambda_ds 0.5000000000000551\nElapsed time [1:58:17], Iteration [3500/6000], D/latent_real 4.4489816e-06\nElapsed time [1:58:17], Iteration [3500/6000], D/latent_fake 3.4955981e-12\nElapsed time [1:58:17], Iteration [3500/6000], D/latent_reg 1.8419665e-05\nElapsed time [1:58:17], Iteration [3500/6000], D/ref_real 4.4123353e-06\nElapsed time [1:58:17], Iteration [3500/6000], D/ref_fake 4.5671822e-18\nElapsed time [1:58:17], Iteration [3500/6000], D/ref_reg 1.8420711e-05\nElapsed time [1:58:17], Iteration [3500/6000], G/latent_adv 26.39033\nElapsed time [1:58:17], Iteration [3500/6000], G/latent_sty 0.024623169\nElapsed time [1:58:17], Iteration [3500/6000], G/latent_ds 2.0222964\nElapsed time [1:58:17], Iteration [3500/6000], G/latent_cyc 28.361814\nElapsed time [1:58:17], Iteration [3500/6000], G/ref_adv 40.008335\nElapsed time [1:58:17], Iteration [3500/6000], G/ref_sty 825.08575\nElapsed time [1:58:17], Iteration [3500/6000], G/ref_ds 0.13139339\nElapsed time [1:58:17], Iteration [3500/6000], G/ref_cyc 27.434856\nElapsed time [1:58:17], Iteration [3500/6000], G/lambda_ds 0.4166666666667309\nElapsed time [2:14:58], Iteration [4000/6000], D/latent_real 3.6162882e-07\nElapsed time [2:14:58], Iteration [4000/6000], D/latent_fake 5.1923527e-10\nElapsed time [2:14:58], Iteration [4000/6000], D/latent_reg 3.846259e-05\nElapsed time [2:14:58], Iteration [4000/6000], D/ref_real 3.6288532e-07\nElapsed time [2:14:58], Iteration [4000/6000], D/ref_fake 5.942721e-08\nElapsed time [2:14:58], Iteration [4000/6000], D/ref_reg 3.843496e-05\nElapsed time [2:14:58], Iteration [4000/6000], G/latent_adv 21.378086\nElapsed time [2:14:58], Iteration [4000/6000], G/latent_sty 0.106728025\nElapsed time [2:14:58], Iteration [4000/6000], G/latent_ds 20.904701\nElapsed time [2:14:58], Iteration [4000/6000], G/latent_cyc 42.642372\nElapsed time [2:14:58], Iteration [4000/6000], G/ref_adv 17.064861\nElapsed time [2:14:58], Iteration [4000/6000], G/ref_sty 51.672703\nElapsed time [2:14:58], Iteration [4000/6000], G/ref_ds 2.3570015\nElapsed time [2:14:58], Iteration [4000/6000], G/ref_cyc 42.118973\nElapsed time [2:14:58], Iteration [4000/6000], G/lambda_ds 0.33333333333340676\nElapsed time [2:31:26], Iteration [4500/6000], D/latent_real 1.2024437e-06\nElapsed time [2:31:26], Iteration [4500/6000], D/latent_fake 9.023747e-11\nElapsed time [2:31:26], Iteration [4500/6000], D/latent_reg 2.4809478e-05\nElapsed time [2:31:26], Iteration [4500/6000], D/ref_real 1.2103014e-06\nElapsed time [2:31:26], Iteration [4500/6000], D/ref_fake 1.2240717e-15\nElapsed time [2:31:26], Iteration [4500/6000], D/ref_reg 2.4771667e-05\nElapsed time [2:31:26], Iteration [4500/6000], G/latent_adv 23.126303\nElapsed time [2:31:26], Iteration [4500/6000], G/latent_sty 0.05421421\nElapsed time [2:31:26], Iteration [4500/6000], G/latent_ds 22.202696\nElapsed time [2:31:26], Iteration [4500/6000], G/latent_cyc 26.787922\nElapsed time [2:31:26], Iteration [4500/6000], G/ref_adv 34.32032\nElapsed time [2:31:26], Iteration [4500/6000], G/ref_sty 1543.9764\nElapsed time [2:31:26], Iteration [4500/6000], G/ref_ds 0.5546016\nElapsed time [2:31:26], Iteration [4500/6000], G/ref_cyc 26.969646\nElapsed time [2:31:26], Iteration [4500/6000], G/lambda_ds 0.2500000000000826\nElapsed time [2:47:51], Iteration [5000/6000], D/latent_real 3.7615814e-06\nElapsed time [2:47:51], Iteration [5000/6000], D/latent_fake 6.1679136e-19\nElapsed time [2:47:51], Iteration [5000/6000], D/latent_reg 1.4114717e-05\nElapsed time [2:47:51], Iteration [5000/6000], D/ref_real 3.7170662e-06\nElapsed time [2:47:51], Iteration [5000/6000], D/ref_fake 1.3349664e-15\nElapsed time [2:47:51], Iteration [5000/6000], D/ref_reg 1.4105939e-05\nElapsed time [2:47:51], Iteration [5000/6000], G/latent_adv 42.000294\nElapsed time [2:47:51], Iteration [5000/6000], G/latent_sty 0.03513376\nElapsed time [2:47:51], Iteration [5000/6000], G/latent_ds 2.6810305\nElapsed time [2:47:51], Iteration [5000/6000], G/latent_cyc 16.486334\nElapsed time [2:47:51], Iteration [5000/6000], G/ref_adv 34.27865\nElapsed time [2:47:51], Iteration [5000/6000], G/ref_sty 1753.359\nElapsed time [2:47:51], Iteration [5000/6000], G/ref_ds 0.62547046\nElapsed time [2:47:51], Iteration [5000/6000], G/ref_cyc 17.005342\nElapsed time [2:47:51], Iteration [5000/6000], G/lambda_ds 0.16666666666674457\nElapsed time [3:04:31], Iteration [5500/6000], D/latent_real 9.165446e-07\nElapsed time [3:04:31], Iteration [5500/6000], D/latent_fake 9.630188e-10\nElapsed time [3:04:31], Iteration [5500/6000], D/latent_reg 2.2647702e-05\nElapsed time [3:04:31], Iteration [5500/6000], D/ref_real 9.2337206e-07\nElapsed time [3:04:31], Iteration [5500/6000], D/ref_fake 1.4190508e-07\nElapsed time [3:04:31], Iteration [5500/6000], D/ref_reg 2.254527e-05\nElapsed time [3:04:31], Iteration [5500/6000], G/latent_adv 20.79223\nElapsed time [3:04:31], Iteration [5500/6000], G/latent_sty 0.026682168\nElapsed time [3:04:31], Iteration [5500/6000], G/latent_ds 11.481898\nElapsed time [3:04:31], Iteration [5500/6000], G/latent_cyc 28.555185\nElapsed time [3:04:31], Iteration [5500/6000], G/ref_adv 15.918713\nElapsed time [3:04:31], Iteration [5500/6000], G/ref_sty 1588.65\nElapsed time [3:04:31], Iteration [5500/6000], G/ref_ds 0.9953184\nElapsed time [3:04:31], Iteration [5500/6000], G/ref_cyc 29.72444\nElapsed time [3:04:31], Iteration [5500/6000], G/lambda_ds 0.08333333333341\nElapsed time [3:21:09], Iteration [6000/6000], D/latent_real 3.292037e-05\nElapsed time [3:21:09], Iteration [6000/6000], D/latent_fake 2.396092e-10\nElapsed time [3:21:09], Iteration [6000/6000], D/latent_reg 3.2086697e-05\nElapsed time [3:21:09], Iteration [6000/6000], D/ref_real 3.259703e-05\nElapsed time [3:21:09], Iteration [6000/6000], D/ref_fake 7.805121e-13\nElapsed time [3:21:09], Iteration [6000/6000], D/ref_reg 3.2122865e-05\nElapsed time [3:21:09], Iteration [6000/6000], G/latent_adv 22.151058\nElapsed time [3:21:09], Iteration [6000/6000], G/latent_sty 0.03772318\nElapsed time [3:21:09], Iteration [6000/6000], G/latent_ds 4.0022397\nElapsed time [3:21:09], Iteration [6000/6000], G/latent_cyc 45.717335\nElapsed time [3:21:09], Iteration [6000/6000], G/ref_adv 27.884796\nElapsed time [3:21:09], Iteration [6000/6000], G/ref_sty 1092.5536\nElapsed time [3:21:09], Iteration [6000/6000], G/ref_ds 1.0627509\nElapsed time [3:21:09], Iteration [6000/6000], G/ref_cyc 45.289036\nElapsed time [3:21:09], Iteration [6000/6000], G/lambda_ds 7.683496850915961e-14\n" ] ], [ [ "# Results\n\nIn this first cell we show an image where the rows represent a source image and the columns the style they are trying to mimic. We can see in this case that that the image still highly resembles the source image but has obtained some characteristics depending on the style of our reference. In most cases this style is mostly about the thickness of the lines, but it does vary slightly in other ways.", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as pyplot\nfor i in range(4):\n\tpyplot.subplot(5,5,2+i)\n\tpyplot.axis('off')\n\tpyplot.imshow(np.reshape(inputs[i]['x_ref'],[28,28]), cmap='gray_r')\nfor i in range(4):\n\tpyplot.subplot(5, 5, 5*(i+1) + 1)\n\tpyplot.axis('off')\n\tpyplot.imshow(np.reshape(inputs[i]['x_src'], [28,28]), cmap='gray_r')\n\tfor j in range(4):\n\t\tpyplot.subplot(5, 5, 5*(i+1) + j +2)\n\t\tpyplot.axis('off')\n\t\tpyplot.imshow(np.reshape(solv.nets['generator'](inputs[i]['x_src'],solv.nets['style_encoder'](inputs[j]['x_ref'],inputs[j]['y_ref'])).numpy(), [28,28]), cmap='gray_r')\npyplot.show()\n\n#left is source and top is the target trying to mimic its font", "_____no_output_____" ] ], [ [ "Below we generate random styles and see the output it generates. We notice that it is quite likely the images are distorted in this case, compared to when using the style of an already existing image it seems it would usually have a good quality.", "_____no_output_____" ] ], [ [ "for i in range(5):\n\tpyplot.subplot(5,5,1+i)\n\tpyplot.axis('off')\n\tpyplot.imshow(np.reshape(solv.nets['generator'](inputs[0]['x_src'],tf.random.normal((1,24))).numpy(), [28,28]), cmap='gray_r')", "_____no_output_____" ] ], [ [ "Here we can see the process of how the image transforms into the target. In these small images there is not too much that is changing but we can still appreciate the process.", "_____no_output_____" ] ], [ [ "s1=solv.nets['style_encoder'](inputs[3]['x_src'],inputs[3]['y_src'])\ns2=solv.nets['style_encoder'](inputs[3]['x_ref'],inputs[3]['y_ref'])\nfor i in range(5):\n pyplot.subplot(5,5,1+i)\n pyplot.axis('off')\n s=(1-i/5)*s1+i/5*s2\n pyplot.imshow(np.reshape(solv.nets['generator'](inputs[3]['x_src'],s).numpy(), [28,28]), cmap='gray_r')", "_____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" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# git-bakup", "_____no_output_____" ] ], [ [ "USER='tonybutzer'\nAPI_TOKEN='ATOKEN'\nGIT_API_URL='https://api.github.com'\n\ndef get_api(url):\n try:\n request = urllib2.Request(GIT_API_URL + url)\n base64string = base64.encodestring('%s/token:%s' % (USER, API_TOKEN)).replace('\\n', '')\n request.add_header(\"Authorization\", \"Basic %s\" % base64string)\n result = urllib2.urlopen(request)\n result.close()\n except:\n print ('Failed to get api request from %s' % url)", "_____no_output_____" ], [ "!curl \"https://api.github.com/users/tonybutzer/repos?per_page=1000\" | grep -w clone_url | grep -o '[^\"]\\+://.\\+.git' >myrepos.txt", " % Total % Received % Xferd Average Speed Time Time Time Current\n Dload Upload Total Spent Left Speed\n100 170k 0 170k 0 0 224k 0 --:--:-- --:--:-- --:--:-- 223k\n" ], [ "%%bash\n\nmkdir -p ~/repo\n\n\nfor i in `cat myrepos.txt` ; do\n{\necho $i\n(cd ~/repo; git clone $i)\n}; done", "Process is interrupted.\n" ], [ "! ls ~/repo", "active-fire\r\n" ] ] ]

[ "markdown", "code" ]

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "Euler Problem 94\n================\n\nIt is easily proved that no equilateral triangle exists with integral length sides and integral area. However, the almost equilateral triangle 5-5-6 has an area of 12 square units.\n\nWe shall define an almost equilateral triangle to be a triangle for which two sides are equal and the third differs by no more than one unit.\n\nFind the sum of the perimeters of all almost equilateral triangles with integral side lengths and area and whose perimeters do not exceed one billion (1,000,000,000).", "_____no_output_____" ] ], [ [ "a, b, p, s = 1, 0, 0, 0\nwhile p <= 10**9:\n s += p\n a, b = 2*a + 3*b, a + 2*b\n p = 4*a*a\n\na, b, p = 1, 1, 0\nwhile p <= 10**9:\n s += p\n a, b = 2*a + 3*b, a + 2*b\n p = 2*a*a\n\nprint(s)\n", "518408346\n" ] ], [ [ "**Discussion:** ", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import pandas as pd\nimport matplotlib.pyplot as plt\nimport numpy as np\nimport re\nimport spacy\nimport pickle\nimport time\nfrom collections import defaultdict\nimport sys\npath = \"../data/\"\nplt.style.use(\"seaborn-whitegrid\")\nplt.rcParams['figure.figsize'] = [8.0, 6.0]\nplt.rcParams['figure.dpi'] = 140\nplt.rcParams[\"axes.labelsize\"] = 14\n\n#np.random.seed(250)\nnlp = spacy.load(\"en_core_sci_lg\", disable=['ner', 'parser'])", "_____no_output_____" ], [ "import numpy as np\nimport spacy\nfrom tqdm import tqdm\nfrom collections import defaultdict\n\npath = \"../data/\"\n\ndef tokenize(string):\n doc = nlp.make_doc(string)\n words = [doc[i].text.lower() for i in range(len(doc)) if \n doc[i].is_alpha and not doc[i].is_stop]\n mn = 2\n ngrams= [' '.join(words[i:i+n]) for n in range(1, mn+1) for i in range(len(words)-n+1)]\n #words = [token.text.lower() for token in doc if token.is_alpha and not token.is_stop and len(token.text) > 1 ]\n return ngrams\n\ndef tokenization(train_data):\n tokenized_texts = []\n #print(\"Tokenization....\")\n for _, row in train_data.iterrows():\n #text = str(row['Abstract']) + str(row[\"Title\"])\n text = str(row['review'])\n words = tokenize(text)\n tokenized_texts.append(words)\n return tokenized_texts\n\n# TFIDF (Term frequency and inverse document frequency)\ndef get_word_stat(tokenized_texts):\n '''Words counts in documents\n finds in how many documents this word\n is present\n '''\n texts_number = len(tokenized_texts)\n #print(\"Word Stat....\")\n word2text_count = defaultdict(int)\n for text in tokenized_texts:\n uniquewords = set(text)\n for word in uniquewords:\n word2text_count[word] +=1\n return word2text_count\n\ndef get_doc_tfidf(words, word2text_count, N):\n num_words = len(words)\n word2tfidf = defaultdict(int)\n for word in words:\n if word2text_count[word] > 0:\n idf = np.log(N/(word2text_count[word]))\n word2tfidf[word] += (1/num_words) * idf\n else:\n word2tfidf[word] = 1\n return word2tfidf\n\ndef create_pmi_dict(tokenized_texts, targets, min_count=5):\n #print(\"PMI dictionary ....\")\n np.seterr(divide = 'ignore')\n # words count\n d = {0:defaultdict(int), 1:defaultdict(int), 'tot':defaultdict(int)}\n for idx, words in enumerate(tokenized_texts):\n target = targets[idx]\n for w in words:\n d[ target ][w] += 1\n Dictionary = set(list(d[0].keys()) + list(d[1].keys()))\n d['tot'] = {w:d[0][w] + d[1][w] for w in Dictionary}\n # pmi calculation\n N_0 = sum(d[0].values())\n N_1 = sum(d[1].values())\n d[0] = {w: -np.log((v/N_0 + 10**(-15)) / (0.5 * d['tot'][w]/(N_0 + N_1))) / np.log(v/N_0 + 10**(-15))\n for w, v in d[0].items() if d['tot'][w] > min_count}\n d[1] = {w: -np.log((v/N_1+ 10**(-15)) / (0.5 * d['tot'][w]/(N_0 + N_1))) / np.log(v/N_1 + 10**(-15))\n for w, v in d[1].items() if d['tot'][w] > min_count}\n del d['tot']\n return d\n\n\ndef calc_collinearity(word, words_dict, n=10):\n new_word_emb = nlp(word).vector\n pmi_new = 0\n max_pmis_words = sorted(list(words_dict.items()), key=lambda x: x[1], reverse=True)[:n]\n for w, pmi in max_pmis_words:\n w_emb = nlp(w).vector\n cos_similarity = \\\n np.dot(w_emb, new_word_emb)/(np.linalg.norm(w_emb) * np.linalg.norm(new_word_emb) + 1e-12)\n pmi_new += cos_similarity * pmi\n return pmi_new / n\n\n\ndef create_tot_pmitfidf(words, words_pmis, word2tfidf):\n tot_pmitfidf0 = []\n tot_pmitfidf1 = []\n for word in words:\n if word in words_pmis[0]:\n tot_pmitfidf0.append( words_pmis[0][word] * word2tfidf[word] )\n else:\n pmi0idf = pmiidf_net.forward( nlp(word).vector )\n #pmi0 = calc_collinearity(word, words_pmis[0])\n tot_pmitfidf0.append( pmi0 )\n if word in words_pmis[1]:\n tot_pmitfidf1.append( words_pmis[1][word] * word2tfidf[word] )\n else:\n pmi1 = calc_collinearity(word, words_pmis[1])\n tot_pmitfidf1.append( pmi1 )\n\n return tot_pmitfidf0, tot_pmitfidf1\n\n\ndef classify_pmi_based(words_pmis, word2text_count, tokenized_test_texts, N):\n results = np.zeros(len(tokenized_test_texts))\n for idx, words in enumerate(tokenized_test_texts):\n word2tfidf = get_doc_tfidf(words, word2text_count, N)\n # PMI - determines significance of the word for the class\n # TFIDF - determines significance of the word for the document\n #tot_pmi0, tot_pmi1 = create_tot_pmitfidf(words, words_pmis, word2tfidf)\n tot_pmi0 = [ words_pmis[0][w] * word2tfidf[w] for w in set(words) if w in words_pmis[0] ]\n tot_pmi1 = [ words_pmis[1][w] * word2tfidf[w] for w in set(words) if w in words_pmis[1] ]\n pmi0 = np.sum(tot_pmi0)\n pmi1 = np.sum(tot_pmi1)\n diff = pmi1 - pmi0\n if diff > 0.001:\n results[idx] = 1\n return results", "_____no_output_____" ], [ "data_raw = pd.read_csv(path + 'IMDB_Dataset.csv')\nindices = np.random.permutation(data_raw.index)\ndata = data_raw.loc[indices]\ndata = data_raw.sample(frac=1)\ndata = data.replace(to_replace=['negative', 'positive'], value=[0, 1])", "_____no_output_____" ], [ "idx = int(data.shape[0] * 0.1)\ntest_data = data.iloc[:idx]\ntrain_data = data.iloc[idx:]\ntargets_train = train_data[\"sentiment\"].values\ntargets_test = test_data[\"sentiment\"].values", "_____no_output_____" ], [ "tokenized_texts = tokenization(train_data)\ntokenized_test_texts = tokenization(test_data)\nN = len(tokenized_texts)", "_____no_output_____" ], [ "word2text_count = get_word_stat(tokenized_texts)\nwords_pmis = create_pmi_dict(tokenized_texts, targets_train, min_count=5)\nresults = classify_pmi_based(words_pmis, word2text_count, tokenized_test_texts, N)\nprecision = np.sum( np.logical_and(results, targets_test) ) / np.sum(results)\nrecall = np.sum( np.logical_and(results, targets_test) ) / np.sum(targets_test)\nF1 = 2 * (recall * precision)/(recall + precision)\naccuracy = (results == targets_test).mean()", "_____no_output_____" ], [ "print(\"Accuracy: \", accuracy)\nprint(\"Precision: \", precision)\nprint(\"Recall: \", recall)\nprint(\"F1: \", F1)", "Accuracy: 0.9183098591549296\nPrecision: 0.8944591029023746\nRecall: 0.9495798319327731\nF1: 0.921195652173913\n" ], [ "print(\"Accuracy: \", accuracy)\nprint(\"Precision: \", precision)\nprint(\"Recall: \", recall)\nprint(\"F1: \", F1)", "Accuracy: 0.9211267605633803\nPrecision: 0.8981481481481481\nRecall: 0.9509803921568627\nF1: 0.9238095238095237\n" ], [ "print(\"Accuracy: \", accuracy)\nprint(\"Precision: \", precision)\nprint(\"Recall: \", recall)\nprint(\"F1: \", F1)", "Accuracy: 0.9274647887323944\nPrecision: 0.9190672153635117\nRecall: 0.938375350140056\nF1: 0.9286209286209285\n" ], [ "scores = {\"accuracies\":[], \"precisions\":[], \"recalls\":[], \"F1s\":[], \"size\":[]}\ndict_size = [i for i in np.arange(0.02, 1, 0.01)]\nfor i in dict_size:\n part = tokenized_texts[:int(N * i)]\n scores[\"size\"].append(len(part))\n word2text_count = get_word_stat(part)\n words_pmis = create_pmi_dict(part, targets_train, min_count=5)\n\n results = classify_pmi_based(words_pmis, word2text_count, tokenized_test_texts, N)\n\n precision = np.sum( np.logical_and(results, targets_test) ) / np.sum(results)\n recall = np.sum( np.logical_and(results, targets_test) ) / np.sum(targets_test)\n F1 = 2 * (recall * precision)/(recall + precision)\n\n accuracy = (results == targets_test).mean()\n scores[\"accuracies\"].append( accuracy )\n scores[\"precisions\"].append( precision )\n scores[\"recalls\"].append( recall )\n scores[\"F1s\"].append( F1 )", "_____no_output_____" ], [ "fig, axs = plt.subplots(1, 1)\naxs.plot(scores[\"size\"], scores[\"accuracies\"], label=\"Accuracy\")\naxs.plot(scores[\"size\"], scores[\"precisions\"], label=\"Precision\")\naxs.plot(scores[\"size\"], scores[\"recalls\"], label=\"Recall\")\naxs.plot(scores[\"size\"], scores[\"F1s\"], label=\"F1\")\naxs.legend(title=\"Scores\");\naxs.set(xlabel=\"Dictionary size\");\nplt.savefig(\"score_dict_size_IMDB.png\")", "_____no_output_____" ], [ "with open(\"scores_IMDB_ngrams.p\", \"wb\") as file:\n pickle.dump(scores, file)", "_____no_output_____" ], [ "def create_dict(tokenized_texts, targets, min_count=5):\n np.seterr(divide = 'ignore')\n # words count\n d = {0:defaultdict(int), 1:defaultdict(int), 'tot':defaultdict(int)}\n for idx, words in enumerate(tokenized_texts):\n target = targets[idx]\n for w in words:\n d[ target ][w] += 1\n return d", "_____no_output_____" ], [ "dictionary_imdb = create_dict(tokenized_texts, targets_train, min_count=5)", "_____no_output_____" ], [ "with open(\"dict_IMDB.p\", \"wb\") as file:\n pickle.dump(dictionary_imdb, file)", "_____no_output_____" ] ] ]

[ "code" ]

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Now You Code 4: Temperature Conversion\n\nWrite a python program which will convert temperatures from Celcius to Fahrenheight.\nThe program should take a temperature in degrees Celcius as input and output a temperature in degrees Fahrenheight.\n\nExample:\n\n```\nEnter the temperature in Celcius: 100\n100 Celcius is 212 Fahrenheight\n```\n\nHINT: Use the web to find the formula to convert from Celcius to Fahrenheight.\n", "_____no_output_____" ], [ "## Step 1: Problem Analysis\n\nInputs: celcius and fahrenhieght\n\nOutputs: celcius to farenhieght \n\nAlgorithm (Steps in Program):\n\n\n", "_____no_output_____" ] ], [ [ "celcius = float(input(\"enter the temperature in celcius: \")) \nfahrenhieght=(celcius*9/5)+32\nprint(\"fahrenhieght equals \" \"%.2f\" %fahrenhieght) ", "enter the temperature in celcius: 100\nfahrenhieght equals 212.00\n" ] ], [ [ "## Step 3: Questions\n\n1. Why does the program still run when you enter a negative number for temperature? Is this an error? because we didn't put in a string that wont accept negative numbers, no its not unless we programmed it to.\n\n2. Would it be difficult to write a program which did the opposite (conversion F to C)? Explain. no because you would right the same thing you did for C to F but just a different formula \n\n3. Did you store the conversion in a variable before printing it on the last line? I argue this makes your program easier to understand. Why? because checking each line makes sure that your code is running smoothly unless changes are needed to be made.\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" ]

[ [ [ "from keras.models import Sequential\nfrom keras.layers import Dense\nimport numpy as np", "Using TensorFlow backend.\n" ], [ "np.random.seed(7)", "_____no_output_____" ], [ "dataset = np.loadtxt('pima-indians-diabetes.data', delimiter=',')\nX = dataset[:, 0:8]\nY = dataset[:, 8]", "_____no_output_____" ], [ "model = Sequential()\nmodel.add(Dense(12, input_dim=8, activation='relu'))\nmodel.add(Dense(8, activation='relu'))\nmodel.add(Dense(1, activation='sigmoid'))\nmodel.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])", "_____no_output_____" ], [ "model.fit(X, Y, validation_split=0.33, epochs=150, batch_size=10)", "Train on 514 samples, validate on 254 samples\nEpoch 1/150\n514/514 [==============================] - 0s - loss: 6.0377 - acc: 0.3852 - val_loss: 4.3065 - val_acc: 0.5118\nEpoch 2/150\n514/514 [==============================] - 0s - loss: 2.4940 - acc: 0.5409 - val_loss: 1.9934 - val_acc: 0.5118\nEpoch 3/150\n514/514 [==============================] - 0s - loss: 1.5921 - acc: 0.5603 - val_loss: 1.7380 - val_acc: 0.5197\nEpoch 4/150\n514/514 [==============================] - 0s - loss: 1.3562 - acc: 0.5720 - val_loss: 1.4103 - val_acc: 0.5354\nEpoch 5/150\n514/514 [==============================] - 0s - loss: 1.1636 - acc: 0.5895 - val_loss: 1.3189 - val_acc: 0.5039\nEpoch 6/150\n514/514 [==============================] - 0s - loss: 1.0265 - acc: 0.5642 - val_loss: 1.1053 - val_acc: 0.5433\nEpoch 7/150\n514/514 [==============================] - 0s - loss: 0.9333 - acc: 0.5856 - val_loss: 1.0059 - val_acc: 0.5827\nEpoch 8/150\n514/514 [==============================] - 0s - loss: 0.8754 - acc: 0.6109 - val_loss: 0.9774 - val_acc: 0.5669\nEpoch 9/150\n514/514 [==============================] - 0s - loss: 0.8073 - acc: 0.6187 - val_loss: 0.9216 - val_acc: 0.6142\nEpoch 10/150\n514/514 [==============================] - 0s - loss: 0.7870 - acc: 0.6167 - val_loss: 0.8556 - val_acc: 0.6063\nEpoch 11/150\n514/514 [==============================] - 0s - loss: 0.7494 - acc: 0.6323 - val_loss: 0.8728 - val_acc: 0.6181\nEpoch 12/150\n514/514 [==============================] - 0s - loss: 0.7397 - acc: 0.6498 - val_loss: 0.8721 - val_acc: 0.6496\nEpoch 13/150\n514/514 [==============================] - 0s - loss: 0.6973 - acc: 0.6576 - val_loss: 0.8084 - val_acc: 0.6339\nEpoch 14/150\n514/514 [==============================] - 0s - loss: 0.7233 - acc: 0.6459 - val_loss: 0.8645 - val_acc: 0.6102\nEpoch 15/150\n514/514 [==============================] - 0s - loss: 0.6903 - acc: 0.6712 - val_loss: 0.7862 - val_acc: 0.6417\nEpoch 16/150\n514/514 [==============================] - 0s - loss: 0.6798 - acc: 0.6712 - val_loss: 0.7577 - val_acc: 0.6220\nEpoch 17/150\n514/514 [==============================] - 0s - loss: 0.6971 - acc: 0.6595 - val_loss: 0.7647 - val_acc: 0.6417\nEpoch 18/150\n514/514 [==============================] - 0s - loss: 0.6533 - acc: 0.6712 - val_loss: 0.7759 - val_acc: 0.6181\nEpoch 19/150\n514/514 [==============================] - 0s - loss: 0.6643 - acc: 0.6907 - val_loss: 0.7405 - val_acc: 0.6417\nEpoch 20/150\n514/514 [==============================] - 0s - loss: 0.6687 - acc: 0.6751 - val_loss: 0.7224 - val_acc: 0.6535\nEpoch 21/150\n514/514 [==============================] - 0s - loss: 0.6264 - acc: 0.6829 - val_loss: 0.7075 - val_acc: 0.6535\nEpoch 22/150\n514/514 [==============================] - 0s - loss: 0.6632 - acc: 0.6907 - val_loss: 0.7352 - val_acc: 0.6614\nEpoch 23/150\n514/514 [==============================] - 0s - loss: 0.6391 - acc: 0.6770 - val_loss: 0.7112 - val_acc: 0.6732\nEpoch 24/150\n514/514 [==============================] - 0s - loss: 0.6227 - acc: 0.6907 - val_loss: 0.7241 - val_acc: 0.6457\nEpoch 25/150\n514/514 [==============================] - 0s - loss: 0.6485 - acc: 0.6693 - val_loss: 0.6954 - val_acc: 0.6496\nEpoch 26/150\n514/514 [==============================] - 0s - loss: 0.6084 - acc: 0.7023 - val_loss: 0.7477 - val_acc: 0.6457\nEpoch 27/150\n514/514 [==============================] - 0s - loss: 0.6227 - acc: 0.6946 - val_loss: 0.7741 - val_acc: 0.6535\nEpoch 28/150\n514/514 [==============================] - 0s - loss: 0.6088 - acc: 0.7004 - val_loss: 0.6983 - val_acc: 0.6850\nEpoch 29/150\n514/514 [==============================] - 0s - loss: 0.6352 - acc: 0.6712 - val_loss: 0.6995 - val_acc: 0.6732\nEpoch 30/150\n514/514 [==============================] - 0s - loss: 0.6027 - acc: 0.6946 - val_loss: 0.6734 - val_acc: 0.6457\nEpoch 31/150\n514/514 [==============================] - 0s - loss: 0.6113 - acc: 0.6984 - val_loss: 0.7503 - val_acc: 0.6535\nEpoch 32/150\n514/514 [==============================] - 0s - loss: 0.5939 - acc: 0.6984 - val_loss: 0.6500 - val_acc: 0.6969\nEpoch 33/150\n514/514 [==============================] - 0s - loss: 0.5920 - acc: 0.6887 - val_loss: 0.7098 - val_acc: 0.6732\nEpoch 34/150\n514/514 [==============================] - 0s - loss: 0.6464 - acc: 0.6790 - val_loss: 0.6538 - val_acc: 0.6850\nEpoch 35/150\n514/514 [==============================] - 0s - loss: 0.5878 - acc: 0.7276 - val_loss: 0.6414 - val_acc: 0.6929\nEpoch 36/150\n514/514 [==============================] - 0s - loss: 0.5998 - acc: 0.7160 - val_loss: 0.6339 - val_acc: 0.6929\nEpoch 37/150\n514/514 [==============================] - 0s - loss: 0.5907 - acc: 0.7023 - val_loss: 0.6194 - val_acc: 0.7087\nEpoch 38/150\n514/514 [==============================] - 0s - loss: 0.5708 - acc: 0.7276 - val_loss: 0.6212 - val_acc: 0.7047\nEpoch 39/150\n514/514 [==============================] - 0s - loss: 0.6052 - acc: 0.6984 - val_loss: 0.6567 - val_acc: 0.6969\nEpoch 40/150\n514/514 [==============================] - 0s - loss: 0.6039 - acc: 0.6965 - val_loss: 0.6598 - val_acc: 0.6732\nEpoch 41/150\n514/514 [==============================] - 0s - loss: 0.5963 - acc: 0.6984 - val_loss: 0.6576 - val_acc: 0.6850\nEpoch 42/150\n514/514 [==============================] - 0s - loss: 0.6217 - acc: 0.7257 - val_loss: 0.6581 - val_acc: 0.6850\nEpoch 43/150\n514/514 [==============================] - 0s - loss: 0.5799 - acc: 0.7062 - val_loss: 0.6287 - val_acc: 0.7008\nEpoch 44/150\n514/514 [==============================] - 0s - loss: 0.5630 - acc: 0.7198 - val_loss: 0.6621 - val_acc: 0.6693\nEpoch 45/150\n514/514 [==============================] - 0s - loss: 0.5967 - acc: 0.7062 - val_loss: 0.6510 - val_acc: 0.6929\nEpoch 46/150\n514/514 [==============================] - 0s - loss: 0.6639 - acc: 0.7062 - val_loss: 0.6187 - val_acc: 0.7165\nEpoch 47/150\n514/514 [==============================] - 0s - loss: 0.5891 - acc: 0.7004 - val_loss: 0.6079 - val_acc: 0.7362\nEpoch 48/150\n514/514 [==============================] - 0s - loss: 0.5949 - acc: 0.7121 - val_loss: 0.6315 - val_acc: 0.7087\nEpoch 49/150\n514/514 [==============================] - 0s - loss: 0.5747 - acc: 0.7179 - val_loss: 0.6226 - val_acc: 0.6929\nEpoch 50/150\n514/514 [==============================] - 0s - loss: 0.5634 - acc: 0.7179 - val_loss: 0.6289 - val_acc: 0.7008\nEpoch 51/150\n514/514 [==============================] - 0s - loss: 0.5804 - acc: 0.7101 - val_loss: 0.6843 - val_acc: 0.6850\nEpoch 52/150\n514/514 [==============================] - 0s - loss: 0.5792 - acc: 0.7101 - val_loss: 0.6947 - val_acc: 0.6693\nEpoch 53/150\n514/514 [==============================] - 0s - loss: 0.6082 - acc: 0.7062 - val_loss: 0.6342 - val_acc: 0.6969\nEpoch 54/150\n514/514 [==============================] - 0s - loss: 0.5952 - acc: 0.7160 - val_loss: 0.6449 - val_acc: 0.7047\nEpoch 55/150\n514/514 [==============================] - 0s - loss: 0.5592 - acc: 0.7335 - val_loss: 0.6223 - val_acc: 0.7362\nEpoch 56/150\n514/514 [==============================] - 0s - loss: 0.5631 - acc: 0.7374 - val_loss: 0.6191 - val_acc: 0.7008\nEpoch 57/150\n514/514 [==============================] - 0s - loss: 0.5892 - acc: 0.6946 - val_loss: 0.6368 - val_acc: 0.7441\nEpoch 58/150\n514/514 [==============================] - 0s - loss: 0.6010 - acc: 0.7121 - val_loss: 0.5870 - val_acc: 0.7441\nEpoch 59/150\n514/514 [==============================] - 0s - loss: 0.5910 - acc: 0.6946 - val_loss: 0.6003 - val_acc: 0.7047\nEpoch 60/150\n514/514 [==============================] - 0s - loss: 0.5902 - acc: 0.6965 - val_loss: 0.5918 - val_acc: 0.7283\nEpoch 61/150\n514/514 [==============================] - 0s - loss: 0.5518 - acc: 0.7451 - val_loss: 0.6086 - val_acc: 0.7165\nEpoch 62/150\n514/514 [==============================] - 0s - loss: 0.5456 - acc: 0.7432 - val_loss: 0.6439 - val_acc: 0.6850\nEpoch 63/150\n514/514 [==============================] - 0s - loss: 0.5788 - acc: 0.7257 - val_loss: 0.6009 - val_acc: 0.7402\nEpoch 64/150\n514/514 [==============================] - 0s - loss: 0.5712 - acc: 0.7315 - val_loss: 0.6342 - val_acc: 0.7008\nEpoch 65/150\n514/514 [==============================] - 0s - loss: 0.5477 - acc: 0.7393 - val_loss: 0.6095 - val_acc: 0.7244\nEpoch 66/150\n514/514 [==============================] - 0s - loss: 0.5357 - acc: 0.7529 - val_loss: 0.6311 - val_acc: 0.7165\nEpoch 67/150\n514/514 [==============================] - 0s - loss: 0.5707 - acc: 0.7179 - val_loss: 0.6251 - val_acc: 0.7126\nEpoch 68/150\n514/514 [==============================] - 0s - loss: 0.5493 - acc: 0.7198 - val_loss: 0.6012 - val_acc: 0.7165\nEpoch 69/150\n514/514 [==============================] - 0s - loss: 0.5646 - acc: 0.7140 - val_loss: 0.5979 - val_acc: 0.7165\nEpoch 70/150\n514/514 [==============================] - 0s - loss: 0.5684 - acc: 0.7140 - val_loss: 0.6257 - val_acc: 0.6929\nEpoch 71/150\n514/514 [==============================] - 0s - loss: 0.5675 - acc: 0.7354 - val_loss: 0.6413 - val_acc: 0.6850\nEpoch 72/150\n514/514 [==============================] - 0s - loss: 0.5641 - acc: 0.7198 - val_loss: 0.6093 - val_acc: 0.6969\nEpoch 73/150\n514/514 [==============================] - 0s - loss: 0.5663 - acc: 0.7374 - val_loss: 0.5954 - val_acc: 0.7126\nEpoch 74/150\n514/514 [==============================] - 0s - loss: 0.5756 - acc: 0.7510 - val_loss: 0.6067 - val_acc: 0.7402\nEpoch 75/150\n514/514 [==============================] - 0s - loss: 0.5591 - acc: 0.7179 - val_loss: 0.5778 - val_acc: 0.7283\nEpoch 76/150\n514/514 [==============================] - 0s - loss: 0.5493 - acc: 0.7568 - val_loss: 0.5987 - val_acc: 0.7244\nEpoch 77/150\n514/514 [==============================] - 0s - loss: 0.5427 - acc: 0.7315 - val_loss: 0.6136 - val_acc: 0.7087\nEpoch 78/150\n514/514 [==============================] - 0s - loss: 0.5495 - acc: 0.7471 - val_loss: 0.6016 - val_acc: 0.7165\nEpoch 79/150\n514/514 [==============================] - 0s - loss: 0.5748 - acc: 0.7237 - val_loss: 0.5972 - val_acc: 0.7087\nEpoch 80/150\n514/514 [==============================] - 0s - loss: 0.5570 - acc: 0.7121 - val_loss: 0.5873 - val_acc: 0.7520\nEpoch 81/150\n514/514 [==============================] - 0s - loss: 0.5633 - acc: 0.7335 - val_loss: 0.6893 - val_acc: 0.6811\nEpoch 82/150\n514/514 [==============================] - 0s - loss: 0.5575 - acc: 0.7451 - val_loss: 0.6777 - val_acc: 0.6417\nEpoch 83/150\n514/514 [==============================] - 0s - loss: 0.5510 - acc: 0.7529 - val_loss: 0.6004 - val_acc: 0.7205\nEpoch 84/150\n514/514 [==============================] - 0s - loss: 0.5339 - acc: 0.7549 - val_loss: 0.6571 - val_acc: 0.6929\nEpoch 85/150\n514/514 [==============================] - 0s - loss: 0.5481 - acc: 0.7374 - val_loss: 0.6087 - val_acc: 0.7047\nEpoch 86/150\n514/514 [==============================] - 0s - loss: 0.5507 - acc: 0.7335 - val_loss: 0.5764 - val_acc: 0.7441\nEpoch 87/150\n514/514 [==============================] - 0s - loss: 0.5370 - acc: 0.7315 - val_loss: 0.5848 - val_acc: 0.7205\nEpoch 88/150\n514/514 [==============================] - 0s - loss: 0.5514 - acc: 0.7276 - val_loss: 0.6322 - val_acc: 0.7126\nEpoch 89/150\n514/514 [==============================] - 0s - loss: 0.5583 - acc: 0.7374 - val_loss: 0.6930 - val_acc: 0.6929\nEpoch 90/150\n514/514 [==============================] - 0s - loss: 0.5517 - acc: 0.7432 - val_loss: 0.6209 - val_acc: 0.6929\nEpoch 91/150\n514/514 [==============================] - 0s - loss: 0.5502 - acc: 0.7276 - val_loss: 0.5909 - val_acc: 0.7283\nEpoch 92/150\n514/514 [==============================] - 0s - loss: 0.5587 - acc: 0.7490 - val_loss: 0.6103 - val_acc: 0.7165\nEpoch 93/150\n514/514 [==============================] - 0s - loss: 0.5625 - acc: 0.7257 - val_loss: 0.7228 - val_acc: 0.6850\nEpoch 94/150\n514/514 [==============================] - 0s - loss: 0.5532 - acc: 0.7296 - val_loss: 0.6420 - val_acc: 0.6732\nEpoch 95/150\n514/514 [==============================] - 0s - loss: 0.6471 - acc: 0.7179 - val_loss: 0.7324 - val_acc: 0.6811\nEpoch 96/150\n514/514 [==============================] - 0s - loss: 0.5711 - acc: 0.7471 - val_loss: 0.5692 - val_acc: 0.7205\nEpoch 97/150\n514/514 [==============================] - 0s - loss: 0.6085 - acc: 0.7198 - val_loss: 0.6056 - val_acc: 0.6969\nEpoch 98/150\n514/514 [==============================] - 0s - loss: 0.5433 - acc: 0.7393 - val_loss: 0.5737 - val_acc: 0.7205\nEpoch 99/150\n514/514 [==============================] - 0s - loss: 0.5472 - acc: 0.7432 - val_loss: 0.5880 - val_acc: 0.7441\nEpoch 100/150\n514/514 [==============================] - 0s - loss: 0.5965 - acc: 0.7374 - val_loss: 0.6206 - val_acc: 0.7008\nEpoch 101/150\n514/514 [==============================] - 0s - loss: 0.5803 - acc: 0.7237 - val_loss: 0.5956 - val_acc: 0.7047\nEpoch 102/150\n514/514 [==============================] - 0s - loss: 0.5510 - acc: 0.7374 - val_loss: 0.6116 - val_acc: 0.7087\nEpoch 103/150\n514/514 [==============================] - 0s - loss: 0.5517 - acc: 0.7121 - val_loss: 0.5846 - val_acc: 0.7244\nEpoch 104/150\n514/514 [==============================] - 0s - loss: 0.5510 - acc: 0.7354 - val_loss: 0.5874 - val_acc: 0.7283\nEpoch 105/150\n514/514 [==============================] - 0s - loss: 0.5433 - acc: 0.7276 - val_loss: 0.6032 - val_acc: 0.6969\nEpoch 106/150\n514/514 [==============================] - 0s - loss: 0.5386 - acc: 0.7451 - val_loss: 0.5885 - val_acc: 0.7205\nEpoch 107/150\n514/514 [==============================] - 0s - loss: 0.5253 - acc: 0.7451 - val_loss: 0.6288 - val_acc: 0.7047\nEpoch 108/150\n514/514 [==============================] - 0s - loss: 0.5470 - acc: 0.7237 - val_loss: 0.5788 - val_acc: 0.7283\nEpoch 109/150\n514/514 [==============================] - 0s - loss: 0.5319 - acc: 0.7529 - val_loss: 0.6064 - val_acc: 0.7244\nEpoch 110/150\n514/514 [==============================] - 0s - loss: 0.5320 - acc: 0.7568 - val_loss: 0.5848 - val_acc: 0.7283\nEpoch 111/150\n514/514 [==============================] - 0s - loss: 0.5412 - acc: 0.7354 - val_loss: 0.7940 - val_acc: 0.6457\nEpoch 112/150\n514/514 [==============================] - 0s - loss: 0.5516 - acc: 0.7043 - val_loss: 0.5837 - val_acc: 0.7165\nEpoch 113/150\n514/514 [==============================] - 0s - loss: 0.5393 - acc: 0.7257 - val_loss: 0.5716 - val_acc: 0.7205\nEpoch 114/150\n514/514 [==============================] - 0s - loss: 0.5412 - acc: 0.7296 - val_loss: 0.5985 - val_acc: 0.7087\nEpoch 115/150\n514/514 [==============================] - 0s - loss: 0.5306 - acc: 0.7607 - val_loss: 0.5860 - val_acc: 0.7087\nEpoch 116/150\n514/514 [==============================] - 0s - loss: 0.5840 - acc: 0.7335 - val_loss: 0.6217 - val_acc: 0.7205\nEpoch 117/150\n514/514 [==============================] - 0s - loss: 0.5462 - acc: 0.7549 - val_loss: 0.5881 - val_acc: 0.7087\nEpoch 118/150\n514/514 [==============================] - 0s - loss: 0.5441 - acc: 0.7140 - val_loss: 0.6008 - val_acc: 0.7126\nEpoch 119/150\n514/514 [==============================] - 0s - loss: 0.5695 - acc: 0.7432 - val_loss: 0.6690 - val_acc: 0.6890\nEpoch 120/150\n514/514 [==============================] - 0s - loss: 0.5343 - acc: 0.7568 - val_loss: 0.5713 - val_acc: 0.7244\nEpoch 121/150\n514/514 [==============================] - 0s - loss: 0.5569 - acc: 0.7179 - val_loss: 0.5848 - val_acc: 0.7244\nEpoch 122/150\n514/514 [==============================] - 0s - loss: 0.5825 - acc: 0.7257 - val_loss: 0.6380 - val_acc: 0.6969\nEpoch 123/150\n514/514 [==============================] - 0s - loss: 0.5548 - acc: 0.7451 - val_loss: 0.5774 - val_acc: 0.7283\nEpoch 124/150\n514/514 [==============================] - 0s - loss: 0.5418 - acc: 0.7374 - val_loss: 0.5629 - val_acc: 0.7402\nEpoch 125/150\n514/514 [==============================] - 0s - loss: 0.5259 - acc: 0.7432 - val_loss: 0.5900 - val_acc: 0.7047\nEpoch 126/150\n514/514 [==============================] - 0s - loss: 0.5528 - acc: 0.7257 - val_loss: 0.6274 - val_acc: 0.7205\nEpoch 127/150\n514/514 [==============================] - 0s - loss: 0.5297 - acc: 0.7412 - val_loss: 0.6106 - val_acc: 0.7441\nEpoch 128/150\n514/514 [==============================] - 0s - loss: 0.5692 - acc: 0.7218 - val_loss: 0.7843 - val_acc: 0.6417\nEpoch 129/150\n514/514 [==============================] - 0s - loss: 0.5344 - acc: 0.7412 - val_loss: 0.5721 - val_acc: 0.7205\nEpoch 130/150\n514/514 [==============================] - 0s - loss: 0.5273 - acc: 0.7568 - val_loss: 0.5664 - val_acc: 0.7126\nEpoch 131/150\n514/514 [==============================] - 0s - loss: 0.5289 - acc: 0.7296 - val_loss: 0.5726 - val_acc: 0.7087\nEpoch 132/150\n514/514 [==============================] - 0s - loss: 0.5304 - acc: 0.7490 - val_loss: 0.5856 - val_acc: 0.7323\nEpoch 133/150\n514/514 [==============================] - 0s - loss: 0.5126 - acc: 0.7451 - val_loss: 0.5728 - val_acc: 0.7087\nEpoch 134/150\n514/514 [==============================] - 0s - loss: 0.5151 - acc: 0.7607 - val_loss: 0.5663 - val_acc: 0.7283\nEpoch 135/150\n514/514 [==============================] - 0s - loss: 0.5099 - acc: 0.7588 - val_loss: 0.5630 - val_acc: 0.7283\nEpoch 136/150\n514/514 [==============================] - 0s - loss: 0.5170 - acc: 0.7432 - val_loss: 0.5635 - val_acc: 0.7165\nEpoch 137/150\n514/514 [==============================] - 0s - loss: 0.5151 - acc: 0.7374 - val_loss: 0.5601 - val_acc: 0.7205\nEpoch 138/150\n514/514 [==============================] - 0s - loss: 0.5089 - acc: 0.7529 - val_loss: 0.5590 - val_acc: 0.7283\nEpoch 139/150\n514/514 [==============================] - 0s - loss: 0.5137 - acc: 0.7451 - val_loss: 0.5604 - val_acc: 0.7165\nEpoch 140/150\n514/514 [==============================] - 0s - loss: 0.5300 - acc: 0.7257 - val_loss: 0.5730 - val_acc: 0.7402\nEpoch 141/150\n514/514 [==============================] - 0s - loss: 0.5072 - acc: 0.7646 - val_loss: 0.5581 - val_acc: 0.7598\nEpoch 142/150\n514/514 [==============================] - 0s - loss: 0.5404 - acc: 0.7607 - val_loss: 0.6411 - val_acc: 0.7008\nEpoch 143/150\n514/514 [==============================] - 0s - loss: 0.5328 - acc: 0.7296 - val_loss: 0.5693 - val_acc: 0.7402\nEpoch 144/150\n514/514 [==============================] - 0s - loss: 0.5143 - acc: 0.7626 - val_loss: 0.5983 - val_acc: 0.7362\nEpoch 145/150\n514/514 [==============================] - 0s - loss: 0.5388 - acc: 0.7451 - val_loss: 0.5664 - val_acc: 0.7283\nEpoch 146/150\n514/514 [==============================] - 0s - loss: 0.5115 - acc: 0.7374 - val_loss: 0.5625 - val_acc: 0.7283\nEpoch 147/150\n514/514 [==============================] - 0s - loss: 0.5020 - acc: 0.7549 - val_loss: 0.5858 - val_acc: 0.7126\nEpoch 148/150\n514/514 [==============================] - 0s - loss: 0.5241 - acc: 0.7374 - val_loss: 0.6635 - val_acc: 0.6890\nEpoch 149/150\n514/514 [==============================] - 0s - loss: 0.5241 - acc: 0.7490 - val_loss: 0.5511 - val_acc: 0.7480\nEpoch 150/150\n514/514 [==============================] - 0s - loss: 0.5221 - acc: 0.7607 - val_loss: 0.5785 - val_acc: 0.7087\n" ], [ "from sklearn.model_selection import train_test_split", "_____no_output_____" ], [ "seed = 7\ndataset = np.loadtxt('pima-indians-diabetes.data', delimiter=',')\nX = dataset[:, 0:8]\nY = dataset[:, 8]\nX_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.33, random_state=seed)", "_____no_output_____" ], [ "model = Sequential()\nmodel.add(Dense(12, input_dim=8, activation='relu'))\nmodel.add(Dense(8, activation='relu'))\nmodel.add(Dense(1, activation='sigmoid'))\nmodel.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])", "_____no_output_____" ], [ "model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=150, batch_size=10)", "Train on 514 samples, validate on 254 samples\nEpoch 1/150\n514/514 [==============================] - 0s - loss: 5.4229 - acc: 0.6304 - val_loss: 5.6229 - val_acc: 0.6378\nEpoch 2/150\n514/514 [==============================] - 0s - loss: 5.3085 - acc: 0.6401 - val_loss: 5.5007 - val_acc: 0.6417\nEpoch 3/150\n514/514 [==============================] - 0s - loss: 5.2346 - acc: 0.6459 - val_loss: 5.4103 - val_acc: 0.6378\nEpoch 4/150\n514/514 [==============================] - 0s - loss: 5.1298 - acc: 0.6323 - val_loss: 5.3088 - val_acc: 0.6378\nEpoch 5/150\n514/514 [==============================] - 0s - loss: 5.0421 - acc: 0.6401 - val_loss: 5.2040 - val_acc: 0.6378\nEpoch 6/150\n514/514 [==============================] - 0s - loss: 4.9210 - acc: 0.6323 - val_loss: 4.8707 - val_acc: 0.6378\nEpoch 7/150\n514/514 [==============================] - 0s - loss: 4.5375 - acc: 0.6498 - val_loss: 4.4086 - val_acc: 0.6220\nEpoch 8/150\n514/514 [==============================] - 0s - loss: 4.0227 - acc: 0.6128 - val_loss: 3.7692 - val_acc: 0.6063\nEpoch 9/150\n514/514 [==============================] - 0s - loss: 3.6753 - acc: 0.5992 - val_loss: 3.4825 - val_acc: 0.6260\nEpoch 10/150\n514/514 [==============================] - 0s - loss: 3.4646 - acc: 0.6420 - val_loss: 3.5302 - val_acc: 0.6378\nEpoch 11/150\n514/514 [==============================] - 0s - loss: 3.4321 - acc: 0.6245 - val_loss: 3.2797 - val_acc: 0.6496\nEpoch 12/150\n514/514 [==============================] - 0s - loss: 3.1151 - acc: 0.6342 - val_loss: 2.3772 - val_acc: 0.6378\nEpoch 13/150\n514/514 [==============================] - 0s - loss: 1.6956 - acc: 0.5486 - val_loss: 1.0138 - val_acc: 0.5866\nEpoch 14/150\n514/514 [==============================] - 0s - loss: 1.0465 - acc: 0.6051 - val_loss: 0.8784 - val_acc: 0.6220\nEpoch 15/150\n514/514 [==============================] - 0s - loss: 0.9950 - acc: 0.5953 - val_loss: 0.8158 - val_acc: 0.6417\nEpoch 16/150\n514/514 [==============================] - 0s - loss: 0.8745 - acc: 0.6498 - val_loss: 0.8259 - val_acc: 0.6339\nEpoch 17/150\n514/514 [==============================] - 0s - loss: 0.8333 - acc: 0.6770 - val_loss: 0.7215 - val_acc: 0.6890\nEpoch 18/150\n514/514 [==============================] - 0s - loss: 0.7670 - acc: 0.6673 - val_loss: 0.8371 - val_acc: 0.6772\nEpoch 19/150\n514/514 [==============================] - 0s - loss: 0.7134 - acc: 0.6887 - val_loss: 0.7544 - val_acc: 0.6693\nEpoch 20/150\n514/514 [==============================] - 0s - loss: 0.6996 - acc: 0.6770 - val_loss: 0.6842 - val_acc: 0.6693\nEpoch 21/150\n514/514 [==============================] - 0s - loss: 0.6600 - acc: 0.7043 - val_loss: 0.7152 - val_acc: 0.6772\nEpoch 22/150\n514/514 [==============================] - 0s - loss: 0.6240 - acc: 0.6907 - val_loss: 0.8918 - val_acc: 0.6535\nEpoch 23/150\n514/514 [==============================] - 0s - loss: 0.6177 - acc: 0.7121 - val_loss: 0.6894 - val_acc: 0.6693\nEpoch 24/150\n514/514 [==============================] - 0s - loss: 0.6216 - acc: 0.7043 - val_loss: 0.7710 - val_acc: 0.6535\nEpoch 25/150\n514/514 [==============================] - 0s - loss: 0.6004 - acc: 0.7179 - val_loss: 0.7013 - val_acc: 0.6654\nEpoch 26/150\n514/514 [==============================] - 0s - loss: 0.6341 - acc: 0.7004 - val_loss: 0.7162 - val_acc: 0.6575\nEpoch 27/150\n514/514 [==============================] - 0s - loss: 0.6112 - acc: 0.7062 - val_loss: 0.6918 - val_acc: 0.6693\nEpoch 28/150\n514/514 [==============================] - 0s - loss: 0.6237 - acc: 0.7004 - val_loss: 0.6774 - val_acc: 0.6811\nEpoch 29/150\n514/514 [==============================] - 0s - loss: 0.5800 - acc: 0.7140 - val_loss: 0.7153 - val_acc: 0.6614\nEpoch 30/150\n514/514 [==============================] - 0s - loss: 0.5761 - acc: 0.7335 - val_loss: 0.6456 - val_acc: 0.6929\nEpoch 31/150\n514/514 [==============================] - 0s - loss: 0.6061 - acc: 0.6965 - val_loss: 0.7567 - val_acc: 0.6457\nEpoch 32/150\n514/514 [==============================] - 0s - loss: 0.6030 - acc: 0.7082 - val_loss: 0.6793 - val_acc: 0.6654\nEpoch 33/150\n514/514 [==============================] - 0s - loss: 0.5941 - acc: 0.7354 - val_loss: 0.7651 - val_acc: 0.6575\nEpoch 34/150\n514/514 [==============================] - 0s - loss: 0.5608 - acc: 0.7257 - val_loss: 0.6485 - val_acc: 0.6969\nEpoch 35/150\n514/514 [==============================] - 0s - loss: 0.5502 - acc: 0.7374 - val_loss: 0.7316 - val_acc: 0.6575\nEpoch 36/150\n514/514 [==============================] - 0s - loss: 0.5767 - acc: 0.7510 - val_loss: 0.6455 - val_acc: 0.6890\nEpoch 37/150\n514/514 [==============================] - 0s - loss: 0.5594 - acc: 0.7160 - val_loss: 0.6402 - val_acc: 0.6929\nEpoch 38/150\n514/514 [==============================] - 0s - loss: 0.5439 - acc: 0.7276 - val_loss: 0.6381 - val_acc: 0.7047\nEpoch 39/150\n514/514 [==============================] - 0s - loss: 0.5635 - acc: 0.7218 - val_loss: 0.7396 - val_acc: 0.6732\nEpoch 40/150\n514/514 [==============================] - 0s - loss: 0.5515 - acc: 0.7393 - val_loss: 0.7263 - val_acc: 0.6614\nEpoch 41/150\n514/514 [==============================] - 0s - loss: 0.5921 - acc: 0.7082 - val_loss: 0.6538 - val_acc: 0.6850\nEpoch 42/150\n514/514 [==============================] - 0s - loss: 0.5501 - acc: 0.7315 - val_loss: 0.6771 - val_acc: 0.6772\nEpoch 43/150\n514/514 [==============================] - 0s - loss: 0.5532 - acc: 0.7529 - val_loss: 0.6349 - val_acc: 0.7165\nEpoch 44/150\n514/514 [==============================] - 0s - loss: 0.5652 - acc: 0.7374 - val_loss: 0.6378 - val_acc: 0.7283\nEpoch 45/150\n514/514 [==============================] - 0s - loss: 0.5577 - acc: 0.7257 - val_loss: 0.6409 - val_acc: 0.6969\nEpoch 46/150\n514/514 [==============================] - 0s - loss: 0.5257 - acc: 0.7626 - val_loss: 0.7328 - val_acc: 0.6772\nEpoch 47/150\n514/514 [==============================] - 0s - loss: 0.5445 - acc: 0.7374 - val_loss: 0.6554 - val_acc: 0.6890\nEpoch 48/150\n514/514 [==============================] - 0s - loss: 0.5227 - acc: 0.7335 - val_loss: 0.6995 - val_acc: 0.6693\nEpoch 49/150\n514/514 [==============================] - 0s - loss: 0.5438 - acc: 0.7451 - val_loss: 0.6356 - val_acc: 0.7008\nEpoch 50/150\n514/514 [==============================] - 0s - loss: 0.5326 - acc: 0.7393 - val_loss: 0.6192 - val_acc: 0.7362\nEpoch 51/150\n514/514 [==============================] - 0s - loss: 0.5323 - acc: 0.7374 - val_loss: 0.6228 - val_acc: 0.7244\nEpoch 52/150\n514/514 [==============================] - 0s - loss: 0.5473 - acc: 0.7335 - val_loss: 0.6297 - val_acc: 0.7008\nEpoch 53/150\n514/514 [==============================] - 0s - loss: 0.5417 - acc: 0.7393 - val_loss: 0.7364 - val_acc: 0.6772\nEpoch 54/150\n514/514 [==============================] - 0s - loss: 0.5593 - acc: 0.7354 - val_loss: 0.6551 - val_acc: 0.7087\nEpoch 55/150\n514/514 [==============================] - 0s - loss: 0.5665 - acc: 0.7101 - val_loss: 0.7917 - val_acc: 0.6732\nEpoch 56/150\n514/514 [==============================] - 0s - loss: 0.5598 - acc: 0.7276 - val_loss: 0.7145 - val_acc: 0.6732\nEpoch 57/150\n514/514 [==============================] - 0s - loss: 0.5683 - acc: 0.7121 - val_loss: 0.6474 - val_acc: 0.6929\nEpoch 58/150\n514/514 [==============================] - 0s - loss: 0.5130 - acc: 0.7490 - val_loss: 0.6193 - val_acc: 0.7205\nEpoch 59/150\n514/514 [==============================] - 0s - loss: 0.5131 - acc: 0.7549 - val_loss: 0.6713 - val_acc: 0.6850\nEpoch 60/150\n514/514 [==============================] - 0s - loss: 0.5356 - acc: 0.7412 - val_loss: 0.6392 - val_acc: 0.6654\nEpoch 61/150\n514/514 [==============================] - 0s - loss: 0.5237 - acc: 0.7510 - val_loss: 0.6218 - val_acc: 0.7205\nEpoch 62/150\n514/514 [==============================] - 0s - loss: 0.5149 - acc: 0.7529 - val_loss: 0.6698 - val_acc: 0.6890\nEpoch 63/150\n514/514 [==============================] - 0s - loss: 0.5303 - acc: 0.7490 - val_loss: 0.6416 - val_acc: 0.7087\nEpoch 64/150\n514/514 [==============================] - 0s - loss: 0.5160 - acc: 0.7510 - val_loss: 0.6206 - val_acc: 0.7126\nEpoch 65/150\n514/514 [==============================] - 0s - loss: 0.5177 - acc: 0.7354 - val_loss: 0.6191 - val_acc: 0.7244\nEpoch 66/150\n514/514 [==============================] - 0s - loss: 0.5431 - acc: 0.7412 - val_loss: 0.6633 - val_acc: 0.6614\nEpoch 67/150\n514/514 [==============================] - 0s - loss: 0.5224 - acc: 0.7451 - val_loss: 0.6463 - val_acc: 0.6890\nEpoch 68/150\n514/514 [==============================] - 0s - loss: 0.5173 - acc: 0.7607 - val_loss: 0.6123 - val_acc: 0.7323\nEpoch 69/150\n514/514 [==============================] - 0s - loss: 0.5175 - acc: 0.7529 - val_loss: 0.6173 - val_acc: 0.7047\nEpoch 70/150\n514/514 [==============================] - 0s - loss: 0.5393 - acc: 0.7529 - val_loss: 0.6327 - val_acc: 0.6890\nEpoch 71/150\n514/514 [==============================] - 0s - loss: 0.5392 - acc: 0.7588 - val_loss: 0.6250 - val_acc: 0.7283\nEpoch 72/150\n514/514 [==============================] - 0s - loss: 0.5603 - acc: 0.7393 - val_loss: 0.6474 - val_acc: 0.6969\nEpoch 73/150\n514/514 [==============================] - 0s - loss: 0.5235 - acc: 0.7490 - val_loss: 0.6080 - val_acc: 0.7244\nEpoch 74/150\n514/514 [==============================] - 0s - loss: 0.5251 - acc: 0.7549 - val_loss: 0.6376 - val_acc: 0.6969\nEpoch 75/150\n514/514 [==============================] - 0s - loss: 0.5034 - acc: 0.7568 - val_loss: 0.6201 - val_acc: 0.7323\nEpoch 76/150\n514/514 [==============================] - 0s - loss: 0.5240 - acc: 0.7412 - val_loss: 0.6070 - val_acc: 0.7402\nEpoch 77/150\n514/514 [==============================] - 0s - loss: 0.5021 - acc: 0.7665 - val_loss: 0.6797 - val_acc: 0.6732\nEpoch 78/150\n514/514 [==============================] - 0s - loss: 0.4974 - acc: 0.7782 - val_loss: 0.6169 - val_acc: 0.7283\nEpoch 79/150\n514/514 [==============================] - 0s - loss: 0.5128 - acc: 0.7490 - val_loss: 0.6092 - val_acc: 0.7362\nEpoch 80/150\n514/514 [==============================] - 0s - loss: 0.5172 - acc: 0.7490 - val_loss: 0.6262 - val_acc: 0.7047\nEpoch 81/150\n514/514 [==============================] - 0s - loss: 0.4978 - acc: 0.7471 - val_loss: 0.6047 - val_acc: 0.7362\nEpoch 82/150\n514/514 [==============================] - 0s - loss: 0.5027 - acc: 0.7704 - val_loss: 0.6434 - val_acc: 0.6772\nEpoch 83/150\n514/514 [==============================] - 0s - loss: 0.5026 - acc: 0.7529 - val_loss: 0.6077 - val_acc: 0.7244\nEpoch 84/150\n514/514 [==============================] - 0s - loss: 0.5383 - acc: 0.7529 - val_loss: 0.6169 - val_acc: 0.7205\nEpoch 85/150\n514/514 [==============================] - 0s - loss: 0.4882 - acc: 0.7685 - val_loss: 0.6228 - val_acc: 0.7165\nEpoch 86/150\n514/514 [==============================] - 0s - loss: 0.4995 - acc: 0.7529 - val_loss: 0.6546 - val_acc: 0.6811\nEpoch 87/150\n514/514 [==============================] - 0s - loss: 0.5166 - acc: 0.7646 - val_loss: 0.6051 - val_acc: 0.7283\nEpoch 88/150\n514/514 [==============================] - 0s - loss: 0.5017 - acc: 0.7510 - val_loss: 0.6108 - val_acc: 0.7126\nEpoch 89/150\n514/514 [==============================] - 0s - loss: 0.5304 - acc: 0.7626 - val_loss: 0.6625 - val_acc: 0.6850\nEpoch 90/150\n514/514 [==============================] - 0s - loss: 0.5193 - acc: 0.7490 - val_loss: 0.6013 - val_acc: 0.7283\nEpoch 91/150\n514/514 [==============================] - 0s - loss: 0.4943 - acc: 0.7743 - val_loss: 0.6079 - val_acc: 0.7283\nEpoch 92/150\n514/514 [==============================] - 0s - loss: 0.5184 - acc: 0.7529 - val_loss: 0.6213 - val_acc: 0.7126\nEpoch 93/150\n514/514 [==============================] - 0s - loss: 0.5042 - acc: 0.7451 - val_loss: 0.6311 - val_acc: 0.6811\nEpoch 94/150\n514/514 [==============================] - 0s - loss: 0.4954 - acc: 0.7607 - val_loss: 0.6025 - val_acc: 0.7323\nEpoch 95/150\n514/514 [==============================] - 0s - loss: 0.4917 - acc: 0.7529 - val_loss: 0.6167 - val_acc: 0.7205\nEpoch 96/150\n514/514 [==============================] - 0s - loss: 0.4894 - acc: 0.7490 - val_loss: 0.6030 - val_acc: 0.7441\nEpoch 97/150\n514/514 [==============================] - 0s - loss: 0.5054 - acc: 0.7471 - val_loss: 0.6028 - val_acc: 0.7283\nEpoch 98/150\n514/514 [==============================] - 0s - loss: 0.4863 - acc: 0.7840 - val_loss: 0.6134 - val_acc: 0.7244\nEpoch 99/150\n514/514 [==============================] - 0s - loss: 0.4870 - acc: 0.7471 - val_loss: 0.6442 - val_acc: 0.7087\nEpoch 100/150\n514/514 [==============================] - 0s - loss: 0.5013 - acc: 0.7704 - val_loss: 0.6560 - val_acc: 0.7008\nEpoch 101/150\n514/514 [==============================] - 0s - loss: 0.5459 - acc: 0.7374 - val_loss: 0.6787 - val_acc: 0.6890\nEpoch 102/150\n514/514 [==============================] - 0s - loss: 0.5184 - acc: 0.7626 - val_loss: 0.6015 - val_acc: 0.7244\nEpoch 103/150\n514/514 [==============================] - 0s - loss: 0.4902 - acc: 0.7724 - val_loss: 0.6074 - val_acc: 0.7244\nEpoch 104/150\n514/514 [==============================] - 0s - loss: 0.4989 - acc: 0.7763 - val_loss: 0.6312 - val_acc: 0.7126\nEpoch 105/150\n514/514 [==============================] - 0s - loss: 0.5064 - acc: 0.7549 - val_loss: 0.6232 - val_acc: 0.7205\nEpoch 106/150\n514/514 [==============================] - 0s - loss: 0.4992 - acc: 0.7588 - val_loss: 0.6248 - val_acc: 0.7283\nEpoch 107/150\n514/514 [==============================] - 0s - loss: 0.5326 - acc: 0.7451 - val_loss: 0.6476 - val_acc: 0.6890\nEpoch 108/150\n514/514 [==============================] - 0s - loss: 0.4882 - acc: 0.7802 - val_loss: 0.6395 - val_acc: 0.6654\nEpoch 109/150\n514/514 [==============================] - 0s - loss: 0.5056 - acc: 0.7743 - val_loss: 0.5995 - val_acc: 0.7362\nEpoch 110/150\n514/514 [==============================] - 0s - loss: 0.4828 - acc: 0.7782 - val_loss: 0.6349 - val_acc: 0.7087\nEpoch 111/150\n514/514 [==============================] - 0s - loss: 0.4801 - acc: 0.7821 - val_loss: 0.6462 - val_acc: 0.6732\nEpoch 112/150\n514/514 [==============================] - 0s - loss: 0.4938 - acc: 0.7549 - val_loss: 0.6039 - val_acc: 0.7244\nEpoch 113/150\n514/514 [==============================] - 0s - loss: 0.5066 - acc: 0.7802 - val_loss: 0.6126 - val_acc: 0.7283\nEpoch 114/150\n514/514 [==============================] - 0s - loss: 0.5020 - acc: 0.7490 - val_loss: 0.6032 - val_acc: 0.7283\nEpoch 115/150\n514/514 [==============================] - 0s - loss: 0.4987 - acc: 0.7549 - val_loss: 0.6063 - val_acc: 0.7165\nEpoch 116/150\n514/514 [==============================] - 0s - loss: 0.5207 - acc: 0.7646 - val_loss: 0.6163 - val_acc: 0.7087\nEpoch 117/150\n514/514 [==============================] - 0s - loss: 0.4993 - acc: 0.7471 - val_loss: 0.6008 - val_acc: 0.7283\nEpoch 118/150\n514/514 [==============================] - 0s - loss: 0.4954 - acc: 0.7685 - val_loss: 0.5984 - val_acc: 0.7362\nEpoch 119/150\n514/514 [==============================] - 0s - loss: 0.4874 - acc: 0.7743 - val_loss: 0.6070 - val_acc: 0.7362\nEpoch 120/150\n514/514 [==============================] - 0s - loss: 0.4909 - acc: 0.7568 - val_loss: 0.5990 - val_acc: 0.7402\nEpoch 121/150\n514/514 [==============================] - 0s - loss: 0.4781 - acc: 0.7588 - val_loss: 0.6695 - val_acc: 0.6890\nEpoch 122/150\n514/514 [==============================] - 0s - loss: 0.4861 - acc: 0.7665 - val_loss: 0.6198 - val_acc: 0.7402\nEpoch 123/150\n514/514 [==============================] - 0s - loss: 0.4786 - acc: 0.7626 - val_loss: 0.6440 - val_acc: 0.6811\nEpoch 124/150\n514/514 [==============================] - 0s - loss: 0.4830 - acc: 0.7665 - val_loss: 0.6391 - val_acc: 0.6811\nEpoch 125/150\n514/514 [==============================] - 0s - loss: 0.4707 - acc: 0.7685 - val_loss: 0.6373 - val_acc: 0.7205\nEpoch 126/150\n514/514 [==============================] - 0s - loss: 0.4953 - acc: 0.7802 - val_loss: 0.5981 - val_acc: 0.7441\nEpoch 127/150\n514/514 [==============================] - 0s - loss: 0.4689 - acc: 0.7957 - val_loss: 0.5971 - val_acc: 0.7362\nEpoch 128/150\n514/514 [==============================] - 0s - loss: 0.4938 - acc: 0.7685 - val_loss: 0.8264 - val_acc: 0.6850\nEpoch 129/150\n514/514 [==============================] - 0s - loss: 0.5091 - acc: 0.7626 - val_loss: 0.6702 - val_acc: 0.6772\nEpoch 130/150\n514/514 [==============================] - 0s - loss: 0.4939 - acc: 0.7665 - val_loss: 0.6371 - val_acc: 0.7205\nEpoch 131/150\n514/514 [==============================] - 0s - loss: 0.4974 - acc: 0.7607 - val_loss: 0.6433 - val_acc: 0.6890\nEpoch 132/150\n514/514 [==============================] - 0s - loss: 0.4957 - acc: 0.7743 - val_loss: 0.7286 - val_acc: 0.6772\nEpoch 133/150\n514/514 [==============================] - 0s - loss: 0.5178 - acc: 0.7743 - val_loss: 0.6052 - val_acc: 0.7323\nEpoch 134/150\n514/514 [==============================] - 0s - loss: 0.4986 - acc: 0.7626 - val_loss: 0.5918 - val_acc: 0.7520\nEpoch 135/150\n514/514 [==============================] - 0s - loss: 0.4705 - acc: 0.7704 - val_loss: 0.5899 - val_acc: 0.7441\nEpoch 136/150\n514/514 [==============================] - 0s - loss: 0.5097 - acc: 0.7724 - val_loss: 0.6217 - val_acc: 0.7126\nEpoch 137/150\n514/514 [==============================] - 0s - loss: 0.4939 - acc: 0.7704 - val_loss: 0.6085 - val_acc: 0.7402\nEpoch 138/150\n514/514 [==============================] - 0s - loss: 0.4722 - acc: 0.7626 - val_loss: 0.6014 - val_acc: 0.7402\nEpoch 139/150\n514/514 [==============================] - 0s - loss: 0.4921 - acc: 0.7626 - val_loss: 0.5945 - val_acc: 0.7362\nEpoch 140/150\n514/514 [==============================] - 0s - loss: 0.4802 - acc: 0.7802 - val_loss: 0.6175 - val_acc: 0.7047\nEpoch 141/150\n514/514 [==============================] - 0s - loss: 0.4786 - acc: 0.7685 - val_loss: 0.6166 - val_acc: 0.7244\nEpoch 142/150\n514/514 [==============================] - 0s - loss: 0.4773 - acc: 0.7529 - val_loss: 0.5986 - val_acc: 0.7283\nEpoch 143/150\n514/514 [==============================] - 0s - loss: 0.4803 - acc: 0.7626 - val_loss: 0.6455 - val_acc: 0.7205\nEpoch 144/150\n514/514 [==============================] - 0s - loss: 0.4776 - acc: 0.7607 - val_loss: 0.5955 - val_acc: 0.7362\nEpoch 145/150\n514/514 [==============================] - 0s - loss: 0.4808 - acc: 0.7626 - val_loss: 0.6175 - val_acc: 0.7323\nEpoch 146/150\n514/514 [==============================] - 0s - loss: 0.4659 - acc: 0.7743 - val_loss: 0.6024 - val_acc: 0.7362\nEpoch 147/150\n514/514 [==============================] - 0s - loss: 0.4818 - acc: 0.7840 - val_loss: 0.6829 - val_acc: 0.7008\nEpoch 148/150\n514/514 [==============================] - 0s - loss: 0.5261 - acc: 0.7451 - val_loss: 0.6686 - val_acc: 0.7126\nEpoch 149/150\n514/514 [==============================] - 0s - loss: 0.4856 - acc: 0.7763 - val_loss: 0.6530 - val_acc: 0.7244\nEpoch 150/150\n514/514 [==============================] - 0s - loss: 0.4944 - acc: 0.7918 - val_loss: 0.6109 - val_acc: 0.7402\n" ] ], [ [ "## Cross Validation", "_____no_output_____" ] ], [ [ "from keras.models import Sequential\nfrom keras.layers import Dense\nfrom sklearn.model_selection import StratifiedKFold\nimport numpy as np", "_____no_output_____" ], [ "seed = 7\nnp.random.seed(seed)", "_____no_output_____" ], [ "dataset = np.loadtxt('pima-indians-diabetes.data', delimiter=',')\nX = dataset[:, 0:8]\nY = dataset[:, 8]\nX.shape", "_____no_output_____" ], [ "kfold = StratifiedKFold(n_splits=5, shuffle=True, random_state=seed)", "_____no_output_____" ], [ "cvscores = []\nfor train, test in kfold.split(X, Y):\n model = Sequential()\n model.add(Dense(12, input_dim=8, activation='relu'))\n model.add(Dense(8, activation='relu'))\n model.add(Dense(1, activation='sigmoid'))\n model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])\n model.fit(X[train], Y[train], epochs=150, batch_size=10, verbose=0)\n scores = model.evaluate(X[test], Y[test], verbose=0)\n print('%s: %.2f%%' % (model.metrics_names[1], scores[1] * 100))\n cvscores.append(scores[1] * 100)\nprint('%.2f%% (+/- %.2f%%' % (np.mean(cvscores), np.std(cvscores)))", "acc: 72.73%\nacc: 74.03%\nacc: 75.97%\nacc: 65.36%\nacc: 68.63%\n71.34% (+/- 3.84%\n" ], [ "X.shape", "_____no_output_____" ], [ "Y.shape", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import frontmatter\nimport re\nimport urllib\nimport requests\n\nDIRNAME = os.path.abspath('')\nWWW_SHIPS_DATA_PATH = os.path.join(DIRNAME, \"../../www/views/ships\")\nWWW_IMG_PATH = os.path.join(DIRNAME, \"../../www/assets/img\")", "_____no_output_____" ], [ "for ship_filename in os.listdir(WWW_SHIPS_DATA_PATH):\n md_match = re.match(r\"(.*)\\.md\", ship_filename)\n if not md_match:\n continue\n ship_slug = md_match.groups()[0]\n ship_frontmatter = frontmatter.load(os.path.join(WWW_SHIPS_DATA_PATH, ship_filename))\n wikipedia_url = ship_frontmatter.metadata.get(\"wikipediaUrl\")\n photo_path = ship_frontmatter.metadata.get(\"photo\", \"\")\n if photo_path != \"\" or not wikipedia_url:\n continue\n parsed = urllib.parse.urlparse(wikipedia_url)\n country, title = re.match(r\".*([a-z]{2,3})\\.wikipedia\\.org\\/wiki\\/(.*)\", wikipedia_url).groups()\n json_url = f\"http://{country}.wikipedia.org/w/api.php?action=query&titles={title}&prop=pageimages&format=json&pithumbsize=500\"\n data = requests.get(json_url).json()\n keys = list(data[\"query\"][\"pages\"])\n image_url = data[\"query\"][\"pages\"][keys[0]].get(\"thumbnail\", {}).get(\"source\")\n if not image_url:\n print(f\"could not find image url in {json_url}\")\n continue\n img_ext = image_url[-10:].split(\".\")[-1]\n img_filename = f\"{ship_slug}.{img_ext}\"\n with open(os.path.join(f\"{WWW_IMG_PATH}\", img_filename), \"wb\") as f:\n f.write(requests.get(image_url).content)\n print(f\"wrote image www/assets/js/{img_filename}\")\n ship_frontmatter.metadata[\"photo\"] = f\"/img/{img_filename}\"\n with open(os.path.join(WWW_SHIPS_DATA_PATH, ship_filename), \"w\") as f:\n f.write(frontmatter.dumps(ship_frontmatter))\n print(f\"wrote back md to www/views/ships/{ship_filename}\")\n ", "wrote image www/assets/js/caribbean-princess-9215490.jpg\nwrote back md to www/views/ships/caribbean-princess-9215490.md\nwrote image www/assets/js/star-breeze-8807997.jpg\nwrote back md to www/views/ships/star-breeze-8807997.md\nwrote image www/assets/js/veendam-9102992.jpg\nwrote back md to www/views/ships/veendam-9102992.md\nwrote image www/assets/js/zuiderdam-9221279.JPG\nwrote back md to www/views/ships/zuiderdam-9221279.md\ncould not find image url in http://en.wikipedia.org/w/api.php?action=query&titles=MV_Nova_Star&prop=pageimages&format=json&pithumbsize=500\nwrote image www/assets/js/amsterdam-9188037.JPG\nwrote back md to www/views/ships/amsterdam-9188037.md\nwrote image www/assets/js/star-pride-8707343.JPG\nwrote back md to www/views/ships/star-pride-8707343.md\nwrote image www/assets/js/pride-of-york-8501957.jpg\nwrote back md to www/views/ships/pride-of-york-8501957.md\nwrote image www/assets/js/spirit-of-france-9533816.JPEG\nwrote back md to www/views/ships/spirit-of-france-9533816.md\nwrote image www/assets/js/wind-surf-8700785.jpg\nwrote back md to www/views/ships/wind-surf-8700785.md\ncould not find image url in http://en.wikipedia.org/w/api.php?action=query&titles=Le_Champlain&prop=pageimages&format=json&pithumbsize=500\nwrote image www/assets/js/coral-princess-9229659.jpg\nwrote back md to www/views/ships/coral-princess-9229659.md\nwrote image www/assets/js/sky-princess-9802396.jpg\nwrote back md to www/views/ships/sky-princess-9802396.md\nwrote image www/assets/js/emerald-princess-9333151.jpg\nwrote back md to www/views/ships/emerald-princess-9333151.md\nwrote image www/assets/js/pride-of-bruges-8503797.jpg\nwrote back md to www/views/ships/pride-of-bruges-8503797.md\nwrote image www/assets/js/crown-princess-9293399.jpg\nwrote back md to www/views/ships/crown-princess-9293399.md\nwrote image www/assets/js/spirit-of-britain-9524231.JPG\nwrote back md to www/views/ships/spirit-of-britain-9524231.md\nwrote image www/assets/js/norbay-9056595.jpg\nwrote back md to www/views/ships/norbay-9056595.md\ncould not find image url in http://en.wikipedia.org/w/api.php?action=query&titles=Le_Bougainville&prop=pageimages&format=json&pithumbsize=500\nwrote image www/assets/js/seabourn-sojourn-9417098.jpg\nwrote back md to www/views/ships/seabourn-sojourn-9417098.md\nwrote image www/assets/js/pride-of-hull-9208629.jpg\nwrote back md to www/views/ships/pride-of-hull-9208629.md\ncould not find image url in http://fr.wikipedia.org/w/api.php?action=query&titles=Baltic_Princess_(ferry)&prop=pageimages&format=json&pithumbsize=500\nwrote image www/assets/js/pride-of-kent-9015266.jpg\nwrote back md to www/views/ships/pride-of-kent-9015266.md\nwrote image www/assets/js/pride-of-burgundy-9015254.JPG\nwrote back md to www/views/ships/pride-of-burgundy-9015254.md\nwrote image www/assets/js/european-highlander-9244116.jpg\nwrote back md to www/views/ships/european-highlander-9244116.md\nwrote image www/assets/js/pride-of-canterbury-9007295.jpg\nwrote back md to www/views/ships/pride-of-canterbury-9007295.md\nwrote image www/assets/js/norbank-9056583.JPEG\nwrote back md to www/views/ships/norbank-9056583.md\nwrote image www/assets/js/visborg-9763655.jpg\nwrote back md to www/views/ships/visborg-9763655.md\ncould not find image url in http://en.wikipedia.org/w/api.php?action=query&titles=AIDAmira&prop=pageimages&format=json&pithumbsize=500\nwrote image www/assets/js/finbo-cargo-9181106.jpg\nwrote back md to www/views/ships/finbo-cargo-9181106.md\nwrote image www/assets/js/national-geographic-explorer-8019356.jpg\nwrote back md to www/views/ships/national-geographic-explorer-8019356.md\nwrote image www/assets/js/pride-of-rotterdam-9208617.jpg\nwrote back md to www/views/ships/pride-of-rotterdam-9208617.md\ncould not find image url in http://fr.wikipedia.org/w/api.php?action=query&titles=Viking_ADCC&prop=pageimages&format=json&pithumbsize=500\ncould not find image url in http://de.wikipedia.org/w/api.php?action=query&titles=Victoria_Seaways&prop=pageimages&format=json&pithumbsize=500\nwrote image www/assets/js/european-seaway-9007283.JPG\nwrote back md to www/views/ships/european-seaway-9007283.md\nwrote image www/assets/js/wind-star-8420878.jpg\nwrote back md to www/views/ships/wind-star-8420878.md\nwrote image www/assets/js/sun-princess-9000259.jpg\nwrote back md to www/views/ships/sun-princess-9000259.md\n" ] ] ]

[ "code" ]

[ [ "code", "code" ] ]

[ "OLDAP-2.3" ]

[ "OLDAP-2.3" ]

[ "OLDAP-2.3" ]

[ [ [ "import numpy as np\nimport cv2\nimport matplotlib.pyplot as plt\nfrom keras import models\nimport keras.backend as K\nimport tensorflow as tf\nfrom sklearn.metrics import f1_score\nimport requests\nimport xmltodict\nimport json", "Using TensorFlow backend.\n" ], [ "plateCascade = cv2.CascadeClassifier('indian_license_plate.xml')", "_____no_output_____" ], [ "#detect the plate and return car + plate image\ndef plate_detect(img):\n plateImg = img.copy()\n roi = img.copy()\n plateRect = plateCascade.detectMultiScale(plateImg,scaleFactor = 1.2, minNeighbors = 7)\n for (x,y,w,h) in plateRect:\n roi_ = roi[y:y+h, x:x+w, :]\n plate_part = roi[y:y+h, x:x+w, :]\n cv2.rectangle(plateImg,(x+2,y),(x+w-3, y+h-5),(0,255,0),3)\n return plateImg, plate_part", "_____no_output_____" ], [ "#normal function to display \ndef display_img(img):\n img_ = cv2.cvtColor(img,cv2.COLOR_BGR2RGB)\n plt.imshow(img_)\n plt.show()", "_____no_output_____" ], [ "#test image is used for detecting plate\ninputImg = cv2.imread('test.jpeg')\ninpImg, plate = plate_detect(inputImg)\ndisplay_img(inpImg)", "_____no_output_____" ], [ "def find_contours(dimensions, img) :\n\n #finding all contours in the image using \n #retrieval mode: RETR_TREE\n #contour approximation method: CHAIN_APPROX_SIMPLE\n cntrs, _ = cv2.findContours(img.copy(), cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)\n\n #Approx dimensions of the contours\n lower_width = dimensions[0]\n upper_width = dimensions[1]\n lower_height = dimensions[2]\n upper_height = dimensions[3]\n \n #Check largest 15 contours for license plate character respectively\n cntrs = sorted(cntrs, key=cv2.contourArea, reverse=True)[:15]\n \n ci = cv2.imread('contour.jpg')\n \n x_cntr_list = []\n target_contours = []\n img_res = []\n for cntr in cntrs :\n #detecting contour in binary image and returns the coordinates of rectangle enclosing it\n intX, intY, intWidth, intHeight = cv2.boundingRect(cntr)\n \n #checking the dimensions of the contour to filter out the characters by contour's size\n if intWidth > lower_width and intWidth < upper_width and intHeight > lower_height and intHeight < upper_height :\n x_cntr_list.append(intX) \n char_copy = np.zeros((44,24))\n #extracting each character using the enclosing rectangle's coordinates.\n char = img[intY:intY+intHeight, intX:intX+intWidth]\n char = cv2.resize(char, (20, 40))\n cv2.rectangle(ci, (intX,intY), (intWidth+intX, intY+intHeight), (50,21,200), 2)\n plt.imshow(ci, cmap='gray')\n char = cv2.subtract(255, char)\n char_copy[2:42, 2:22] = char\n char_copy[0:2, :] = 0\n char_copy[:, 0:2] = 0\n char_copy[42:44, :] = 0\n char_copy[:, 22:24] = 0\n img_res.append(char_copy) # List that stores the character's binary image (unsorted)\n \n #return characters on ascending order with respect to the x-coordinate\n \n plt.show()\n #arbitrary function that stores sorted list of character indeces\n indices = sorted(range(len(x_cntr_list)), key=lambda k: x_cntr_list[k])\n img_res_copy = []\n for idx in indices:\n img_res_copy.append(img_res[idx])# stores character images according to their index\n img_res = np.array(img_res_copy)\n\n return img_res", "_____no_output_____" ], [ "def segment_characters(image) :\n\n #pre-processing cropped image of plate\n #threshold: convert to pure b&w with sharpe edges\n #erod: increasing the backgroung black\n #dilate: increasing the char white\n img_lp = cv2.resize(image, (333, 75))\n img_gray_lp = cv2.cvtColor(img_lp, cv2.COLOR_BGR2GRAY)\n _, img_binary_lp = cv2.threshold(img_gray_lp, 200, 255, cv2.THRESH_BINARY+cv2.THRESH_OTSU)\n img_binary_lp = cv2.erode(img_binary_lp, (3,3))\n img_binary_lp = cv2.dilate(img_binary_lp, (3,3))\n\n LP_WIDTH = img_binary_lp.shape[0]\n LP_HEIGHT = img_binary_lp.shape[1]\n img_binary_lp[0:3,:] = 255\n img_binary_lp[:,0:3] = 255\n img_binary_lp[72:75,:] = 255\n img_binary_lp[:,330:333] = 255\n\n #estimations of character contours sizes of cropped license plates\n dimensions = [LP_WIDTH/6,\n LP_WIDTH/2,\n LP_HEIGHT/10,\n 2*LP_HEIGHT/3]\n plt.imshow(img_binary_lp, cmap='gray')\n plt.show()\n cv2.imwrite('contour.jpg',img_binary_lp)\n\n #getting contours\n char_list = find_contours(dimensions, img_binary_lp)\n\n return char_list", "_____no_output_____" ], [ "char = segment_characters(plate)", "_____no_output_____" ], [ "for i in range(10):\n plt.subplot(1, 10, i+1)\n plt.imshow(char[i], cmap='gray')\n plt.axis('off')", "_____no_output_____" ], [ "#It is the harmonic mean of precision and recall\n#Output range is [0, 1]\n#Works for both multi-class and multi-label classification\n\ndef f1score(y, y_pred):\n return f1_score(y, tf.math.argmax(y_pred, axis=1), average='micro') \n\ndef custom_f1score(y, y_pred):\n return tf.py_function(f1score, (y, y_pred), tf.double)", "_____no_output_____" ], [ " model = models.load_model('license_plate_character.pkl', custom_objects= {'custom_f1score': custom_f1score})", "_____no_output_____" ], [ "def fix_dimension(img):\n new_img = np.zeros((28,28,3))\n for i in range(3):\n new_img[:,:,i] = img\n return new_img\n \ndef show_results():\n dic = {}\n characters = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ'\n for i,c in enumerate(characters):\n dic[i] = c\n\n output = []\n for i,ch in enumerate(char): \n img_ = cv2.resize(ch, (28,28), interpolation=cv2.INTER_AREA)\n img = fix_dimension(img_)\n img = img.reshape(1,28,28,3)\n y_ = model.predict_classes(img)[0]\n character = dic[y_] #\n output.append(character) \n \n plate_number = ''.join(output)\n \n return plate_number\n\nfinal_plate = show_results()\nprint(final_plate)", "IMH20EE7598\n" ], [ "def get_vehicle_info(plate_number):\n r = requests.get(\"http://www.regcheck.org.uk/api/reg.asmx/CheckIndia?RegistrationNumber={0}&username=licenseguy\".format(str(plate_number)))\n data = xmltodict.parse(r.content)\n jdata = json.dumps(data)\n df = json.loads(jdata)\n df1 = json.loads(df['Vehicle']['vehicleJson'])\n return df1\n", "_____no_output_____" ], [ "if len(final_plate) > 10:\n final_plate = final_plate[-10:]\n print(final_plate)", "_____no_output_____" ], [ "get_vehicle_info(final_plate)", "_____no_output_____" ] ] ]

[ "code" ]

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<a href=\"https://cognitiveclass.ai\"><img src = \"https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/Logos/organization_logo/organization_logo.png\" width = 400> </a>\n\n<h1 align=center><font size = 5>Convolutional Neural Networks with Keras</font></h1>\n", "_____no_output_____" ], [ "In this lab, we will learn how to use the Keras library to build convolutional neural networks. We will also use the popular MNIST dataset and we will compare our results to using a conventional neural network.\n", "_____no_output_____" ], [ "<h2>Convolutional Neural Networks with Keras</h2>\n\n<h3>Objective for this Notebook<h3> \n<h5> 1. How to use the Keras library to build convolutional neural networks.</h5>\n<h5> 2. Convolutional Neural Network with One Convolutional and Pooling Layers.</h5>\n<h5> 3. Convolutional Neural Network with Two Convolutional and Pooling Layers.</h5>\n", "_____no_output_____" ], [ "## Table of Contents\n\n<div class=\"alert alert-block alert-info\" style=\"margin-top: 20px\">\n\n<font size = 3>\n \n1. <a href=\"#item41\">Import Keras and Packages</a> \n2. <a href=\"#item42\">Convolutional Neural Network with One Convolutional and Pooling Layers</a> \n3. <a href=\"#item43\">Convolutional Neural Network with Two Convolutional and Pooling Layers</a> \n\n</font>\n</div>\n", "_____no_output_____" ], [ "<a id='item41'></a>\n", "_____no_output_____" ], [ "## Import Keras and Packages\n", "_____no_output_____" ], [ "Let's start by importing the keras libraries and the packages that we would need to build a neural network.\n", "_____no_output_____" ] ], [ [ "import tensorflow.keras\nfrom tensorflow.keras.models import Sequential\nfrom tensorflow.keras.layers import Dense\nfrom tensorflow.keras.utils import to_categorical", "_____no_output_____" ] ], [ [ "When working with convolutional neural networks in particular, we will need additional packages.\n", "_____no_output_____" ] ], [ [ "from tensorflow.keras.layers import Conv2D # to add convolutional layers\nfrom tensorflow.keras.layers import MaxPooling2D # to add pooling layers\nfrom tensorflow.keras.layers import Flatten # to flatten data for fully connected layers", "_____no_output_____" ] ], [ [ "<a id='item42'></a>\n", "_____no_output_____" ], [ "## Convolutional Layer with One set of convolutional and pooling layers\n", "_____no_output_____" ] ], [ [ "# import data\nfrom tensorflow.keras.datasets import mnist\n\n# load data\n(X_train, y_train), (X_test, y_test) = mnist.load_data()\n\n# reshape to be [samples][pixels][width][height]\nX_train = X_train.reshape(X_train.shape[0], 28, 28, 1).astype('float32')\nX_test = X_test.reshape(X_test.shape[0], 28, 28, 1).astype('float32')", "_____no_output_____" ] ], [ [ "Let's normalize the pixel values to be between 0 and 1\n", "_____no_output_____" ] ], [ [ "X_train = X_train / 255 # normalize training data\nX_test = X_test / 255 # normalize test data", "_____no_output_____" ] ], [ [ "Next, let's convert the target variable into binary categories\n", "_____no_output_____" ] ], [ [ "y_train = to_categorical(y_train)\ny_test = to_categorical(y_test)\n\nnum_classes = y_test.shape[1] # number of categories", "_____no_output_____" ] ], [ [ "Next, let's define a function that creates our model. Let's start with one set of convolutional and pooling layers.\n", "_____no_output_____" ] ], [ [ "def convolutional_model():\n \n # create model\n model = Sequential()\n model.add(Conv2D(16, (5, 5), strides=(1, 1), activation='relu', input_shape=(28, 28, 1)))\n model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))\n \n model.add(Flatten())\n model.add(Dense(100, activation='relu'))\n model.add(Dense(num_classes, activation='softmax'))\n \n # compile model\n model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])\n return model", "_____no_output_____" ] ], [ [ "Finally, let's call the function to create the model, and then let's train it and evaluate it.\n", "_____no_output_____" ] ], [ [ "# build the model\nmodel = convolutional_model()\n\n# fit the model\nmodel.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=10, batch_size=200, verbose=2)\n\n# evaluate the model\nscores = model.evaluate(X_test, y_test, verbose=0)\nprint(\"Accuracy: {} \\n Error: {}\".format(scores[1], 100-scores[1]*100))", "WARNING:tensorflow:From /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages/tensorflow/python/ops/init_ops.py:1251: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\nInstructions for updating:\nCall initializer instance with the dtype argument instead of passing it to the constructor\nTrain on 60000 samples, validate on 10000 samples\nEpoch 1/10\n60000/60000 - 43s - loss: 0.2902 - acc: 0.9203 - val_loss: 0.1027 - val_acc: 0.9695\nEpoch 2/10\n60000/60000 - 43s - loss: 0.0866 - acc: 0.9751 - val_loss: 0.0647 - val_acc: 0.9785\nEpoch 3/10\n60000/60000 - 43s - loss: 0.0591 - acc: 0.9827 - val_loss: 0.0489 - val_acc: 0.9847\nEpoch 4/10\n60000/60000 - 43s - loss: 0.0458 - acc: 0.9862 - val_loss: 0.0415 - val_acc: 0.9867\nEpoch 5/10\n60000/60000 - 43s - loss: 0.0355 - acc: 0.9892 - val_loss: 0.0371 - val_acc: 0.9876\nEpoch 6/10\n60000/60000 - 44s - loss: 0.0295 - acc: 0.9911 - val_loss: 0.0378 - val_acc: 0.9870\nEpoch 7/10\n60000/60000 - 43s - loss: 0.0235 - acc: 0.9926 - val_loss: 0.0358 - val_acc: 0.9877\nEpoch 8/10\n60000/60000 - 43s - loss: 0.0195 - acc: 0.9942 - val_loss: 0.0363 - val_acc: 0.9882\nEpoch 9/10\n60000/60000 - 43s - loss: 0.0163 - acc: 0.9953 - val_loss: 0.0353 - val_acc: 0.9880\nEpoch 10/10\n60000/60000 - 43s - loss: 0.0133 - acc: 0.9962 - val_loss: 0.0331 - val_acc: 0.9888\nAccuracy: 0.9887999892234802 \n Error: 1.1200010776519775\n" ] ], [ [ "* * *\n", "_____no_output_____" ], [ "<a id='item43'></a>\n", "_____no_output_____" ], [ "## Convolutional Layer with two sets of convolutional and pooling layers\n", "_____no_output_____" ], [ "Let's redefine our convolutional model so that it has two convolutional and pooling layers instead of just one layer of each.\n", "_____no_output_____" ] ], [ [ "def convolutional_model():\n \n # create model\n model = Sequential()\n model.add(Conv2D(16, (5, 5), activation='relu', input_shape=(28, 28, 1)))\n model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))\n \n model.add(Conv2D(8, (2, 2), activation='relu'))\n model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))\n \n model.add(Flatten())\n model.add(Dense(100, activation='relu'))\n model.add(Dense(num_classes, activation='softmax'))\n \n # Compile model\n model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])\n return model", "_____no_output_____" ] ], [ [ "Now, let's call the function to create our new convolutional neural network, and then let's train it and evaluate it.\n", "_____no_output_____" ] ], [ [ "# build the model\nmodel = convolutional_model()\n\n# fit the model\nmodel.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=10, batch_size=200, verbose=2)\n\n# evaluate the model\nscores = model.evaluate(X_test, y_test, verbose=0)\nprint(\"Accuracy: {} \\n Error: {}\".format(scores[1], 100-scores[1]*100))", "Train on 60000 samples, validate on 10000 samples\nEpoch 1/10\n60000/60000 - 47s - loss: 0.4901 - acc: 0.8633 - val_loss: 0.1385 - val_acc: 0.9570\nEpoch 2/10\n60000/60000 - 47s - loss: 0.1185 - acc: 0.9642 - val_loss: 0.0848 - val_acc: 0.9728\nEpoch 3/10\n60000/60000 - 47s - loss: 0.0831 - acc: 0.9740 - val_loss: 0.0633 - val_acc: 0.9813\nEpoch 4/10\n60000/60000 - 47s - loss: 0.0657 - acc: 0.9795 - val_loss: 0.0661 - val_acc: 0.9783\nEpoch 5/10\n60000/60000 - 47s - loss: 0.0566 - acc: 0.9830 - val_loss: 0.0514 - val_acc: 0.9843\nEpoch 6/10\n60000/60000 - 47s - loss: 0.0496 - acc: 0.9845 - val_loss: 0.0476 - val_acc: 0.9868\nEpoch 7/10\n60000/60000 - 47s - loss: 0.0432 - acc: 0.9869 - val_loss: 0.0478 - val_acc: 0.9857\nEpoch 8/10\n60000/60000 - 47s - loss: 0.0400 - acc: 0.9873 - val_loss: 0.0497 - val_acc: 0.9848\nEpoch 9/10\n60000/60000 - 47s - loss: 0.0364 - acc: 0.9887 - val_loss: 0.0406 - val_acc: 0.9873\nEpoch 10/10\n60000/60000 - 47s - loss: 0.0325 - acc: 0.9899 - val_loss: 0.0373 - val_acc: 0.9883\nAccuracy: 0.9883000254631042 \n Error: 1.1699974536895752\n" ] ], [ [ "### Thank you for completing this lab!\n\nThis notebook was created by [Alex Aklson](https://www.linkedin.com/in/aklson?cm_mmc=Email_Newsletter-_-Developer_Ed%2BTech-_-WW_WW-_-SkillsNetwork-Courses-IBMDeveloperSkillsNetwork-DL0101EN-SkillsNetwork-20718188&cm_mmca1=000026UJ&cm_mmca2=10006555&cm_mmca3=M12345678&cvosrc=email.Newsletter.M12345678&cvo_campaign=000026UJ&cm_mmc=Email_Newsletter-_-Developer_Ed%2BTech-_-WW_WW-_-SkillsNetwork-Courses-IBMDeveloperSkillsNetwork-DL0101EN-SkillsNetwork-20718188&cm_mmca1=000026UJ&cm_mmca2=10006555&cm_mmca3=M12345678&cvosrc=email.Newsletter.M12345678&cvo_campaign=000026UJ). I hope you found this lab interesting and educational. Feel free to contact me if you have any questions!\n", "_____no_output_____" ], [ "## Change Log\n\n| Date (YYYY-MM-DD) | Version | Changed By | Change Description |\n| ----------------- | ------- | ---------- | ----------------------------------------------------------- |\n| 2020-09-21 | 2.0 | Srishti | Migrated Lab to Markdown and added to course repo in GitLab |\n\n<hr>\n\n## <h3 align=\"center\"> © IBM Corporation 2020. All rights reserved. <h3/>\n", "_____no_output_____" ], [ "This notebook is part of a course on **Coursera** called _Introduction to Deep Learning & Neural Networks with Keras_. If you accessed this notebook outside the course, you can take this course online by clicking [here](https://cocl.us/DL0101EN_Coursera_Week4_LAB1).\n", "_____no_output_____" ], [ "<hr>\n\nCopyright © 2019 [IBM Developer Skills Network](https://cognitiveclass.ai?utm_source=bducopyrightlink&utm_medium=dswb&utm_campaign=bdu&cm_mmc=Email_Newsletter-_-Developer_Ed%2BTech-_-WW_WW-_-SkillsNetwork-Courses-IBMDeveloperSkillsNetwork-DL0101EN-SkillsNetwork-20718188&cm_mmca1=000026UJ&cm_mmca2=10006555&cm_mmca3=M12345678&cvosrc=email.Newsletter.M12345678&cvo_campaign=000026UJ&cm_mmc=Email_Newsletter-_-Developer_Ed%2BTech-_-WW_WW-_-SkillsNetwork-Courses-IBMDeveloperSkillsNetwork-DL0101EN-SkillsNetwork-20718188&cm_mmca1=000026UJ&cm_mmca2=10006555&cm_mmca3=M12345678&cvosrc=email.Newsletter.M12345678&cvo_campaign=000026UJ&cm_mmc=Email_Newsletter-_-Developer_Ed%2BTech-_-WW_WW-_-SkillsNetwork-Courses-IBMDeveloperSkillsNetwork-DL0101EN-SkillsNetwork-20718188&cm_mmca1=000026UJ&cm_mmca2=10006555&cm_mmca3=M12345678&cvosrc=email.Newsletter.M12345678&cvo_campaign=000026UJ&cm_mmc=Email_Newsletter-_-Developer_Ed%2BTech-_-WW_WW-_-SkillsNetwork-Courses-IBMDeveloperSkillsNetwork-DL0101EN-SkillsNetwork-20718188&cm_mmca1=000026UJ&cm_mmca2=10006555&cm_mmca3=M12345678&cvosrc=email.Newsletter.M12345678&cvo_campaign=000026UJ). This notebook and its source code are released under the terms of the [MIT License](https://bigdatauniversity.com/mit-license?cm_mmc=Email_Newsletter-_-Developer_Ed%2BTech-_-WW_WW-_-SkillsNetwork-Courses-IBMDeveloperSkillsNetwork-DL0101EN-SkillsNetwork-20718188&cm_mmca1=000026UJ&cm_mmca2=10006555&cm_mmca3=M12345678&cvosrc=email.Newsletter.M12345678&cvo_campaign=000026UJ&cm_mmc=Email_Newsletter-_-Developer_Ed%2BTech-_-WW_WW-_-SkillsNetwork-Courses-IBMDeveloperSkillsNetwork-DL0101EN-SkillsNetwork-20718188&cm_mmca1=000026UJ&cm_mmca2=10006555&cm_mmca3=M12345678&cvosrc=email.Newsletter.M12345678&cvo_campaign=000026UJ).\n", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# 多元函数微分法及其应用\n\n只有一个自变量的函数叫做一元函数。在很多实际问题中往往牵涉到多方面的因素，反映到数学上，就是一个变量依赖于多个变量的情形。这就提出了多元函数以及多元函数的微分和积分问题。本章将在一元函数微分学的基础上，讨论多元函数的微分法及其应用。讨论中我们以二元函数为主，因为从一元函数到二元函数会产生新的问题，而从二元函数到二元以上的多元函数则可以类推。\n\n本节包括以下内容：\n1. 多元函数的基本概念\n2. 偏导数\n3. 全微分\n4. 多元复合函数的求导法则\n5. 隐函数的求导公式\n6. 多元函数微分学的几何应用\n7. 方向导数和梯度\n8. 多元函数的极值及其求法\n9. 二元函数的泰勒公式\n10. 最小二乘法", "_____no_output_____" ], [ "### 1. 多元函数的基本概念\n\n#### 1.1 平面点集 n维空间\n\n在讨论一元函数时，一些概念、理论和方法，都是基于 $\\mathbb{R}^1$ 中的点集、两点间的距离、区间和邻域等概念。为了将一元函数微积分推广到多元的情形，首先需要将上述概念加以推广，同时还需涉及一些其他概念。为此先引入平面点集的一些基本概念，将有关概念从 $\\mathbb{R}^1$ 中的情形推广到 $\\mathbb{R}^2$ 中；然后引入 $n$ 维空间，以便推广到一般的 $\\mathbb{R}^n$ 中。\n\n**平面点集**：由平面解析几何知道，当在平面上引入一个直角坐标系后，平面上的点 $P$ 与有序二元实数组 $(x,y)$ 之间就建立了一一对应。于是，我们常把有序实数组 $(x,y)$ 与平面上的点 $P$ 视作是等同的。这种建立了坐标系的平面称为坐标平面。二元有序实数组 $(x,y)$ 的全体，即 $\\mathbb{R}^2=\\mathbb{R} \\times \\mathbb{R}=\\{(x,y)|x,y\\in\\mathbb{R}\\}$ 就表示坐标平面。\n\n坐标平面上具有某种性质 $P$ 的点的集合，称为平面点集，记作 $E=\\{(x,y)|(x,y)具有性质P\\}$。\n\n现在我们来引入 $\\mathbb{R}^2$ 中邻域的概念。\n\n设 $P_0(x_0,y_0)$ 是 $xOy$ 平面上的一个点，$\\delta$ 是某一正数。与点 $P_0(x_0,y_0)$ 距离小于 $\\delta$ 的点 $P(x,y)$ 的全体，称为点 $P_0$ 的 $\\delta$ 邻域，记作 $U(P_0,\\delta)$，即\n$$ U(P_0,\\delta)=\\{P||PP_0|<\\delta\\} $$\n也就是\n$$ U(P_0,\\delta)=\\{(x,y)|\\sqrt{(x-x_0)^2+(y-y_0)^2}<\\delta\\} $$\n\n点 $P_0$ 的去心 $\\delta$ 邻域，记作 $\\mathring{U}(P_0, \\delta)$，即\n$$\\mathring{U}(P_0, \\delta)=\\{P|0<|PP_0|<\\delta\\}$$\n\n在几何上，$U(P_0,\\delta)$ 就是 $xOy$ 平面上以点 $P_0(x_0,y_0)$ 为中心，$\\delta>0$ 为半径的圆内部的点 $P(x,y)$ 的全体。\n\n如果不需要强调邻域的半径 $\\delta$，则用 $U(P_0)$ 表示点 $P_0$ 的某个邻域，点 $P_0$ 的去心邻域记作 $\\mathring{U}(P_0)$。\n\n下面利用邻域来描述点和点集之间的关系。\n\n任意一点 $P \\in \\mathbb{R}^2$ 与任意一个点集 $E \\subset \\mathbb{R}^2$ 之间必有以下三种关系中的一种：\n\n1. **内点**：如果存在点 $P$ 的某个邻域 $U(P)$，使得 $U(P) \\subset E$，则称 $P$ 为 $E$ 的内点。\n2. **外点**：如果存在点 $P$ 的某个邻域 $U(P)$，使得 $U(P) \\cap E = \\emptyset$，则称 $P$ 为 $E$ 的外点。\n3. **边界点**：如果点 $P$ 的任一邻域内既含有属于 $E$ 的点，又含有不属于 $E$ 的点，则称 $P$ 为 $E$ 的边界点。\n\n$E$ 的所有边界点的全体，称为 $E$ 的**边界**，记作 $\\partial E$。\n\n$E$ 的内点必属于 $E$；$E$ 的外点不定不属于 $E$；而 $E$ 的边界点可能属于 $E$，也可能不属于 $E$。\n\n根据点集所属点的特征，再来定义一些重要的平面点集。\n\n1. **开集**：如果点集 $E$ 的点都是 $E$ 的内点，则称 $E$ 为开集。\n2. **闭集**：如果点集 $E$ 的边界 $\\partial E \\subset E$，则称 $E$ 为闭集。\n3. **连通集**：如果点集 $E$ 内的任何两点，都可用折线联结起来，且该折线上的点都属于 $E$，则称 $E$ 为连通集。\n4. **区域（开区域）**：连通的开集称为区域或开区域。\n5. **闭区域**：开区域连通它的边界一起所构成的点集称为闭区域。\n6. **有界集**：对于平面点集 $E$，如果存在某一正数 $r$，使得 $E \\subset U(O,r)$，其中 $O$ 是坐标原点，则称 $E$ 为有界集。\n7. **无界集**：一个集合如果不是有界集，就称这集合为无界集。\n\n**n维空间**：设 $n$ 为取定的一个正整数，我们用 $\\mathbb{R}^n$ 表示 $n$ 元有序实数组 $(x_1, x_2,...,x_n)$ 的全体构成的集合，即\n$$ \\mathbb{R}^n = \\mathbb{R} \\times \\mathbb{R} \\times \\cdots \\times \\mathbb{R} = \\{(x_1, x_2,..., x_n)|x_i \\in \\mathbb{R}, i=1,2,...,n\\} $$\n$\\mathbb{R}^n$ 中的元素 $(x_1, x_2,...,x_n)$ 有时也用单个字母 $x$ 来表示，即 $x=(x_1,x_2,...,x_n)$。当所有的 $x_i(i=1,2,...,n)$ 都为零时，称这样的元素为 $\\mathbb{R}^n$ 中的零元，即为 $0$ 或 $O$。在解析几何中，通过直角坐标系，$\\mathbb{R}^2$（或 $\\mathbb{R}^3$）中的元素分别与平面（或空间）中的点或向量建立一一对应，因而 $\\mathbb{R}^n$ 中的元素 $x=(x_1,x_2,...,x_n)$ 也称为 $\\mathbb{R}^n$ 中的一个点或一个 $n$ 维向量，$x_i$ 称为点 $x$ 的第 $i$ 个坐标或 $n$ 维向量 $x$ 的第 $i$ 个分量。特别地，$\\mathbb{R}^n$ 中的零元 $0$ 称为 $\\mathbb{R}^n$ 中的坐标原点或 $n$ 维零向量。\n\n为了在集合 $\\mathbb{R}^n$ 中的元素之间建立联系，在 $\\mathbb{R}^n$ 中定义线性运算如下：\n\n设 $x=(x_1,x_2,...,x_n), y=(y_1,y_2,...,y_n)$ 为 $\\mathbb{R}^n$ 中任意两个元素，$\\lambda \\in \\mathbb{R}$，规定：\n$$\n\\begin{split}\n& x+y=(x_1+y_1, x_2+y_2,...,x_n+y_n) \\\\\n& \\lambda x = (\\lambda x_1, \\lambda x_2, ..., \\lambda x_n)\n\\end{split}\n$$\n这样定义了线性运算的集合 $\\mathbb{R}^n$ 称为 $n$ 维空间。\n\n$\\mathbb{R}^n$ 中点 $x=(x_1,x_2,...,x_n)$ 和点 $y=(y_1,y_2,...,y_n)$ 间的距离，记作 $\\rho(x,y)$，规定\n$$ \\rho(x,y) = \\sqrt{(x_1-y_1)^2+(x_2-y_2)^2+\\cdots+(x_n-y_n)^2} $$\n显然，$n=1,2,3$ 时，上述规定与数轴上、直角坐标系下平面及空间中两点间的距离一致。\n\n$\\mathbb{R}^n$ 中元素 $x=(x_1,x_2,...,x_n)$ 与零元 $0$ 之间的距离 $\\rho(x, 0)$ 记作 $||x||$（在 $\\mathbb{R}^1, \\mathbb{R}^2, \\mathbb{R}^3$ 中，通常将 $||x||$ 记作 $|x|$），即\n$$ ||x|| = \\sqrt{x_1^2+x_2^2+\\cdots+x_n^2} $$\n采用这一记号，结合向量的线性运算，便得\n$$ ||x-y|| = \\sqrt{(x_1-y_1)^2+(x_2-y_2)^2+\\cdots+(x_n-y_n)^2} = \\rho(x,y) $$\n\n在 $n$ 维空间 $\\mathbb{R}^n$ 中定义了距离以后，就可以定义 $\\mathbb{R}^n$ 中变元的极限：\n\n设 $x=(x_1,x_2,...,x_n), a=(a_1,a_2,...,a_n) \\in \\mathbb{R}^n$。 如果\n$$ ||x-a|| \\rightarrow 0 $$\n则称变元 $x$ 在 $\\mathbb{R}^n$ 中趋于固定元 $a$，记作 $x \\rightarrow a$，显然\n$$ x \\rightarrow a \\Leftrightarrow x_1 \\rightarrow a_1, x_2 \\rightarrow a_2, \\cdots, x_n \\rightarrow a_n $$\n\n在 $\\mathbb{R}^n$ 中线性运算和距离的引入，使得前面讨论过的有关平面点集的一系列概念，可以方便地引入到 $n(n \\geq 3)$ 维平面中来，例如：\n\n设 $a=(a_1,a_2,...,a_n) \\in \\mathbb{R}^n$，$\\delta$ 是某一正数，则 $n$ 维空间内的点集\n$$ U(a, \\delta) = \\{x|x \\in \\mathbb{R}^n, \\rho(x,a) < \\delta \\} $$\n就定义为 $\\mathbb{R}^n$ 中点 $a$ 的邻域。以邻域为基础，可以定义点集的内点、外电、边界点以及开集、闭集、区域等一系列概念，这里不再赘述。", "_____no_output_____" ], [ "#### 1.2 多元函数概念\n\n**定义1 设 $D$ 是 $\\mathbb{R}^2$ 的一个非空子集，称映射 $f:D \\rightarrow \\mathbb{R}$ 为定义在 $D$ 上的二元函数，通常记为\n$$ z=f(x,y), (x,y) \\in D $$\n或\n$$ z=f(P), P \\in D $$\n其中点集 $D$ 称为该函数的定义域，$x,y$ 称为自变量，$z$ 称为因变量。\n**\n\n上述定义中，与自变量 $x,y$ 的一对值（即二元有序实数组）$(x,y)$ 相对应的因变量 $z$ 的值，也称为 $f$ 在点 $(x,y)$ 处的函数值，记作 $f(x,y)$，即 $z=f(x,y)$。函数值 $f(x,y)$ 全体所构成的集合称为函数 $f$ 的值域，记作 $f(D)$，即\n$$ f(D)=\\{z|z=f(x,y),(x,y) \\in D\\} $$\n\n与一元函数的情形相仿，记号 $f$ 和 $f(x,y)$ 的意义是有区别的，但习惯上常用记号 $f(x,y),(x,y) \\in D$ 或 $z=f(x,y),(x,y) \\in D$ 来表示 $D$ 上的二元函数 $f$。表示二元函数的记号 $f$ 也是可以任意选取的，例如也可以记为 $z=\\phi(x,y), z=z(x,y)$ 等。\n\n类似地可以定义三元函数 $u=f(x,y,z),(x,y,z) \\in D$ 以及三元以上的函数。一般的，把定义1中的平面点集 $D$ 换成 $n$ 维空间 $\\mathbb{R}^n$ 内的点集 $D$，映射 $f:D \\rightarrow \\mathbb{R}$ 就称为定义在 $D$ 上的** $n$ 元函数**，通常记为\n$$ u=f(x_1,x_2,\\cdots,x_n),(x_1,x_2,\\cdots,x_n) \\in D $$\n或简记为\n$$ u=f(x),x=(x_1,x_2,\\cdots,x_n) \\in D $$\n也可记为\n$$ u=f(P),P(x_1,x_2,\\cdots,x_n) \\in D $$\n\n在 $n=2$ 或 $n=3$ 时，习惯上将点 $(x_1, x_2)$ 与点 $(x_1, x_2, x_3)$ 分别写成 $(x,y)$ 与 $(x,y,z)$。这时，若用字母表示 $\\mathbb{R}^2$ 或 $\\mathbb{R}^3$ 中的点，即写成 $P(x,y)$ 或 $M(x,y,z)$，则相应的二元函数及三元函数也可简记为 $z=f(P)$ 及 $u=f(M)$。\n\n当 $n=1$ 时，$n$ 元函数就是一元函数。当 $n \\geq 2$ 时，$n$ 元函数统称为**多元函数**。\n\n关于多元函数的定义域，与一元函数相类似，我们作如下约定：在一般地讨论用算式表达的多元函数 $u=f(x)$ 时，就以使这个算式有意义的变元 $x$ 的值所组成的点集为这个**多元函数的自然定义域**。因而，对这类函数，它的定义域不再特别标出。\n\n设函数 $z=f(x,y)$ 的定义域为 $D$。对于任意取定的点 $P(x,y) \\in D$，对应的函数值为 $z=f(x,y)$。这样，以 $x$ 为横坐标，$y$ 为纵坐标，$z=f(x,y)$ 为竖坐标在空间就确定一点 $M(x,y,z)$。当 $(x,y)$ 遍取 $D$ 上的一切点时，得到一个空间点集\n$$ \\{(x,y,z)|z=f(x,y), (x,y) \\in D\\} $$\n这个点集称为**二元函数 $z=f(x,y)$ 的图形**。通常我们也说二元函数的图形是一张曲面。", "_____no_output_____" ], [ "#### 1.3 多元函数的极限\n\n先讨论二元函数 $z=f(x,y)$ 当 $(x,y) \\rightarrow (x_0,y_0)$，即 $P(x,y) \\rightarrow P_0(x_0,y_0)$ 时的极限。\n\n这里 $P \\rightarrow P_0$ 表示点 $P$ 以任何方式趋于点 $P_0$，也就是点 $P$ 与点 $P_0$ 间的距离趋于零，即\n$$ |PP_0| = \\sqrt{(x-x_0)^2 + (y-y_0)^2} \\rightarrow 0 $$\n\n与一元函数的极限概念类似，如果在 $P(x,y) \\rightarrow P_0(x_0, y_0)$ 的过程中，对应的函数值 $f(x,y)$ 无限接近于一个确定的常数 $A$，就说 $A$ 是函数 $f(x,y)$ 当 $(x,y) \\rightarrow (x_0,y_0)$ 时的极限。下面用 $\\epsilon - \\delta$ 语言描述这个极限概念。\n\n**定义2 设二元函数 $f(P)=f(x,y)$ 的定义域为 $D$，$P_0(x_0,y_0)$ 是 $D$ 的聚点。如果存在常数 $A$，对于任意给定的正数 $\\epsilon$，总存在正数 $\\delta$，使得当点 $P(x,y) \\in D \\cap \\mathring{U}(P_0,\\delta) $ 时，都有\n$$ |f(P)-A| = |f(x,y)-A| < \\epsilon $$\n成立，那么就称常数 $A$ 为函数 $f(x,y)$ 当 $(x,y) \\rightarrow (x_0,y_0) $ 时的极限，记作\n$$ \\lim_{(x,y) \\rightarrow (x_0, y_0)}f(x,y)=A 或 f(x,y) \\rightarrow A((x,y) \\rightarrow (x_0,y_0)) $$\n也记作\n$$ \\lim_{P \\rightarrow P_0}f(P)=A 或 f(P) \\rightarrow A(P \\rightarrow P_0) $$\n**\n\n为了区别于一元函数的极限，我们把二元函数的极限叫做**二重极限**。\n\n必须注意，所谓二重极限存在，是指 $P(x,y)$ 以任何方式趋于 $P_0(x_0,y_0)$ 时，$f(x,y)$ 都无限接近于 $A$。因此，如果 $P(x,y)$ 以某一种特殊方式，例如沿着一条定直线或定曲线趋于 $P_0(x_0,y_0)$ 时，即使 $f(x,y)$ 无限接近于某一确定值，我们还不能由此断定函数的极限存在。但是反过来，如果当 $P(x,y)$ 以不同的方式趋于 $P_0(x_0,y_0)$ 时，$f(x,y)$ 趋于不同的值，那么就可以判定这函数的极限不存在。\n\n以上关于二元函数的极限概念，可相应地推广到 $n$ 元函数 $u=f(P)$ 上去。\n\n关于多元函数的极限运算，有与一元函数类似的运算法则。", "_____no_output_____" ], [ "#### 1.4 多元函数的连续性\n\n**定义3 设二元函数 $f(P)=f(x,y)$ 的定义域为 $D$，$P_0(x_0,y_0)$ 为 $D$ 的聚点，且 $P_0 \\in D$，如果\n$$ \\lim_{(x,y) \\rightarrow (x_0, y_0)}f(x,y) = f(x_0,y_0) $$\n则称函数 $f(x,y)$ 在点 $P_0(x_0,y_0)$ 连续。\n设函数 $f(x,y)$ 在 $D$ 上有定义，$D$ 内的每一点都是函数定义域的聚点。如果函数 $f(x,y)$ 在 $D$ 的每一点都连续，那么就称函数 $f(x,y)$ 在 $D$ 上连续，或者称 $f(x,y)$ 是 $D$ 上的连续函数。\n**\n\n以上关于二元函数的连续性概念，可相应地推广到 $n$ 元函数 $f(P)$ 上去。\n\n**定义4 设函数 $f(x,y)$ 的定义域为 $D$，$P_0(x_0,y_0)$ 是 $D$ 的聚点。如果函数 $f(x,y)$ 在点 $P_0(x_0,y_0)$ 不连续，则称 $P_0(x_0,y_0)$ 为函数 $f(x,y)$ 的间断点。**\n\n前面已经指出：一元函数中关于极限的运算法则，对于多元函数仍然适用。根据多元函数的极限运算法则，可以证明多元连续函数的和、差、积仍为连续函数；连续函数的商在分母不为零处仍连续；多元连续函数的复合函数也是连续函数。\n\n与一元初等函数相类似。多元初等函数是指可用一个式子表示的多元函数，这个式子是由常数及具有不同自变量的一元基本初等函数经过有限次的四则运算和符合运算而得到的。\n\n一切多元初等函数在其定义区域内是连续的。所谓定义区域是指包含在定义域内的区域或闭区域。\n\n由多元初等函数的连续性，如果要求它在点 $P_0$ 处的极限，而该点又在此函数的定义区域内，则极限值就是函数在该点的函数值，即\n$$ \\lim_{P \\rightarrow P_0}f(P) = f(P_0) $$\n\n与闭区间上一元连续函数的性质相类似，在有界闭区域上连续的多元函数具有如下性质：\n\n**性质1（有界性与最大值最小值定理）**：在有界闭区域 $D$ 上的多元连续函数，必定在 $D$ 上有界，且能取得它的最大值和最小值。\n\n**性质2（介值定理）**：在有界闭区域 $D$ 上的多元连续函数必取得介于最大值和最小值之间的任何值。\n\n**性质3（一致连续性定理）**：在有界闭区域 $D$ 上的多元连续函数必定在 $D$ 上一致连续。", "_____no_output_____" ], [ "### 2. 偏导数\n\n#### 2.1 偏导数的定义及其计算方法\n\n在研究一元函数时，我们从研究函数的变化率引入了导数概念。对于多元函数同样需要讨论它的变化率。但多元函数的自变量不止一个，因变量和自变量的关系要比一元函数复杂得多。在这一节里，我们首先考虑多元函数关于其中一个自变量的变化率。以二元函数 $f(x,y)$ 为例，如果只有自变量 $x$ 变化，而自变量 $y$ 固定（即看做常量），这时它就是 $x$ 的一元函数，这函数对 $x$ 的导数，就称为二元函数 $z=f(x,y)$ 对于 $x$ 的**偏导数**，即有如下定义：\n\n**定义 设函数 $z=f(x,y)$ 在点 $(x_0,y_0)$ 的某一邻域内有定义，当 $y$ 固定在 $y_0$ 而 $x$ 在 $x_0$ 处有增量 $\\Delta x$ 时，相应的函数有增量\n$$ f(x_0+\\Delta x,y_0) - f(x_0,y_0) $$\n如果\n$$ \\lim_{\\Delta x \\rightarrow 0} \\frac{f(x_0+\\Delta x,y_0) - f(x_0,y_0)}{\\Delta x} $$\n存在，则称此极限为函数 $z=f(x,y)$ 在点 $(x_0,y_0)$ 处对 $x$ 的偏导数，记作\n$$ \\frac{\\partial z}{\\partial x}|_{\\begin{split}x=x_0\\\\y=y_0\\end{split}}, \\frac{\\partial f}{\\partial x}|_{\\begin{split}x=x_0\\\\y=y_0\\end{split}}, z_x|_{\\begin{split}x=x_0\\\\y=y_0\\end{split}} 或 f_x(x_0,y_0) $$\n**\n\n如果函数 $z=f(x,y)$ 在区域 $D$ 内每一点 $(x,y)$ 处对 $x$ 的偏导数都存在，那么这个偏导数就是 $x,y$ 的函数，它就称为函数 $z=f(x,y)$ 对自变量 $x$ 的偏导函数，记作\n$$ \\frac{\\partial z}{\\partial x},\\frac{\\partial f}{\\partial x},z_x 或 f_x(x,y) $$\n\n类似地，可以定义函数 $z=f(x,y)$ 对自变量 $y$ 的偏导函数，记作\n$$ \\frac{\\partial z}{\\partial y},\\frac{\\partial f}{\\partial y},z_y 或 f_y(x,y) $$\n\n就像一元函数的导函数一样，在不至于混淆的地方也把偏导函数简称为偏导数。至于实际求 $z=f(x,y)$ 的偏导数，并不需要新的方法，因为这里只有一个自变量在变动，另一个自变量是看做固定的，所以仍旧是一元函数的微分法问题。偏导数的概念还可推广到二元以上的函数。\n\n二元函数 $z=f(x,y)$ 在点 $(x_0,y_0)$ 的偏导数有下述几何意义：\n\n设 $M_0(x_0,y_0,f(x_0,y_0))$ 为曲面 $z=f(x,y)$ 上的一点，过 $M_0$ 作平面 $y=y_0$，截此平面得一曲线，此曲线在平面 $y=y_0$ 上的方程为 $z=f(x, y_0)$，则导数 $\\frac{d}{dx}f(x,y_0)|_{x=x_0}$，即偏导数 $f_x(x_0,y_0)$，就是这曲线在点 $M_0$ 处的切线对 $x$ 轴的斜率。同样，偏导数 $f_y(x_0,y_0)$ 的几何意义是曲面被平面 $x=x_0$ 所截得的曲线在点 $M_0$ 处的切线对 $y$ 轴的斜率。\n\n我们已经知道，如果一元函数在某点具有导数，则它在该点必定连续。但对于多元函数来说，即使各偏导数在某点都存在，也不能保证函数在该点连续。这是因为各偏导数存在只能保证点 $P$ 沿着平行于坐标轴的方向趋于 $P_0$ 时，函数值 $f(P)$ 趋于 $f(P_0)$，但不能保证点 $P$ 按任何方式趋于 $P_0$ 时，函数值 $f(P)$ 都趋于 $f(P_0)$。", "_____no_output_____" ], [ "#### 2.2 高阶偏导数\n\n设函数 $z=f(x,y)$ 在区域 $D$ 内具有偏导数\n$$ \\frac{\\partial z}{\\partial x}=f_x(x,y), \\frac{\\partial z}{\\partial y}=f_y(x,y) $$\n那么在 $D$ 内 $f_x(x,y),f_y(x,y)$ 都是 $x,y$ 的函数。如果这两个函数的偏导数也存在，则称它们是函数 $z=f(x,y)$ 的二阶偏导数。按照对变量求导次序的不同有下列四个二阶偏导数：\n$$\n\\begin{split}\n\\frac{\\partial}{\\partial x}(\\frac{\\partial z}{\\partial x})=\\frac{\\partial^2 z}{\\partial x^2}=f_{xx}(x,y) \\\\\n\\frac{\\partial}{\\partial y}(\\frac{\\partial z}{\\partial x})=\\frac{\\partial^2 z}{\\partial x \\partial y}=f_{xy}(x,y) \\\\\n\\frac{\\partial}{\\partial x}(\\frac{\\partial z}{\\partial y})=\\frac{\\partial^2 z}{\\partial y \\partial x}=f_{yx}(x,y) \\\\\n\\frac{\\partial}{\\partial y}(\\frac{\\partial z}{\\partial y})=\\frac{\\partial^2 z}{\\partial y^2}=f_{yy}(x,y) \\\\\n\\end{split}\n$$\n\n其中第二、三两个偏导数称为**混合偏导数**。同样可得三阶、四阶、...以及 $n$ 阶偏导数。二阶及二阶以上的偏导数统称为**高阶偏导数**。\n\n**定理 如果函数 $z=f(x,y)$ 的两个二阶混合偏导数 $\\frac{\\partial^2 z}{\\partial y \\partial x}$ 及 $\\frac{\\partial^2 z}{\\partial x \\partial y}$ 在区域 $D$ 内连续，那么在该区域内这两个二阶混合偏导数必相等。**\n\n换句话说，二阶混合偏导数在连续的条件下与求导的次序无关。\n\n对于二元以上的函数，也可以类似地定义高阶偏导数。而且高阶混合偏导数在偏导数连续的条件下也与求导的次序无关。", "_____no_output_____" ], [ "### 3. 全微分\n\n#### 3.1 全微分的定义\n\n由偏导数的定义知道，二元函数对某个自变量的偏导数表示当另一个自变量固定时，因变量相对于该自变量的变化率。根据一元函数微分学中增量与微分的关系，可得\n$$\n\\begin{split}\nf(x+\\Delta x, y) - f(x,y) \\approx f_x(x,y)\\Delta x \\\\\nf(x, y+\\Delta y) - f(x,y) \\approx f_y(x,y)\\Delta y\n\\end{split}\n$$\n上面两式的左端分别叫做二元函数对 $x$ 和对 $y$ 的**偏增量**，而右端分别叫做二元函数对 $x$ 和对 $y$ 的**偏微分**。\n\n在实际问题中，有时需要研究多元函数中各个自变量都取得增量时因变量所获得的增量，即所谓全增量的问题。下面以二元函数为例进行讨论。\n\n设函数 $z=f(x,y)$ 在点 $P(x,y)$ 的某邻域内有定义，$P'(x+\\Delta x,y+\\Delta y)$ 为这邻域内的任意一点，则称这两点的函数值之差 $f(x+\\Delta x, y+\\Delta y)-f(x,y)$ 为函数在点 $P$ 对应于自变量增量 $\\Delta x, \\Delta y$ 的**全增量**，记作 $\\Delta z$，即\n$$ \\Delta z = f(x+\\Delta x, y+\\Delta y)-f(x,y) $$\n\n一般说来，计算全增量 $\\Delta z$ 比较复杂。与一元函数的情形一样，我们希望用自变量的增量 $\\Delta x, \\Delta y$ 的线性函数来近似地代替函数的全增量 $\\Delta z$，从而引入如下定义。\n\n**定义 设函数 $z=f(x,y)$ 在点 $(x,y)$ 的某邻域内有定义，如果函数在点 $(x,y)$ 的全增量\n$$ \\Delta z = f(x+\\Delta x, y+\\Delta y) - f(x,y) $$\n可表示为\n$$ \\Delta z = A\\Delta x + B\\Delta y + o(\\rho) $$\n其中 $A, B$ 不依赖于 $\\Delta x, \\Delta y$ 而仅与 $x,y$ 有关，$\\rho=\\sqrt{(\\Delta x)^2+(\\Delta y)^2}$，则称函数 $z=f(x,y)$ 在点 $(x,y)$ 可微分，而 $A\\Delta x + B\\Delta y$ 称为函数 $z=f(x,y)$ 在点 $(x,y)$ 的全微分，记作 $dz$，即\n$$ dz=A\\Delta x + B\\Delta y $$\n**\n\n如果函数在区域 $D$ 内各点处都可微分，那么称这函数**在 $D$ 内可微分**。\n\n在第二节中曾指出，多元函数在某点的偏导数存在，并不能保证函数在该点连续。但是，由上述定义可知，如果函数 $z=f(x,y)$ 在点 $(x,y)$ 可微分，那么这函数在该点必定连续。事实上，由\n$$ \\lim_{\\rho \\rightarrow 0}\\Delta z=0 $$\n从而\n$$ \\lim_{(\\Delta x, \\Delta y) \\rightarrow (0,0)}f(x+\\Delta x, y+\\Delta y) = \\lim_{\\rho \\rightarrow 0}[f(x,y)+\\Delta z] = f(x,y) $$\n因此函数 $z=f(x,y)$ 在点 $(x,y)$ 处连续。\n\n下面讨论函数 $z=f(x,y)$ 在点 $(x,y)$ 可微分的条件。\n\n**定理1（必要条件） 如果函数 $z=f(x,y)$ 在点 $(x,y)$ 可微分，则该函数在点 $(x,y)$ 的偏导数 $\\frac{\\partial z}{\\partial x}, \\frac{\\partial z}{\\partial y}$ 必定存在，且函数 $z=f(x,y)$ 在点 $(x,y)$ 的全微分为\n$$ dz=\\frac{\\partial z}{\\partial x}\\Delta x + \\frac{\\partial z}{\\partial y}\\Delta y $$\n**\n\n一元函数在某点的导数存在是微分存在的充分必要条件。但对于多元函数来说，情形就不同了。当函数的各偏导数都存在时，虽然能形式地写出 $\\frac{\\partial z}{\\partial x}\\Delta x + \\frac{\\partial z}{\\partial y}\\Delta y$，但它与 $\\Delta z$ 之差并不一定是较 $\\rho$ 高阶的无穷小，因此它不一定是函数的全微分。换句话说，各偏导数的存在只是全微分存在的必要条件而不是充分条件。\n\n**定理2（充分条件） 如果函数 $z=f(x,y)$ 的偏导数 $\\frac{\\partial z}{\\partial x}, \\frac{\\partial z}{\\partial y}$ 在点 $(x,y)$ 连续，则函数在该点可微分。**\n\n以上关于二元函数全微分的定义及可微分的必要条件和充分条件，可以完全类似地推广到三元和三元以上的多元函数。\n\n习惯上，我们将自变量的增量 $\\Delta x, \\Delta y$ 分别记作 $dx, dy$，并分别称为自变量 $x,y$ 的微分。这样，函数 $z=f(x,y)$ 的全微分就可写成\n$$ dz=\\frac{\\partial z}{\\partial x}dx + \\frac{\\partial z}{\\partial y}dy $$\n\n通常把二元函数的全微分等于它的两个偏微分之和这件事称为二元函数的微分符合**叠加原理**。叠加原理也适用于二元以上的函数的情形。", "_____no_output_____" ], [ "#### 3.2 全微分在近似计算中的应用\n\n由二元函数的全微分的定义及关于全微分存在的充分条件可知，当二元函数 $z=f(x,y)$ 在点 $P(x,y)$ 的两个偏导数 $f_x(x,y), f_y(x,y)$ 连续，并且 $|\\Delta x|, |\\Delta y|$ 都较小时，就有近似等式\n$$ \\Delta z \\approx dz = f_x(x,y)\\Delta x + f_y(x,y)\\Delta y $$\n上式也可以写成\n$$ f(x+\\Delta x, y+\\Delta y) \\approx f(x,y) + f_x(x,y)\\Delta x + f_y(x,y)\\Delta y $$\n\n与一元函数的情形相类似，可以利用上述两个式子对二元函数作近似计算和误差估计。", "_____no_output_____" ], [ "#### 3.3 二元函数实例\n\n考虑函数\n$$f(x,y) = \\left\\{\n\\begin{aligned}\n& \\frac{xy}{x^2+y^2}, & x^2+y^2 \\neq 0 \\\\\n& 0, & x^2+y^2 = 0\n\\end{aligned}\n\\right.\n$$\n\n其曲面图像如下所示：", "_____no_output_____" ] ], [ [ "import numpy as np\nimport matplotlib.pyplot as plt\nfrom matplotlib import cm\nfrom mpl_toolkits.mplot3d import Axes3D\n\n%matplotlib inline\n\[email protected]\ndef f(x, y):\n return x * y / (x ** 2 + y ** 2)\n\nstep = 0.05\nx_min, x_max = -1, 1\ny_min, y_max = -1, 1\nx_range, y_range = np.arange(x_min, x_max + step, step), np.arange(y_min, y_max + step, step)\nx_mat, y_mat = np.meshgrid(x_range, y_range)\nz = f(x_mat.reshape(-1), y_mat.reshape(-1)).reshape(x_mat.shape)\nfig = plt.figure(figsize=(12, 6))\nax1 = fig.add_subplot(1, 2, 1, projection='3d', elev=50, azim=-50)\nax1.plot_surface(x_mat, y_mat, z, cmap=cm.jet, rstride=1, cstride=1, edgecolor='none',alpha=.8)\nax1.set_xlabel('$x$')\nax1.set_ylabel('$y$')\nax1.set_zlabel('$z$')\nplt.show()", "_____no_output_____" ] ], [ [ "重点考察 $(0, 0)$ 这个点：\n\n**极限**\n\n显然当点 $P(x, y)$ 沿 $x$ 轴趋于点 $(0,0)$ 时\n$$ \\lim_{\\begin{split}(x,y)\\rightarrow (0,0) \\\\ y=0 \\end{split}}f(x,y) = \\lim_{x \\rightarrow 0}f(x,0) =\\lim_{x \\rightarrow 0}0 = 0$$\n又当点 $P(x, y)$ 沿 $y$ 轴趋于点 $(0,0)$ 时\n$$ \\lim_{\\begin{split}(x,y)\\rightarrow (0,0) \\\\ x=0 \\end{split}}f(x,y) = \\lim_{y \\rightarrow 0}f(0,y) =\\lim_{y \\rightarrow 0}0 = 0$$\n虽然点 $P(x,y)$ 以上述两种特殊方式（沿 $x$ 轴或沿 $y$ 轴）趋于原点时函数的极限存在并且相等，但是**极限 $\\lim_{(x,y) \\rightarrow (0,0)}f(x,y)$ 并不存在**。这是因为当点 $P(x,y)$ 沿着直线 $y=kx$ 趋于点 $(0,0)$ 时，有\n$$ \\lim_{\\begin{split}(x,y)\\rightarrow (0,0) \\\\ y=kx \\end{split}}\\frac{xy}{x^2+y^2} = \\lim_{x \\rightarrow 0}\\frac{kx^2}{x^2+k^2x^2} = \\frac{k}{1+k^2}$$\n显然它是随着 $k$ 的值的不同而改变的。\n\n**连续性**\n\n由于极限不存在，所以点 $(0,0)$ 是该函数的一个间断点。故**该函数在点 $(0, 0)$ 不连续**。\n\n**偏导数**\n$$\n\\begin{split}\nf_x(0,0) = \\lim_{\\Delta x \\rightarrow 0} \\frac{f(0+\\Delta x,0) - f(0,0)}{\\Delta x} = \\lim_{\\Delta x \\rightarrow 0}0 = 0 \\\\\nf_y(0,0) = \\lim_{\\Delta y \\rightarrow 0} \\frac{f(0,0+\\Delta y) - f(0,0)}{\\Delta y} = \\lim_{\\Delta y \\rightarrow 0}0 = 0 \\\\\n\\end{split}\n$$\n对一元函数，可导则一定连续（连续不一定可导）。在这里可以发现，**对多元函数，可偏导无法推出连续**。", "_____no_output_____" ], [ "为了讨论全微分的情况，修改函数为\n$$f(x,y) = \\left\\{\n\\begin{aligned}\n& \\frac{xy}{\\sqrt{x^2+y^2}}, & x^2+y^2 \\neq 0 \\\\\n& 0, & x^2+y^2 = 0\n\\end{aligned}\n\\right.\n$$\n\n其曲面图像如下所示：", "_____no_output_____" ] ], [ [ "@np.vectorize\ndef f(x, y):\n return x * y / np.sqrt(x ** 2 + y ** 2)\n\nstep = 0.05\nx_min, x_max = -1, 1\ny_min, y_max = -1, 1\nx_range, y_range = np.arange(x_min, x_max + step, step), np.arange(y_min, y_max + step, step)\nx_mat, y_mat = np.meshgrid(x_range, y_range)\nz = f(x_mat.reshape(-1), y_mat.reshape(-1)).reshape(x_mat.shape)\nfig = plt.figure(figsize=(12, 6))\nax1 = fig.add_subplot(1, 2, 1, projection='3d', elev=50, azim=-50)\nax1.plot_surface(x_mat, y_mat, z, cmap=cm.jet, rstride=1, cstride=1, edgecolor='none',alpha=.8)\nax1.set_xlabel('$x$')\nax1.set_ylabel('$y$')\nax1.set_zlabel('$z$')\nplt.show()", "_____no_output_____" ] ], [ [ "**微分**\n\n易证上述函数在点 $(0,0)$ 处极限存在且等于函数值 $0$，从而在该点连续。并且该点两个偏导也都等于 $0$。\n$$ \\Delta z - [f_x(0,0) + f_y(0,0)] = f(0+\\Delta x, 0+\\Delta y) - f(0,0) - [0+0] = \\frac{\\Delta x \\cdot \\Delta y}{\\sqrt{(\\Delta x)^+(\\Delta y)^2}} $$\n根据定义，如果在 $(0,0)$ 可微，则上式必须是 $\\rho=\\sqrt{(\\Delta x)^+(\\Delta y)^2}$ 的高阶无穷小。考虑点 $P'(\\Delta x, \\Delta y)$ 沿着直线 $y=kx$ 趋于 $(0,0)$\n$$ \\frac{\\frac{\\Delta x \\cdot \\Delta y}{\\sqrt{(\\Delta x)^2+(\\Delta y)^2}}}{\\rho} = \\frac{\\Delta x \\cdot \\Delta y}{(\\Delta x)^2+(\\Delta y)^2} = \\frac{\\Delta x \\cdot k \\Delta x}{(\\Delta x)^2+(k \\Delta x)^2} = \\frac{k}{1+k^2}$$\n它不能随 $\\rho \\rightarrow 0$ 而趋于 $0$，所以上式不是 $\\rho$ 的高阶无穷小，因此函数在点 $(0,0)$ 处的全微分并不存在，即函数在点 $(0,0)$ 处是不可微分的（尽管在该点连续，偏导也都存在）。", "_____no_output_____" ], [ "### 4. 多元复合函数的求导法则\n\n#### 4.1 一元函数与多元函数复合的情形\n\n**定理1 如果函数 $u=\\phi(t)$ 及 $v=\\psi(t)$ 都在点 $t$ 可导，函数 $z=f(u,v)$ 在对应点 $(u, v)$ 具有连续偏导数，则复合函数 $z=f[\\phi(t), \\psi(t)]$ 在点 $t$ 可导，且有\n$$ \\frac{dz}{dt}=\\frac{\\partial z}{\\partial u}\\frac{du}{dt}+\\frac{\\partial z}{\\partial v}\\frac{dv}{dt} $$\n**\n\n用同样的方法，可把定理推广到复合函数的中间变量多于两个的情形。上述公式中的导数 $\\frac{dz}{dt}$ 称为 **全导数**。\n\n#### 4.2 多元函数与多元函数复合的情形\n\n**定理2 如果函数 $u=\\phi(x,y)$ 及 $v=\\psi(x,y)$ 都在点 $(x,y)$ 具有对 $x$ 及对 $y$ 的偏导数，函数 $z=f(u,v)$ 在对应点 $(u,v)$ 具有连续偏导数，则复合函数 $z=f[\\phi(x,y),\\psi(x,y)]$ 在点 $(x,y)$ 的两个偏导数都存在，且有\n$$\n\\begin{split}\n\\frac{\\partial z}{\\partial x} = \\frac{\\partial z}{\\partial u}\\frac{\\partial u}{\\partial x} + \\frac{\\partial z}{\\partial v}\\frac{\\partial v}{\\partial x} \\\\\n\\frac{\\partial z}{\\partial y} = \\frac{\\partial z}{\\partial u}\\frac{\\partial u}{\\partial y} + \\frac{\\partial z}{\\partial v}\\frac{\\partial v}{\\partial y}\n\\end{split}\n$$\n**\n\n#### 4.3 其他情形\n\n**定理3 如果函数 $u=\\phi(x,y)$ 在点 $(x,y)$ 具有对 $x$ 及对 $y$ 的偏导数，函数 $v=\\psi(y)$ 在点 $y$ 可导，函数 $z=f(u,v)$ 在对应点 $(u,v)$ 具有连续偏导数，则复合函数 $z=f[\\phi(x,y), \\psi(y)]$ 在点 $(x,y)$ 的两个偏导数都存在，且有\n$$\n\\begin{split}\n\\frac{\\partial z}{\\partial x} &= \\frac{\\partial z}{\\partial u}\\frac{\\partial u}{\\partial x} \\\\\n\\frac{\\partial z}{\\partial y} &= \\frac{\\partial z}{\\partial u}\\frac{\\partial u}{\\partial y} + \\frac{\\partial z}{\\partial v}\\frac{dv}{dy}\n\\end{split}\n$$\n**\n\n上述情形实际上是情形 $2$ 的一种特例，即在情形 $2$ 中，如变量 $v$ 与 $x$ 无关，从而 $\\frac{\\partial v}{\\partial x}=0$；在 $v$ 对 $y$ 求导时，由于 $v=\\psi(y)$ 是一元函数，故 $\\frac{\\partial v}{\\partial y}$ 换成了 $\\frac{dv}{dy}$，这就得上述结果。\n\n**全微分形式不变性** 设函数 $z=f(u,v)$ 具有连续偏导数，则有全微分\n$$ dz = \\frac{\\partial z}{\\partial u}du + \\frac{\\partial z}{\\partial v}dv $$\n如果 $u,v$ 又是中间变量，即 $u=\\phi(x,y), v=\\psi(x,y)$，且这两个函数也具有连续偏导数，则复合函数\n$$ z = f[\\phi(x,y), \\psi(x,y)] $$\n的全微分为\n$$\n\\begin{split}\ndz &= \\frac{\\partial z}{\\partial x}dx + \\frac{\\partial z}{\\partial y}dy \\\\\n&= (\\frac{\\partial z}{\\partial u}\\frac{\\partial u}{\\partial x} + \\frac{\\partial z}{\\partial v}\\frac{\\partial v}{\\partial x})dx + (\\frac{\\partial z}{\\partial u}\\frac{\\partial u}{\\partial y} + \\frac{\\partial z}{\\partial v}\\frac{\\partial v}{\\partial y})dy \\\\\n&= \\frac{\\partial z}{\\partial u}(\\frac{\\partial u}{\\partial x}dx + \\frac{\\partial u}{\\partial y}dy) + \\frac{\\partial z}{\\partial v}(\\frac{\\partial v}{\\partial x}dx + \\frac{\\partial v}{\\partial y}dy) \\\\\n&= \\frac{\\partial z}{\\partial u}du + \\frac{\\partial z}{\\partial v}dv\n\\end{split}\n$$\n由此可见，无论 $u,v$ 是自变量还是中间变量，函数 $z=f(u,v)$ 的全微分形式是一样的。这个性质叫做**全微分形式不变性**。", "_____no_output_____" ], [ "### 5. 隐函数的求导公式\n\n#### 5.1 一个方程的情形\n\n**隐函数存在定理1 设函数 $F(x,y)$ 在点 $P(x_0, y_0)$ 的某一邻域内具有连续偏导数，且 $F(x_0, y_0)=0, F_y(x_0, y_0) \\neq 0$，则方程 $F(x,y)=0$ 在点 $(x_0, y_0)$ 的某一邻域内恒能唯一确定一个连续且有连续导数的函数 $y=f(x)$，它满足条件 $y_0=f(x_0)$，并有\n$$ \\frac{dy}{dx} = -\\frac{F_x}{F_y} $$\n**\n\n**隐函数存在定理2 设函数 $F(x,y,z)$ 在点 $P(x_0, y_0, z_0)$ 的某一邻域内具有连续偏导数，且 $F(x_0, y_0, z_0)=0, F_z(x_0, y_0) \\neq 0$，则方程 $F(x,y,z)=0$ 在点 $(x_0, y_0, z_0)$ 的某一邻域内恒能唯一确定一个连续且有连续导数的函数 $z=f(x,y)$，它满足条件 $z_0=f(x_0, y_0)$，并有\n$$ \\frac{\\partial z}{\\partial x} = -\\frac{F_x}{F_z}, \\frac{\\partial z}{\\partial y} = -\\frac{F_y}{F_z} $$\n**\n\n上述公式就是隐函数的求导公式。\n\n#### 5.2 方程组的情形\n\n略。", "_____no_output_____" ], [ "### 6. 多元函数微分学的几何应用\n\n略", "_____no_output_____" ], [ "### 7. 方向导数与梯度\n\n#### 7.1 方向导数\n\n偏导数反映的是函数沿坐标轴方向的变化率，但有时我们需要考虑函数沿任一指定方向的变化率问题。\n\n设 $l$ 是 $xOy$ 平面上以 $P_0(x_0,y_0)$ 为始点的一条射线，$e_l=(cos \\alpha, cos \\beta)$ 是与 $l$ 同方向的单位向量。射线 $l$ 的参数方程为\n$$\n\\begin{split}\nx = x_0 + tcos\\alpha \\\\\ny = y_0 + tcos\\beta \\\\\n(t \\geq 0)\n\\end{split}\n$$\n\n设函数 $z=f(x,y)$ 在点 $P_0(x_0,y_0)$ 为某个邻域 $U(P_0)$ 内有定义，$P(x_0+tcos\\alpha, y_0+tcos\\beta)$ 为 $l$ 上的另一点，且 $P \\in U(P_0)$。如果函数增量 $f(x_0+tcos\\alpha, y_0+cos\\beta) - f(x_0, y_0)$ 与 $P$ 到 $P_0$ 的距离 $|PP_0|=t$ 的比值\n$$ \\frac{f(x_0+tcos\\alpha, y_0+cos\\beta) - f(x_0, y_0)}{t} $$\n当 $P$ 沿着 $l$ 趋于 $P_0$（即 $t \\rightarrow 0^+$）时的极限存在，则称此极限为函数 $f(x,y)$ 在点 $P_0$ 沿方向 $l$ 的**方向导数**，记作 $\\frac{\\partial f}{\\partial l}|_{(x_0,y_0)}$，即\n$$ \\frac{\\partial f}{\\partial l}|_{(x_0,y_0)} = \\lim_{t \\rightarrow 0^+}\\frac{f(x_0+tcos\\alpha, y_0+cos\\beta) - f(x_0, y_0)}{t} $$\n\n从方向导数的定义可知，方向导数 $\\frac{\\partial f}{\\partial l}|_{(x_0,y_0)}$ 就是函数 $f(x,y)$ 在点 $P_0(x_0,y_0)$ 处沿方向 $l$ 的变化率。若函数 $f(x,y)$ 在点 $P_0(x_0,y_0)$ 的偏导数存在，$e_l = i = (1,0)$，则\n$$ \\frac{\\partial f}{\\partial l}|_{(x_0,y_0)} = \\lim_{t \\rightarrow 0^+}\\frac{f(x_0+tcos\\alpha, y_0+cos\\beta) - f(x_0, y_0)}{t} = f_x(x,y) $$\n又若 $e_l = j = (0,1)$，则\n$$ \\frac{\\partial f}{\\partial l}|_{(x_0,y_0)} = \\lim_{t \\rightarrow 0^+}\\frac{f(x_0+tcos\\alpha, y_0+cos\\beta) - f(x_0, y_0)}{t} = f_y(x,y) $$\n但反之，若 $e_l = i$，$\\frac{\\partial z}{\\partial l}|_{(x_0,y_0)}$ 存在，则 $\\frac{\\partial z}{\\partial x}|_{(x_0,y_0)}$ 未必存在。例如，$z=\\sqrt{x^2+y^2}$ 在点 $O(0,0)$ 处沿 $l=i$ 方向的方向导数 $\\frac{\\partial z}{\\partial l}|_{(0,0)}=1$，而偏导数 $\\frac{\\partial z}{\\partial x}|_{(0,0)}$ 不存在（对比极限的条件，$t \\rightarrow 0^+$ 和 $t \\rightarrow 0$）。\n\n**定理 如果函数 $f(x,y)$ 在点 $P_0(x_0,y_0)$ 可微分，那么函数在该点沿任一方向 $l$ 的方向导数存在，且有\n$$ \\frac{\\partial f}{\\partial l}|_{x_0,y_0}=f_x(x_0,y_0)cos\\alpha + f_y(x_0,y_0)cos\\beta $$\n其中 $cos\\alpha,cos\\beta$ 是方向 $l$ 的方向余弦。\n**", "_____no_output_____" ], [ "#### 7.2 梯度\n\n与方向导数有关联的一个概念是函数的梯度。在二元函数的情形，设函数 $f(x,y)$ 在平面区域 $D$ 内具有一阶连续偏导数，则对于每一点 $P_0(x_0,y_0) \\in D$，都可定出一个向量\n$$ f_x(x_0,y_0)i + f_y(x_0,y_0)j $$\n这向量称为函数 $f(x,y)$ 在点 $P_0(x_0,y_0)$ 的**梯度**，记作 $grad\\,f(x_0,y_0)$ 或 $\\nabla f(x_0,y_0)$，即\n$$ grad\\,f(x_0,y_0) = \\nabla f(x_0,y_0) = f_x(x_0,y_0)i + f_y(x_0,y_0)j $$\n其中 $\\nabla = \\frac{\\partial}{\\partial x}i + \\frac{\\partial}{\\partial y}j$ 称为（二维的）**向量微分算子**或**Nabla算子**，$\\nabla f = \\frac{\\partial f}{\\partial x}i + \\frac{\\partial f}{\\partial y}j$。\n\n如果函数 $f(x,y)$ 在点 $P_0(x_0,y_0)$ 可微分，$e_l=(cos\\alpha,cos\\beta)$ 是与方向 $l$ 同向的单位向量，则\n$$\n\\begin{split}\n\\frac{\\partial f}{\\partial l}|_{(x_0,y_0)} &= f_x(x_0,y_0)cos\\alpha + f_y(x_0,y_0)cos\\beta \\\\\n&= grad \\, f(x_0,y_0) \\cdot e_l \\\\\n&= |grad \\, f(x_0,y_0)|cos \\theta\n\\end{split}\n$$\n其中 $\\theta$ 是梯度和方向 $e_l$ 的夹角。\n\n这一关系式表明了函数在一点的梯度与函数在这点的方向导数间的关系。特别，由这关系可知：\n\n(1) 当 $\\theta=0$，即方向 $e_l$ 与梯度 $grad \\, f(x_0,y_0)$ 的方向相同时，函数 $f(x,y)$ 增加最快。此时，函数在这个方向的方向导数达到最大值，这个最大值就是梯度 $grad \\, f(x_0,y_0)$ 的模，即\n$$ \\frac{\\partial f}{\\partial l}|_{(x_0,y_0)} = |grad \\, f(x_0,y_0)| $$\n这个结果也表示：函数 $f(x,y)$ 在一点的梯度 $grad \\, f$ 是这样一个向量，它的方向是函数在这点的方向导数取得最大值的方向，它的模就等于方向导数的最大值。\n\n(2) 当 $\\theta=\\pi$，即方向 $e_l$ 与梯度 $grad \\, f(x_0,y_0)$ 的方向相反时，函数 $f(x,y)$ 减少最快。函数在这个方向的方向导数达到最小值，即\n$$ \\frac{\\partial f}{\\partial l}|_{(x_0,y_0)} = -|grad \\, f(x_0,y_0)| $$\n\n(3) 当 $\\theta=\\frac{\\pi}{2}$，即方向 $e_l$ 与梯度 $grad \\, f(x_0,y_0)$ 的方向正交时，函数的变化率为零，即\n$$ \\frac{\\partial f}{\\partial l}|_{(x_0,y_0)} = |grad \\, f(x_0,y_0)|cos\\theta = 0 $$\n\n我们知道，一般说来二元函数 $z=f(x,y)$ 在几何上表示一个曲面，这曲面被平面 $z=c(c是常数)$ 所截得的曲线 $L$ 的方程为\n$$\\left\\{\n\\begin{aligned}\n& z=f(x,y) \\\\\n& z=c\n\\end{aligned}\n\\right.\n$$\n这条曲线 $L$ 在 $xOy$ 面上的投影是一条平面曲线 $L^*$，它在 $xOy$ 平面直角坐标系中的方程为\n$$ f(x,y) = c $$\n对于曲线 $L^*$ 上的一切点，已给函数的函数值都是 $c$，所以我们称平面曲线 $L^*$ 为函数 $z=f(x,y)$ 的**等值线**。\n\n若 $f_x,f_y$ 不同时为零，则等值线 $f(x,y)=c$ 上任一点 $P_0(x_0,y_0)$ 处的一个单位法向量为\n$$\n\\begin{split}\nn &= \\frac{1}{\\sqrt{f^2_x(x_0,y_0)+f^2_y(x_0,y_0)}}(f_x(x_0,y_0),f_y(x_0,y_0)) \\\\\n&= \\frac{\\nabla f(x_0,y_0)}{|\\nabla f(x_0,y_0)|}\n\\end{split}\n$$\n这表明函数 $f(x,y)$ 在这一点 $(x_0,y_0)$ 的梯度 $\\nabla f(x_0,y_0)$ 的方向就是等值线 $f(x,y)=c$ 在这点的法线方向 $n$，而梯度的模 $|\\nabla f(x_0,y_0)|$ 就是沿这个法线方向的方向导数 $\\frac{\\partial f}{\\partial n}$，于是有\n$$ \\nabla f(x_0,y_0) = \\frac{\\partial f}{\\partial n}n $$", "_____no_output_____" ], [ "### 8. 多元函数的极值及其求法\n\n#### 8.1 多元函数的极值及最大值、最小值\n\n**定义 设函数 $z=f(x,y)$ 的定义域为 $D$，$P_0(x_0,y_0)$ 为 $D$ 的内点。若存在 $P_0$ 的某个邻域 $U(P_0) \\subset D$，使得对于该邻域内异于 $P_0$ 的任何点 $(x,y)$，都有\n$$ f(x,y) < f(x_0,y_0) $$\n则称函数 $f(x,y)$ 在点 $(x_0,y_0)$ 有极大值 $f(x_0,y_0)$，点 $(x_0,y_0)$ 称为函数 $f(x,y)$ 的极大值点；若对于该邻域内异于 $P_0$ 的任何点 $(x,y)$，都有\n$$ f(x,y) > f(x_0,y_0) $$\n则称函数 $f(x,y)$ 在点 $(x_0,y_0)$ 有极小值 $f(x_0,y_0)$，点 $(x_0,y_0)$ 称为函数 $f(x,y)$ 的极小值点。极大值、极小值统称为极值。使得函数取得极值的点称为极值点。\n**\n\n**定理1（必要条件） 设函数 $z=f(x,y)$ 在点 $(x_0,y_0)$ 具有偏导数，且在点 $(x_0,y_0)$ 处有极值，则有\n$$ f_x(x_0,y_0)=0, f_y(x_0,y_0)=0 $$\n**\n\n仿照一元函数，凡是能使 $ f_x(x_0,y_0)=0, f_y(x_0,y_0)=0 $ 同时成立的点 $(x_0,y_0)$ 称为函数 $ z=f(x,y) $ 的**驻点**。从定理1可知，具有偏导数的函数的极值点必定是驻点。但函数的驻点不一定是极值点。例如，点 $(0,0)$ 是函数 $z=xy$ 的驻点，但函数在该点并无极值。\n\n**定理2（充分条件） 设函数 $z=f(x,y)$ 在点 $(x_0,y_0)$ 的某邻域内连续且有一阶及二阶连续偏导数，又 $f_x(x_0,y_0),f_y(x_0,y_0)$，令\n$$ f_{xx}(x_0,y_0)=A, f_{xy}(x_0,y_0)=B, f_{yy}(x_0,y_0)=c $$\n则 $f(x,y)$ 在 $(x_0,y_0)$ 处是否取得极值的条件如下**：\n\n(1) $AC-B^2>0$ 时具有极值，且当 $A<0$ 时有极大值，当 $A>0$ 时有极小值；\n\n(2) $AC-B^2<0$ 时没有极值；\n\n(3) $AC-B^2=0$ 时可能有极值，也可能没有极值，还需另作讨论；", "_____no_output_____" ], [ "#### 8.2 条件极值 拉格朗日乘数法\n\n上面所讨论的极值问题，对于函数的自变量，除了限制在函数的定义域内以外，并无其他条件，所以有时候称为**无条件极值**。但在实际问题中，有时会遇到对函数的自变量还有附加条件的极值问题，这种极值称为**条件极值**。对于有些实际问题，可以把条件极值化为无条件极值求解；但在很多情形下，将条件极值化为无条件极值并不这样简单。另有一种直接寻求条件极值的方法，可以不必先把问题化到无条件极值的问题。这就是下面要介绍的**拉格朗日乘数法**。\n\n现在先来寻求函数\n$$ z=f(x,y) $$\n在条件\n$$ \\varphi(x,y)=0 $$\n下取得极值的必要条件。\n\n如果函数在 $(x_0,y_0)$ 取得所求的极值，那么首先有\n$$ \\varphi(x_0,y_0)=0 $$\n我们假定在 $(x_0,y_0)$ 的某一邻域内 $f(x,y)$ 与 $\\varphi(x,y)$ 均有连续的一阶偏导数，且 $\\varphi(x_0,y_0) \\neq 0$。由隐函数存在定理可知，约束条件确定一个连续且有连续导数的函数 $y=\\psi(x)$，将其带入原函数，结果得到一个变量 $x$ 的函数\n$$ z=f[x,\\psi(x)] $$\n于是原函数在 $(x_0,y_0)$ 取得所求的极值，也就是相当于上述函数在 $x=x_0$ 取得极值。由一元可导函数取得极值的必要条件知道\n$$ \\frac{dz}{dx}|_{x=x_0}=f_x(x_0,y_0)+f_y(x_0,y_0)\\frac{dy}{dx}|_{x=x_0}=0 $$\n而对约束条件用隐函数求导公式，有\n$$ \\frac{dy}{dx}|_{x=x_0} = -\\frac{\\varphi_x(x_0,y_0)}{\\varphi_y(x_0,y_0)} $$\n结合上面两式，可得\n$$ f_x(x_0,y_0) - f_y(x_0,y_0)\\frac{\\varphi_x(x_0,y_0)}{\\varphi_y(x_0,y_0)} = 0 $$\n\n设 $ \\frac{f_y(x_0,y_0)}{\\varphi_y(x_0,y_0)} = \\lambda$，上述必要条件就变为\n$$\\left\\{\n\\begin{aligned}\n& f_x(x_0,y_0) + \\lambda \\varphi_x(x_0,y_0) = 0 \\\\\n& f_y(x_0,y_0) + \\lambda \\varphi_y(x_0,y_0) = 0 \\\\\n& \\varphi(x_0,y_0) = 0\n\\end{aligned}\n\\right.\n$$\n\n若引进辅助函数\n$$ L(x,y) = f(x,y)+\\lambda \\varphi(x,y) $$\n则不难看出，方程组的前两式就是\n$$ L_x(x_0,y_0)=0, L_y(x_0,y_0)=0 $$\n函数 $L(x,y)$ 称为**拉格朗日函数**，参数 $\\lambda$ 称为**拉格朗日乘子**。\n\n由以上讨论，我们得到一下结论。\n\n**拉格朗日乘数法** 要找函数 $z=f(x,y)$ 在附加条件 $\\varphi(x,y)=0$ 下的可能极值点，可以先作拉格朗日函数\n$$ L(x,y) = f(x,y) + \\lambda \\varphi(x,y) $$\n其中 $\\lambda$ 为参数，求其对 $x$ 与 $y$ 的一阶偏导数，并使之为零，然后与约束条件联立起来\n$$\\left\\{\n\\begin{aligned}\n& f_x(x,y) + \\lambda \\varphi_x(x,y) = 0 \\\\\n& f_y(x,y) + \\lambda \\varphi_y(x,y) = 0 \\\\\n& \\varphi(x,y) = 0\n\\end{aligned}\n\\right.\n$$\n由这方程组解出 $x,y,\\lambda$，这样得到的 $(x,y)$ 就是函数 $f(x,y)$ 在附加条件 $\\varphi(x,y)=0$ 下的可能极值点。\n\n这方法还可以推广到自变量多于两个而条件多于一个的情形。例如，要求函数\n$$ u=f(x,y,z,t) $$\n在附加条件\n$$ \\varphi(x,y,z,t)=0, \\psi(x,y,z,t)=0 $$\n下的极值，可以先作拉格朗日函数\n$$ L(x,y,z,t) = f(x,y,z,t) + \\lambda \\varphi(x,y,z,t) + \\mu \\psi(x,y,z,t) $$\n其中 $\\lambda, \\mu$ 均为参数，求其一阶偏导数，并使之为零，然后与两个附加条件联立起来求解，这样得出的 $(x,y,z,t)$ 就是函数 $f(x,y,z,t)$ 在附加条件下的可能极值点。\n\n至于如何确定所求得的点是否极值点，在实际问题中往往可根据问题本身的性质来判定。", "_____no_output_____" ], [ "### 9. 二元函数的泰勒公式\n\n略。", "_____no_output_____" ], [ "### 10. 最小二乘法\n\n略。", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "## Data Import and Check", "_____no_output_____" ] ], [ [ "import pandas as pd\nimport numpy as np \nimport matplotlib.pyplot as plt\nimport seaborn as sns; sns.set()\nfrom sklearn.metrics import roc_auc_score\nfrom sklearn.metrics import roc_curve\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.metrics import classification_report\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import accuracy_score\nfrom scipy import stats\nimport statsmodels.api as sm\nfrom scipy.stats import mannwhitneyu\nimport matplotlib.gridspec as gridspec", "_____no_output_____" ] ], [ [ "* I import data and drop duplicates\n* I had tried to set the user id as index. Expectedly, it did not work as a user can have multiple trips. However the user - trip combination did not work either which revealed the entire rows duplicated\n* Once the duplicates are removed, the count of use -trip combinations reveal they constitute a unique key", "_____no_output_____" ] ], [ [ "hoppi = pd.read_csv('C:/Users/gurkaali/Documents/Info/Ben/Hop/WatchesTable.csv', sep=\",\")\nhoppi.drop_duplicates(inplace = True)\nhoppi.groupby(['user_id', 'trip_id'])['user_id']\\\n .count() \\\n .reset_index(name='count')\\\n .sort_values(['count'], ascending = False)\\\n .head(5)", "_____no_output_____" ] ], [ [ "Now that I am sure, I can set the index:", "_____no_output_____" ] ], [ [ "hoppi.set_index(['user_id', 'trip_id'], inplace = True)", "_____no_output_____" ] ], [ [ "Pandas has great features for date calculations. I set the related field types as datetime in case I need those features", "_____no_output_____" ] ], [ [ "hoppi['departure_date'] = pd.to_datetime(hoppi['departure_date'], format = '%m/%d/%y')\nhoppi['return_date'] = pd.to_datetime(hoppi['return_date'], format = '%m/%d/%y')\nhoppi['first_search_dt'] = pd.to_datetime(hoppi['first_search_dt'], format = '%m/%d/%y %H:%M')\nhoppi['watch_added_dt'] = pd.to_datetime(hoppi['watch_added_dt'], format = '%m/%d/%y %H:%M')\nhoppi['latest_status_change_dt'] = pd.to_datetime(hoppi['latest_status_change_dt'], format = '%m/%d/%y %H:%M')\nhoppi['first_buy_dt'] = pd.to_datetime(hoppi['first_buy_dt'], format = '%m/%d/%y %H:%M')\nhoppi['last_notif_dt'] = pd.to_datetime(hoppi['last_notif_dt'], format = '%m/%d/%y %H:%M')\nhoppi['forecast_last_warning_date'] = pd.to_datetime(hoppi['forecast_last_warning_date'], format = '%m/%d/%y')\nhoppi['forecast_last_danger_date'] = pd.to_datetime(hoppi['forecast_last_danger_date'], format = '%m/%d/%y')", "_____no_output_____" ] ], [ [ "The explanations in the assignment do not cover all fields but field names and the content enable further data verification\n* Stay should be the difference between departure and return dates. Based on that assumption, the query below should return no records i.e. the 1st item in the tuple returned by shape should be 0:", "_____no_output_____" ] ], [ [ "hoppi['stay2'] = pd.to_timedelta(hoppi['stay'], unit = 'D')\nhoppi['stay_check'] = hoppi['return_date'] - hoppi['departure_date']\nhoppi.loc[(hoppi['stay_check'] != hoppi['stay2']) & (hoppi['return_date'].isnull() == False), \\\n ['stay2', 'stay_check', 'return_date', 'departure_date']].shape", "_____no_output_____" ] ], [ [ "The following date fields must not be before the first search date. Therefore the queries below should reveal no records\n* watch_added_dt\n* latest_status_change_dt\n* first_buy_dt\n* last_notif_dt\n* forecast_last_warning_date\n* forecast_last_danger_date", "_____no_output_____" ] ], [ [ "hoppi.loc[(hoppi['watch_added_dt'] < hoppi['first_search_dt']), ['first_search_dt', 'watch_added_dt']].shape", "_____no_output_____" ], [ "hoppi.loc[(hoppi['latest_status_change_dt'] < hoppi['first_search_dt']), ['first_search_dt', 'latest_status_change_dt']].shape", "_____no_output_____" ] ], [ [ "33 records have a first buy suggestion datetime earlier than the user's first search.", "_____no_output_____" ] ], [ [ "hoppi.loc[(hoppi['first_buy_dt'] < hoppi['first_search_dt']), ['first_search_dt', 'first_buy_dt']].shape", "_____no_output_____" ] ], [ [ "While the difference is just minutes in most cases, I don't have an explanation to justify it. Given the limited number of cases, I prefer removing them", "_____no_output_____" ] ], [ [ "hoppi.loc[(hoppi['first_buy_dt'] < hoppi['first_search_dt']), ['first_search_dt', 'first_buy_dt']].head()", "_____no_output_____" ], [ "hoppi = hoppi.loc[~(hoppi['first_buy_dt'] < hoppi['first_search_dt'])]", "_____no_output_____" ] ], [ [ "There are also 2 records where the last notification is done before the user's first search. I remove those as well", "_____no_output_____" ] ], [ [ "hoppi.loc[(hoppi['last_notif_dt'] < hoppi['first_search_dt']), ['first_search_dt', 'last_notif_dt']]", "_____no_output_____" ], [ "hoppi = hoppi.loc[~(hoppi['last_notif_dt'] < hoppi['first_search_dt'])]", "_____no_output_____" ] ], [ [ "Same checks on last warning and last danger dates show 362K + and 98K + suspicious records. As the quantitiy is large and descriptions sent with the assignment do not contain details on these 2 fields, I prefer to keep them while taking a note here in case something provides with additional argument to delete them during analyses.", "_____no_output_____" ] ], [ [ "hoppi.loc[(hoppi['forecast_last_warning_date'] < hoppi['first_search_dt']), \\\n ['first_search_dt', 'forecast_last_warning_date']].shape", "_____no_output_____" ], [ "hoppi.loc[(hoppi['forecast_last_danger_date'] < hoppi['first_search_dt']), \\\n ['first_search_dt', 'forecast_last_danger_date']].shape", "_____no_output_____" ] ], [ [ "### Check outliers", "_____no_output_____" ], [ "I reshape the columns in a way that will make working with seaborn easier:", "_____no_output_____" ] ], [ [ "hoppi_box_components = [hoppi[['first_advance']].assign(measurement_type = 'first_advance').reset_index(). \\\n rename(columns = {'first_advance': 'measurement'}),\n hoppi[['watch_advance']].assign(measurement_type = 'watch_advance').reset_index(). \\\n rename(columns = {'watch_advance': 'measurement'}),\n hoppi[['current_advance']].assign(measurement_type = 'current_advance').reset_index(). \\\n rename(columns = {'current_advance': 'measurement'})]\nhoppi_box = pd.concat(hoppi_box_components)", "_____no_output_____" ], [ "sns.set(font = 'DejaVu Sans', style = 'white')\nax = sns.boxplot(x=\"measurement_type\", y=\"measurement\",\n data=hoppi_box, palette=[\"#FA6866\", \"#01AAE4\", \"#505050\"], #Hopper colors\n linewidth = 0.5)", "_____no_output_____" ] ], [ [ "While several observations look like outliers on the boxplots, the histograms below show that the data is highly skewed. Therefore I do not consider them as outliers", "_____no_output_____" ] ], [ [ "f, axes = plt.subplots(1, 3, figsize=(15, 5), sharex=True)\nsns.distplot(hoppi['first_advance'], kde=False, color=\"#FA6866\", ax=axes[0])\nsns.distplot(hoppi.loc[hoppi['watch_advance'].isnull() == False, 'watch_advance'], kde=False, color=\"#01AAE4\", ax=axes[1])\nsns.distplot(hoppi.loc[hoppi['current_advance'].isnull() == False, 'current_advance'], kde=False, color=\"#505050\", ax=axes[2])", "_____no_output_____" ] ], [ [ "### Question 1", "_____no_output_____" ], [ "Given the business model of Hopper, we should understand who is more likely to buy a ticket eventually. Logistic Regression constitutes a convenient way of conducting such analysis. It runs faster than SVN and is easier to interpret, making it ideal for a task like this one:", "_____no_output_____" ], [ "I prepare categorical variables for trip types:", "_____no_output_____" ] ], [ [ "one_hot_trip_type = pd.get_dummies(hoppi['trip_type'])\nhoppi2 = hoppi.join(one_hot_trip_type)", "_____no_output_____" ] ], [ [ "I believe the city / airport distinction in origin and destination fields refer to the fact that some airports are more central such as the difference between Toronto Billy Bishop and Pearson airports. I also checked some airport codes, they do corresponds to cities where there are multiple airports with one or more being city airports", "_____no_output_____" ] ], [ [ "origin_cols = hoppi2['origin'].str.split(\"/\", n = 1, expand = True) \nhoppi2['origin_code'] = origin_cols[1]\nhoppi2['origin_type'] = origin_cols[0]\n\ndestination_cols = hoppi2['destination'].str.split(\"/\", n = 1, expand = True) \nhoppi2['destination_code'] = destination_cols[1]\nhoppi2['destination_type'] = destination_cols[0]", "_____no_output_____" ], [ "one_hot_destination_type = pd.get_dummies(hoppi2['destination_type'])\nhoppi3 = hoppi2.join(one_hot_destination_type)\nhoppi3.rename(columns={\"airport\": \"destination_airport\", \"city\": \"destination_city\"}, inplace = True)", "_____no_output_____" ], [ "one_hot_origin_type = pd.get_dummies(hoppi3['origin_type'])\nhoppi4 = hoppi3.join(one_hot_origin_type)\nhoppi4.rename(columns={\"airport\": \"origin_airport\", \"city\": \"origin_city\"}, inplace = True)", "_____no_output_____" ] ], [ [ "I prepare categorical variables for whether a watch is placed or not:", "_____no_output_____" ] ], [ [ "hoppi4.loc[hoppi3['watch_added_dt'].isnull() == True, 'watch_bin'] = 0\nhoppi4.loc[hoppi3['watch_added_dt'].isnull() == False, 'watch_bin'] = 1", "_____no_output_____" ] ], [ [ "Given the user - trip combination being unique across the data file, we do not have information on the changes for a user who has updated his trip status. As the data looks like covering the last status of a trip, I prefer to focus analyses on concluded queries i.e. trips either expired or booked. I exclude:\n* actives: because their result is yet to be seen. The user can end up booking before departure\n* shopped: because a user can make several searches on the same itinerary with alternative options each ending up as a new record in the database. I consider a search once the suer starts following the trip price\n* inactive: because some have departure in the future so their result cannot be concluded. I also exclude those with departure in the past as it falls in the same category as the shopped trips as the user stopped following the trip.\nI assign a new column for records I take into account in my analyses further below:", "_____no_output_____" ] ], [ [ "hoppi4.loc[hoppi3['status_latest'] == 'expired', 'result'] = 0\nhoppi4.loc[hoppi3['status_latest'] == 'booked', 'result'] = 1", "_____no_output_____" ] ], [ [ "A person might be prompted to buy once the price falls because it makes sense or maybe he buys as soon as it starts increasing to avoid further increase. Whatever the case, it makes sense to compare the price at different time points with respect to the original price at first search. For that, I create columns to measure price difference between the last price, the first time a buy recommended, the lowest price vs the the very first price:", "_____no_output_____" ] ], [ [ "hoppi4['dif_last_first'] = hoppi4['last_total'] - hoppi4['first_total']\nhoppi4['dif_buy_first'] = hoppi4['first_buy_total'] - hoppi4['first_total']\nhoppi4['dif_lowest_first'] = hoppi4['lowest_total'] - hoppi4['first_total']", "_____no_output_____" ] ], [ [ "I create a categorical variable for the last recommendation as well to check whether a buy recommendation makes user to book:", "_____no_output_____" ] ], [ [ "one_hot_last_rec = pd.get_dummies(hoppi4['last_rec']) # this create s 2 columns: buy and wait\nhoppi5 = hoppi4.join(one_hot_last_rec)\nhoppi5.loc[hoppi5['last_rec'].isnull(), 'buy'] = np.nan # originally null values are given 0. I undo that manipulation here", "_____no_output_____" ] ], [ [ "I make a table with rows containing certain results that I want to focus on i.e. expired and booked", "_____no_output_____" ] ], [ [ "hoppi6 = hoppi5.loc[hoppi5['result'].isnull() == False, \n ['round_trip', \n 'destination_city', 'origin_city',\n 'weekend',\n 'filter_no_lcc', 'filter_non_stop', 'filter_short_layover', 'status_updates', \n 'watch_bin', 'total_notifs', 'total_buy_notifs', 'buy',\n 'dif_last_first', 'dif_buy_first', 'dif_lowest_first', 'first_advance', 'result']]\nhoppi6.info()", "<class 'pandas.core.frame.DataFrame'>\nMultiIndex: 45237 entries, (e42e7c15cde08c19905ee12200fad7cb5af36d1fe3a3310b5f94f95c47ae51cd, 05d59806e67fa9a5b2747bc1b24842189bba0c45e49d3714549fc5df9838ed20) to (d414b1c72a16512dbd7b3859c9c9f574633578acef74d120490625d9010103c7, 3a363a2456b6b7605347e06d2879162b3008004370f73a68f52523330ccd38a6)\nData columns (total 17 columns):\nround_trip 45237 non-null uint8\ndestination_city 45237 non-null uint8\norigin_city 45237 non-null uint8\nweekend 45237 non-null int64\nfilter_no_lcc 45237 non-null int64\nfilter_non_stop 45237 non-null int64\nfilter_short_layover 45237 non-null int64\nstatus_updates 45237 non-null int64\nwatch_bin 45237 non-null float64\ntotal_notifs 44800 non-null float64\ntotal_buy_notifs 44800 non-null float64\nbuy 44800 non-null float64\ndif_last_first 44800 non-null float64\ndif_buy_first 44133 non-null float64\ndif_lowest_first 44800 non-null float64\nfirst_advance 45237 non-null int64\nresult 45237 non-null float64\ndtypes: float64(8), int64(6), uint8(3)\nmemory usage: 12.8+ MB\n" ] ], [ [ "Some rows have null values such as the price difference between the buy moment and the first price as some users may not have gt the buy recommendation yet. To cover these features, I get only non-null rows:", "_____no_output_____" ] ], [ [ "df = hoppi6.dropna()\ndf.info()", "<class 'pandas.core.frame.DataFrame'>\nMultiIndex: 44133 entries, (e42e7c15cde08c19905ee12200fad7cb5af36d1fe3a3310b5f94f95c47ae51cd, 05d59806e67fa9a5b2747bc1b24842189bba0c45e49d3714549fc5df9838ed20) to (d414b1c72a16512dbd7b3859c9c9f574633578acef74d120490625d9010103c7, 3a363a2456b6b7605347e06d2879162b3008004370f73a68f52523330ccd38a6)\nData columns (total 17 columns):\nround_trip 44133 non-null uint8\ndestination_city 44133 non-null uint8\norigin_city 44133 non-null uint8\nweekend 44133 non-null int64\nfilter_no_lcc 44133 non-null int64\nfilter_non_stop 44133 non-null int64\nfilter_short_layover 44133 non-null int64\nstatus_updates 44133 non-null int64\nwatch_bin 44133 non-null float64\ntotal_notifs 44133 non-null float64\ntotal_buy_notifs 44133 non-null float64\nbuy 44133 non-null float64\ndif_last_first 44133 non-null float64\ndif_buy_first 44133 non-null float64\ndif_lowest_first 44133 non-null float64\nfirst_advance 44133 non-null int64\nresult 44133 non-null float64\ndtypes: float64(8), int64(6), uint8(3)\nmemory usage: 12.6+ MB\n" ], [ "X = df[['round_trip', \n 'destination_city', 'origin_city',\n 'weekend',\n 'filter_non_stop', 'filter_short_layover', 'status_updates', 'filter_no_lcc', \n 'watch_bin', 'total_notifs', 'buy', 'total_buy_notifs', \n 'dif_lowest_first',\n 'dif_last_first', \n 'dif_buy_first', \n 'first_advance']] \ny = df['result']\nprint(X.shape, y.shape)", "(44133, 16) (44133,)\n" ], [ "X_train, X_test , y_train, y_test = train_test_split(X, y, test_size=0.8, random_state=1)\nlogit_model=sm.Logit(y_train, X_train)\nresult=logit_model.fit(maxiter = 1000)\nprint(result.summary2())", "Optimization terminated successfully.\n Current function value: 0.041569\n Iterations 12\n Results: Logit\n=====================================================================\nModel: Logit Pseudo R-squared: 0.852 \nDependent Variable: result AIC: 765.7809\nDate: 2019-06-20 12:20 BIC: 879.1482\nNo. Observations: 8826 Log-Likelihood: -366.89 \nDf Model: 15 LL-Null: -2484.6 \nDf Residuals: 8810 LLR p-value: 0.0000 \nConverged: 1.0000 Scale: 1.0000 \nNo. Iterations: 12.0000 \n---------------------------------------------------------------------\n Coef. Std.Err. z P>|z| [0.025 0.975]\n---------------------------------------------------------------------\nround_trip -0.7625 0.2318 -3.2898 0.0010 -1.2167 -0.3082\ndestination_city 0.4061 0.2259 1.7976 0.0722 -0.0367 0.8489\norigin_city 0.3157 0.2161 1.4605 0.1441 -0.1079 0.7393\nweekend 0.8184 0.2740 2.9873 0.0028 0.2815 1.3554\nfilter_non_stop 0.1179 0.2647 0.4455 0.6559 -0.4009 0.6368\nfilter_short_layover 0.7755 0.4312 1.7983 0.0721 -0.0697 1.6207\nstatus_updates 0.1079 0.0666 1.6196 0.1053 -0.0227 0.2386\nfilter_no_lcc -0.2027 0.8046 -0.2519 0.8011 -1.7796 1.3743\nwatch_bin -5.3110 0.4588 -11.5750 0.0000 -6.2103 -4.4117\ntotal_notifs -1.0820 0.1657 -6.5284 0.0000 -1.4069 -0.7572\nbuy 2.6103 0.4266 6.1188 0.0000 1.7742 3.4464\ntotal_buy_notifs -0.3924 0.1985 -1.9773 0.0480 -0.7814 -0.0034\ndif_lowest_first 0.0125 0.0035 3.6156 0.0003 0.0057 0.0194\ndif_last_first -0.0067 0.0014 -4.7947 0.0000 -0.0095 -0.0040\ndif_buy_first -0.0135 0.0042 -3.1968 0.0014 -0.0217 -0.0052\nfirst_advance 0.1581 0.0095 16.6355 0.0000 0.1395 0.1768\n=====================================================================\n\n" ], [ "lr = LogisticRegression()\nlr.fit(X_train, y_train)\ny_pred = lr.predict(X_test)", "C:\\ProgramData\\Anaconda3\\envs\\operational\\lib\\site-packages\\sklearn\\linear_model\\logistic.py:433: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n FutureWarning)\n" ], [ "accuracy_score(y_test, y_pred)", "_____no_output_____" ], [ "sum(y_train)/len(y_train)", "_____no_output_____" ], [ "print(classification_report(y_test, y_pred))", " precision recall f1-score support\n\n 0.0 0.99 1.00 0.99 32564\n 1.0 0.97 0.87 0.92 2743\n\n micro avg 0.99 0.99 0.99 35307\n macro avg 0.98 0.93 0.96 35307\nweighted avg 0.99 0.99 0.99 35307\n\n" ] ], [ [ "#### Data driven insights:\nThe model shows a good level of accuracy. However given the imbalance of data (only 8% of data corresponds to an actual booking) it is crucial to check recall which also shows a high value i.e. false negatives are limited.\nNow that we know the model looks robust, we can make the following data-driven insights:\n1. City travelers, regardless their origin and destination, are not necessarily more likely to end up booking. These are the people likely to be business travelers. When we look at the weekend travelers which I use as a substitute as pleasure travelers, people are significantly more likely to end up booking. It looks like people are more sensitive to buy recommendations when it is a personal travel\n2. Among filters, only those who filter for short layover are more likely to book whereas the significance is not as powerfull (at a p = 0.10 level)\n3. Buy recommendations significantly impact booking behavior which indicates that the algorithm makes sense to the customer\n4. Price fluctuations have significant impact on users' booking behavior:\n * The lowest and first price difference is significant with a positive relationship with the booking behavior showing that people are more likely to buy when there is a price drop after their first query\n * When the last price or the price of the buy recommendation are higher than the first price, users are less likely to book.\n * The above 2 points show that the algorithm leads to the expected user behavior.\n5. Those who sign up for a price watch are more likely to book. This feature might be an indicator that the user is seriously making a plan. To be concrete, someone who is looking at options for a dream vacation in case he wins the lottery would not be setting the watch on whereas someone who took days off at work next month would do so.", "_____no_output_____" ], [ "### Question 2", "_____no_output_____" ], [ "#### Most \"watched\" itineraries", "_____no_output_____" ], [ "I would like to see the watch selected cases w.r.t. the itinerary i.e. NY to MTL would be considered the same as MTL to NY", "_____no_output_____" ] ], [ [ "hoppi5.loc[(hoppi5['watch_bin'] == 1.0) & (hoppi5['result'] == 0)].info()", "<class 'pandas.core.frame.DataFrame'>\nMultiIndex: 40873 entries, (e42e7c15cde08c19905ee12200fad7cb5af36d1fe3a3310b5f94f95c47ae51cd, 05d59806e67fa9a5b2747bc1b24842189bba0c45e49d3714549fc5df9838ed20) to (d414b1c72a16512dbd7b3859c9c9f574633578acef74d120490625d9010103c7, 3a363a2456b6b7605347e06d2879162b3008004370f73a68f52523330ccd38a6)\nData columns (total 58 columns):\ntrip_type 40873 non-null object\norigin 40873 non-null object\ndestination 40873 non-null object\ndeparture_date 40873 non-null datetime64[ns]\nreturn_date 34064 non-null datetime64[ns]\nstay 34064 non-null float64\nweekend 40873 non-null int64\nfilter_no_lcc 40873 non-null int64\nfilter_non_stop 40873 non-null int64\nfilter_short_layover 40873 non-null int64\nfilter_name 40873 non-null object\nstatus_updates 40873 non-null int64\nfirst_search_dt 40873 non-null datetime64[ns]\nwatch_added_dt 40873 non-null datetime64[ns]\nlatest_status_change_dt 40873 non-null datetime64[ns]\nstatus_latest 40873 non-null object\ntotal_notifs 40713 non-null float64\ntotal_buy_notifs 40713 non-null float64\nfirst_rec 40713 non-null object\nfirst_total 40713 non-null float64\nlast_rec 40713 non-null object\nlast_total 40713 non-null float64\nfirst_buy_dt 40674 non-null datetime64[ns]\nfirst_buy_total 40674 non-null float64\nlowest_total 40713 non-null float64\nlast_notif_dt 38897 non-null datetime64[ns]\nforecast_first_target_price 40187 non-null float64\nforecast_first_good_price 40187 non-null float64\nforecast_last_target_price 40687 non-null float64\nforecast_last_good_price 40687 non-null float64\nforecast_last_warning_date 40450 non-null datetime64[ns]\nforecast_last_danger_date 37664 non-null datetime64[ns]\nforecast_min_target_price 40687 non-null float64\nforecast_max_target_price 40687 non-null float64\nforecast_min_good_price 40687 non-null float64\nforecast_max_good_price 40687 non-null float64\nwatch_advance 40873 non-null float64\nfirst_advance 40873 non-null int64\ncurrent_advance 0 non-null float64\nstay2 34064 non-null timedelta64[ns]\nstay_check 34064 non-null timedelta64[ns]\none_way 40873 non-null uint8\nround_trip 40873 non-null uint8\norigin_code 40873 non-null object\norigin_type 40873 non-null object\ndestination_code 40873 non-null object\ndestination_type 40873 non-null object\ndestination_airport 40873 non-null uint8\ndestination_city 40873 non-null uint8\norigin_airport 40873 non-null uint8\norigin_city 40873 non-null uint8\nwatch_bin 40873 non-null float64\nresult 40873 non-null float64\ndif_last_first 40713 non-null float64\ndif_buy_first 40674 non-null float64\ndif_lowest_first 40713 non-null float64\nbuy 40713 non-null float64\nwait 40873 non-null uint8\ndtypes: datetime64[ns](9), float64(23), int64(6), object(11), timedelta64[ns](2), uint8(7)\nmemory usage: 23.9+ MB\n" ], [ "pareto_watch_0 = hoppi5.loc[(hoppi5['watch_bin'] == 1.0) & (hoppi5['result'] == 0.0), ['origin_code', 'destination_code']]\npareto_watch_0.loc[pareto_watch_0['origin_code'] < pareto_watch_0['destination_code'], \\\n 'itinerary'] = \\\n pareto_watch_0['origin_code'] + pareto_watch_0['destination_code']\npareto_watch_0.loc[pareto_watch_0['origin_code'] > pareto_watch_0['destination_code'], \\\n 'itinerary'] = \\\n pareto_watch_0['destination_code'] + pareto_watch_0['origin_code']", "_____no_output_____" ], [ "pareto_watch_0.info()", "<class 'pandas.core.frame.DataFrame'>\nMultiIndex: 40873 entries, (e42e7c15cde08c19905ee12200fad7cb5af36d1fe3a3310b5f94f95c47ae51cd, 05d59806e67fa9a5b2747bc1b24842189bba0c45e49d3714549fc5df9838ed20) to (d414b1c72a16512dbd7b3859c9c9f574633578acef74d120490625d9010103c7, 3a363a2456b6b7605347e06d2879162b3008004370f73a68f52523330ccd38a6)\nData columns (total 3 columns):\norigin_code 40873 non-null object\ndestination_code 40873 non-null object\nitinerary 40873 non-null object\ndtypes: object(3)\nmemory usage: 39.9+ MB\n" ], [ "pareto_watch = pareto_watch_0 \\\n .groupby(['itinerary']) \\\n .size().reset_index() \\\n .rename(columns = {0: 'count'}) \\\n .sort_values(['count'], ascending = False)", "_____no_output_____" ], [ "pareto_watch.set_index('itinerary', inplace = True)\npareto_watch['cumulative_sum'] = pareto_watch['count'].cumsum()\npareto_watch['cumulative_perc'] = 100 * pareto_watch['cumulative_sum'] / pareto_watch['count'].sum()", "_____no_output_____" ], [ "pareto_watch.loc[pareto_watch['cumulative_perc'] <= 80].shape[0]\npareto_watch.shape[0]\nprint('All observations where the user watched the price but did not book, cover ',\n pareto_watch.shape[0],\n 'itineraries. Out of these, ',\n pareto_watch.loc[pareto_watch['cumulative_perc'] <= 80].shape[0],\n ' constitute 80% of the whole observation set. That is around ',\n round(100 * pareto_watch.loc[pareto_watch['cumulative_perc'] <= 80].shape[0] / pareto_watch.shape[0], 1),\n '% of the whole set.')", "All observations where the user watched the price but did not book, cover 11697 itineraries. Out of these, 4236 constitute 80% of the whole observation set. That is around 36.2 % of the whole set.\n" ] ], [ [ "The list gives the biggest airports. This result reassures that it is additionally critical to make reliable estimations for these itineraries. The top 10 itineraries consist only of US destinations showing the importance of the US market. As we have seen in the previous question, a user setting the watch on is a good estimator of an actual booking. Therefore accuracy of price estimations is extra important for the US market. This information could be handy for the data scientists developing algorithms e.g. they can give extra weight to the accuracy of US flights", "_____no_output_____" ], [ "#### Watch vs the Moment the First Search is Done", "_____no_output_____" ], [ "* Here I am looking whether there is significant difference between users with a watch and without in terms of the following two:\n * the first price found \n * the number of days left to departure as of first search", "_____no_output_____" ] ], [ [ "dfw = hoppi5.loc[hoppi5['result'].isnull() == False, ['first_advance', 'first_total', 'watch_bin', 'result']]\ndfw = dfw.dropna()", "_____no_output_____" ], [ "dfw.groupby('watch_bin').agg({'first_advance': np.mean, 'first_total': np.mean})", "_____no_output_____" ] ], [ [ "* As the data is skewed using non-parametrical tests makes more sense. I use the Mann Whitney test for that purpose\n* The test reveal significant difference between watched and non-watched itineraries at 0.1 level in terms of the number of days between the departure and the first search. Those who place a watch have a week less time left to their departure compared to the rest. Users may be using hopper as an assistant when they feel like they missed the time window where they could shop for different offers. For those users more frequent notifications can be planned", "_____no_output_____" ] ], [ [ "stat, p = mannwhitneyu(dfw.loc[dfw['watch_bin'] == 1, 'first_advance'], \n dfw.loc[dfw['watch_bin'] == 0, 'first_advance'])\nprint('Statistics=%.3f, p=%.3f' % (stat, p))", "Statistics=41274130.000, p=0.074\n" ] ], [ [ "* The test on same user groups (those watching vs those who don't) show that they differ in terms of the price they get at their first search. The difference is highly significant given the p-value. \n* Those who watch have a trip cost of USD125 more on average.\n* There might be a growth opportunity in budget passengers. When the user makes a first search which reveals a relatively cheap price, Hopper can suggest watching for the same trip with additional services such as business class. If that suggesiton can be supported with a statement like \"business flights for this flight can get as close as $X to the economy fares, why don't you watch?\" the user can be convinced to shop for more.", "_____no_output_____" ] ], [ [ "stat, p = mannwhitneyu(dfw.loc[dfw['watch_bin'] == 1, 'first_total'], \n dfw.loc[dfw['watch_bin'] == 0, 'first_total'])\nprint('Statistics=%.3f, p=%.3f' % (stat, p))", "Statistics=29810391.000, p=0.000\n" ] ], [ [ "### Question 3", "_____no_output_____" ], [ "Chart 1: What is the situation as of now compared to PY?\n* Note that from the current advance field in the data, I see that we are on April 10th 2018\n* Expired: Watch is on + Current Date > Departure Date\n* Inactive: Watch is off + Current Date can be before or after Current Date\n* Active: Watch is on + Current Date <= Departure Date\n* Shopped: Watch is on or off + Current Date can be before or after Current Date; the latter of first search and watch date added is equal to the latest_status_change\n* Booked: ", "_____no_output_____" ], [ "#### Chart 1: Number of Incoming / Outgoing / Converted Searches Through Time", "_____no_output_____" ], [ "On a daily basis, I'd like to see the number of \n* new searches of trips (incoming), \n* end of validity trips i.e. trips with departure date passing by\n* converted searches i.e. booked trips\n\n<br>\nIdeally I would like to see the number for these for a given day / time window as well as for the same period prior year (more on this in Q4). howeveer the data covers first searches over a period from 2018 start to April 10th. For illustrative purposes I show the count of the 2 KPIs above throughout the year", "_____no_output_____" ], [ "It is good practive to create date range and join data onto that as the data source may not have data for every day:", "_____no_output_____" ] ], [ [ "date_range = pd.date_range(start='1/1/2018', end='04/10/2018', freq='D')\ndf_date = pd.DataFrame(date_range, columns = ['date_range'])\ndf_date.set_index('date_range', inplace = True)", "_____no_output_____" ] ], [ [ "incoming traffic counts the number of first time searches each day:", "_____no_output_____" ] ], [ [ "hoppi5['first_search_dt_dateonly'] = hoppi5['first_search_dt'].dt.date\nincoming_traffic = hoppi5.groupby(['first_search_dt_dateonly']) \\\n .size().reset_index() \\\n .rename(columns = {0: 'count'}) \nincoming_traffic.set_index('first_search_dt_dateonly', inplace = True)", "_____no_output_____" ] ], [ [ "outgoing traffic counts the number of trips with departure within the same day, each day. Until a trip is considered 'outgoing' there is a chance that it can be converted to booking:", "_____no_output_____" ] ], [ [ "outgoing_traffic = hoppi5.groupby(['departure_date']) \\\n .size().reset_index() \\\n .rename(columns = {0: 'count'}) \noutgoing_traffic.set_index('departure_date', inplace = True)", "_____no_output_____" ] ], [ [ "converted traffic is the numbe rof bookings that took place each day i.e. conversions:", "_____no_output_____" ] ], [ [ "hoppi5['latest_status_change_dt_dateonly'] = hoppi5['first_search_dt'].dt.date\nconverted_traffic = hoppi5.loc[hoppi5['status_latest'] == 'booked'].groupby(['latest_status_change_dt_dateonly']) \\\n .size().reset_index() \\\n .rename(columns = {0: 'count'}) \nconverted_traffic.set_index('latest_status_change_dt_dateonly', inplace = True)", "_____no_output_____" ] ], [ [ "I join counts on the date range index created above:", "_____no_output_____" ] ], [ [ "df_chart1 = pd.merge(df_date, incoming_traffic, left_index = True, right_index = True, how='left')\ndf_chart1.rename(columns = {'count': 'incoming_count'}, inplace = True)\ndf_chart2 = pd.merge(df_chart1, outgoing_traffic, left_index = True, right_index = True, how='left')\ndf_chart2.rename(columns = {'count': 'outgoing_count'}, inplace = True)\ndf_chart3 = pd.merge(df_chart2, converted_traffic, left_index = True, right_index = True, how='left')\ndf_chart3.rename(columns = {'count': 'converted_count'}, inplace = True)\ndf_chart3['day'] = df_chart3.index.dayofyear", "_____no_output_____" ], [ "df_chart3_components = [df_chart3[['incoming_count', 'day']].assign(count_type = 'incoming').reset_index(). \\\n rename(columns = {'incoming_count': 'count'}),\n df_chart3[['outgoing_count', 'day']].assign(count_type = 'outgoing').reset_index(). \\\n rename(columns = {'outgoing_count': 'count'}),\n df_chart3[['converted_count', 'day']].assign(count_type = 'converted').reset_index(). \\\n rename(columns = {'converted_count': 'count'})]\ndf_chart4 = pd.concat(df_chart3_components)", "_____no_output_____" ] ], [ [ "I plot the chart here below. Note that the data collection seems to have started as of 2018 start. Therefore the outgoing count do not reflect the reality in the early periods of the chart. Also the number of trips whose departure is in the future at a given time could be shown as well. That would show the pool of trips that could be converted at a given time.", "_____no_output_____" ] ], [ [ "sns.set_style('dark')\nfig, ax1 = plt.subplots(figsize=(15,10))\nax2 = ax1.twinx()\nsns.lineplot(x=df_chart3['day'], \n y=df_chart3['incoming_count'],\n color='#6FC28B',\n marker = \"X\",\n ax=ax1)\nsns.lineplot(x=df_chart3['day'], \n y=df_chart3['outgoing_count'],\n color='#FA6866',\n marker=\"v\",\n ax=ax1)\nsns.lineplot(x=df_chart3['day'], \n y=df_chart3['converted_count'],\n color='#F0A02A',\n marker=\"o\",\n ax=ax2)\nfig.legend(['Incoming #', 'Expiring #', 'Converted #'])\nax1.set(xlabel='Day of Year', ylabel='Incoming and Expiring Search Count')\nax2.set(ylabel='Converted Search Count')\nplt.title('Number of Incoming / Expiring / Converted Searches by Day', fontsize = 14)\nplt.show()", "_____no_output_____" ] ], [ [ "### Chart 2: KPIs Affecting Conversion - Categorical KPIs", "_____no_output_____" ], [ "Categorical variables that turned out to have an impact on conversion are worth following daily. As I sugegsted for the 1st chart, it makes more sense to compare these with prior year same period figures.\nIn this chart we follow the % of people who\n* look for a round trip\n* look for a weekend trip\n* look for a short layover\n* have an ongoing watch\n* have an received a buy suggestion\n\n<br>\nNote that these are all categories that help estimate conversion", "_____no_output_____" ] ], [ [ "df_chart_perc1 = hoppi5.loc[hoppi5['departure_date'] >= '04-10-2018'].describe() # describe() gives the mean per vcategory. \n# As they were binary, it gives the %\ndf_chart_perc2 = df_chart_perc1.loc[['mean'], ['round_trip', 'weekend', 'filter_short_layover', 'watch_bin', 'buy']]\ndf_chart_perc2 = df_chart_perc2.transpose().reset_index() # transpose to make it convenient for seaborn notation\ndf_chart_perc2['mean'] = df_chart_perc2['mean'] * 100 # percentages in absolute numbers\ndf_chart_perc2.rename(columns = {'mean':'percentage'}, inplace=True)", "_____no_output_____" ], [ "sns.set_style('white')\nfig, ax = plt.subplots(figsize=(15,8))\nsns.barplot(x=\"index\",\n y=\"percentage\", \n palette=[\"#FA6866\", \"#01AAE4\", \"#505050\", \"#AAAAAA\", \"#F67096\"],\n data=df_chart_perc2,\n ax=ax)\nax.set(xlabel='Trip Categories', ylabel='% of Qualified Trips')\nplt.title(\"Percentage of Trips\",fontsize=14)\nplt.show()", "_____no_output_____" ] ], [ [ "### Chart 3: KPIs Affecting Conversion - Ordinal KPIs", "_____no_output_____" ], [ "In a similar vein to the Chart2, I look at KPIs having an impact on conversion here as well. This time I check ordinal variables. Again, it would make more sense to compare with prior year same period figures.\nIn this chart we follow \n* average difference between the lowest price found and the first price\n* average difference between the last price and the first price\n* average difference between the price when a buy recommendation was made and the first price\n* average number of days between the first search and the departure date\n\n<br>\nNote that these are all categories that help estimate conversion as well.", "_____no_output_____" ] ], [ [ "df_chart_abs1 = hoppi5.loc[hoppi5['departure_date'] >= '04-10-2018'].describe()\ndf_chart_abs2 = df_chart_abs1.loc[['mean'], ['dif_lowest_first',\n 'dif_last_first', 'dif_buy_first',\n 'first_advance']]\ndf_chart_abs3 = df_chart_abs2.transpose().reset_index()\ndf_chart_abs3.rename(columns = {'mean':'average'}, inplace=True)", "_____no_output_____" ], [ "fig, (ax1, ax2) = plt.subplots(1, 2, gridspec_kw={'width_ratios': [3, 1]}, figsize=(15,8))\n\nsns.barplot(x=df_chart_abs3.loc[df_chart_abs3['index'] != 'first_advance']['index'], \n y=df_chart_abs3.loc[df_chart_abs3['index'] != 'first_advance']['average'],\n palette=[\"#FA6866\", \"#01AAE4\", \"#505050\"],\n ax=ax1)\nsns.barplot(x=df_chart_abs3.loc[df_chart_abs3['index'] == 'first_advance']['index'], \n y=df_chart_abs3.loc[df_chart_abs3['index'] == 'first_advance']['average'],\n color='#AAAAAA',\n ax=ax2)\nax1.set(xlabel='KPIs', ylabel='Average Difference in Prices ($)')\nax2.set(ylabel='Average Number of Days')\nax1.set_title('Trips by Absolute Numbers', fontsize = 14)\nplt.show()", "_____no_output_____" ] ], [ [ "### Question 4", "_____no_output_____" ], [ "* I should have the history on every trip - user. That would enable insight in 2 main fields:\n * Communication Strategy: Recording of every update related to a user - trip allows:\n * to understand after how many notifications...\n * a booking takes place\n * a watch is canceled\n * Year on Year (YoY) comparisons: A record with active status now could have been inactive at a certain point of time as the user can enable / disable watch at different times. As a result, a YoY comparison of records in terms of their statuses is not possible with the information in hand\n* I would also like to have this data for a longer period to identify returning customers, frequent flyers for whom I could develop new features", "_____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" ]

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Background \n\nThis project deals with artificial advertising data set, indicating whether or not a particular internet user clicked on an Advertisement. This dataset can be explored to train a model that can predict whether or not the new users will click on an ad based on their various low-level features.\n\nThis data set contains the following features:\n\n* 'Daily Time Spent on Site': consumer time on site in minutes\n* 'Age': cutomer age in years\n* 'Area Income': Avg. Income of geographical area of consumer\n* 'Daily Internet Usage': Avg. minutes a day consumer is on the internet\n* 'Ad Topic Line': Headline of the advertisement\n* 'City': City of consumer\n* 'Male': Whether or not consumer was male\n* 'Country': Country of consumer\n* 'Timestamp': Time at which consumer clicked on Ad or closed window\n* 'Clicked on Ad': 0 or 1 indicated clicking on Ad\n\n# Dataset overview", "_____no_output_____" ] ], [ [ "import numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nimport seaborn as sns\n%matplotlib inline\nsns.set_style('white')", "_____no_output_____" ], [ "df_ad = pd.read_csv('Data/advertising.csv')", "_____no_output_____" ], [ "df_ad.head(3)", "_____no_output_____" ], [ "df_ad.info()", "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 1000 entries, 0 to 999\nData columns (total 10 columns):\nDaily Time Spent on Site 1000 non-null float64\nAge 1000 non-null int64\nArea Income 1000 non-null float64\nDaily Internet Usage 1000 non-null float64\nAd Topic Line 1000 non-null object\nCity 1000 non-null object\nMale 1000 non-null int64\nCountry 1000 non-null object\nTimestamp 1000 non-null object\nClicked on Ad 1000 non-null int64\ndtypes: float64(3), int64(3), object(4)\nmemory usage: 78.2+ KB\n" ], [ "df_ad.isnull().any()", "_____no_output_____" ], [ "df_ad.describe()", "_____no_output_____" ] ], [ [ "# EDA\n#### Age distribution of the dataset", "_____no_output_____" ] ], [ [ "sns.set_context('notebook',font_scale=1.5)\nsns.distplot(df_ad.Age,bins=30,kde=False,color='red')\nplt.show()", "_____no_output_____" ] ], [ [ "#### pairplot of dataset defined by `Clicked on Ad`", "_____no_output_____" ] ], [ [ "import warnings\nwarnings.filterwarnings('ignore') #### since the target variable is numeric, the joint plot by the target variable generates the warning.", "_____no_output_____" ], [ "sns.pairplot(df_ad,hue='Clicked on Ad')", "_____no_output_____" ] ], [ [ "# Model training: Basic Logistic Regression", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import train_test_split", "_____no_output_____" ], [ "X = df_ad[['Daily Time Spent on Site', 'Age', 'Area Income',\n 'Daily Internet Usage', 'Male']]\ny = df_ad['Clicked on Ad']\nX_train, X_test, y_train, y_test = train_test_split(X,y,random_state=100)", "_____no_output_____" ] ], [ [ "#### training", "_____no_output_____" ] ], [ [ "from sklearn.linear_model import LogisticRegression", "_____no_output_____" ], [ "lr = LogisticRegression().fit(X_train,y_train)", "_____no_output_____" ] ], [ [ "#### Predictions and Evaluations", "_____no_output_____" ] ], [ [ "from sklearn.metrics import classification_report,confusion_matrix", "_____no_output_____" ], [ "y_predict = lr.predict(X_test)", "_____no_output_____" ], [ "pd.DataFrame(confusion_matrix(y_test,y_predict),index=['True 0','True 1'],\n columns=['Predicted 0','Predicted 1'])", "_____no_output_____" ], [ "print(classification_report(y_test,y_predict))", " precision recall f1-score support\n\n 0 0.86 0.92 0.89 119\n 1 0.93 0.86 0.89 131\n\n micro avg 0.89 0.89 0.89 250\n macro avg 0.89 0.89 0.89 250\nweighted avg 0.89 0.89 0.89 250\n\n" ] ], [ [ "# Model training: Optimized Logistic Regression", "_____no_output_____" ] ], [ [ "from sklearn.preprocessing import StandardScaler\nfrom sklearn.model_selection import GridSearchCV\n\nscaler = StandardScaler().fit(X_train)\nX_train_scaled = scaler.transform(X_train)\nx_test_scaled = scaler.transform(X_test)", "_____no_output_____" ] ], [ [ "#### 3-fold CV grid search", "_____no_output_____" ] ], [ [ "grid_param = {'C':[0.01,0.03,0.1,0.3,1,3,10]}\ngrid_lr = GridSearchCV(LogisticRegression(),grid_param,cv=3).fit(X_train_scaled,y_train)", "_____no_output_____" ], [ "print('best regularization parameter: {}'.format(grid_lr.best_params_))\nprint('best CV score: {}'.format(grid_lr.best_score_.round(3)))", "best regularization parameter: {'C': 0.3}\nbest CV score: 0.971\n" ] ], [ [ "#### Predictions and Evaluations", "_____no_output_____" ] ], [ [ "y_predict_2 = grid_lr.predict(x_test_scaled)", "_____no_output_____" ], [ "pd.DataFrame(confusion_matrix(y_test,y_predict_2),index=['True 0','True 1'],\n columns=['Predicted 0','Predicted 1'])", "_____no_output_____" ], [ "print(classification_report(y_test,y_predict_2))", " precision recall f1-score support\n\n 0 0.94 1.00 0.97 119\n 1 1.00 0.94 0.97 131\n\n micro avg 0.97 0.97 0.97 250\n macro avg 0.97 0.97 0.97 250\nweighted avg 0.97 0.97 0.97 250\n\n" ] ], [ [ "A simple logistic regression without any tuning effort shows high classification performance. With standarization and the tuned 'C' parameter, the performance of the same logistic regression model could be improved by much to ~0.97 f1-score. ", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import os\nimport time\nimport numpy as np\nimport matplotlib\nimport matplotlib.pyplot as plt\nfrom matplotlib import pylab\n%config InlineBackend.figure_format = 'retina'\nmatplotlib.rcParams['figure.dpi'] = 80\ntextsize = 'x-large'\nparams = {'legend.fontsize': 'x-large',\n 'figure.figsize': (10, 8),\n 'axes.labelsize': textsize,\n 'axes.titlesize': textsize,\n 'xtick.labelsize': textsize,\n 'ytick.labelsize': textsize}\npylab.rcParams.update(params)\nplt.ion()\n\nimport Corrfunc\nfrom Corrfunc.io import read_lognormal_catalog\nfrom Corrfunc.theory.DDsmu import DDsmu\nfrom Corrfunc.theory.DD import DD\nfrom Corrfunc.theory.xi import xi\nfrom Corrfunc.utils import compute_amps\nfrom Corrfunc.utils import evaluate_xi\nfrom Corrfunc.utils import qq_analytic\nfrom Corrfunc.bases import spline\n\n%load_ext autoreload\n%autoreload 2", "/Users/ksf/code/nyu/research/Corrfunc/Corrfunc/__init__.py\n" ] ], [ [ "# Demo Notebook: The Continuous-Function Estimator\n## Tophat and Spline bases on a periodic box\n\nHello! In this notebook we'll show you how to use the continuous-function estimator to estimate the 2-point correlation function (2pcf) with a method that produces, well, continuous correlation functions.", "_____no_output_____" ], [ "## Load in data", "_____no_output_____" ], [ "We'll demonstrate with a low-density lognormal simulation box, which we've included with the code. We'll show here the box with 3e-4 ($h^{-1}$Mpc)$^{-3}$, but if you're only running with a single thread, you will want to run this notebook with the 1e-4 ($h^{-1}$Mpc)$^{-3}$ box for speed. (The code is extremely parallel, so when you're running for real, you'll definitely want to bump up the number of threads.)", "_____no_output_____" ] ], [ [ "x, y, z = read_lognormal_catalog(n='3e-4')\nboxsize = 750.0\nnd = len(x)\nprint(\"Number of data points:\",nd)", "Number of data points: 125342\n" ] ], [ [ "We'll also want a random catalog, that's a bit bigger than our data:", "_____no_output_____" ] ], [ [ "nr = 3*nd\nx_rand = np.random.uniform(0, boxsize, nr)\ny_rand = np.random.uniform(0, boxsize, nr)\nz_rand = np.random.uniform(0, boxsize, nr)\nprint(\"Number of random points:\",nr)", "Number of random points: 376026\n" ], [ "print(x)\nprint(x_rand)", "[1.13136184e+00 4.30035293e-01 2.08324015e-01 ... 7.49666077e+02\n 7.49922791e+02 7.49938477e+02]\n[567.62600303 166.85340522 461.79238824 ... 577.65066275 9.85155819\n 581.1525008 ]\n" ] ], [ [ "Let's first compute the regular correlation function. We'll need some radial bins. We'll also need to tell Corrfunc that we're working with a periodic box, and the number of parallel threads. Then we can go ahead and compute the real-space correlation function xi(r) from the pair counts, DD(r) (documentation here: https://corrfunc.readthedocs.io/en/master/api/Corrfunc.theory.html)", "_____no_output_____" ] ], [ [ "rmin = 40.0\nrmax = 150.0\nnbins = 22\nr_edges = np.linspace(rmin, rmax, nbins+1)\nr_avg = 0.5*(r_edges[1:]+r_edges[:-1])\n\nperiodic = True\nnthreads = 1", "_____no_output_____" ], [ "dd_res = DD(1, nthreads, r_edges, x, y, z, boxsize=boxsize, periodic=periodic)\ndr_res = DD(0, nthreads, r_edges, x, y, z, X2=x_rand, Y2=y_rand, Z2=z_rand, boxsize=boxsize, periodic=periodic)\nrr_res = DD(1, nthreads, r_edges, x_rand, y_rand, z_rand, boxsize=boxsize, periodic=periodic)", "_____no_output_____" ] ], [ [ "We can use these pair counts to compute the Landy-Szalay 2pcf estimator (Landy & Szalay 1993). Let's define a function, as we'll want to reuse this:", "_____no_output_____" ] ], [ [ "def landy_szalay(nd, nr, dd, dr, rr):\n # Normalize the pair counts\n dd = dd/(nd*nd)\n dr = dr/(nd*nr)\n rr = rr/(nr*nr)\n xi_ls = (dd-2*dr+rr)/rr\n return xi_ls", "_____no_output_____" ] ], [ [ "Let's unpack the pair counts from the Corrfunc results object, and plot the resulting correlation function: \n\n(Note that if you use weights, you need to multiply by the 'weightavg' column.)", "_____no_output_____" ] ], [ [ "dd = np.array([x['npairs'] for x in dd_res], dtype=float)\ndr = np.array([x['npairs'] for x in dr_res], dtype=float)\nrr = np.array([x['npairs'] for x in rr_res], dtype=float)\nxi_ls = landy_szalay(nd, nr, dd, dr, rr)", "_____no_output_____" ], [ "plt.figure(figsize=(8,5))\nplt.plot(r_avg, xi_ls, marker='o', ls='None', color='grey', label='Standard estimator')\nplt.xlabel(r'r ($h^{-1}$Mpc)')\nplt.ylabel(r'$\\xi$(r)')\nplt.legend()", "_____no_output_____" ] ], [ [ "Great, we can even see the baryon acoustic feauture at ~100 $h^{-1}$Mpc!", "_____no_output_____" ], [ "## Continuous-function estimator: Tophat basis", "_____no_output_____" ], [ "Now we'll use the continuous-function estimator to compute the same correlation function, but in a continuous representation. First we'll use a tophat basis, to achieve the equivalent (but more correct!) result.\n\nWe need to give the name of the basis as 'proj_type'. We also need to choose the number of components, 'nprojbins'. In this case, we want the components to be a tophat for each bin, so this will just be 'nbins'. ", "_____no_output_____" ] ], [ [ "proj_type = 'tophat'\nnprojbins = nbins", "_____no_output_____" ] ], [ [ "Currently the continuous-function estimator is only implemented in DD(s,mu) ('DDsmu'), the redshift-space correlation function which divides the transverse direction s from the line-of-sight direction mu. But we can simply set the number of mu bins to 1, and mumax to 1 (the max of cosine), to achieve the equivalent of DD in real space.", "_____no_output_____" ] ], [ [ "nmubins = 1\nmumax = 1.0", "_____no_output_____" ] ], [ [ "Then we just need to give Corrfunc all this info, and unpack the continuous results! The first returned object is still the regular Corrfunc results object (we could have just used this in our above demo of the standard result).", "_____no_output_____" ] ], [ [ "dd_res, dd_proj, _ = DDsmu(1, nthreads, r_edges, mumax, nmubins, x, y, z, \n boxsize=boxsize, periodic=periodic, proj_type=proj_type, nprojbins=nprojbins)", "_____no_output_____" ], [ "dr_res, dr_proj, _ = DDsmu(0, nthreads, r_edges, mumax, nmubins, x, y, z, X2=x_rand, Y2=y_rand, Z2=z_rand,\n boxsize=boxsize, periodic=periodic, proj_type=proj_type, nprojbins=nprojbins)", "_____no_output_____" ], [ "rr_res, rr_proj, qq_proj = DDsmu(1, nthreads, r_edges, mumax, nmubins, x_rand, y_rand, z_rand,\n boxsize=boxsize, periodic=periodic, proj_type=proj_type, nprojbins=nprojbins)", "_____no_output_____" ] ], [ [ "We can now compute the amplitudes of the correlation function from these continuous pair counts. The compute_amps function uses the Landy-Szalay formulation of the estimator, but adapted for continuous bases. (Note that you have to pass some values twice, as this is flexible enough to translate to cross-correlations between two datasets and two random catalogs.)", "_____no_output_____" ] ], [ [ "amps = compute_amps(nprojbins, nd, nd, nr, nr, dd_proj, dr_proj, dr_proj, rr_proj, qq_proj)", "Computing amplitudes (Corrfunc/utils.py)\n" ] ], [ [ "With these amplitudes, we can evaluate our correlation function at any set of radial separations! Let's make a fine-grained array and evaluate. We need to pass 'nprojbins' and 'proj_type'. Because we will be evaluating our tophat function at the new separations, we also need to give it the original bins.", "_____no_output_____" ] ], [ [ "r_fine = np.linspace(rmin, rmax, 2000)", "_____no_output_____" ], [ "xi_proj = evaluate_xi(nprojbins, amps, len(r_fine), r_fine, nbins, r_edges, proj_type)", "Evaluating xi (Corrfunc/utils.py)\n" ] ], [ [ "Let's check out the results, compared with the standard estimator!", "_____no_output_____" ] ], [ [ "plt.figure(figsize=(8,5))\nplt.plot(r_fine, xi_proj, color='steelblue', label='Tophat estimator')\nplt.plot(r_avg, xi_ls, marker='o', ls='None', color='grey', label='Standard estimator')\nplt.xlabel(r'r ($h^{-1}$Mpc)')\nplt.ylabel(r'$\\xi$(r)')\nplt.legend()", "_____no_output_____" ] ], [ [ "We can see that we're getting \"the same\" result, but continously, with the hard bin edges made clear.", "_____no_output_____" ], [ "### Analytically computing the random term", "_____no_output_____" ], [ "Because we're working with a periodic box, we don't actually need a random catalog. We can analytically compute the RR term, as well as the QQ matrix.\n\nWe'll need the volume of the box, and the same info about our basis function as before:", "_____no_output_____" ] ], [ [ "volume = boxsize**3\nrr_ana, qq_ana = qq_analytic(rmin, rmax, nd, volume, nprojbins, nbins, r_edges, proj_type)", "Evaluating qq_analytic (Corrfunc/utils.py)\n" ] ], [ [ "We also don't need to use the Landy-Szalay estimator (we don't have a DR term!). To get the amplitudes we can just use the naive estimator, $\\frac{\\text{DD}}{\\text{RR}}-1$. In our formulation, the RR term in the demoninator becomes the inverse QQ term, so we have QQ$^{-1}$ $\\cdot$ (DD-RR). ", "_____no_output_____" ] ], [ [ "numerator = dd_proj - rr_ana\namps_ana, *_ = np.linalg.lstsq(qq_ana, numerator, rcond=None) # Use linalg.lstsq instead of actually computing inverse!", "_____no_output_____" ] ], [ [ "Now we can go ahead and evaluate the correlation function at our fine separations.", "_____no_output_____" ] ], [ [ "xi_ana = evaluate_xi(nbins, amps_ana, len(r_fine), r_fine, nbins, r_edges, proj_type)", "Evaluating xi (Corrfunc/utils.py)\n" ] ], [ [ "We'll compare this to computing the analytic correlation function with standard Corrfunc:", "_____no_output_____" ] ], [ [ "xi_res = Corrfunc.theory.xi(boxsize, nthreads, r_edges, x, y, z)\nxi_theory = np.array([x['xi'] for x in xi_res], dtype=float)", "_____no_output_____" ], [ "plt.figure(figsize=(8,5))\nplt.plot(r_fine, xi_ana, color='blue', label='Tophat basis')\nplt.plot(r_avg, xi_theory, marker='o', ls='None', color='grey', label='Standard Estimator')\nplt.xlabel(r'r ($h^{-1}$Mpc)')\nplt.ylabel(r'$\\xi$(r)')\nplt.legend()", "_____no_output_____" ] ], [ [ "Once again, the standard and continuous correlation functions line up exactly. The correlation function looks smoother, as we didn't have to deal with a non-exact random catalog to estimate the window function.", "_____no_output_____" ], [ "## Continuous-function estimator: Cubic spline basis", "_____no_output_____" ], [ "Now we can make things more interesting! Let's choose a cubic spline basis. Luckily, this capability comes with the continuous-function version of Corrfunc!\n\nWe need to choose the parameters for our spline. Here, we'll choose a cubic spline. If we used a linear spline, we'd get a piecewise function; if we used a zeroth-order spline, we'd recover our tophat bases from above.\n\nWe'll take the min and max of the same separation values we used, and choose half the number of components as our previous bins (these will be related to the 'knots' in the spline). We'll also need the number of radial bins at which to evaluate our functions; the code will interpolate between these.\n\nThen we'll write our basis to a file. For any set of basis functions that is read from a file, 'proj_type' must be set to 'general_r'.", "_____no_output_____" ] ], [ [ "proj_type = 'generalr'\nkwargs = {'order': 3} # 3: cubic spline\nprojfn = 'quadratic_spline.dat'\nnprojbins = int(nbins/2)\nspline.write_bases(rmin, rmax, nprojbins, projfn, ncont=1000, **kwargs)", "_____no_output_____" ] ], [ [ "Let's check out the basis functions:", "_____no_output_____" ] ], [ [ "bases = np.loadtxt(projfn)\nbases.shape\nr = bases[:,0]\nplt.figure(figsize=(8,5))\nfor i in range(1, len(bases[0])):\n plt.plot(r, bases[:,i], color='red', alpha=0.5)\nplt.xlabel(r'r ($h^{-1}$Mpc)')", "_____no_output_____" ] ], [ [ "The bases on the ends are different so that they have the same normalization.", "_____no_output_____" ], [ "We'll use the analytic version of the estimator, making sure to pass the basis file:", "_____no_output_____" ] ], [ [ "dd_res_spline, dd_spline, _ = DDsmu(1, nthreads, r_edges, mumax, nmubins, x, y, z, \n boxsize=boxsize, periodic=periodic, proj_type=proj_type, nprojbins=nprojbins, projfn=projfn)", "_____no_output_____" ], [ "volume = boxsize**3\n# nbins and r_edges won't be used here because we passed projfn, but they're needed for compatibility. (TODO: fix!)\nrr_ana_spline, qq_ana_spline = qq_analytic(rmin, rmax, nd, volume, nprojbins, nbins, r_edges, proj_type, projfn=projfn)\n\nnumerator = dd_spline - rr_ana_spline\namps_ana_spline, *_ = np.linalg.lstsq(qq_ana_spline, numerator, rcond=None) # Use linalg.lstsq instead of actually computing inverse!\n\nxi_ana_spline = evaluate_xi(nprojbins, amps_ana_spline, len(r_fine), r_fine, nbins, r_edges, proj_type, projfn=projfn)", "Evaluating qq_analytic (Corrfunc/utils.py)\nEvaluating xi (Corrfunc/utils.py)\n" ] ], [ [ "Let's compare the results:", "_____no_output_____" ] ], [ [ "plt.figure(figsize=(8,5))\nplt.plot(r_fine, xi_ana_spline, color='red', label='Cubic spline basis')\nplt.plot(r_fine, xi_ana, color='blue', label='Tophat basis')\nplt.plot(r_avg, xi_theory, marker='o', ls='None', color='grey', label='Standard estimator')\nplt.xlabel(r'r ($h^{-1}$Mpc)')\nplt.ylabel(r'$\\xi$(r)')\nplt.legend()", "_____no_output_____" ] ], [ [ "We can see that the spline basis function produced a completely smooth correlation function; no hard-edged bins! It also captured that baryon acoustic feature (which we expect to be a smooth peak).\n\nThis basis function is a bit noisy and likely has some non-physical features - but so does the tophat / standard basis! In the next notebook, we'll use a physically motivated basis function.", "_____no_output_____" ], [ "Finally, Remember to clean up the basis function file:", "_____no_output_____" ] ], [ [ "os.remove(projfn)", "_____no_output_____" ] ], [ [ "The below ipython magic line will convert this notebook to a regular old python script.", "_____no_output_____" ] ], [ [ "#!jupyter nbconvert --to script example_theory.ipynb", "_____no_output_____" ] ] ]

[ "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" ]

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

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ "CC-BY-4.0" ]

[ [ [ "# Data Analysis - FIB-SEM Datasets\n* Goal: identify changes occurred across different time points", "_____no_output_____" ] ], [ [ "import os, sys, glob\nimport re\nimport numpy as np\nimport pandas as pd\nfrom scipy.stats import ttest_ind\nimport matplotlib.pyplot as plt\nimport pprint", "_____no_output_____" ] ], [ [ "## 01 Compile data into single .csv file for each label", "_____no_output_____" ] ], [ [ "mainpath = 'D:\\PerlmutterData'\nfolder = 'segmentation_compiled_export'\ndata_folder = 'data'\n\npath = os.path.join(mainpath, folder, data_folder)\nprint(path)\n\nfolders = ['cell_membrane', 'nucleus', 'mito', 'cristae', 'inclusion', 'ER']", "D:\\PerlmutterData\\segmentation_compiled_export\\data\n" ], [ "target_list = glob.glob(os.path.join(path, 'compile', '*.csv'))\ntarget_list = [os.path.basename(x) for x in target_list]\ntarget_list = [os.path.splitext(x)[0] for x in target_list]\nprint(target_list)", "['cell_membrane', 'cristae', 'ER', 'inclusion', 'mito', 'nodes', 'nucleus', 'points', 'segments_s']\n" ], [ "file_meta = {\n 'data_d00_batch01_loc01': 0,\n 'data_d00_batch02_loc02': 0,\n 'data_d00_batch02_loc03': 0,\n 'data_d07_batch01_loc01': 7, \n 'data_d07_batch02_loc01': 7,\n 'data_d07_batch02_loc02': 7, \n 'data_d14_batch01_loc01': 14, \n 'data_d17_batch01_loc01': 17,\n 'data_d21_batch01_loc01': 21,\n}\n", "_____no_output_____" ], [ "for i in folders:\n file_list = glob.glob(os.path.join(path, 'raw', i, '*.csv'))\n \n if not i in target_list:\n df = pd.DataFrame()\n\n for j in file_list: \n data_temp = pd.read_csv(j, header = 1)\n \n filename_tmp = os.path.basename(j)\n \n # add filename\n data_temp['filename'] = filename_tmp\n \n # add day\n filename_noext = os.path.splitext(filename_tmp)[0]\n pattern = re.compile(\"data_d[0-9][0-9]_batch[0-9][0-9]_loc[0-9][0-9]\")\n original_filename = pattern.search(filename_noext).group(0)\n day_tmp = file_meta[original_filename]\n \n data_temp['day'] = day_tmp\n \n df = df.append(data_temp, ignore_index = True)\n \n display(df)\n df.to_csv(os.path.join(path, 'compile', i + '.csv'))\n ", "_____no_output_____" ] ], [ [ "## 02 Load data", "_____no_output_____" ], [ "### 02-01 Calculate mean and tota|l volumn for mito, cristate, ER and inclusion", "_____no_output_____" ] ], [ [ "df_mito = pd.read_csv(os.path.join(path, 'compile', 'mito' + '.csv'))\ndf_mito['Volume3d_µm^3'] = df_mito['Volume3d']/1e9\ndf_mito['Area3d_µm^2'] = df_mito['Area3d']/1e6\n\ndf_mito_sum_grouped = df_mito.groupby(['day', 'filename']).sum().reset_index()\ndf_mito_mean_grouped = df_mito.groupby(['day', 'filename']).mean().reset_index()\n\ndf_cristae = pd.read_csv(os.path.join(path, 'compile', 'cristae' + '.csv'))\ndf_cristae['Volume3d_µm^3'] = df_cristae['Volume3d']/1e9\ndf_cristae['Area3d_µm^2'] = df_cristae['Area3d']/1e6\n\ndf_cristae_sum_grouped = df_cristae.groupby(['day', 'filename']).sum().reset_index()\ndf_cristae_mean_grouped = df_cristae.groupby(['day', 'filename']).mean().reset_index()\n\ndf_ER = pd.read_csv(os.path.join(path, 'compile', 'ER' + '.csv'))\ndf_ER['Volume3d_µm^3'] = df_ER['Volume3d']/1e9\ndf_ER['Area3d_µm^2'] = df_ER['Area3d']/1e6\n\ndf_ER_sum_grouped = df_ER.groupby(['day', 'filename']).sum().reset_index()\ndf_ER_mean_grouped = df_ER.groupby(['day', 'filename']).mean().reset_index()\n\ndf_inclusion = pd.read_csv(os.path.join(path, 'compile', 'inclusion' + '.csv'))\ndf_inclusion['Volume3d_µm^3'] = df_inclusion['Volume3d']/1e9\ndf_inclusion['Area3d_µm^2'] = df_inclusion['Area3d']/1e6\n\ndf_inclusion_sum_grouped = df_inclusion.groupby(['day', 'filename']).sum().reset_index()\ndf_inclusion_mean_grouped = df_inclusion.groupby(['day', 'filename']).mean().reset_index()", "_____no_output_____" ] ], [ [ "### 02-02 Calculate the total volume for cell membrane and nucleus", "_____no_output_____" ] ], [ [ "df_nucleus = pd.read_csv(os.path.join(path, 'compile', 'nucleus' + '.csv'))\ndf_nucleus['Volume3d_µm^3'] = df_nucleus['Volume3d']/1e9\ndf_nucleus['Area3d_µm^2'] = df_nucleus['Area3d']/1e6\ndf_nucleus_sum_grouped = df_nucleus.groupby(['day', 'filename']).sum().reset_index()", "_____no_output_____" ], [ "df_cell_membrane = pd.read_csv(os.path.join(path, 'compile', 'cell_membrane' + '.csv'))\ndf_cell_membrane['Volume3d_µm^3'] = df_cell_membrane['Volume3d']/1e9\ndf_cell_membrane['Area3d_µm^2'] = df_cell_membrane['Area3d']/1e6\ndf_cell_membrane_sum_grouped = df_cell_membrane.groupby(['day', 'filename']).sum().reset_index()", "_____no_output_____" ], [ "df_cell_membrane_sum_grouped", "_____no_output_____" ] ], [ [ "### 02-03 Calculate the volume of cytoplasm", "_____no_output_____" ] ], [ [ "df_cyto = pd.DataFrame()\ndf_cyto['filename'] = df_cell_membrane_sum_grouped['filename']\ndf_cyto['Volume3d_µm^3'] = df_cell_membrane_sum_grouped['Volume3d_µm^3'] - df_nucleus_sum_grouped['Volume3d_µm^3']\ndisplay(df_cyto)", "_____no_output_____" ] ], [ [ "### 02-03 Omit unhealthy data or data with poor quality ", "_____no_output_____" ] ], [ [ "omit_data = ['data_d00_batch02_loc02', \n 'data_d17_batch01_loc01_01', \n 'data_d17_batch01_loc01_02']\nfor omit in omit_data: \n df_mito = df_mito.loc[df_mito['filename']!= omit+ '_mito.csv']\n df_mito_sum_grouped = df_mito_sum_grouped.loc[df_mito_sum_grouped['filename']!=omit+ '_mito.csv']\n df_mito_mean_grouped = df_mito_mean_grouped.loc[df_mito_mean_grouped['filename']!=omit+ '_mito.csv']\n df_cristae = df_cristae.loc[df_cristae['filename']!=omit+ '_cristae.csv']\n df_cristae_sum_grouped = df_cristae_sum_grouped.loc[df_cristae_sum_grouped['filename']!=omit+ '_cristae.csv']\n df_cristae_mean_grouped = df_cristae_mean_grouped.loc[df_cristae_mean_grouped['filename']!=omit+ '_cristae.csv']\n df_ER = df_ER.loc[df_ER['filename']!=omit+ '_ER.csv']\n df_ER_sum_grouped = df_ER_sum_grouped.loc[df_ER_sum_grouped['filename']!=omit+ '_ER.csv']\n df_ER_mean_grouped = df_ER_mean_grouped.loc[df_ER_mean_grouped['filename']!=omit+ '_ER.csv']\n df_inclusion = df_inclusion.loc[df_inclusion['filename']!=omit+'_inclusion.csv']\n df_inclusion_sum_grouped = df_inclusion_sum_grouped.loc[df_inclusion_sum_grouped['filename']!=omit+'_inclusion.csv']\n df_inclusion_mean_grouped = df_inclusion_mean_grouped.loc[df_inclusion_mean_grouped['filename']!=omit+'_inclusion.csv']\n df_nucleus = df_nucleus.loc[df_nucleus['filename']!=omit+'_nucleus.csv']\n df_nucleus_sum_grouped = df_nucleus_sum_grouped.loc[df_nucleus_sum_grouped['filename']!=omit+'_nucleus.csv']\n df_cell_membrane = df_cell_membrane.loc[df_cell_membrane['filename']!=omit+'_cell_membrane.csv']\n df_cell_membrane_sum_grouped = df_cell_membrane_sum_grouped.loc[df_cell_membrane_sum_grouped['filename']!=omit+'_cell_membrane.csv']\n df_cyto = df_cyto.loc[df_cyto['filename']!=omit+'_cell_membrane.csv']\n", "_____no_output_____" ], [ "df_mito = df_mito.reset_index(drop=True)\ndf_mito_sum_grouped = df_mito_sum_grouped.reset_index(drop=True)\ndf_mito_mean_grouped = df_mito_mean_grouped.reset_index(drop=True)\ndf_cristae = df_cristae.reset_index(drop=True)\ndf_cristae_sum_grouped = df_cristae_sum_grouped.reset_index(drop=True)\ndf_cristae_mean_grouped = df_cristae_mean_grouped.reset_index(drop=True)\ndf_ER = df_ER.reset_index(drop=True)\ndf_ER_sum_grouped = df_ER_sum_grouped.reset_index(drop=True)\ndf_ER_mean_grouped = df_ER_mean_grouped.reset_index(drop=True)\ndf_inclusion = df_inclusion.reset_index(drop=True)\ndf_inclusion_sum_grouped = df_inclusion_sum_grouped.reset_index(drop=True)\ndf_inclusion_mean_grouped = df_inclusion_mean_grouped.reset_index(drop=True)\ndf_nucleus = df_nucleus.reset_index(drop=True)\ndf_nucleus_sum_grouped = df_nucleus_sum_grouped.reset_index(drop=True)\ndf_cell_membrane = df_cell_membrane.reset_index(drop=True)\ndf_cell_membrane_sum_grouped = df_cell_membrane_sum_grouped.reset_index(drop=True)\ndf_cyto = df_cyto.reset_index(drop=True)", "_____no_output_____" ], [ "df_mito.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'mito.csv'))\ndf_mito_sum_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'mito_sum_volume.csv'))\ndf_mito_mean_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'mito_mean_volume.csv'))\n\ndf_cristae.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'cristae.csv'))\ndf_cristae_sum_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'cristae_sum_volume.csv'))\ndf_cristae_mean_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'cristae_mean_volume.csv'))\n\ndf_ER.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'ER.csv'))\ndf_ER_sum_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'ER_sum_volume.csv'))\ndf_ER_mean_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'ER_mean_volume.csv'))\n\ndf_inclusion.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'inclusion.csv'))\ndf_inclusion_sum_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'inclusion_sum_volume.csv'))\ndf_inclusion_mean_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'inclusion_mean_volume.csv'))\n\ndf_nucleus.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'nucleus.csv'))\ndf_nucleus_sum_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'nucleus_sum_volume.csv'))\n\ndf_cell_membrane.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'cell_membrane_volume.csv'))\ndf_cell_membrane_sum_grouped.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'cell_membrane_sum_volume.csv'))\ndf_cyto.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'cytoplasm_sum_volume.csv'))", "_____no_output_____" ], [ "df_mito_sum_grouped", "_____no_output_____" ] ], [ [ "### 02-04 Compile total volume of mito, cristate, ER and inclusion into one table\n1. raw value\n2. normalized by the total volume of cytoplasm", "_____no_output_____" ] ], [ [ "df_sum_compiled = pd.DataFrame()\ndf_sum_compiled[['filename', 'day']] = df_cell_membrane_sum_grouped[['filename', 'day']]\ndf_sum_compiled['day'] = df_sum_compiled['day'].astype('int8')\ndf_sum_compiled[['mito_Volume3d_µm^3', 'mito_Area3d_µm^2']] = df_mito_sum_grouped[['Volume3d_µm^3', 'Area3d_µm^2']]\ndf_sum_compiled[['cristae_Volume3d_µm^3', 'cristae_Area3d_µm^2']] = df_cristae_sum_grouped[['Volume3d_µm^3', 'Area3d_µm^2']]\ndf_sum_compiled[['ER_Volume3d_µm^3', 'ER_Area3d_µm^2']] = df_ER_sum_grouped[['Volume3d_µm^3', 'Area3d_µm^2']]\n\ndf_inclusion_sum_tmp = df_inclusion_sum_grouped[['Volume3d_µm^3', 'Area3d_µm^2']]\ndf_inclusion_sum_fill = pd.DataFrame([[0, 0]], columns = ['Volume3d_µm^3', 'Area3d_µm^2'])\ndf_inclusion_sum_tmp = df_inclusion_sum_fill.append(df_inclusion_sum_tmp, ignore_index = True)\n\ndf_sum_compiled[['inclusion_Volume3d_µm^3', 'inclusion_Area3d_µm^2']] = df_inclusion_sum_tmp\ndf_sum_compiled", "_____no_output_____" ], [ "fig, ax = plt.subplots(nrows=2, ncols=2, figsize=(10, 10))\nidx = 0\nfor i in range(2):\n for j in range(2):\n ax[i, j].bar(df_sum_compiled.index, \n df_sum_compiled.iloc[:, idx +2], \n tick_label=['0', '0', '7', '7', '7', '14', '21'])\n ax[i, j].set_title(df_sum_compiled.columns[idx+2])\n ax[i, j].set_xlabel('Day')\n idx += 1\n # ax[i].set_ylabel('Total Volume ($µm^3$)')\n \nfig.tight_layout(pad=3.0)\n\n\nmainpath = 'D:\\PerlmutterData'\nfolder = 'segmentation_compiled_export'\ndata_folder = 'data'\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'mito_cristae_totoal_volume_area.png'))\nplt.show()", "_____no_output_____" ], [ "fig, ax = plt.subplots(nrows=2, ncols=2, figsize=(10, 10))\nidx = 4\nfor i in range(2):\n for j in range(2):\n ax[i, j].bar(df_sum_compiled.index, \n df_sum_compiled.iloc[:, idx +2], \n tick_label=['0', '0', '7', '7', '7', '14', '21'])\n ax[i, j].set_title(df_sum_compiled.columns[idx+2])\n ax[i, j].set_xlabel('Day')\n idx += 1\n # ax[i].set_ylabel('Total Volume ($µm^3$)')\n \nfig.tight_layout(pad=3.0)\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'ER_inclusion_totoal_volume_area.png'))\nplt.show()", "_____no_output_____" ], [ "df_sum_compiled_normalized = pd.DataFrame()\ndf_sum_compiled_normalized[['filename', 'day']] = df_cell_membrane_sum_grouped[['filename', 'day']]\ncal_tmp = df_sum_compiled.iloc[:, 2:].div(df_cyto['Volume3d_µm^3'], axis=0)\ndf_sum_compiled_normalized = pd.concat([df_sum_compiled_normalized, cal_tmp], axis=1)\ndf_sum_compiled_normalized", "_____no_output_____" ], [ "fig, ax = plt.subplots(nrows=2, ncols=2, figsize=(10, 10))\nidx = 0\nfor i in range(2):\n for j in range(2):\n ax[i, j].bar(df_sum_compiled_normalized.index, \n df_sum_compiled_normalized.iloc[:, idx +2], \n tick_label=['0', '0', '7', '7', '7', '14', '21'])\n ax[i, j].set_title(df_sum_compiled_normalized.columns[idx+2])\n ax[i, j].set_xlabel('Day')\n idx += 1\n # ax[i].set_ylabel('Total Volume ($µm^3$)')\n \nfig.tight_layout(pad=3.0)\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'mito_cristae_normalized_totoal_volume_area.png'))\nplt.show()", "_____no_output_____" ], [ "fig, ax = plt.subplots(nrows=2, ncols=2, figsize=(10, 10))\nidx = 4\nfor i in range(2):\n for j in range(2):\n ax[i, j].bar(df_sum_compiled_normalized.index, \n df_sum_compiled_normalized.iloc[:, idx +2], \n tick_label=['0', '0', '7', '7', '7', '14', '21'])\n ax[i, j].set_title(df_sum_compiled_normalized.columns[idx+2])\n ax[i, j].set_xlabel('Day')\n idx += 1\n # ax[i].set_ylabel('Total Volume ($µm^3$)')\n \nfig.tight_layout(pad=3.0)\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'ER_inclusion_normalized_totoal_volume_area.png'))\nplt.show()", "_____no_output_____" ] ], [ [ "### 02-05 Compile mean volume of mito, cristate, ER and inclusion into one table\n1. raw value\n2. normalized by the total volume of cytoplasm", "_____no_output_____" ] ], [ [ "df_mean_compiled = pd.DataFrame()\ndf_mean_compiled[['filename', 'day']] = df_cell_membrane_sum_grouped[['filename', 'day']]\ndf_mean_compiled['day'] = df_mean_compiled['day'].astype('int8')\ndf_mean_compiled[['mito_Volume3d_µm^3', 'mito_Area3d_µm^2']] = df_mito_mean_grouped[['Volume3d_µm^3', 'Area3d_µm^2']]\ndf_mean_compiled[['cristae_Volume3d_µm^3', 'cristae_Area3d_µm^2']] = df_cristae_mean_grouped[['Volume3d_µm^3', 'Area3d_µm^2']]\ndf_mean_compiled[['ER_Volume3d_µm^3', 'ER_Area3d_µm^2']] = df_ER_mean_grouped[['Volume3d_µm^3', 'Area3d_µm^2']]\n\ndf_inclusion_mean_tmp = df_inclusion_mean_grouped[['Volume3d_µm^3', 'Area3d_µm^2']]\ndf_inclusion_mean_fill = pd.DataFrame([[0, 0]], columns = ['Volume3d_µm^3', 'Area3d_µm^2'])\ndf_inclusion_mean_tmp = df_inclusion_mean_fill.append(df_inclusion_mean_tmp, ignore_index = True)\n\ndf_mean_compiled[['inclusion_Volume3d_µm^3', 'inclusion_Area3d_µm^2']] = df_inclusion_mean_tmp\ndf_mean_compiled", "_____no_output_____" ], [ "fig, ax = plt.subplots(nrows=2, ncols=2, figsize=(10, 10))\nidx = 0\nfor i in range(2):\n for j in range(2):\n ax[i, j].bar(df_mean_compiled.index, \n df_mean_compiled.iloc[:, idx +2], \n tick_label=['0', '0', '7', '7', '7', '14', '21'])\n ax[i, j].set_title(df_mean_compiled.columns[idx+2])\n ax[i, j].set_xlabel('Day')\n idx += 1\n # ax[i].set_ylabel('Total Volume ($µm^3$)')\n \nfig.tight_layout(pad=3.0)\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'mito_cristae_mean_volume_area.png'))\nplt.show()", "_____no_output_____" ], [ "fig, ax = plt.subplots(nrows=2, ncols=2, figsize=(10, 10))\nidx = 4\nfor i in range(2):\n for j in range(2):\n ax[i, j].bar(df_mean_compiled.index, \n df_mean_compiled.iloc[:, idx +2], \n tick_label=['0', '0', '7', '7', '7', '14', '21'])\n ax[i, j].set_title(df_mean_compiled.columns[idx+2])\n ax[i, j].set_xlabel('Day')\n idx += 1\n # ax[i].set_ylabel('Total Volume ($µm^3$)')\n \nfig.tight_layout(pad=3.0)\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'ER_inclusion_mean_volume_area.png'))\nplt.show()", "_____no_output_____" ], [ "'''\ndf_mean_compiled_normalized = pd.DataFrame()\ndf_mean_compiled_normalized[['filename', 'day']] = df_cell_membrane_sum_grouped[['filename', 'day']]\ncal_tmp = df_mean_compiled.iloc[:, 2:].div(df_cyto['Volume3d_µm^3'], axis=0)\ndf_mean_compiled_normalized = pd.concat([df_mean_compiled_normalized, cal_tmp], axis=1)\ndf_mean_compiled_normalized\n'''", "_____no_output_____" ], [ "'''\nfig, ax = plt.subplots(nrows=8, ncols=1, figsize=(5, 30))\n\nfor i in range(8):\n ax[i].bar(df_mean_compiled_normalized.index, \n df_mean_compiled_normalized.iloc[:, i +2], \n tick_label=['0', '0', '0', '7', '7', '7', '14', '17', '17', '21'])\n ax[i].set_title(df_mean_compiled_normalized.columns[i+2])\n ax[i].set_xlabel('Day')\n \nfig.tight_layout(pad=3.0)\nplt.show()\n'''", "_____no_output_____" ] ], [ [ "### 02-06 Distribution", "_____no_output_____" ] ], [ [ "# mito\nmaxval = df_mito['Volume3d_µm^3'].max()\nminval = df_mito['Volume3d_µm^3'].min()\nprint(maxval)\nprint(minval)\nbins = np.linspace(minval + (maxval - minval)* 0, minval + (maxval - minval)* 1, num = 100)\n# bins = np.linspace(500000000, minval + (maxval - minval)* 1, num = 50)\ndays = [0, 7, 14, 21]\n\nnrows = 4\nncols = 1\nfig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=(10, 15))\n\nfor i, day in enumerate(days):\n df_tmp = df_mito.loc[df_mito['day'] == day, :]\n axes[i%nrows].hist(df_tmp['Volume3d_µm^3'], bins= bins, log=True, density = True)\n axes[i%nrows].set_xlim([0, maxval])\n axes[i%nrows].set_ylim([0, 1])\n axes[i%nrows].set_title('Day ' + str(day))\n \nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'distribution_mito_volume.png'))\nplt.show()", "46.7896\n0.005008\n" ], [ "# cristae\nmaxval = df_cristae['Area3d_µm^2'].max()\nminval = df_cristae['Area3d_µm^2'].min()\nprint(maxval)\nprint(minval)\nbins = np.linspace(minval + (maxval - minval)* 0, minval + (maxval - minval)* 1, num = 100)\n# bins = np.linspace(500000000, minval + (maxval - minval)* 1, num = 50)\ndays = [0, 7, 14, 21]\n\nnrows = 4\nncols = 1\nfig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=(10, 15))\n\nfor i, day in enumerate(days):\n df_tmp = df_cristae.loc[df_cristae['day'] == day, :]\n axes[i%nrows].hist(df_tmp['Area3d_µm^2'], bins= bins, log=True, density = True)\n axes[i%nrows].set_xlim([0, maxval])\n axes[i%nrows].set_ylim([0, 0.1])\n axes[i%nrows].set_title('Day ' + str(day))\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'distribution_cristae_volume.png'))\nplt.show()", "1311.66\n0.0003004190000000001\n" ], [ "# ER\nmaxval = df_ER['Volume3d_µm^3'].max()\nminval = df_ER['Volume3d_µm^3'].min()\nprint(maxval)\nprint(minval)\nfactor = 0.03\n\nbins = np.linspace(minval + (maxval - minval)* 0, minval + (maxval - minval)* factor, num = 100)\ndays = [0, 7, 14, 21]\n\n\nnrows = 4\nncols = 1\nfig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=(10, 15))\n\nfor i, day in enumerate(days):\n df_tmp = df_ER.loc[df_ER['day'] == day, :]\n axes[i%nrows].hist(df_tmp['Volume3d_µm^3'], bins= bins, log=True, density = True)\n axes[i%nrows].set_xlim([0, minval + (maxval - minval)* factor])\n axes[i%nrows].set_ylim([0, 100])\n axes[i%nrows].set_title('Day ' + str(day))\n \nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'distribution_ER_volume.png'))\nplt.show()", "86.4886\n1e-06\n" ], [ "# inclusion\nmaxval = df_inclusion['Volume3d_µm^3'].max()\nminval = df_inclusion['Volume3d_µm^3'].min()\nprint(maxval)\nprint(minval)\nfactor = 1\n\nbins = np.linspace(minval + (maxval - minval)* 0, minval + (maxval - minval)* factor, num = 100)\ndays = [0, 7, 14, 21]\n\nnrows = 4\nncols = 1\nfig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=(10, 15))\n\nfor i, day in enumerate(days):\n df_tmp = df_inclusion.loc[df_inclusion['day'] == day, :]\n axes[i%nrows].hist(df_tmp['Volume3d_µm^3'], bins= bins, log=True, density = True)\n axes[i%nrows].set_xlim([0, minval + (maxval - minval)* factor])\n axes[i%nrows].set_ylim([0, 1])\n axes[i%nrows].set_title('Day ' + str(day))\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'distribution_inclusion_volume.png'))\nplt.show()", "261.036\n1e-06\n" ] ], [ [ "# 03 Load Data from Auto Skeletonization of Mitocondria", "_____no_output_____" ], [ "## 03-01", "_____no_output_____" ] ], [ [ "mainpath = 'D:\\PerlmutterData'\nfolder = 'segmentation_compiled_export'\ndata_folder = 'data'\n\npath = os.path.join(mainpath, folder, data_folder)\nprint(path)\n\nfolders = ['skeleton_output']\nsubcat = ['nodes', 'points', 'segments_s']", "D:\\PerlmutterData\\segmentation_compiled_export\\data\n" ], [ "target_list = glob.glob(os.path.join(path, 'compile', '*.csv'))\ntarget_list = [os.path.basename(x) for x in target_list]\ntarget_list = [os.path.splitext(x)[0] for x in target_list]\nprint(target_list)", "['cell_membrane', 'cristae', 'ER', 'inclusion', 'mito', 'nodes', 'nucleus', 'points', 'segments_s']\n" ], [ "file_meta = {\n 'data_d00_batch01_loc01': 0,\n 'data_d00_batch02_loc02': 0,\n 'data_d00_batch02_loc03': 0,\n 'data_d07_batch01_loc01': 7, \n 'data_d07_batch02_loc01': 7,\n 'data_d07_batch02_loc02': 7, \n 'data_d14_batch01_loc01': 14, \n 'data_d17_batch01_loc01': 17,\n 'data_d21_batch01_loc01': 21,\n}", "_____no_output_____" ], [ "for i in subcat:\n file_list = glob.glob(os.path.join(path, 'raw', 'skeleton_output', '*', i + '.csv'))\n # print(file_list)\n \n if not i in target_list:\n df = pd.DataFrame()\n\n for j in file_list: \n data_temp = pd.read_csv(j, header = 0)\n \n foldername_tmp = os.path.dirname(j)\n foldername_tmp = os.path.basename(foldername_tmp)\n \n \n # add day\n pattern = re.compile(\"data_d[0-9][0-9]_batch[0-9][0-9]_loc[0-9][0-9]\")\n original_foldername = pattern.search(foldername_tmp).group(0)\n day_tmp = file_meta[original_foldername]\n data_temp['day'] = day_tmp\n # add filename\n data_temp['filename'] = original_foldername\n \n df = df.append(data_temp, ignore_index = True)\n \n display(df)\n df.to_csv(os.path.join(path, 'compile', i + '.csv'))", "_____no_output_____" ], [ "df_points = pd.read_csv(os.path.join(path, 'compile', 'points' + '.csv'))\ndf_segments = pd.read_csv(os.path.join(path, 'compile', 'segments_s' + '.csv'))\ndf_nodes = pd.read_csv(os.path.join(path, 'compile', 'nodes' + '.csv'))", "_____no_output_____" ], [ "df_points", "_____no_output_____" ], [ "for omit in omit_data: \n df_points = df_points.loc[df_points['filename']!= omit]\n df_segments = df_segments.loc[df_segments['filename']!=omit]\n df_nodes = df_nodes.loc[df_nodes['filename']!=omit]", "_____no_output_____" ], [ "# points\nmaxval = df_points['thickness'].max()\nminval = df_points['thickness'].min()\ndays = [0, 7, 14, 21]\nbins = np.linspace(minval + (maxval - minval)* 0, minval + (maxval - minval)* 1, num = 20)\n\nnrows = 4\nncols = 1\nfig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=(10, 15))\n\nfor i, day in enumerate(days):\n df_tmp = df_points.loc[df_points['day'] == day, :]\n axes[i%nrows].hist(df_tmp['thickness'], bins= bins, log=False, density = True)\n axes[i%nrows].set_ylim([0, 0.008])\n axes[i%nrows].set_title('Day ' + str(day))\n \nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'distribution_points_thickness.png'))\nplt.show()", "_____no_output_____" ], [ "# segments\nmaxval = df_segments['thickness'].max()\nminval = df_segments['thickness'].min()\ndays = [0, 7, 14, 21]\nbins = np.linspace(minval + (maxval - minval)* 0, minval + (maxval - minval)* 1, num = 20)\n\nnrows = 4\nncols = 1\nfig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=(10, 15))\n\nfor i, day in enumerate(days):\n df_tmp = df_segments.loc[df_segments['day'] == day, :]\n axes[i%nrows].hist(df_tmp['thickness'], bins= bins, log=False, density = True)\n axes[i%nrows].set_ylim([0, 0.008])\n axes[i%nrows].set_title('Day ' + str(day))\n \nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'distribution_segments_thickness.png'))\nplt.show()", "_____no_output_____" ], [ "df_segments_count_grouped = df_nodes.groupby(['day', 'filename', 'Coordination Number']).count().reset_index()\ndf_segments_count_grouped", "_____no_output_____" ], [ "filename = df_segments_count_grouped['filename'].unique()\nprint(filename)", "['data_d00_batch01_loc01' 'data_d00_batch02_loc03'\n 'data_d07_batch01_loc01' 'data_d07_batch02_loc01'\n 'data_d07_batch02_loc02' 'data_d14_batch01_loc01'\n 'data_d17_batch01_loc01' 'data_d21_batch01_loc01']\n" ], [ "days = [0, 7, 14, 21]\n\nnrows = 4\nncols = 1\nfig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=(10, 15))\n\nfor i, day in enumerate(days):\n df_tmp = df_segments_count_grouped.loc[df_segments_count_grouped['day'] == day]\n \n x = df_tmp['Coordination Number']\n y = df_tmp['Node ID']\n x_pos = [str(i) for i in x]\n \n axes[i%nrows].bar(x_pos, y)\n axes[i%nrows].set_title('Day ' + str(day))\n \nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'coordination_number.png'))\nplt.show()", "_____no_output_____" ] ], [ [ "# 04 Average of total size", "_____no_output_____" ] ], [ [ "mito_mean = df_mito_sum_grouped[['day', 'Volume3d_µm^3']].groupby(['day']).mean().reset_index()\nmito_sem = df_mito_sum_grouped[['day', 'Volume3d_µm^3']].groupby(['day']).sem().reset_index()\nmito_sem = mito_sem.fillna(0)\n\nmito_mean.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'mito_mean.csv'))\nmito_sem.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'mito_sem.csv'))\n\nfig = plt.figure(figsize=(5, 5))\nx = ['Day 0', 'Day 7', 'Day 14', 'Day 21']\nplt.bar(x, mito_mean['Volume3d_µm^3'], yerr= mito_sem['Volume3d_µm^3'], error_kw=dict(capsize=10))\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'mito_mean_barplot.png'))\nplt.show()", "_____no_output_____" ], [ "cristae_mean = df_cristae_sum_grouped[['day', 'Volume3d_µm^3']].groupby(['day']).mean().reset_index()\ncristae_sem = df_cristae_sum_grouped[['day', 'Volume3d_µm^3']].groupby(['day']).sem().reset_index()\ncristae_sem = cristae_sem.fillna(0)\n\ncristae_mean.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'cristae_mean.csv'))\ncristae_sem.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'cristae_sem.csv'))\n\nfig = plt.figure(figsize=(5, 5))\nx = ['Day 0', 'Day 7', 'Day 14', 'Day 21']\nplt.bar(x, cristae_mean['Volume3d_µm^3'], yerr= cristae_sem['Volume3d_µm^3'], error_kw=dict(capsize=10))\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'cristae_mean_barplot.png'))\nplt.show()", "_____no_output_____" ], [ "ER_mean = df_ER_sum_grouped[['day', 'Volume3d_µm^3']].groupby(['day']).mean().reset_index()\nER_sem = df_ER_sum_grouped[['day', 'Volume3d_µm^3']].groupby(['day']).sem().reset_index()\nER_sem = ER_sem.fillna(0)\n\nER_mean.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'ER_mean.csv'))\nER_sem.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'ER_sem.csv'))\n\nfig = plt.figure(figsize=(5, 5))\nx = ['Day 0', 'Day 7', 'Day 14', 'Day 21']\nplt.bar(x, ER_mean['Volume3d_µm^3'], yerr= ER_sem['Volume3d_µm^3'], error_kw=dict(capsize=10))\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'ER_mean_barplot.png'))\nplt.show()", "_____no_output_____" ], [ "inclusion_mean = df_inclusion_sum_grouped[['day', 'Volume3d_µm^3']].groupby(['day']).mean().reset_index()\ninclusion_sem = df_inclusion_sum_grouped[['day', 'Volume3d_µm^3']].groupby(['day']).sem().reset_index()\ninclusion_sem = inclusion_sem.fillna(0)\n\ninclusion_mean.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'inclusion_mean.csv'))\ninclusion_sem.to_csv(os.path.join(mainpath, folder, data_folder, 'spreadsheet', 'inclusion_sem.csv'))\n\nfig = plt.figure(figsize=(5, 5))\nx = ['Day 0', 'Day 7', 'Day 14', 'Day 21']\nplt.bar(x, inclusion_mean['Volume3d_µm^3'], yerr= inclusion_sem['Volume3d_µm^3'], error_kw=dict(capsize=10))\n\nplt.savefig(os.path.join(mainpath, folder, data_folder, 'plots', 'inclusion_mean_barplot.png'))\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", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Harvesting data from Home\n\nThis is an example of how my original recipe for [harvesting data from The Bulletin](Harvesting-data-from-the-Bulletin.ipynb) can be modified for other journals.\n\nIf you'd like a pre-harvested dataset of all the Home covers (229 images in a 3.3gb zip file), open this link using your preferred BitTorrent client: <magnet:?xt=urn:btih:7888BCEA44E5FF5670931A3394369E5018BFC32B&dn=home-quarterly.zip>", "_____no_output_____" ] ], [ [ "# Let's import the libraries we need.\nimport requests\nfrom bs4 import BeautifulSoup\nimport time\nimport json\nimport os\nimport re", "_____no_output_____" ], [ "# Create a directory for this journal\n# Edit as necessary for a new journal\ndata_dir = '../../data/Trove/Home'\nos.makedirs(data_dir, exist_ok=True)", "_____no_output_____" ] ], [ [ "## Getting the issue data\n\nEach issue of a digitised journal like has it's own unique identifier. You've probably noticed them in the urls of Trove resources. They look something like this `nla.obj-362409353`. Once we have the identifier for an issue we can easily download the contents, but how do we get a complete list of identifiers?\n\nThe [harvesting data from the Bulletin](Harvesting-data-from-the-Bulletin.ipynb) notebook explains how we can find a url that lists all the available issues of a journal.\n\nThis is the url we need to start harvesting issue metadata about *Home*. You could easily modify this to get metadata from another journal by changing the identifier.\n\n```\nhttps://nla.gov.au/nla.obj-362409353/browse?startIdx=0&rows=20&op=c\n```", "_____no_output_____" ] ], [ [ "# This is just the url we found above, with a slot into which we can insert the startIdx value\n# If you want to download data from another journal, just change the nla.obj identifier to point to the journal.\nstart_url = 'https://nla.gov.au/nla.obj-362409353/browse?startIdx={}&rows=20&op=c'", "_____no_output_____" ], [ "# The initial startIdx value\nstart = 0\n# Number of results per page\nn = 20\nissues = []\n# If there aren't 20 results on the page then we've reached the end, so continue harvesting until that happens.\nwhile n == 20:\n # Get the browse page\n response = requests.get(start_url.format(start))\n # Beautifulsoup turns the HTML into an easily navigable structure\n soup = BeautifulSoup(response.text, 'lxml')\n # Find all the divs containing issue details and loop through them\n details = soup.find_all(class_='l-item-info')\n for detail in details:\n issue = {}\n # Get the issue id\n issue['id'] = detail.dt.a.string\n rows = detail.find_all('dd')\n # Get the issue details\n issue['details'] = rows[2].p.string\n # Get the number of pages\n issue['pages'] = re.search(r'^(\\d+)', detail.find('a', class_=\"browse-child\").text, flags=re.MULTILINE).group(1)\n issues.append(issue)\n print(issue)\n time.sleep(0.2)\n # Increment the startIdx\n start += n\n # Set n to the number of results on the current page\n n = len(details)\n \n ", "_____no_output_____" ], [ "len(issues)", "_____no_output_____" ], [ "# Save the harvested results as a JSON file in case we need them later on\nwith open('{}/home_issues.json'.format(data_dir), 'w') as outfile:\n json.dump(issues, outfile)", "_____no_output_____" ], [ "# Open the saved JSON file\nwith open('{}/home_issues.json'.format(data_dir), 'r') as infile:\n issues = json.load(infile)", "_____no_output_____" ] ], [ [ "## Cleaning up the metadata\n\nSo far we've just grabbed the complete issue details as a single string. It would be good to parse this string so that we have the dates, volume and issue numbers in separate fields. As is always the case, there's a bit of variation in the way this information is recorded. The code below tries out different combinations and then saves the structured data in a Python list.\n\nI had to modify the code I used with the *Bulletin* due to slight variations in the way the issue data was recorded. For example, issue dates for *Home* use the full names of months, while the *Bulletin* records used abbreviations. It's likely that there will be other variations between journals, so you might have to adjust this code. ", "_____no_output_____" ] ], [ [ "\nimport arrow\nfrom arrow.parser import ParserError\nissues_data = []\n# Loop through the issues\nfor issue in issues:\n issue_data = {}\n issue_data['id'] = issue['id']\n issue_data['pages'] = int(issue['pages'])\n print(issue['details'])\n try:\n # This pattern looks for details in the form: Vol. 2 No. 3 (2 Jul 1878)\n details = re.search(r'(.*)Vol. (\\d+) No\\.* (\\d+) \\((.+)\\)', issue['details'].strip())\n issue_data['label'] = details.group(1).strip()\n issue_data['volume'] = details.group(2)\n issue_data['number'] = details.group(3)\n date = details.group(4)\n except AttributeError:\n try:\n # This pattern looks for details in the form: No. 3 (2 Jul 1878)\n details = re.search(r'No. (\\d+) \\((.+)\\)', issue['details'].strip())\n issue_data['label'] = ''\n issue_data['volume'] = ''\n issue_data['number'] = details.group(1)\n date = details.group(2)\n except AttributeError:\n try:\n # This pattern looks for details in the form: Bulletin Christmas Edition (2 Jul 1878)\n details = re.search(r'(.*) \\((.+)\\)', issue['details'].strip())\n issue_data['label'] = details.group(1)\n issue_data['volume'] = ''\n issue_data['number'] = ''\n date = details.group(2)\n except AttributeError:\n # This pattern looks for details in the form: Bulletin 1878 Jul 3\n details = re.search(r'Bulletin (.+)', issue['details'].strip())\n date_str = details.group(1)\n # Date is wrong way round, split and reverse\n date = ' '.join(reversed(date_str.split()))\n issue_data['label'] = ''\n issue_data['volume'] = ''\n issue_data['number'] = ''\n # Normalise months\n date = date.replace('Sept', 'Sep').replace('Sepember', 'September').replace('July August', 'July').replace('September October', 'September').replace(' ', ' ')\n # Convert date to ISO format\n try:\n issue_data['date'] = arrow.get(date, 'D MMMM YYYY').isoformat()[:-15]\n except ParserError:\n issue_data['date'] = arrow.get(date, 'D MMM YYYY').isoformat()[:-15]\n issues_data.append(issue_data)\n ", "_____no_output_____" ] ], [ [ "## Save as CSV\n\nNow the issues data is in a nice, structured form, we can load it into a Pandas dataframe. This allows us to do things like find the total number of pages digitised.\n\nWe can also save the metadata as a CSV.", "_____no_output_____" ] ], [ [ "import pandas as pd\n# Convert issues metadata into a dataframe\ndf = pd.DataFrame(issues_data, columns=['id', 'label', 'volume', 'number', 'date', 'pages'])", "_____no_output_____" ], [ "# Find the total number of pages\ndf['pages'].sum()", "_____no_output_____" ], [ "# Save metadata as a CSV.\ndf.to_csv('{}/home_issues.csv'.format(data_dir), index=False)", "_____no_output_____" ] ], [ [ "## Download front covers\n\nOptions for downloading images, PDFs and text are described in the [harvesting data from the Bulletin](Harvesting-data-from-the-Bulletin.ipynb) notebook. In this recipe we'll just download the fromt covers (because they're awesome).\n\nThe code below checks to see if an image has already been saved before downloading it, so if the process is interrupted you can just run it again to pick up where it stopped. If more issues are added to Trove you could run it again to pick up any new images.", "_____no_output_____" ] ], [ [ "import zipfile\nimport io\n# Prepare a directory to save the images into\noutput_dir = data_dir + '/images'\nos.makedirs(output_dir, exist_ok=True)\n# Loop through the issue metadata\nfor issue in issues_data:\n print(issue['id'])\n id = issue['id']\n # Check to see if the first page of this issue has already been downloaded\n if not os.path.exists('{}/{}-1.jpg'.format(output_dir, id)):\n url = 'https://nla.gov.au/{}/download?downloadOption=zip&firstPage=0&lastPage=0'.format(id)\n # Get the file\n r = requests.get(url)\n # The image is in a zip, so we need to extract the contents into the output directory\n z = zipfile.ZipFile(io.BytesIO(r.content))\n z.extractall(output_dir)\n time.sleep(1)", "_____no_output_____" ] ] ]

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

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# \"Old Skool Image Classification\"\n> \"A blog on how to manuallly create features from an Image for classification task.\"\n\n- toc: true\n- branch: master\n- badges: true\n- comments: false\n- categories: [CV, image classification, feature engineering, pyTorch, CIFAR10]\n- image: images/blog1.png\n- hide: false\n- search_exclude: true", "_____no_output_____" ], [ "## Introduction\n\nThe objective of the current notebook is to give a glimpse of some of the methods for feature extraction that were prevelent before the advent of Deep Neural Networks in the Computer Vision domain.\n\nIn the current Notebook we shall see how to do the same using the Python language with CIFAR10 dataset. The goal is to extract several features from the provided images and finally perform Image Classification using a Multi Layer Perceptron.\n\n## Texture Analysis\n\nOn the preface of the book _Image Processing: Dealing with Textures_ {% fn 1 %}\n, the authors provided a very captivating definition of Texture: __Texture is what makes life beautiful; texture is what makes life interesting and texture is what makes life possible. Texture is what makes Mozart’s music beautiful, the masterpieces of the art of the Renaissance classical and the facades of Barcelona’s buildings attractive.\"__.(Not so helpful eh!) \n\nSo, what is Texture? Technically, its the 'variation of Data on a smaller scale than the scale of interest'.\n\nClassical Visual Computing comprises of two main branches when it comes to analyzing Texture of an Image [[1]](#1):\n1. Structural \n - Local Binary Pattern\n - Gabor Wavelets\n - Fourier Co-efficients\n2. Statistical \n - Co-Occurance Matrix\n - Orientation Histogram\n\n\n## Workflow\n\n- We shall start by downloading the [CIFAR-10](https://www.cs.toronto.edu/~kriz/cifar.html) dataset.\n\n- Next, we manually craft the features to obtain Texture Metrics.\n\n- After creating functions to obtain textual features from an image, we create a loop to extract the same from all the images in Training and Test dataset.\n\n- Next, we save the Training and Test set extracted features as serialized file. \n\n- Then we shall use the created features as co-variates against the label for each image and train a Softmax classifier on the Training Set. \n\n- Eventually, we evaluate the classifier on the Test set. ", "_____no_output_____" ] ], [ [ "%matplotlib inline\n\nfrom torchvision import datasets\nimport PIL\nfrom skimage.feature import local_binary_pattern, greycomatrix, greycoprops\nfrom skimage.filters import gabor\n\nimport torch\nfrom torch import nn \nfrom torch.utils.data import TensorDataset, DataLoader\nimport torch.nn.functional as F\nimport numpy as np\n\nimport matplotlib.pyplot as plt\nimport tqdm\nfrom tqdm import notebook\n\nfrom pathlib import Path\nimport pickle\nimport time", "_____no_output_____" ] ], [ [ "### Data Loading", "_____no_output_____" ] ], [ [ "#collapse\n#collapse-output\ntrainDset = datasets.CIFAR10(root=\"./cifar10/\", train=True, download=True)\ntestDset = datasets.CIFAR10(root = \"./cifar10/\", train=False, download=True)", "Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./cifar10/cifar-10-python.tar.gz\n" ], [ "# Looking at a single image", "_____no_output_____" ], [ "#collapse\nimg = trainDset[0][0] # PIL Image\nimg_grey = img.convert('L') # convert to Grey-scale\nimg_arr = np.array(img_grey) # convert to numpy array\nplt.imshow(img)", "_____no_output_____" ] ], [ [ "## Local Binary Patterns(LBP)\n\nLBP is helpful in extracting \"local\" structure of the image. It does so by encoding the local neighbourhood after they have been maximally simplified, i.e. binarized. In case, we want to perform LBP on a coloured image, we need to do so individually on each channel(Red/Blue/Green).", "_____no_output_____" ] ], [ [ "#collapse\nfeat_lbp = local_binary_pattern(img_arr, 8,1,'uniform')\nfeat_lbp = np.uint8( (feat_lbp/feat_lbp.max())*255) # converting to unit 8\nlbp_img = PIL.Image.fromarray(feat_lbp) # Convert from array\n\nplt.imshow(lbp_img, cmap = 'gray')", "_____no_output_____" ], [ "# Energy, Entropy\n\ndef get_lbp(img):\n \"\"\"Function to implement Local Binary Pattern\"\"\"\n lbp_hist, _ = np.histogram(img, 8)\n lbp_hist = np.array(lbp_hist, dtype = float)\n lbp_prob = np.divide(lbp_hist, np.sum(lbp_hist))\n lbp_prob = np.where(np.isclose(0, lbp_prob), 0.0000000001, lbp_prob) # to avoid log(0)\n lbp_energy = np.sum(lbp_prob**2)\n lbp_entropy = -np.sum(np.multiply(lbp_prob, np.log2(lbp_prob)))\n return lbp_energy, lbp_entropy", "\n" ] ], [ [ "## Co Occurence Matrix\n\nIntiutively, if we were to extract information of a pixel in an image and also record its neighbouring pixels and their intensities, we will be able to capture both spatial and relative information. This is where Co-Occurance matrix are useful. They extract the representation of joint probability of chosen set of pixels having certain values. \n\nOnce we have the co-occurance matrix, we can start calculating the feature matrics such as:\n- $\\textbf{Energy} = \\sum_{m=0}^{G-1}\\sum_{n=0}^{G-1}p^2\\left(m,n\\right)$\n\n- $\\textbf{Entropy} = \\sum_{m=0}^{G-1}\\sum_{n=0}^{G-1}p\\left(m,n\\right)\\cdot log \\left(p\\left(m,n\\right)\\right)$\n\n- $\\textbf{Contrast} = \\frac{1}{(G-1)^2}\\sum_{m=0}^{G-1}\\sum_{n=0}^{G-1}(m-n)^2\\cdot p(m,n)$\n- $\\textbf{Homogeneity} = \\sum_{m=0}^{G-1}\\sum_{n=0}^{G-1} \\frac{p(m,n)}{1+|m-n|}$\n\nWhere, $m,n$ are the neighbouring pixels and $G$ is the total number of grey levels we use. $G=256$ for an 8-bit gray-scale image", "_____no_output_____" ] ], [ [ "def creat_cooccur(img_arr, *args, **kwargs):\n \"\"\"Implements extraction of features from Co-Occurance Matrix\"\"\"\n gCoMat = greycomatrix(img_arr, [2], [0], 256, symmetric=True, normed=True)\n contrast = greycoprops(gCoMat, prop='contrast')\n dissimilarity = greycoprops(gCoMat, prop='dissimilarity')\n homogeneity = greycoprops(gCoMat, prop='homogeneity')\n energy = greycoprops(gCoMat, prop='energy')\n correlation = greycoprops(gCoMat, prop = 'correlation')\n return contrast[0][0], dissimilarity[0][0], homogeneity[0][0], energy[0][0], correlation[0][0]", "_____no_output_____" ], [ "#collapse\ngCoMat = greycomatrix(img_arr, [2], [0], 256, symmetric=True, normed=True)\ncontrast = greycoprops(gCoMat, prop='contrast')\ndissimilarity = greycoprops(gCoMat, prop='dissimilarity')\nhomogeneity = greycoprops(gCoMat, prop='homogeneity')\nenergy = greycoprops(gCoMat, prop='energy')\ncorrelation = greycoprops(gCoMat, prop = 'correlation')\nprint(energy[0][0])", "_____no_output_____" ] ], [ [ "## [Gabor Filter](https://en.wikipedia.org/wiki/Gabor_filter#Applications_of_2-D_Gabor_filters_in_image_processing)\n\n", "_____no_output_____" ] ], [ [ "gf_real, gf_img = gabor(img_arr, frequency=0.6)\ngf =(gf_real**2 + gf_img**2)//2\n# Displaying the filter response\nfig, ax = plt.subplots(1,3) \nax[0].imshow(gf_real,cmap='gray')\nax[1].imshow(gf_img,cmap='gray')\nax[2].imshow(gf,cmap='gray')", "_____no_output_____" ], [ "def get_gabor(img, N, *args, **kwargs):\n \"\"\"Gabor Feature extraction\"\"\"\n gf_real, gf_img = gabor(img_arr, frequency=0.6)\n gf =(gf_real**2 + gf_img**2)//2\n gabor_hist, _ = np.histogram(gf, N)\n gabor_hist = np.array(gabor_hist, dtype = float)\n gabor_prob = np.divide(gabor_hist, np.sum(gabor_hist))\n # To discard pixels resulting in 0 probability\n gabor_prob = np.where(np.isclose(0, gabor_prob), 0.0000000001, gabor_prob)\n gabor_energy = np.sum(gabor_prob**2)\n gabor_entropy = np.sum(np.multiply(gabor_prob, np.log2(gabor_prob)))\n return gabor_energy, gabor_entropy", "_____no_output_____" ] ], [ [ "## Feature Extraction", "_____no_output_____" ] ], [ [ "# Generate Training Data\n# Extract features from all images\n\nlabel = []\n\nfeatLength = 2+5+2 # LBP, Co-occurance, Gabor\ntrainFeats = np.zeros((len(trainDset), featLength))\ntestFeats = np.zeros((len(testDset), featLength))", "_____no_output_____" ], [ "label = [trainDset[tr][1] for tr in tqdm.tqdm_notebook(range(len(trainFeats)))]\n\ntrainLabel = np.array(label)", "_____no_output_____" ], [ "for tr in tqdm.tqdm_notebook(range(len(trainFeats))):\n img = trainDset[tr][0]\n img_grey = img.convert('L')\n img_arr = np.array(img_grey.getdata()).reshape(img.size[1], img.size[0])\n # LBP \n feat_lbp = local_binary_pattern(img_arr, 5,2,'uniform').reshape(img.size[0]*img.size[1])\n feat_lbp = np.uint8((feat_lbp/feat_lbp.max())*255) # converting to unit 8\n lbp_energy, lbp_entropy = get_lbp(feat_lbp)\n # Co-Occurance\n gCoMat = greycomatrix(img_arr, [2], [0], 256, True,True)\n featglcm = np.array(creat_cooccur(img_arr))\n # Gabor\n gabor_energy, gabor_entropy = get_gabor(img_arr, 8)\n\n # Concat features\n concat_feat = np.concatenate(([lbp_energy, lbp_entropy], featglcm, [gabor_energy, gabor_entropy]), axis=0)\n trainFeats[tr,:] = concat_feat\n label.append(trainDset[tr][1])\ntrainLabel = np.array(label)", "_____no_output_____" ], [ "label = []\n\nfor ts in tqdm.tqdm_notebook(range(len(testDset))):\n img = testDset[ts][0]\n img_grey = img.convert('L')\n img_arr = np.array(img_grey.getdata()).reshape(img.size[1], img.size[0])\n # LBP \n feat_lbp = local_binary_pattern(img_arr, 5,2,'uniform').reshape(img.size[0]*img.size[1])\n lbp_energy, lbp_entropy = get_lbp(feat_lbp)\n # Co-Occurance\n gCoMat = greycomatrix(img_arr, [2], [0], 256, True,True)\n featglcm = np.array(creat_cooccur(img_arr))\n # Gabor\n gabor_energy, gabor_entropy = get_gabor(img_arr, 8)\n\n # Concat features\n concat_feat = np.concatenate(([lbp_energy, lbp_entropy], featglcm, [gabor_energy, gabor_entropy]), axis=0)\n\n testFeats[ts,:] = concat_feat\n label.append(testDset[ts][1]) \n\ntestLabel = np.array(label)", "_____no_output_____" ] ], [ [ "### Normalize Features", "_____no_output_____" ] ], [ [ "# Normalizing the train features to the range [0,1]\ntrMaxs = np.amax(trainFeats,axis=0) #Finding maximum along each column\ntrMins = np.amin(trainFeats,axis=0) #Finding maximum along each column\ntrMaxs_rep = np.tile(trMaxs,(50000,1)) #Repeating the maximum value along the rows\ntrMins_rep = np.tile(trMins,(50000,1)) #Repeating the minimum value along the rows\ntrainFeatsNorm = np.divide(trainFeats-trMins_rep,trMaxs_rep) #Element-wise division\n# Normalizing the test features\ntsMaxs_rep = np.tile(trMaxs,(10000,1)) #Repeating the maximum value along the rows\ntsMins_rep = np.tile(trMins,(10000,1)) #Repeating the maximum value along the rows\ntestFeatsNorm = np.divide(testFeats-tsMins_rep,tsMaxs_rep) #Element-wise division", "_____no_output_____" ] ], [ [ "### Save Data", "_____no_output_____" ] ], [ [ "with open(\"TrainFeats.pckl\", \"wb\") as f:\n pickle.dump(trainFeatsNorm, f)\nwith open(\"TrainLabel.pckl\", \"wb\") as f:\n pickle.dump(trainLabel, f)\n\nwith open(\"TestFeats.pckl\", \"wb\") as f:\n pickle.dump(testFeatsNorm, f)\nwith open(\"TestLabel.pckl\", \"wb\") as f:\n pickle.dump(testLabel, f)\nprint(\"files Saved!\")", "_____no_output_____" ] ], [ [ "## Classification with SoftMax Regression", "_____no_output_____" ], [ "### Data Preparation\n", "_____no_output_____" ] ], [ [ "##########################\n### SETTINGS\n##########################\n\n# Device\nDEVICE = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n\n# Hyperparameters\nrandom_seed = 123\nlearning_rate = 0.01\nnum_epochs = 100\nbatch_size = 64\n\n# Architecture\nnum_features = 9\nnum_classes = 10\n\n##########################\n### CIFAR10 DATASET\n##########################\n## Converting Numpy array to Torch-Tensor\n\ntrainLabels = torch.from_numpy(trainLabel)\ntrainDataset = TensorDataset(torch.from_numpy(trainFeats),trainLabels)\n\ntestLabels = torch.from_numpy(testLabel)\ntestDataset = TensorDataset(torch.from_numpy(testFeats), testLabels)\n\n## Creating DataLoader\n\ntrain_loader = DataLoader(trainDataset, batch_size=batch_size, shuffle=True)\n\ntest_loader = DataLoader(testDataset,batch_size=batch_size,shuffle=False)", "_____no_output_____" ] ], [ [ "### Define Model", "_____no_output_____" ] ], [ [ "##########################\n### MODEL\n##########################\n\nclass SoftmaxRegression(torch.nn.Module):\n\n def __init__(self, num_features, num_classes):\n super(SoftmaxRegression, self).__init__()\n self.linear = torch.nn.Linear(num_features, num_classes)\n \n # self.linear.weight.detach().zero_()\n # self.linear.bias.detach().zero_()\n \n def forward(self, x):\n logits = self.linear(x)\n probas = F.softmax(logits, dim=1)\n return logits, probas\n\nmodel = SoftmaxRegression(num_features=num_features,\n num_classes=num_classes)\n\nmodel.to(DEVICE)\n\n##########################\n### COST AND OPTIMIZER\n##########################\n\noptimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)", "_____no_output_____" ] ], [ [ "### Define Training Route", "_____no_output_____" ] ], [ [ "# Manual seed for deterministic data loader\ntorch.manual_seed(random_seed)\n\n\ndef compute_accuracy(model, data_loader):\n correct_pred, num_examples = 0, 0\n \n for features, targets in data_loader:\n features = features.float().view(-1, 9).to(DEVICE)\n targets = targets.to(DEVICE)\n logits, probas = model(features)\n _, predicted_labels = torch.max(probas, 1)\n num_examples += targets.size(0)\n correct_pred += (predicted_labels == targets).sum()\n \n return correct_pred.float() / num_examples * 100\n \n\nstart_time = time.time()\nepoch_costs = []\nfor epoch in range(num_epochs):\n avg_cost = 0.\n for batch_idx, (features, targets) in enumerate(train_loader):\n \n features = features.float().view(-1, 9).to(DEVICE)\n targets = targets.to(DEVICE)\n \n ### FORWARD AND BACK PROP\n logits, probas = model(features)\n \n # note that the PyTorch implementation of\n # CrossEntropyLoss works with logits, not\n # probabilities\n cost = F.cross_entropy(logits, targets)\n optimizer.zero_grad()\n cost.backward()\n avg_cost += cost\n \n ### UPDATE MODEL PARAMETERS\n optimizer.step()\n \n ### LOGGING\n if not batch_idx % 50:\n print ('Epoch: %03d/%03d | Batch %03d/%03d | Cost: %.4f' \n %(epoch+1, num_epochs, batch_idx, \n len(trainDataset)//batch_size, cost))\n \n with torch.set_grad_enabled(False):\n avg_cost = avg_cost/len(trainDataset)\n epoch_costs.append(avg_cost)\n print('Epoch: %03d/%03d training accuracy: %.2f%%' % (\n epoch+1, num_epochs, \n compute_accuracy(model, train_loader)))\n print('Time elapsed: %.2f min' % ((time.time() - start_time)/60))", "_____no_output_____" ] ], [ [ "### Model Performance", "_____no_output_____" ] ], [ [ "%matplotlib inline\nimport matplotlib\nimport matplotlib.pyplot as plt\n\n\nplt.plot(epoch_costs)\nplt.ylabel('Avg Cross Entropy Loss\\n(approximated by averaging over minibatches)')\nplt.xlabel('Epoch')\nplt.show()", "_____no_output_____" ], [ "print(f'Train accuracy: {(compute_accuracy(model, train_loader)): .2f}%')\nprint(f'Train accuracy: {(compute_accuracy(model, test_loader)): .2f}%')", "Train accuracy: 24.92%\nTrain accuracy: 25.29%\n" ] ], [ [ "## Comments\n\n- This was a demonstration of how we can use manually crafted features in Image Classification tasks.\n- The model can be improved in several ways:\n - Tweaking the parameters to modify features generated for **_LBP, Co-Occurance Matrix and Gabor Filter_**\n - Extending the parameters for Red,Blue and Green channels.\n - Modifying the Learning rate, Epochs.\n - Trying a different Algorithm such as Multi Layer Perceptron.\n- The results aren't great but offer a glimpse of manually creating features from images.", "_____no_output_____" ], [ "{{\"Maria Petrou, Pedro Garcia Sevilla. _Image Processing: Dealing with Texture_. John Wiley & Sons, Ltd (2006)\" | fndetail: 1 }}", "_____no_output_____" ] ], [ [ "", "_____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", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ] ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "# TensorFlow 2.0+ Low Level APIs Convert Example\n\nThis example demonstrates the workflow to build a model using\nTensorFlow 2.0+ low-level APIs and convert it to Core ML \n`.mlmodel` format using the `coremltools.converters.tensorflow` converter.\nFor more example, refer `test_tf_2x.py` file.\n\nNote: \n\n- This notebook was tested with following dependencies:\n\n```\ntensorflow==2.0.0\ncoremltools==3.1\n```\n\n- Models from TensorFlow 2.0+ is supported only for `minimum_ios_deployment_target>=13`.\nYou can also use `tfcoreml.convert()` instead of \n`coremltools.converters.tensorflow.convert()` to convert your model.", "_____no_output_____" ] ], [ [ "import tensorflow as tf\nimport numpy as np\nimport coremltools\n\nprint(tf.__version__)\nprint(coremltools.__version__)", "WARNING: Logging before flag parsing goes to stderr.\nW1101 14:02:33.174557 4762860864 __init__.py:74] TensorFlow version 2.0.0 detected. Last version known to be fully compatible is 1.14.0 .\n" ] ], [ [ "## Using Low-Level APIs", "_____no_output_____" ] ], [ [ "# construct a toy model with low level APIs\nroot = tf.train.Checkpoint()\nroot.v1 = tf.Variable(3.)\nroot.v2 = tf.Variable(2.)\nroot.f = tf.function(lambda x: root.v1 * root.v2 * x)\n\n# save the model\nsaved_model_dir = './tf_model'\ninput_data = tf.constant(1., shape=[1, 1])\nto_save = root.f.get_concrete_function(input_data)\ntf.saved_model.save(root, saved_model_dir, to_save)\n\ntf_model = tf.saved_model.load(saved_model_dir)\nconcrete_func = tf_model.signatures[tf.saved_model.DEFAULT_SERVING_SIGNATURE_DEF_KEY]", "W1101 14:02:33.537978 4762860864 deprecation.py:506] From /Volumes/data/venv-py36/lib/python3.6/site-packages/tensorflow_core/python/ops/resource_variable_ops.py:1781: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.\nInstructions for updating:\nIf using Keras pass *_constraint arguments to layers.\n" ], [ "# convert model into Core ML format\nmodel = coremltools.converters.tensorflow.convert(\n [concrete_func],\n inputs={'x': (1, 1)},\n outputs=['Identity']\n)\n\nassert isinstance(model, coremltools.models.MLModel)", "0 assert nodes deleted\n['Func/StatefulPartitionedCall/input/_2:0', 'StatefulPartitionedCall/mul/ReadVariableOp:0', 'statefulpartitionedcall_args_1:0', 'Func/StatefulPartitionedCall/input/_3:0', 'StatefulPartitionedCall/mul:0', 'StatefulPartitionedCall/ReadVariableOp:0', 'statefulpartitionedcall_args_2:0']\n6 nodes deleted\n0 nodes deleted\n0 nodes deleted\n2 identity nodes deleted\n0 disconnected nodes deleted\n[SSAConverter] Converting function main ...\n[SSAConverter] [1/3] Converting op type: 'Placeholder', name: 'x', output_shape: (1, 1).\n[SSAConverter] [2/3] Converting op type: 'Const', name: 'StatefulPartitionedCall/mul'.\n[SSAConverter] [3/3] Converting op type: 'Mul', name: 'Identity', output_shape: (1, 1).\n" ] ], [ [ "## Using Control Flow", "_____no_output_____" ] ], [ [ "# construct a TensorFlow 2.0+ model with tf.function()\n\[email protected](input_signature=[tf.TensorSpec([], tf.float32)])\ndef control_flow(x):\n if x <= 0:\n return 0.\n else:\n return x * 3.\n\nto_save = tf.Module()\nto_save.control_flow = control_flow\n\nsaved_model_dir = './tf_model'\ntf.saved_model.save(to_save, saved_model_dir)\ntf_model = tf.saved_model.load(saved_model_dir)\nconcrete_func = tf_model.signatures[tf.saved_model.DEFAULT_SERVING_SIGNATURE_DEF_KEY]", "_____no_output_____" ], [ "# convert model into Core ML format\nmodel = coremltools.converters.tensorflow.convert(\n [concrete_func],\n inputs={'x': (1,)},\n outputs=['Identity']\n)\n\nassert isinstance(model, coremltools.models.MLModel)", "0 assert nodes deleted\n['PartitionedCall/cond/then/_2/Identity_1:0', 'PartitionedCall/LessEqual/y:0', 'PartitionedCall/cond/else/_3/mul/y:0', 'Func/PartitionedCall/cond/then/_2/output/_14:0', 'PartitionedCall/cond/then/_2/Const_1:0']\n2 nodes deleted\nFixing cond at merge location: PartitionedCall/cond/output/_9\nIn an IFF node fp32 != tensor[fp32,1]\n0 nodes deleted\n0 nodes deleted\n2 identity nodes deleted\n0 disconnected nodes deleted\n[SSAConverter] Converting function main ...\n[SSAConverter] [1/7] Converting op type: 'Placeholder', name: 'x', output_shape: (1,).\n[SSAConverter] [2/7] Converting op type: 'Const', name: 'PartitionedCall/LessEqual/y'.\n[SSAConverter] [3/7] Converting op type: 'Const', name: 'Func/PartitionedCall/cond/then/_2/output/_14'.\n[SSAConverter] [4/7] Converting op type: 'Const', name: 'PartitionedCall/cond/else/_3/mul/y'.\n[SSAConverter] [5/7] Converting op type: 'LessEqual', name: 'PartitionedCall/LessEqual', output_shape: (1,).\n[SSAConverter] [6/7] Converting op type: 'Mul', name: 'PartitionedCall/cond/else/_3/mul', output_shape: (1,).\n[SSAConverter] [7/7] Converting op type: 'iff', name: 'Identity'.\n" ], [ "# try with some sample inputs\n\ninputs = [-3.7, 6.17, 0.0, 1984., -5.]\nfor data in inputs:\n out1 = to_save.control_flow(data).numpy()\n out2 = model.predict({'x': np.array([data])})['Identity']\n np.testing.assert_array_almost_equal(out1, out2)", "_____no_output_____" ] ], [ [ "## Using `tf.keras` Subclassing APIs", "_____no_output_____" ] ], [ [ "class MyModel(tf.keras.Model):\n def __init__(self):\n super(MyModel, self).__init__()\n self.dense1 = tf.keras.layers.Dense(4)\n self.dense2 = tf.keras.layers.Dense(5)\n\n @tf.function\n def call(self, input_data):\n return self.dense2(self.dense1(input_data))\n\nkeras_model = MyModel()", "_____no_output_____" ], [ "inputs = np.random.rand(4, 4)\n\n# subclassed model can only be saved as SavedModel format\nkeras_model._set_inputs(inputs)\nsaved_model_dir = './tf_model_subclassing'\nkeras_model.save(saved_model_dir, save_format='tf')\n# convert and validate\nmodel = coremltools.converters.tensorflow.convert(\n saved_model_dir,\n inputs={'input_1': (4, 4)},\n outputs=['Identity']\n)\nassert isinstance(model, coremltools.models.MLModel)\n# verify the prediction matches\nkeras_prediction = keras_model.predict(inputs)\nprediction = model.predict({'input_1': inputs})['Identity']\nnp.testing.assert_array_equal(keras_prediction.shape, prediction.shape)\nnp.testing.assert_almost_equal(keras_prediction.flatten(), prediction.flatten(), decimal=4)", "0 assert nodes deleted\n['my_model/StatefulPartitionedCall/args_3:0', 'Func/my_model/StatefulPartitionedCall/input/_2:0', 'Func/my_model/StatefulPartitionedCall/StatefulPartitionedCall/input/_11:0', 'my_model/StatefulPartitionedCall/args_4:0', 'Func/my_model/StatefulPartitionedCall/input/_4:0', 'Func/my_model/StatefulPartitionedCall/StatefulPartitionedCall/input/_12:0', 'my_model/StatefulPartitionedCall/args_2:0', 'my_model/StatefulPartitionedCall/StatefulPartitionedCall/dense_1/StatefulPartitionedCall/MatMul/ReadVariableOp:0', 'Func/my_model/StatefulPartitionedCall/StatefulPartitionedCall/dense_1/StatefulPartitionedCall/input/_25:0', 'Func/my_model/StatefulPartitionedCall/input/_3:0', 'Func/my_model/StatefulPartitionedCall/StatefulPartitionedCall/input/_13:0', 'Func/my_model/StatefulPartitionedCall/input/_5:0', 'Func/my_model/StatefulPartitionedCall/StatefulPartitionedCall/input/_10:0', 'Func/my_model/StatefulPartitionedCall/StatefulPartitionedCall/dense_1/StatefulPartitionedCall/input/_24:0', 'my_model/StatefulPartitionedCall/args_1:0', 'Func/my_model/StatefulPartitionedCall/StatefulPartitionedCall/dense/StatefulPartitionedCall/input/_18:0', 'my_model/StatefulPartitionedCall/StatefulPartitionedCall/dense/StatefulPartitionedCall/BiasAdd/ReadVariableOp:0', 'my_model/StatefulPartitionedCall/StatefulPartitionedCall/dense/StatefulPartitionedCall/MatMul/ReadVariableOp:0', 'my_model/StatefulPartitionedCall/StatefulPartitionedCall/dense_1/StatefulPartitionedCall/BiasAdd/ReadVariableOp:0', 'Func/my_model/StatefulPartitionedCall/StatefulPartitionedCall/dense/StatefulPartitionedCall/input/_19:0']\n16 nodes deleted\n0 nodes deleted\n0 nodes deleted\n[Op Fusion] fuse_bias_add() deleted 4 nodes.\n2 identity nodes deleted\n2 disconnected nodes deleted\n[SSAConverter] Converting function main ...\n[SSAConverter] [1/3] Converting op type: 'Placeholder', name: 'input_1', output_shape: (4, 4).\n[SSAConverter] [2/3] Converting op type: 'MatMul', name: 'my_model/StatefulPartitionedCall/StatefulPartitionedCall/dense/StatefulPartitionedCall/MatMul', output_shape: (4, 4).\n[SSAConverter] [3/3] Converting op type: 'MatMul', name: 'Identity', output_shape: (4, 5).\n" ] ] ]

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

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

[ "Apache-2.0", "CC-BY-4.0" ]

[ "Apache-2.0", "CC-BY-4.0" ]

[ "Apache-2.0", "CC-BY-4.0" ]

[ [ [ "<table> <tr>\n <td style=\"background-color:#ffffff;\">\n <a href=\"http://qworld.lu.lv\" target=\"_blank\"><img src=\"../images/qworld.jpg\" width=\"25%\" align=\"left\"> </a></td>\n <td style=\"background-color:#ffffff;vertical-align:bottom;text-align:right;\">\n prepared by <a href=\"http://abu.lu.lv\" target=\"_blank\">Abuzer Yakaryilmaz</a> (<a href=\"http://qworld.lu.lv/index.php/qlatvia/\" target=\"_blank\">QLatvia</a>)\n </td> \n</tr></table>", "_____no_output_____" ], [ "<table width=\"100%\"><tr><td style=\"color:#bbbbbb;background-color:#ffffff;font-size:11px;font-style:italic;text-align:right;\">This cell contains some macros. If there is a problem with displaying mathematical formulas, please run this cell to load these macros. </td></tr></table>\n$ \\newcommand{\\bra}[1]{\\langle #1|} $\n$ \\newcommand{\\ket}[1]{|#1\\rangle} $\n$ \\newcommand{\\braket}[2]{\\langle #1|#2\\rangle} $\n$ \\newcommand{\\dot}[2]{ #1 \\cdot #2} $\n$ \\newcommand{\\biginner}[2]{\\left\\langle #1,#2\\right\\rangle} $\n$ \\newcommand{\\mymatrix}[2]{\\left( \\begin{array}{#1} #2\\end{array} \\right)} $\n$ \\newcommand{\\myvector}[1]{\\mymatrix{c}{#1}} $\n$ \\newcommand{\\myrvector}[1]{\\mymatrix{r}{#1}} $\n$ \\newcommand{\\mypar}[1]{\\left( #1 \\right)} $\n$ \\newcommand{\\mybigpar}[1]{ \\Big( #1 \\Big)} $\n$ \\newcommand{\\sqrttwo}{\\frac{1}{\\sqrt{2}}} $\n$ \\newcommand{\\dsqrttwo}{\\dfrac{1}{\\sqrt{2}}} $\n$ \\newcommand{\\onehalf}{\\frac{1}{2}} $\n$ \\newcommand{\\donehalf}{\\dfrac{1}{2}} $\n$ \\newcommand{\\hadamard}{ \\mymatrix{rr}{ \\sqrttwo & \\sqrttwo \\\\ \\sqrttwo & -\\sqrttwo }} $\n$ \\newcommand{\\vzero}{\\myvector{1\\\\0}} $\n$ \\newcommand{\\vone}{\\myvector{0\\\\1}} $\n$ \\newcommand{\\vhadamardzero}{\\myvector{ \\sqrttwo \\\\ \\sqrttwo } } $\n$ \\newcommand{\\vhadamardone}{ \\myrvector{ \\sqrttwo \\\\ -\\sqrttwo } } $\n$ \\newcommand{\\myarray}[2]{ \\begin{array}{#1}#2\\end{array}} $\n$ \\newcommand{\\X}{ \\mymatrix{cc}{0 & 1 \\\\ 1 & 0} } $\n$ \\newcommand{\\Z}{ \\mymatrix{rr}{1 & 0 \\\\ 0 & -1} } $\n$ \\newcommand{\\Htwo}{ \\mymatrix{rrrr}{ \\frac{1}{2} & \\frac{1}{2} & \\frac{1}{2} & \\frac{1}{2} \\\\ \\frac{1}{2} & -\\frac{1}{2} & \\frac{1}{2} & -\\frac{1}{2} \\\\ \\frac{1}{2} & \\frac{1}{2} & -\\frac{1}{2} & -\\frac{1}{2} \\\\ \\frac{1}{2} & -\\frac{1}{2} & -\\frac{1}{2} & \\frac{1}{2} } } $\n$ \\newcommand{\\CNOT}{ \\mymatrix{cccc}{1 & 0 & 0 & 0 \\\\ 0 & 1 & 0 & 0 \\\\ 0 & 0 & 0 & 1 \\\\ 0 & 0 & 1 & 0} } $\n$ \\newcommand{\\norm}[1]{ \\left\\lVert #1 \\right\\rVert } $", "_____no_output_____" ], [ "<h2>Vectors: One Dimensional Lists</h2>\n\nA <b>vector</b> is a list of numbers. \n\nVectors are very useful to describe the state of a system, as we will see in the main tutorial. \n\nA list is a single object in python.\n\nSimilarly, a vector is a single mathematical object. \n\nThe number of elements in a list is its size or length.\n\nSimilarly, the number of entries in a vector is called as the <b>size</b> or <b>dimension</b> of the vector.", "_____no_output_____" ] ], [ [ "# consider the following list with 4 elements \nL = [1,-2,0,5]\nprint(L)", "_____no_output_____" ] ], [ [ "Vectors can be in horizontal or vertical shape.\n\nWe show this list as a <i><u>four dimensional</u></i> <b>row vector</b> (horizontal) or a <b>column vector</b> (vertical):\n\n$$\n u = \\mypar{1~~-2~~0~~-5} ~~~\\mbox{ or }~~~ v =\\mymatrix{r}{1 \\\\ -2 \\\\ 0 \\\\ 5}, ~~~\\mbox{ respectively.}\n$$\n\nRemark that we do not need to use any comma in vector representation.", "_____no_output_____" ], [ "<h3> Multiplying a vector with a number</h3>\n\nA vector can be multiplied by a number.\n\nMultiplication of a vector with a number is also a vector: each entry is multiplied by this number.\n\n$$\n 3 \\cdot v = 3 \\cdot \\mymatrix{r}{1 \\\\ -2 \\\\ 0 \\\\ 5} = \\mymatrix{r}{3 \\\\ -6 \\\\ 0 \\\\ 15}\n ~~~~~~\\mbox{ or }~~~~~~\n (-0.6) \\cdot v = (-0.6) \\cdot \\mymatrix{r}{1 \\\\ -2 \\\\ 0 \\\\ 5} = \\mymatrix{r}{-0.6 \\\\ 1.2 \\\\ 0 \\\\ -3}.\n$$\n\nWe may consider this as enlarging or making smaller the entries of a vector.\n\nWe verify our calculations in python.", "_____no_output_____" ] ], [ [ "# 3 * v\nv = [1,-2,0,5]\nprint(\"v is\",v)\n# we use the same list for the result\nfor i in range(len(v)):\n v[i] = 3 * v[i]\nprint(\"3v is\",v)\n\n# -0.6 * u\n# reinitialize the list v\nv = [1,-2,0,5]\nfor i in range(len(v)):\n v[i] = -0.6 * v[i]\nprint(\"0.6v is\",v)", "_____no_output_____" ] ], [ [ "<h3> Summation of vectors</h3>\n\nTwo vectors (with same dimension) can be summed up.\n\nThe summation of two vectors is a vector: the numbers on the same entries are added up.\n\n$$\n u = \\myrvector{-3 \\\\ -2 \\\\ 0 \\\\ -1 \\\\ 4} \\mbox{ and } v = \\myrvector{-1\\\\ -1 \\\\2 \\\\ -3 \\\\ 5}.\n ~~~~~~~ \\mbox{Then, }~~\n u+v = \\myrvector{-3 \\\\ -2 \\\\ 0 \\\\ -1 \\\\ 4} + \\myrvector{-1\\\\ -1 \\\\2 \\\\ -3 \\\\ 5} =\n \\myrvector{-3+(-1)\\\\ -2+(-1) \\\\0+2 \\\\ -1+(-3) \\\\ 4+5} = \\myrvector{-4\\\\ -3 \\\\2 \\\\ -4 \\\\ 9}.\n$$\n\nWe do the same calculations in Python.", "_____no_output_____" ] ], [ [ "u = [-3,-2,0,-1,4]\nv = [-1,-1,2,-3,5]\nresult=[]\nfor i in range(len(u)):\n result.append(u[i]+v[i])\n\nprint(\"u+v is\",result)\n\n# print the result vector similarly to a column vector\nprint() # print an empty line\nprint(\"the elements of u+v are\")\nfor j in range(len(result)):\n print(result[j])", "_____no_output_____" ] ], [ [ "<h3> Task 1 </h3>\n\nCreate two 7-dimensional vectors $u$ and $ v $ as two different lists in Python having entries randomly picked between $-10$ and $10$. \n\nPrint their entries.", "_____no_output_____" ] ], [ [ "from random import randrange\n#\n# your solution is here\n#\n\n#r=randrange(-10,11) # randomly pick a number from the list {-10,-9,...,-1,0,1,...,9,10}\n", "_____no_output_____" ] ], [ [ "<a href=\"Math20_Vectors_Solutions.ipynb#task1\">click for our solution</a>", "_____no_output_____" ], [ "<h3> Task 2 </h3>\n\nBy using the same vectors, find the vector $ (3 u-2 v) $ and print its entries. Here $ 3u $ and $ 2v $ means $u$ and $v$ are multiplied by $3$ and $2$, respectively.", "_____no_output_____" ] ], [ [ "#\n# your solution is here\n#\n", "_____no_output_____" ] ], [ [ "<a href=\"Math20_Vectors_Solutions.ipynb#task2\">click for our solution</a>", "_____no_output_____" ], [ "<h3> Visualization of vectors </h3>\n\nWe can visualize the vectors with dimension at most 3. \n\nFor simplicity, we give examples of 2-dimensional vectors. \n\nConsider the vector $ v = \\myvector{1 \\\\ 2} $. \n\nA 2-dimensional vector can be represented on the two-dimensional plane by an arrow starting from the origin $ (0,0) $ to the point $ (1,2) $.", "_____no_output_____" ], [ "<img src=\"../images/vector_1_2-small.jpg\" width=\"40%\">", "_____no_output_____" ], [ "We represent the vectors $ 2v = \\myvector{2 \\\\ 4} $ and $ -v = \\myvector{-1 \\\\ -2} $ below.\n\n<img src=\"../images/vectors_2_4_-1_-2.jpg\" width=\"40%\">\n\nAs we can observe, after multiplying by 2, the vector is enlarged, and, after multiplying by $(-1)$, the vector is the same but its direction is opposite.", "_____no_output_____" ], [ "<h3> The length of a vector </h3>\n\nThe length of a vector is the (shortest) distance from the points represented by the entries of vector to the origin point $(0,0)$.\n\nThe length of a vector can be calculated by using Pythagoras Theorem. \n\nWe visualize a vector, its length, and the contributions of each entry to the length. \n\nConsider the vector $ u = \\myrvector{-3 \\\\ 4} $.", "_____no_output_____" ], [ "<img src=\"../images/length_-3_4-small.jpg\" width=\"80%\">", "_____no_output_____" ], [ "The length of $ u $ is denoted as $ \\norm{u} $, and it is calculated as $ \\norm{u} =\\sqrt{(-3)^2+4^2} = 5 $. \n\nHere each entry contributes with its square value. All contributions are summed up. Then, we obtain the square of the length. \n\nThis formula is generalized to any dimension. \n\nWe find the length of the following vector by using Python:\n \n$$\n v = \\myrvector{-1 \\\\ -3 \\\\ 5 \\\\ 3 \\\\ 1 \\\\ 2}\n ~~~~~~~~~~\n \\mbox{and}\n ~~~~~~~~~~\n \\norm{v} = \\sqrt{(-1)^2+(-3)^2+5^2+3^2+1^2+2^2} .\n$$", "_____no_output_____" ], [ "<div style=\"font-style:italic;background-color:#fafafa;font-size:10pt;\"> Remember: There is a short way of writing power operation in Python. \n <ul>\n <li> In its generic form: $ a^x $ can be denoted by $ a ** x $ in Python. </li>\n <li> The square of a number $a$: $ a^2 $ can be denoted by $ a ** 2 $ in Python. </li>\n <li> The square root of a number $ a $: $ \\sqrt{a} = a^{\\frac{1}{2}} = a^{0.5} $ can be denoted by $ a ** 0.5 $ in Python.</li>\n </ul>\n</div>", "_____no_output_____" ] ], [ [ "v = [-1,-3,5,3,1,2]\n\nlength_square=0\nfor i in range(len(v)):\n print(v[i],\":square ->\",v[i]**2) # print each entry and its square value\n length_square = length_square + v[i]**2 # sum up the square of each entry\n\nlength = length_square ** 0.5 # take the square root of the summation of the squares of all entries\nprint(\"the summation is\",length_square)\nprint(\"then the length is\",length)\n\n# for square root, we can also use built-in function math.sqrt\nprint() # print an empty line\nfrom math import sqrt\nprint(\"the square root of\",length_square,\"is\",sqrt(length_square))", "_____no_output_____" ] ], [ [ "<h3> Task 3 </h3>\n\nLet $ u = \\myrvector{1 \\\\ -2 \\\\ -4 \\\\ 2} $ be a four dimensional vector.\n\nVerify that $ \\norm{4 u} = 4 \\cdot \\norm{u} $ in Python. \n\nRemark that $ 4u $ is another vector obtained from $ u $ by multiplying it with 4. ", "_____no_output_____" ] ], [ [ "#\n# your solution is here\n#\n", "_____no_output_____" ] ], [ [ "<a href=\"Math20_Vectors_Solutions.ipynb#task3\">click for our solution</a>", "_____no_output_____" ], [ "<h3> Notes:</h3>\n\nWhen a vector is multiplied by a number, then its length is also multiplied with the same number.\n\nBut, we should be careful with the sign.\n\nConsider the vector $ -3 v $. It has the same length of $ 3v $, but its direction is opposite.\n\nSo, when calculating the length of $ -3 v $, we use absolute value of the number:\n\n$ \\norm{-3 v} = |-3| \\norm{v} = 3 \\norm{v} $.\n\nHere $ |-3| $ is the absolute value of $ -3 $. \n\nThe absolute value of a number is its distance to 0. So, $ |-3| = 3 $.", "_____no_output_____" ], [ "<h3> Task 4 </h3>\n\nLet $ u = \\myrvector{1 \\\\ -2 \\\\ -4 \\\\ 2} $ be a four dimensional vector.\n\nRandomly pick a number $r$ from $ \\left\\{ \\dfrac{1}{10}, \\dfrac{2}{10}, \\cdots, \\dfrac{9}{10} \\right\\} $.\n\nFind the vector $(-r)\\cdot u$ and then its length.", "_____no_output_____" ] ], [ [ "#\n# your solution is here\n#\n", "_____no_output_____" ] ], [ [ "<a href=\"Math20_Vectors_Solutions.ipynb#task4\">click for our solution</a>", "_____no_output_____" ] ] ]

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

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# default_exp data.external", "_____no_output_____" ] ], [ [ "# External data\n\n> Helper functions used to download and extract common time series datasets.", "_____no_output_____" ] ], [ [ "#export\nfrom tsai.imports import *\nfrom tsai.utils import * \nfrom tsai.data.validation import *", "_____no_output_____" ], [ "#export\nfrom sktime.utils.data_io import load_from_tsfile_to_dataframe as ts2df\nfrom sktime.utils.validation.panel import check_X\nfrom sktime.utils.data_io import TsFileParseException", "_____no_output_____" ], [ "#export\nfrom fastai.data.external import *\nfrom tqdm import tqdm\nimport zipfile\nimport tempfile\ntry: from urllib import urlretrieve\nexcept ImportError: from urllib.request import urlretrieve\nimport shutil\nfrom numpy import distutils\nimport distutils", "_____no_output_____" ], [ "#export\ndef decompress_from_url(url, target_dir=None, verbose=False):\n # Download\n try:\n pv(\"downloading data...\", verbose)\n fname = os.path.basename(url)\n tmpdir = tempfile.mkdtemp()\n tmpfile = os.path.join(tmpdir, fname)\n urlretrieve(url, tmpfile)\n pv(\"...data downloaded\", verbose)\n\n # Decompress\n try:\n pv(\"decompressing data...\", verbose)\n if not os.path.exists(target_dir): os.makedirs(target_dir)\n shutil.unpack_archive(tmpfile, target_dir)\n shutil.rmtree(tmpdir)\n pv(\"...data decompressed\", verbose)\n return target_dir\n \n except:\n shutil.rmtree(tmpdir)\n if verbose: sys.stderr.write(\"Could not decompress file, aborting.\\n\")\n\n except:\n shutil.rmtree(tmpdir)\n if verbose:\n sys.stderr.write(\"Could not download url. Please, check url.\\n\")", "_____no_output_____" ], [ "#export\nfrom fastdownload import download_url\ndef download_data(url, fname=None, c_key='archive', force_download=False, timeout=4, verbose=False):\n \"Download `url` to `fname`.\"\n fname = Path(fname or URLs.path(url, c_key=c_key))\n fname.parent.mkdir(parents=True, exist_ok=True)\n if not fname.exists() or force_download: download_url(url, dest=fname, timeout=timeout, show_progress=verbose)\n return fname", "_____no_output_____" ], [ "# export\ndef get_UCR_univariate_list():\n return [\n 'ACSF1', 'Adiac', 'AllGestureWiimoteX', 'AllGestureWiimoteY',\n 'AllGestureWiimoteZ', 'ArrowHead', 'Beef', 'BeetleFly', 'BirdChicken',\n 'BME', 'Car', 'CBF', 'Chinatown', 'ChlorineConcentration',\n 'CinCECGTorso', 'Coffee', 'Computers', 'CricketX', 'CricketY',\n 'CricketZ', 'Crop', 'DiatomSizeReduction',\n 'DistalPhalanxOutlineAgeGroup', 'DistalPhalanxOutlineCorrect',\n 'DistalPhalanxTW', 'DodgerLoopDay', 'DodgerLoopGame',\n 'DodgerLoopWeekend', 'Earthquakes', 'ECG200', 'ECG5000', 'ECGFiveDays',\n 'ElectricDevices', 'EOGHorizontalSignal', 'EOGVerticalSignal',\n 'EthanolLevel', 'FaceAll', 'FaceFour', 'FacesUCR', 'FiftyWords',\n 'Fish', 'FordA', 'FordB', 'FreezerRegularTrain', 'FreezerSmallTrain',\n 'Fungi', 'GestureMidAirD1', 'GestureMidAirD2', 'GestureMidAirD3',\n 'GesturePebbleZ1', 'GesturePebbleZ2', 'GunPoint', 'GunPointAgeSpan',\n 'GunPointMaleVersusFemale', 'GunPointOldVersusYoung', 'Ham',\n 'HandOutlines', 'Haptics', 'Herring', 'HouseTwenty', 'InlineSkate',\n 'InsectEPGRegularTrain', 'InsectEPGSmallTrain', 'InsectWingbeatSound',\n 'ItalyPowerDemand', 'LargeKitchenAppliances', 'Lightning2',\n 'Lightning7', 'Mallat', 'Meat', 'MedicalImages', 'MelbournePedestrian',\n 'MiddlePhalanxOutlineAgeGroup', 'MiddlePhalanxOutlineCorrect',\n 'MiddlePhalanxTW', 'MixedShapesRegularTrain', 'MixedShapesSmallTrain',\n 'MoteStrain', 'NonInvasiveFetalECGThorax1',\n 'NonInvasiveFetalECGThorax2', 'OliveOil', 'OSULeaf',\n 'PhalangesOutlinesCorrect', 'Phoneme', 'PickupGestureWiimoteZ',\n 'PigAirwayPressure', 'PigArtPressure', 'PigCVP', 'PLAID', 'Plane',\n 'PowerCons', 'ProximalPhalanxOutlineAgeGroup',\n 'ProximalPhalanxOutlineCorrect', 'ProximalPhalanxTW',\n 'RefrigerationDevices', 'Rock', 'ScreenType', 'SemgHandGenderCh2',\n 'SemgHandMovementCh2', 'SemgHandSubjectCh2', 'ShakeGestureWiimoteZ',\n 'ShapeletSim', 'ShapesAll', 'SmallKitchenAppliances', 'SmoothSubspace',\n 'SonyAIBORobotSurface1', 'SonyAIBORobotSurface2', 'StarLightCurves',\n 'Strawberry', 'SwedishLeaf', 'Symbols', 'SyntheticControl',\n 'ToeSegmentation1', 'ToeSegmentation2', 'Trace', 'TwoLeadECG',\n 'TwoPatterns', 'UMD', 'UWaveGestureLibraryAll', 'UWaveGestureLibraryX',\n 'UWaveGestureLibraryY', 'UWaveGestureLibraryZ', 'Wafer', 'Wine',\n 'WordSynonyms', 'Worms', 'WormsTwoClass', 'Yoga'\n ]\n\n\ntest_eq(len(get_UCR_univariate_list()), 128)\nUTSC_datasets = get_UCR_univariate_list()\nUCR_univariate_list = get_UCR_univariate_list()", "_____no_output_____" ], [ "#export\ndef get_UCR_multivariate_list():\n return [\n 'ArticularyWordRecognition', 'AtrialFibrillation', 'BasicMotions',\n 'CharacterTrajectories', 'Cricket', 'DuckDuckGeese', 'EigenWorms',\n 'Epilepsy', 'ERing', 'EthanolConcentration', 'FaceDetection',\n 'FingerMovements', 'HandMovementDirection', 'Handwriting', 'Heartbeat',\n 'InsectWingbeat', 'JapaneseVowels', 'Libras', 'LSST', 'MotorImagery',\n 'NATOPS', 'PEMS-SF', 'PenDigits', 'PhonemeSpectra', 'RacketSports',\n 'SelfRegulationSCP1', 'SelfRegulationSCP2', 'SpokenArabicDigits',\n 'StandWalkJump', 'UWaveGestureLibrary'\n ]\n\ntest_eq(len(get_UCR_multivariate_list()), 30)\nMTSC_datasets = get_UCR_multivariate_list()\nUCR_multivariate_list = get_UCR_multivariate_list()\n\nUCR_list = sorted(UCR_univariate_list + UCR_multivariate_list)\nclassification_list = UCR_list\nTSC_datasets = classification_datasets = UCR_list\nlen(UCR_list)", "_____no_output_____" ], [ "#export\ndef get_UCR_data(dsid, path='.', parent_dir='data/UCR', on_disk=True, mode='c', Xdtype='float32', ydtype=None, return_split=True, split_data=True, \n force_download=False, verbose=False):\n dsid_list = [ds for ds in UCR_list if ds.lower() == dsid.lower()]\n assert len(dsid_list) > 0, f'{dsid} is not a UCR dataset'\n dsid = dsid_list[0]\n return_split = return_split and split_data # keep return_split for compatibility. It will be replaced by split_data\n if dsid in ['InsectWingbeat']:\n warnings.warn(f'Be aware that download of the {dsid} dataset is very slow!')\n pv(f'Dataset: {dsid}', verbose)\n full_parent_dir = Path(path)/parent_dir\n full_tgt_dir = full_parent_dir/dsid\n# if not os.path.exists(full_tgt_dir): os.makedirs(full_tgt_dir)\n full_tgt_dir.parent.mkdir(parents=True, exist_ok=True)\n if force_download or not all([os.path.isfile(f'{full_tgt_dir}/{fn}.npy') for fn in ['X_train', 'X_valid', 'y_train', 'y_valid', 'X', 'y']]):\n # Option A\n src_website = 'http://www.timeseriesclassification.com/Downloads'\n decompress_from_url(f'{src_website}/{dsid}.zip', target_dir=full_tgt_dir, verbose=verbose)\n if dsid == 'DuckDuckGeese':\n with zipfile.ZipFile(Path(f'{full_parent_dir}/DuckDuckGeese/DuckDuckGeese_ts.zip'), 'r') as zip_ref:\n zip_ref.extractall(Path(parent_dir))\n if not os.path.exists(full_tgt_dir/f'{dsid}_TRAIN.ts') or not os.path.exists(full_tgt_dir/f'{dsid}_TRAIN.ts') or \\\n Path(full_tgt_dir/f'{dsid}_TRAIN.ts').stat().st_size == 0 or Path(full_tgt_dir/f'{dsid}_TEST.ts').stat().st_size == 0: \n print('It has not been possible to download the required files')\n if return_split:\n return None, None, None, None\n else:\n return None, None, None\n \n pv('loading ts files to dataframe...', verbose)\n X_train_df, y_train = ts2df(full_tgt_dir/f'{dsid}_TRAIN.ts')\n X_valid_df, y_valid = ts2df(full_tgt_dir/f'{dsid}_TEST.ts')\n pv('...ts files loaded', verbose)\n pv('preparing numpy arrays...', verbose)\n X_train_ = []\n X_valid_ = []\n for i in progress_bar(range(X_train_df.shape[-1]), display=verbose, leave=False):\n X_train_.append(stack_pad(X_train_df[f'dim_{i}'])) # stack arrays even if they have different lengths\n X_valid_.append(stack_pad(X_valid_df[f'dim_{i}'])) # stack arrays even if they have different lengths\n X_train = np.transpose(np.stack(X_train_, axis=-1), (0, 2, 1))\n X_valid = np.transpose(np.stack(X_valid_, axis=-1), (0, 2, 1))\n X_train, X_valid = match_seq_len(X_train, X_valid) \n \n np.save(f'{full_tgt_dir}/X_train.npy', X_train)\n np.save(f'{full_tgt_dir}/y_train.npy', y_train)\n np.save(f'{full_tgt_dir}/X_valid.npy', X_valid)\n np.save(f'{full_tgt_dir}/y_valid.npy', y_valid)\n np.save(f'{full_tgt_dir}/X.npy', concat(X_train, X_valid))\n np.save(f'{full_tgt_dir}/y.npy', concat(y_train, y_valid))\n del X_train, X_valid, y_train, y_valid\n delete_all_in_dir(full_tgt_dir, exception='.npy')\n pv('...numpy arrays correctly saved', verbose)\n\n mmap_mode = mode if on_disk else None\n X_train = np.load(f'{full_tgt_dir}/X_train.npy', mmap_mode=mmap_mode)\n y_train = np.load(f'{full_tgt_dir}/y_train.npy', mmap_mode=mmap_mode)\n X_valid = np.load(f'{full_tgt_dir}/X_valid.npy', mmap_mode=mmap_mode)\n y_valid = np.load(f'{full_tgt_dir}/y_valid.npy', mmap_mode=mmap_mode)\n\n if return_split:\n if Xdtype is not None: \n X_train = X_train.astype(Xdtype)\n X_valid = X_valid.astype(Xdtype)\n if ydtype is not None: \n y_train = y_train.astype(ydtype)\n y_valid = y_valid.astype(ydtype)\n if verbose:\n print('X_train:', X_train.shape)\n print('y_train:', y_train.shape)\n print('X_valid:', X_valid.shape)\n print('y_valid:', y_valid.shape, '\\n')\n return X_train, y_train, X_valid, y_valid\n else:\n X = np.load(f'{full_tgt_dir}/X.npy', mmap_mode=mmap_mode)\n y = np.load(f'{full_tgt_dir}/y.npy', mmap_mode=mmap_mode)\n splits = get_predefined_splits(X_train, X_valid)\n if Xdtype is not None: \n X = X.astype(Xdtype)\n if verbose:\n print('X :', X .shape)\n print('y :', y .shape)\n print('splits :', coll_repr(splits[0]), coll_repr(splits[1]), '\\n')\n return X, y, splits\n \n \nget_classification_data = get_UCR_data", "_____no_output_____" ], [ "#hide\nPATH = Path('.')\ndsids = ['ECGFiveDays', 'AtrialFibrillation'] # univariate and multivariate\nfor dsid in dsids:\n print(dsid)\n tgt_dir = PATH/f'data/UCR/{dsid}'\n if os.path.isdir(tgt_dir): shutil.rmtree(tgt_dir)\n test_eq(len(get_files(tgt_dir)), 0) # no file left\n X_train, y_train, X_valid, y_valid = get_UCR_data(dsid)\n test_eq(len(get_files(tgt_dir, '.npy')), 6)\n test_eq(len(get_files(tgt_dir, '.npy')), len(get_files(tgt_dir))) # test no left file/ dir\n del X_train, y_train, X_valid, y_valid\n start = time.time()\n X_train, y_train, X_valid, y_valid = get_UCR_data(dsid)\n elapsed = time.time() - start\n test_eq(elapsed < 1, True)\n test_eq(X_train.ndim, 3)\n test_eq(y_train.ndim, 1)\n test_eq(X_valid.ndim, 3)\n test_eq(y_valid.ndim, 1)\n test_eq(len(get_files(tgt_dir, '.npy')), 6)\n test_eq(len(get_files(tgt_dir, '.npy')), len(get_files(tgt_dir))) # test no left file/ dir\n test_eq(X_train.ndim, 3)\n test_eq(y_train.ndim, 1)\n test_eq(X_valid.ndim, 3)\n test_eq(y_valid.ndim, 1)\n test_eq(X_train.dtype, np.float32)\n test_eq(X_train.__class__.__name__, 'memmap')\n del X_train, y_train, X_valid, y_valid\n X_train, y_train, X_valid, y_valid = get_UCR_data(dsid, on_disk=False)\n test_eq(X_train.__class__.__name__, 'ndarray')\n del X_train, y_train, X_valid, y_valid", "ECGFiveDays\nAtrialFibrillation\n" ], [ "X_train, y_train, X_valid, y_valid = get_UCR_data('natops')", "_____no_output_____" ], [ "dsid = 'natops' \nX_train, y_train, X_valid, y_valid = get_UCR_data(dsid, verbose=True)\nX, y, splits = get_UCR_data(dsid, split_data=False)\ntest_eq(X[splits[0]], X_train)\ntest_eq(y[splits[1]], y_valid)\ntest_eq(X[splits[0]], X_train)\ntest_eq(y[splits[1]], y_valid)\ntest_type(X, X_train)\ntest_type(y, y_train)", "Dataset: NATOPS\nX_train: (180, 24, 51)\ny_train: (180,)\nX_valid: (180, 24, 51)\ny_valid: (180,) \n\n" ], [ "#export\ndef check_data(X, y=None, splits=None, show_plot=True):\n try: X_is_nan = np.isnan(X).sum()\n except: X_is_nan = 'couldn not be checked'\n if X.ndim == 3:\n shape = f'[{X.shape[0]} samples x {X.shape[1]} features x {X.shape[-1]} timesteps]'\n print(f'X - shape: {shape} type: {cls_name(X)} dtype:{X.dtype} isnan: {X_is_nan}')\n else:\n print(f'X - shape: {X.shape} type: {cls_name(X)} dtype:{X.dtype} isnan: {X_is_nan}')\n if not isinstance(X, np.ndarray): warnings.warn('X must be a np.ndarray')\n if X_is_nan: \n warnings.warn('X must not contain nan values')\n if y is not None:\n y_shape = y.shape\n y = y.ravel()\n if isinstance(y[0], str):\n n_classes = f'{len(np.unique(y))} ({len(y)//len(np.unique(y))} samples per class) {L(np.unique(y).tolist())}'\n y_is_nan = 'nan' in [c.lower() for c in np.unique(y)]\n print(f'y - shape: {y_shape} type: {cls_name(y)} dtype:{y.dtype} n_classes: {n_classes} isnan: {y_is_nan}')\n else:\n y_is_nan = np.isnan(y).sum()\n print(f'y - shape: {y_shape} type: {cls_name(y)} dtype:{y.dtype} isnan: {y_is_nan}')\n if not isinstance(y, np.ndarray): warnings.warn('y must be a np.ndarray')\n if y_is_nan: \n warnings.warn('y must not contain nan values')\n if splits is not None:\n _splits = get_splits_len(splits)\n overlap = check_splits_overlap(splits)\n print(f'splits - n_splits: {len(_splits)} shape: {_splits} overlap: {overlap}')\n if show_plot: plot_splits(splits)", "_____no_output_____" ], [ "dsid = 'ECGFiveDays'\nX, y, splits = get_UCR_data(dsid, split_data=False, on_disk=False, force_download=True)\ncheck_data(X, y, splits)\ncheck_data(X[:, 0], y, splits)\ny = y.astype(np.float32)\ncheck_data(X, y, splits)\ny[:10] = np.nan\ncheck_data(X[:, 0], y, splits)\nX, y, splits = get_UCR_data(dsid, split_data=False, on_disk=False, force_download=True)\nsplits = get_splits(y, 3)\ncheck_data(X, y, splits)\ncheck_data(X[:, 0], y, splits)\ny[:5]= np.nan\ncheck_data(X[:, 0], y, splits)\nX, y, splits = get_UCR_data(dsid, split_data=False, on_disk=False, force_download=True)", "X - shape: [884 samples x 1 features x 136 timesteps] type: ndarray dtype:float32 isnan: 0\ny - shape: (884,) type: ndarray dtype:<U1 n_classes: 2 (442 samples per class) ['1', '2'] isnan: False\nsplits - n_splits: 2 shape: [23, 861] overlap: [False]\n" ], [ "#export\n# This code comes from https://github.com/ChangWeiTan/TSRegression. As of Jan 16th, 2021 there's no pip install available.\n\n# The following code is adapted from the python package sktime to read .ts file.\nclass _TsFileParseException(Exception):\n \"\"\"\n Should be raised when parsing a .ts file and the format is incorrect.\n \"\"\"\n pass\n\ndef _load_from_tsfile_to_dataframe2(full_file_path_and_name, return_separate_X_and_y=True, replace_missing_vals_with='NaN'):\n \"\"\"Loads data from a .ts file into a Pandas DataFrame.\n Parameters\n ----------\n full_file_path_and_name: str\n The full pathname of the .ts file to read.\n return_separate_X_and_y: bool\n true if X and Y values should be returned as separate Data Frames (X) and a numpy array (y), false otherwise.\n This is only relevant for data that\n replace_missing_vals_with: str\n The value that missing values in the text file should be replaced with prior to parsing.\n Returns\n -------\n DataFrame, ndarray\n If return_separate_X_and_y then a tuple containing a DataFrame and a numpy array containing the relevant time-series and corresponding class values.\n DataFrame\n If not return_separate_X_and_y then a single DataFrame containing all time-series and (if relevant) a column \"class_vals\" the associated class values.\n \"\"\"\n\n # Initialize flags and variables used when parsing the file\n metadata_started = False\n data_started = False\n\n has_problem_name_tag = False\n has_timestamps_tag = False\n has_univariate_tag = False\n has_class_labels_tag = False\n has_target_labels_tag = False\n has_data_tag = False\n\n previous_timestamp_was_float = None\n previous_timestamp_was_int = None\n previous_timestamp_was_timestamp = None\n num_dimensions = None\n is_first_case = True\n instance_list = []\n class_val_list = []\n line_num = 0\n\n # Parse the file\n # print(full_file_path_and_name)\n with open(full_file_path_and_name, 'r', encoding='utf-8') as file:\n for line in tqdm(file):\n # print(\".\", end='')\n # Strip white space from start/end of line and change to lowercase for use below\n line = line.strip().lower()\n # Empty lines are valid at any point in a file\n if line:\n # Check if this line contains metadata\n # Please note that even though metadata is stored in this function it is not currently published externally\n if line.startswith(\"@problemname\"):\n # Check that the data has not started\n if data_started:\n raise _TsFileParseException(\"metadata must come before data\")\n # Check that the associated value is valid\n tokens = line.split(' ')\n token_len = len(tokens)\n\n if token_len == 1:\n raise _TsFileParseException(\"problemname tag requires an associated value\")\n\n problem_name = line[len(\"@problemname\") + 1:]\n has_problem_name_tag = True\n metadata_started = True\n elif line.startswith(\"@timestamps\"):\n # Check that the data has not started\n if data_started:\n raise _TsFileParseException(\"metadata must come before data\")\n\n # Check that the associated value is valid\n tokens = line.split(' ')\n token_len = len(tokens)\n\n if token_len != 2:\n raise _TsFileParseException(\"timestamps tag requires an associated Boolean value\")\n elif tokens[1] == \"true\":\n timestamps = True\n elif tokens[1] == \"false\":\n timestamps = False\n else:\n raise _TsFileParseException(\"invalid timestamps value\")\n has_timestamps_tag = True\n metadata_started = True\n elif line.startswith(\"@univariate\"):\n # Check that the data has not started\n if data_started:\n raise _TsFileParseException(\"metadata must come before data\")\n\n # Check that the associated value is valid\n tokens = line.split(' ')\n token_len = len(tokens)\n if token_len != 2:\n raise _TsFileParseException(\"univariate tag requires an associated Boolean value\")\n elif tokens[1] == \"true\":\n univariate = True\n elif tokens[1] == \"false\":\n univariate = False\n else:\n raise _TsFileParseException(\"invalid univariate value\")\n\n has_univariate_tag = True\n metadata_started = True\n elif line.startswith(\"@classlabel\"):\n # Check that the data has not started\n if data_started:\n raise _TsFileParseException(\"metadata must come before data\")\n\n # Check that the associated value is valid\n tokens = line.split(' ')\n token_len = len(tokens)\n\n if token_len == 1:\n raise _TsFileParseException(\"classlabel tag requires an associated Boolean value\")\n\n if tokens[1] == \"true\":\n class_labels = True\n elif tokens[1] == \"false\":\n class_labels = False\n else:\n raise _TsFileParseException(\"invalid classLabel value\")\n\n # Check if we have any associated class values\n if token_len == 2 and class_labels:\n raise _TsFileParseException(\"if the classlabel tag is true then class values must be supplied\")\n\n has_class_labels_tag = True\n class_label_list = [token.strip() for token in tokens[2:]]\n metadata_started = True\n elif line.startswith(\"@targetlabel\"):\n # Check that the data has not started\n if data_started:\n raise _TsFileParseException(\"metadata must come before data\")\n\n # Check that the associated value is valid\n tokens = line.split(' ')\n token_len = len(tokens)\n\n if token_len == 1:\n raise _TsFileParseException(\"targetlabel tag requires an associated Boolean value\")\n\n if tokens[1] == \"true\":\n target_labels = True\n elif tokens[1] == \"false\":\n target_labels = False\n else:\n raise _TsFileParseException(\"invalid targetLabel value\")\n\n has_target_labels_tag = True\n class_val_list = []\n metadata_started = True\n # Check if this line contains the start of data\n elif line.startswith(\"@data\"):\n if line != \"@data\":\n raise _TsFileParseException(\"data tag should not have an associated value\")\n\n if data_started and not metadata_started:\n raise _TsFileParseException(\"metadata must come before data\")\n else:\n has_data_tag = True\n data_started = True\n # If the 'data tag has been found then metadata has been parsed and data can be loaded\n elif data_started:\n # Check that a full set of metadata has been provided\n incomplete_regression_meta_data = not has_problem_name_tag or not has_timestamps_tag or not has_univariate_tag or not has_target_labels_tag or not has_data_tag\n incomplete_classification_meta_data = not has_problem_name_tag or not has_timestamps_tag or not has_univariate_tag or not has_class_labels_tag or not has_data_tag\n if incomplete_regression_meta_data and incomplete_classification_meta_data:\n raise _TsFileParseException(\"a full set of metadata has not been provided before the data\")\n\n # Replace any missing values with the value specified\n line = line.replace(\"?\", replace_missing_vals_with)\n\n # Check if we dealing with data that has timestamps\n if timestamps:\n # We're dealing with timestamps so cannot just split line on ':' as timestamps may contain one\n has_another_value = False\n has_another_dimension = False\n\n timestamps_for_dimension = []\n values_for_dimension = []\n\n this_line_num_dimensions = 0\n line_len = len(line)\n char_num = 0\n\n while char_num < line_len:\n # Move through any spaces\n while char_num < line_len and str.isspace(line[char_num]):\n char_num += 1\n\n # See if there is any more data to read in or if we should validate that read thus far\n\n if char_num < line_len:\n\n # See if we have an empty dimension (i.e. no values)\n if line[char_num] == \":\":\n if len(instance_list) < (this_line_num_dimensions + 1):\n instance_list.append([])\n\n instance_list[this_line_num_dimensions].append(pd.Series())\n this_line_num_dimensions += 1\n\n has_another_value = False\n has_another_dimension = True\n\n timestamps_for_dimension = []\n values_for_dimension = []\n\n char_num += 1\n else:\n # Check if we have reached a class label\n if line[char_num] != \"(\" and target_labels:\n class_val = line[char_num:].strip()\n\n # if class_val not in class_val_list:\n # raise _TsFileParseException(\n # \"the class value '\" + class_val + \"' on line \" + str(\n # line_num + 1) + \" is not valid\")\n\n class_val_list.append(float(class_val))\n char_num = line_len\n\n has_another_value = False\n has_another_dimension = False\n\n timestamps_for_dimension = []\n values_for_dimension = []\n\n else:\n\n # Read in the data contained within the next tuple\n\n if line[char_num] != \"(\" and not target_labels:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" does not start with a '('\")\n\n char_num += 1\n tuple_data = \"\"\n\n while char_num < line_len and line[char_num] != \")\":\n tuple_data += line[char_num]\n char_num += 1\n\n if char_num >= line_len or line[char_num] != \")\":\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" does not end with a ')'\")\n\n # Read in any spaces immediately after the current tuple\n\n char_num += 1\n\n while char_num < line_len and str.isspace(line[char_num]):\n char_num += 1\n\n # Check if there is another value or dimension to process after this tuple\n\n if char_num >= line_len:\n has_another_value = False\n has_another_dimension = False\n\n elif line[char_num] == \",\":\n has_another_value = True\n has_another_dimension = False\n\n elif line[char_num] == \":\":\n has_another_value = False\n has_another_dimension = True\n\n char_num += 1\n\n # Get the numeric value for the tuple by reading from the end of the tuple data backwards to the last comma\n\n last_comma_index = tuple_data.rfind(',')\n\n if last_comma_index == -1:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" contains a tuple that has no comma inside of it\")\n\n try:\n value = tuple_data[last_comma_index + 1:]\n value = float(value)\n\n except ValueError:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" contains a tuple that does not have a valid numeric value\")\n\n # Check the type of timestamp that we have\n\n timestamp = tuple_data[0: last_comma_index]\n\n try:\n timestamp = int(timestamp)\n timestamp_is_int = True\n timestamp_is_timestamp = False\n except ValueError:\n timestamp_is_int = False\n\n if not timestamp_is_int:\n try:\n timestamp = float(timestamp)\n timestamp_is_float = True\n timestamp_is_timestamp = False\n except ValueError:\n timestamp_is_float = False\n\n if not timestamp_is_int and not timestamp_is_float:\n try:\n timestamp = timestamp.strip()\n timestamp_is_timestamp = True\n except ValueError:\n timestamp_is_timestamp = False\n\n # Make sure that the timestamps in the file (not just this dimension or case) are consistent\n\n if not timestamp_is_timestamp and not timestamp_is_int and not timestamp_is_float:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" contains a tuple that has an invalid timestamp '\" + timestamp + \"'\")\n\n if previous_timestamp_was_float is not None and previous_timestamp_was_float and not timestamp_is_float:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" contains tuples where the timestamp format is inconsistent\")\n\n if previous_timestamp_was_int is not None and previous_timestamp_was_int and not timestamp_is_int:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" contains tuples where the timestamp format is inconsistent\")\n\n if previous_timestamp_was_timestamp is not None and previous_timestamp_was_timestamp and not timestamp_is_timestamp:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" contains tuples where the timestamp format is inconsistent\")\n\n # Store the values\n\n timestamps_for_dimension += [timestamp]\n values_for_dimension += [value]\n\n # If this was our first tuple then we store the type of timestamp we had\n\n if previous_timestamp_was_timestamp is None and timestamp_is_timestamp:\n previous_timestamp_was_timestamp = True\n previous_timestamp_was_int = False\n previous_timestamp_was_float = False\n\n if previous_timestamp_was_int is None and timestamp_is_int:\n previous_timestamp_was_timestamp = False\n previous_timestamp_was_int = True\n previous_timestamp_was_float = False\n\n if previous_timestamp_was_float is None and timestamp_is_float:\n previous_timestamp_was_timestamp = False\n previous_timestamp_was_int = False\n previous_timestamp_was_float = True\n\n # See if we should add the data for this dimension\n\n if not has_another_value:\n if len(instance_list) < (this_line_num_dimensions + 1):\n instance_list.append([])\n\n if timestamp_is_timestamp:\n timestamps_for_dimension = pd.DatetimeIndex(timestamps_for_dimension)\n\n instance_list[this_line_num_dimensions].append(\n pd.Series(index=timestamps_for_dimension, data=values_for_dimension))\n this_line_num_dimensions += 1\n\n timestamps_for_dimension = []\n values_for_dimension = []\n\n elif has_another_value:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" ends with a ',' that is not followed by another tuple\")\n\n elif has_another_dimension and target_labels:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" ends with a ':' while it should list a class value\")\n\n elif has_another_dimension and not target_labels:\n if len(instance_list) < (this_line_num_dimensions + 1):\n instance_list.append([])\n\n instance_list[this_line_num_dimensions].append(pd.Series(dtype=np.float32))\n this_line_num_dimensions += 1\n num_dimensions = this_line_num_dimensions\n\n # If this is the 1st line of data we have seen then note the dimensions\n\n if not has_another_value and not has_another_dimension:\n if num_dimensions is None:\n num_dimensions = this_line_num_dimensions\n\n if num_dimensions != this_line_num_dimensions:\n raise _TsFileParseException(\"line \" + str(\n line_num + 1) + \" does not have the same number of dimensions as the previous line of data\")\n\n # Check that we are not expecting some more data, and if not, store that processed above\n\n if has_another_value:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" ends with a ',' that is not followed by another tuple\")\n\n elif has_another_dimension and target_labels:\n raise _TsFileParseException(\n \"dimension \" + str(this_line_num_dimensions + 1) + \" on line \" + str(\n line_num + 1) + \" ends with a ':' while it should list a class value\")\n\n elif has_another_dimension and not target_labels:\n if len(instance_list) < (this_line_num_dimensions + 1):\n instance_list.append([])\n\n instance_list[this_line_num_dimensions].append(pd.Series())\n this_line_num_dimensions += 1\n num_dimensions = this_line_num_dimensions\n\n # If this is the 1st line of data we have seen then note the dimensions\n\n if not has_another_value and num_dimensions != this_line_num_dimensions:\n raise _TsFileParseException(\"line \" + str(\n line_num + 1) + \" does not have the same number of dimensions as the previous line of data\")\n\n # Check if we should have class values, and if so that they are contained in those listed in the metadata\n\n if target_labels and len(class_val_list) == 0:\n raise _TsFileParseException(\"the cases have no associated class values\")\n else:\n dimensions = line.split(\":\")\n # If first row then note the number of dimensions (that must be the same for all cases)\n if is_first_case:\n num_dimensions = len(dimensions)\n\n if target_labels:\n num_dimensions -= 1\n\n for dim in range(0, num_dimensions):\n instance_list.append([])\n is_first_case = False\n\n # See how many dimensions that the case whose data in represented in this line has\n this_line_num_dimensions = len(dimensions)\n\n if target_labels:\n this_line_num_dimensions -= 1\n\n # All dimensions should be included for all series, even if they are empty\n if this_line_num_dimensions != num_dimensions:\n raise _TsFileParseException(\"inconsistent number of dimensions. Expecting \" + str(\n num_dimensions) + \" but have read \" + str(this_line_num_dimensions))\n\n # Process the data for each dimension\n for dim in range(0, num_dimensions):\n dimension = dimensions[dim].strip()\n\n if dimension:\n data_series = dimension.split(\",\")\n data_series = [float(i) for i in data_series]\n instance_list[dim].append(pd.Series(data_series))\n else:\n instance_list[dim].append(pd.Series())\n\n if target_labels:\n class_val_list.append(float(dimensions[num_dimensions].strip()))\n\n line_num += 1\n\n # Check that the file was not empty\n if line_num:\n # Check that the file contained both metadata and data\n complete_regression_meta_data = has_problem_name_tag and has_timestamps_tag and has_univariate_tag and has_target_labels_tag and has_data_tag\n complete_classification_meta_data = has_problem_name_tag and has_timestamps_tag and has_univariate_tag and has_class_labels_tag and has_data_tag\n\n if metadata_started and not complete_regression_meta_data and not complete_classification_meta_data:\n raise _TsFileParseException(\"metadata incomplete\")\n elif metadata_started and not data_started:\n raise _TsFileParseException(\"file contained metadata but no data\")\n elif metadata_started and data_started and len(instance_list) == 0:\n raise _TsFileParseException(\"file contained metadata but no data\")\n\n # Create a DataFrame from the data parsed above\n data = pd.DataFrame(dtype=np.float32)\n\n for dim in range(0, num_dimensions):\n data['dim_' + str(dim)] = instance_list[dim]\n\n # Check if we should return any associated class labels separately\n\n if target_labels:\n if return_separate_X_and_y:\n return data, np.asarray(class_val_list)\n else:\n data['class_vals'] = pd.Series(class_val_list)\n return data\n else:\n return data\n else:\n raise _TsFileParseException(\"empty file\")", "_____no_output_____" ], [ "#export\ndef get_Monash_regression_list():\n return sorted([\n \"AustraliaRainfall\", \"HouseholdPowerConsumption1\",\n \"HouseholdPowerConsumption2\", \"BeijingPM25Quality\",\n \"BeijingPM10Quality\", \"Covid3Month\", \"LiveFuelMoistureContent\",\n \"FloodModeling1\", \"FloodModeling2\", \"FloodModeling3\",\n \"AppliancesEnergy\", \"BenzeneConcentration\", \"NewsHeadlineSentiment\",\n \"NewsTitleSentiment\", \"IEEEPPG\", \n #\"BIDMC32RR\", \"BIDMC32HR\", \"BIDMC32SpO2\", \"PPGDalia\" # Cannot be downloaded\n ])\n\nMonash_regression_list = get_Monash_regression_list()\nregression_list = Monash_regression_list\nTSR_datasets = regression_datasets = regression_list\nlen(Monash_regression_list)", "_____no_output_____" ], [ "#export\ndef get_Monash_regression_data(dsid, path='./data/Monash', on_disk=True, mode='c', Xdtype='float32', ydtype=None, split_data=True, force_download=False, \n verbose=False):\n\n dsid_list = [rd for rd in Monash_regression_list if rd.lower() == dsid.lower()]\n assert len(dsid_list) > 0, f'{dsid} is not a Monash dataset'\n dsid = dsid_list[0]\n full_tgt_dir = Path(path)/dsid\n pv(f'Dataset: {dsid}', verbose)\n\n if force_download or not all([os.path.isfile(f'{path}/{dsid}/{fn}.npy') for fn in ['X_train', 'X_valid', 'y_train', 'y_valid', 'X', 'y']]):\n if dsid == 'AppliancesEnergy': id = 3902637\n elif dsid == 'HouseholdPowerConsumption1': id = 3902704\n elif dsid == 'HouseholdPowerConsumption2': id = 3902706\n elif dsid == 'BenzeneConcentration': id = 3902673\n elif dsid == 'BeijingPM25Quality': id = 3902671\n elif dsid == 'BeijingPM10Quality': id = 3902667\n elif dsid == 'LiveFuelMoistureContent': id = 3902716\n elif dsid == 'FloodModeling1': id = 3902694\n elif dsid == 'FloodModeling2': id = 3902696\n elif dsid == 'FloodModeling3': id = 3902698\n elif dsid == 'AustraliaRainfall': id = 3902654\n elif dsid == 'PPGDalia': id = 3902728\n elif dsid == 'IEEEPPG': id = 3902710\n elif dsid == 'BIDMCRR' or dsid == 'BIDM32CRR': id = 3902685\n elif dsid == 'BIDMCHR' or dsid == 'BIDM32CHR': id = 3902676\n elif dsid == 'BIDMCSpO2' or dsid == 'BIDM32CSpO2': id = 3902688\n elif dsid == 'NewsHeadlineSentiment': id = 3902718\n elif dsid == 'NewsTitleSentiment': id = 3902726\n elif dsid == 'Covid3Month': id = 3902690\n\n for split in ['TRAIN', 'TEST']:\n url = f\"https://zenodo.org/record/{id}/files/{dsid}_{split}.ts\"\n fname = Path(path)/f'{dsid}/{dsid}_{split}.ts'\n pv('downloading data...', verbose)\n try:\n download_data(url, fname, c_key='archive', force_download=force_download, timeout=4)\n except:\n warnings.warn(f'Cannot download {dsid} dataset')\n if split_data: return None, None, None, None\n else: return None, None, None\n pv('...download complete', verbose)\n if split == 'TRAIN':\n X_train, y_train = _load_from_tsfile_to_dataframe2(fname)\n X_train = check_X(X_train, coerce_to_numpy=True)\n else:\n X_valid, y_valid = _load_from_tsfile_to_dataframe2(fname)\n X_valid = check_X(X_valid, coerce_to_numpy=True)\n np.save(f'{full_tgt_dir}/X_train.npy', X_train)\n np.save(f'{full_tgt_dir}/y_train.npy', y_train)\n np.save(f'{full_tgt_dir}/X_valid.npy', X_valid)\n np.save(f'{full_tgt_dir}/y_valid.npy', y_valid)\n np.save(f'{full_tgt_dir}/X.npy', concat(X_train, X_valid))\n np.save(f'{full_tgt_dir}/y.npy', concat(y_train, y_valid))\n del X_train, X_valid, y_train, y_valid\n delete_all_in_dir(full_tgt_dir, exception='.npy')\n pv('...numpy arrays correctly saved', verbose)\n\n mmap_mode = mode if on_disk else None\n X_train = np.load(f'{full_tgt_dir}/X_train.npy', mmap_mode=mmap_mode)\n y_train = np.load(f'{full_tgt_dir}/y_train.npy', mmap_mode=mmap_mode)\n X_valid = np.load(f'{full_tgt_dir}/X_valid.npy', mmap_mode=mmap_mode)\n y_valid = np.load(f'{full_tgt_dir}/y_valid.npy', mmap_mode=mmap_mode)\n if Xdtype is not None: \n X_train = X_train.astype(Xdtype)\n X_valid = X_valid.astype(Xdtype)\n if ydtype is not None: \n y_train = y_train.astype(ydtype)\n y_valid = y_valid.astype(ydtype)\n\n if split_data:\n if verbose:\n print('X_train:', X_train.shape)\n print('y_train:', y_train.shape)\n print('X_valid:', X_valid.shape)\n print('y_valid:', y_valid.shape, '\\n')\n return X_train, y_train, X_valid, y_valid\n else:\n X = np.load(f'{full_tgt_dir}/X.npy', mmap_mode=mmap_mode)\n y = np.load(f'{full_tgt_dir}/y.npy', mmap_mode=mmap_mode)\n splits = get_predefined_splits(X_train, X_valid)\n if verbose:\n print('X :', X .shape)\n print('y :', y .shape)\n print('splits :', coll_repr(splits[0]), coll_repr(splits[1]), '\\n')\n return X, y, splits\n\n\nget_regression_data = get_Monash_regression_data", "_____no_output_____" ], [ "dsid = \"Covid3Month\"\nX_train, y_train, X_valid, y_valid = get_Monash_regression_data(dsid, on_disk=False, split_data=True, force_download=True)\nX, y, splits = get_Monash_regression_data(dsid, on_disk=True, split_data=False, force_download=True, verbose=True)\nif X_train is not None: \n test_eq(X_train.shape, (140, 1, 84))\nif X is not None: \n test_eq(X.shape, (201, 1, 84))", "153it [00:00, 1786.53it/s]\n74it [00:00, 1694.84it/s]\n" ], [ "#export\ndef get_forecasting_list():\n return sorted([\n \"Sunspots\", \"Weather\"\n ])\n\nforecasting_time_series = get_forecasting_list()", "_____no_output_____" ], [ "#export\ndef get_forecasting_time_series(dsid, path='./data/forecasting/', force_download=False, verbose=True, **kwargs):\n \n dsid_list = [fd for fd in forecasting_time_series if fd.lower() == dsid.lower()]\n assert len(dsid_list) > 0, f'{dsid} is not a forecasting dataset'\n dsid = dsid_list[0]\n if dsid == 'Weather': full_tgt_dir = Path(path)/f'{dsid}.csv.zip'\n else: full_tgt_dir = Path(path)/f'{dsid}.csv'\n pv(f'Dataset: {dsid}', verbose)\n if dsid == 'Sunspots': url = \"https://storage.googleapis.com/laurencemoroney-blog.appspot.com/Sunspots.csv\"\n elif dsid == 'Weather': url = 'https://storage.googleapis.com/tensorflow/tf-keras-datasets/jena_climate_2009_2016.csv.zip'\n\n try: \n pv(\"downloading data...\", verbose)\n if force_download: \n try: os.remove(full_tgt_dir)\n except OSError: pass\n download_data(url, full_tgt_dir, force_download=force_download, **kwargs)\n pv(f\"...data downloaded. Path = {full_tgt_dir}\", verbose)\n\n if dsid == 'Sunspots': \n df = pd.read_csv(full_tgt_dir, parse_dates=['Date'], index_col=['Date'])\n return df['Monthly Mean Total Sunspot Number'].asfreq('1M').to_frame()\n\n elif dsid == 'Weather':\n # This code comes from a great Keras time-series tutorial notebook (https://www.tensorflow.org/tutorials/structured_data/time_series)\n df = pd.read_csv(full_tgt_dir)\n df = df[5::6] # slice [start:stop:step], starting from index 5 take every 6th record.\n\n date_time = pd.to_datetime(df.pop('Date Time'), format='%d.%m.%Y %H:%M:%S')\n\n # remove error (negative wind)\n wv = df['wv (m/s)']\n bad_wv = wv == -9999.0\n wv[bad_wv] = 0.0\n\n max_wv = df['max. wv (m/s)']\n bad_max_wv = max_wv == -9999.0\n max_wv[bad_max_wv] = 0.0\n\n wv = df.pop('wv (m/s)')\n max_wv = df.pop('max. wv (m/s)')\n\n # Convert to radians.\n wd_rad = df.pop('wd (deg)')*np.pi / 180\n\n # Calculate the wind x and y components.\n df['Wx'] = wv*np.cos(wd_rad)\n df['Wy'] = wv*np.sin(wd_rad)\n\n # Calculate the max wind x and y components.\n df['max Wx'] = max_wv*np.cos(wd_rad)\n df['max Wy'] = max_wv*np.sin(wd_rad)\n\n timestamp_s = date_time.map(datetime.timestamp)\n day = 24*60*60\n year = (365.2425)*day\n\n df['Day sin'] = np.sin(timestamp_s * (2 * np.pi / day))\n df['Day cos'] = np.cos(timestamp_s * (2 * np.pi / day))\n df['Year sin'] = np.sin(timestamp_s * (2 * np.pi / year))\n df['Year cos'] = np.cos(timestamp_s * (2 * np.pi / year))\n df.reset_index(drop=True, inplace=True)\n return df\n else: \n return full_tgt_dir\n except: \n warnings.warn(f\"Cannot download {dsid} dataset\")\n return", "_____no_output_____" ], [ "ts = get_forecasting_time_series(\"sunspots\", force_download=True)\ntest_eq(len(ts), 3235)\nts", "Dataset: Sunspots\ndownloading data...\n...data downloaded. Path = data/forecasting/Sunspots.csv\n" ], [ "ts = get_forecasting_time_series(\"weather\", force_download=True)\ntest_eq(len(ts), 70091)\nts", "Dataset: Weather\ndownloading data...\n...data downloaded. Path = data/forecasting/Weather.csv.zip\n" ], [ "# export\n\nMonash_forecasting_list = ['m1_yearly_dataset',\n 'm1_quarterly_dataset',\n 'm1_monthly_dataset',\n 'm3_yearly_dataset',\n 'm3_quarterly_dataset',\n 'm3_monthly_dataset',\n 'm3_other_dataset',\n 'm4_yearly_dataset',\n 'm4_quarterly_dataset',\n 'm4_monthly_dataset',\n 'm4_weekly_dataset',\n 'm4_daily_dataset',\n 'm4_hourly_dataset',\n 'tourism_yearly_dataset',\n 'tourism_quarterly_dataset',\n 'tourism_monthly_dataset',\n 'nn5_daily_dataset_with_missing_values',\n 'nn5_daily_dataset_without_missing_values',\n 'nn5_weekly_dataset',\n 'cif_2016_dataset',\n 'kaggle_web_traffic_dataset_with_missing_values',\n 'kaggle_web_traffic_dataset_without_missing_values',\n 'kaggle_web_traffic_weekly_dataset',\n 'solar_10_minutes_dataset',\n 'solar_weekly_dataset',\n 'electricity_hourly_dataset',\n 'electricity_weekly_dataset',\n 'london_smart_meters_dataset_with_missing_values',\n 'london_smart_meters_dataset_without_missing_values',\n 'wind_farms_minutely_dataset_with_missing_values',\n 'wind_farms_minutely_dataset_without_missing_values',\n 'car_parts_dataset_with_missing_values',\n 'car_parts_dataset_without_missing_values',\n 'dominick_dataset',\n 'fred_md_dataset',\n 'traffic_hourly_dataset',\n 'traffic_weekly_dataset',\n 'pedestrian_counts_dataset',\n 'hospital_dataset',\n 'covid_deaths_dataset',\n 'kdd_cup_2018_dataset_with_missing_values',\n 'kdd_cup_2018_dataset_without_missing_values',\n 'weather_dataset',\n 'sunspot_dataset_with_missing_values',\n 'sunspot_dataset_without_missing_values',\n 'saugeenday_dataset',\n 'us_births_dataset',\n 'elecdemand_dataset',\n 'solar_4_seconds_dataset',\n 'wind_4_seconds_dataset',\n 'Sunspots', 'Weather']\n\n\nforecasting_list = Monash_forecasting_list", "_____no_output_____" ], [ "# export\n\n## Original code available at: https://github.com/rakshitha123/TSForecasting\n# This repository contains the implementations related to the experiments of a set of publicly available datasets that are used in \n# the time series forecasting research space.\n\n# The benchmark datasets are available at: https://zenodo.org/communities/forecasting. For more details, please refer to our website: \n# https://forecastingdata.org/ and paper: https://arxiv.org/abs/2105.06643.\n\n# Citation: \n# @misc{godahewa2021monash,\n# author=\"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n# title=\"Monash Time Series Forecasting Archive\",\n# howpublished =\"\\url{https://arxiv.org/abs/2105.06643}\",\n# year=\"2021\"\n# }\n\n\n# Converts the contents in a .tsf file into a dataframe and returns it along with other meta-data of the dataset: frequency, horizon, whether the dataset contains missing values and whether the series have equal lengths\n#\n# Parameters\n# full_file_path_and_name - complete .tsf file path\n# replace_missing_vals_with - a term to indicate the missing values in series in the returning dataframe\n# value_column_name - Any name that is preferred to have as the name of the column containing series values in the returning dataframe\ndef convert_tsf_to_dataframe(full_file_path_and_name, replace_missing_vals_with = 'NaN', value_column_name = \"series_value\"):\n col_names = []\n col_types = []\n all_data = {}\n line_count = 0\n frequency = None\n forecast_horizon = None\n contain_missing_values = None\n contain_equal_length = None\n found_data_tag = False\n found_data_section = False\n started_reading_data_section = False\n\n with open(full_file_path_and_name, 'r', encoding='cp1252') as file:\n for line in file:\n # Strip white space from start/end of line\n line = line.strip()\n\n if line:\n if line.startswith(\"@\"): # Read meta-data\n if not line.startswith(\"@data\"):\n line_content = line.split(\" \")\n if line.startswith(\"@attribute\"):\n if (len(line_content) != 3): # Attributes have both name and type\n raise TsFileParseException(\"Invalid meta-data specification.\")\n\n col_names.append(line_content[1])\n col_types.append(line_content[2])\n else:\n if len(line_content) != 2: # Other meta-data have only values\n raise TsFileParseException(\"Invalid meta-data specification.\")\n\n if line.startswith(\"@frequency\"):\n frequency = line_content[1]\n elif line.startswith(\"@horizon\"):\n forecast_horizon = int(line_content[1])\n elif line.startswith(\"@missing\"):\n contain_missing_values = bool(distutils.util.strtobool(line_content[1]))\n elif line.startswith(\"@equallength\"):\n contain_equal_length = bool(distutils.util.strtobool(line_content[1]))\n\n else:\n if len(col_names) == 0:\n raise TsFileParseException(\"Missing attribute section. Attribute section must come before data.\")\n\n found_data_tag = True\n elif not line.startswith(\"#\"):\n if len(col_names) == 0:\n raise TsFileParseException(\"Missing attribute section. Attribute section must come before data.\")\n elif not found_data_tag:\n raise TsFileParseException(\"Missing @data tag.\")\n else:\n if not started_reading_data_section:\n started_reading_data_section = True\n found_data_section = True\n all_series = []\n\n for col in col_names:\n all_data[col] = []\n\n full_info = line.split(\":\")\n\n if len(full_info) != (len(col_names) + 1):\n raise TsFileParseException(\"Missing attributes/values in series.\")\n\n series = full_info[len(full_info) - 1]\n series = series.split(\",\")\n\n if(len(series) == 0):\n raise TsFileParseException(\"A given series should contains a set of comma separated numeric values. At least one numeric value should be there in a series. Missing values should be indicated with ? symbol\")\n\n numeric_series = []\n\n for val in series:\n if val == \"?\":\n numeric_series.append(replace_missing_vals_with)\n else:\n numeric_series.append(float(val))\n\n if (numeric_series.count(replace_missing_vals_with) == len(numeric_series)):\n raise TsFileParseException(\"All series values are missing. A given series should contains a set of comma separated numeric values. At least one numeric value should be there in a series.\")\n\n all_series.append(pd.Series(numeric_series).array)\n\n for i in range(len(col_names)):\n att_val = None\n if col_types[i] == \"numeric\":\n att_val = int(full_info[i])\n elif col_types[i] == \"string\":\n att_val = str(full_info[i])\n elif col_types[i] == \"date\":\n att_val = datetime.strptime(full_info[i], '%Y-%m-%d %H-%M-%S')\n else:\n raise TsFileParseException(\"Invalid attribute type.\") # Currently, the code supports only numeric, string and date types. Extend this as required.\n\n if(att_val == None):\n raise TsFileParseException(\"Invalid attribute value.\")\n else:\n all_data[col_names[i]].append(att_val)\n\n line_count = line_count + 1\n\n if line_count == 0:\n raise TsFileParseException(\"Empty file.\")\n if len(col_names) == 0:\n raise TsFileParseException(\"Missing attribute section.\")\n if not found_data_section:\n raise TsFileParseException(\"Missing series information under data section.\")\n\n all_data[value_column_name] = all_series\n loaded_data = pd.DataFrame(all_data)\n\n return loaded_data, frequency, forecast_horizon, contain_missing_values, contain_equal_length", "_____no_output_____" ], [ "# export\n\ndef get_Monash_forecasting_data(dsid, path='./data/forecasting/', force_download=False, remove_from_disk=False, verbose=True):\n\n pv(f'Dataset: {dsid}', verbose)\n dsid = dsid.lower()\n assert dsid in Monash_forecasting_list, f'{dsid} not available in Monash_forecasting_list'\n\n if dsid == 'm1_yearly_dataset': url = 'https://zenodo.org/record/4656193/files/m1_yearly_dataset.zip'\n elif dsid == 'm1_quarterly_dataset': url = 'https://zenodo.org/record/4656154/files/m1_quarterly_dataset.zip'\n elif dsid == 'm1_monthly_dataset': url = 'https://zenodo.org/record/4656159/files/m1_monthly_dataset.zip'\n elif dsid == 'm3_yearly_dataset': url = 'https://zenodo.org/record/4656222/files/m3_yearly_dataset.zip'\n elif dsid == 'm3_quarterly_dataset': url = 'https://zenodo.org/record/4656262/files/m3_quarterly_dataset.zip'\n elif dsid == 'm3_monthly_dataset': url = 'https://zenodo.org/record/4656298/files/m3_monthly_dataset.zip'\n elif dsid == 'm3_other_dataset': url = 'https://zenodo.org/record/4656335/files/m3_other_dataset.zip'\n elif dsid == 'm4_yearly_dataset': url = 'https://zenodo.org/record/4656379/files/m4_yearly_dataset.zip'\n elif dsid == 'm4_quarterly_dataset': url = 'https://zenodo.org/record/4656410/files/m4_quarterly_dataset.zip'\n elif dsid == 'm4_monthly_dataset': url = 'https://zenodo.org/record/4656480/files/m4_monthly_dataset.zip'\n elif dsid == 'm4_weekly_dataset': url = 'https://zenodo.org/record/4656522/files/m4_weekly_dataset.zip'\n elif dsid == 'm4_daily_dataset': url = 'https://zenodo.org/record/4656548/files/m4_daily_dataset.zip'\n elif dsid == 'm4_hourly_dataset': url = 'https://zenodo.org/record/4656589/files/m4_hourly_dataset.zip'\n elif dsid == 'tourism_yearly_dataset': url = 'https://zenodo.org/record/4656103/files/tourism_yearly_dataset.zip'\n elif dsid == 'tourism_quarterly_dataset': url = 'https://zenodo.org/record/4656093/files/tourism_quarterly_dataset.zip'\n elif dsid == 'tourism_monthly_dataset': url = 'https://zenodo.org/record/4656096/files/tourism_monthly_dataset.zip'\n elif dsid == 'nn5_daily_dataset_with_missing_values': url = 'https://zenodo.org/record/4656110/files/nn5_daily_dataset_with_missing_values.zip'\n elif dsid == 'nn5_daily_dataset_without_missing_values': url = 'https://zenodo.org/record/4656117/files/nn5_daily_dataset_without_missing_values.zip'\n elif dsid == 'nn5_weekly_dataset': url = 'https://zenodo.org/record/4656125/files/nn5_weekly_dataset.zip'\n elif dsid == 'cif_2016_dataset': url = 'https://zenodo.org/record/4656042/files/cif_2016_dataset.zip'\n elif dsid == 'kaggle_web_traffic_dataset_with_missing_values': url = 'https://zenodo.org/record/4656080/files/kaggle_web_traffic_dataset_with_missing_values.zip'\n elif dsid == 'kaggle_web_traffic_dataset_without_missing_values': url = 'https://zenodo.org/record/4656075/files/kaggle_web_traffic_dataset_without_missing_values.zip'\n elif dsid == 'kaggle_web_traffic_weekly': url = 'https://zenodo.org/record/4656664/files/kaggle_web_traffic_weekly_dataset.zip'\n elif dsid == 'solar_10_minutes_dataset': url = 'https://zenodo.org/record/4656144/files/solar_10_minutes_dataset.zip'\n elif dsid == 'solar_weekly_dataset': url = 'https://zenodo.org/record/4656151/files/solar_weekly_dataset.zip'\n elif dsid == 'electricity_hourly_dataset': url = 'https://zenodo.org/record/4656140/files/electricity_hourly_dataset.zip'\n elif dsid == 'electricity_weekly_dataset': url = 'https://zenodo.org/record/4656141/files/electricity_weekly_dataset.zip'\n elif dsid == 'london_smart_meters_dataset_with_missing_values': url = 'https://zenodo.org/record/4656072/files/london_smart_meters_dataset_with_missing_values.zip'\n elif dsid == 'london_smart_meters_dataset_without_missing_values': url = 'https://zenodo.org/record/4656091/files/london_smart_meters_dataset_without_missing_values.zip'\n elif dsid == 'wind_farms_minutely_dataset_with_missing_values': url = 'https://zenodo.org/record/4654909/files/wind_farms_minutely_dataset_with_missing_values.zip'\n elif dsid == 'wind_farms_minutely_dataset_without_missing_values': url = 'https://zenodo.org/record/4654858/files/wind_farms_minutely_dataset_without_missing_values.zip'\n elif dsid == 'car_parts_dataset_with_missing_values': url = 'https://zenodo.org/record/4656022/files/car_parts_dataset_with_missing_values.zip'\n elif dsid == 'car_parts_dataset_without_missing_values': url = 'https://zenodo.org/record/4656021/files/car_parts_dataset_without_missing_values.zip'\n elif dsid == 'dominick_dataset': url = 'https://zenodo.org/record/4654802/files/dominick_dataset.zip'\n elif dsid == 'fred_md_dataset': url = 'https://zenodo.org/record/4654833/files/fred_md_dataset.zip'\n elif dsid == 'traffic_hourly_dataset': url = 'https://zenodo.org/record/4656132/files/traffic_hourly_dataset.zip'\n elif dsid == 'traffic_weekly_dataset': url = 'https://zenodo.org/record/4656135/files/traffic_weekly_dataset.zip'\n elif dsid == 'pedestrian_counts_dataset': url = 'https://zenodo.org/record/4656626/files/pedestrian_counts_dataset.zip'\n elif dsid == 'hospital_dataset': url = 'https://zenodo.org/record/4656014/files/hospital_dataset.zip'\n elif dsid == 'covid_deaths_dataset': url = 'https://zenodo.org/record/4656009/files/covid_deaths_dataset.zip'\n elif dsid == 'kdd_cup_2018_dataset_with_missing_values': url = 'https://zenodo.org/record/4656719/files/kdd_cup_2018_dataset_with_missing_values.zip'\n elif dsid == 'kdd_cup_2018_dataset_without_missing_values': url = 'https://zenodo.org/record/4656756/files/kdd_cup_2018_dataset_without_missing_values.zip'\n elif dsid == 'weather_dataset': url = 'https://zenodo.org/record/4654822/files/weather_dataset.zip'\n elif dsid == 'sunspot_dataset_with_missing_values': url = 'https://zenodo.org/record/4654773/files/sunspot_dataset_with_missing_values.zip'\n elif dsid == 'sunspot_dataset_without_missing_values': url = 'https://zenodo.org/record/4654722/files/sunspot_dataset_without_missing_values.zip'\n elif dsid == 'saugeenday_dataset': url = 'https://zenodo.org/record/4656058/files/saugeenday_dataset.zip'\n elif dsid == 'us_births_dataset': url = 'https://zenodo.org/record/4656049/files/us_births_dataset.zip'\n elif dsid == 'elecdemand_dataset': url = 'https://zenodo.org/record/4656069/files/elecdemand_dataset.zip'\n elif dsid == 'solar_4_seconds_dataset': url = 'https://zenodo.org/record/4656027/files/solar_4_seconds_dataset.zip'\n elif dsid == 'wind_4_seconds_dataset': url = 'https://zenodo.org/record/4656032/files/wind_4_seconds_dataset.zip'\n\n path = Path(path)\n full_path = path/f'{dsid}.tsf'\n if not full_path.exists() or force_download: \n decompress_from_url(url, target_dir=path, verbose=verbose)\n pv(\"converting dataframe to numpy array...\", verbose)\n data, frequency, forecast_horizon, contain_missing_values, contain_equal_length = convert_tsf_to_dataframe(full_path)\n X = to3d(stack_pad(data['series_value']))\n pv(\"...dataframe converted to numpy array\", verbose)\n pv(f'\\nX.shape: {X.shape}', verbose) \n pv(f'freq: {frequency}', verbose) \n pv(f'forecast_horizon: {forecast_horizon}', verbose) \n pv(f'contain_missing_values: {contain_missing_values}', verbose) \n pv(f'contain_equal_length: {contain_equal_length}', verbose=verbose)\n if remove_from_disk: os.remove(full_path)\n return X\n\nget_forecasting_data = get_Monash_forecasting_data", "_____no_output_____" ], [ "dsid = 'm1_yearly_dataset'\nX = get_Monash_forecasting_data(dsid, force_download=True, remove_from_disk=True)\ntest_eq(X.shape, (181, 1, 58))", "Dataset: m1_yearly_dataset\ndownloading data...\n...data downloaded\ndecompressing data...\n...data decompressed\nconverting dataframe to numpy array...\n...dataframe converted to numpy array\n\nX.shape: (181, 1, 58)\nfreq: yearly\nforecast_horizon: 6\ncontain_missing_values: False\ncontain_equal_length: False\n" ], [ "#hide\nfrom tsai.imports import create_scripts\nfrom tsai.export import get_nb_name\nnb_name = get_nb_name()\ncreate_scripts(nb_name);", "_____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", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# default_exp models_resnet_3d", "_____no_output_____" ] ], [ [ "# Models using 3D convolutions\n\n> This module focuses on preparing the data of the UCF101 dataset to be used with the core functions.\n\nRefs.\n[understanding-1d-and-3d-convolution](https://towardsdatascience.com/understanding-1d-and-3d-convolution-neural-network-keras-9d8f76e29610)", "_____no_output_____" ] ], [ [ "#export\nimport torch\nimport torch.nn as nn\nimport torchvision # used to download the model\nimport torch.nn.functional as F\nfrom torch.autograd import Variable\nimport math", "_____no_output_____" ], [ "#export\ndef conv3x3x3(in_channels, out_channels, stride=1):\n # 3x3x3 convolution with padding\n return nn.Conv3d(\n in_channels,\n out_channels,\n kernel_size=3,\n stride=stride,\n padding=1,\n bias=False)\n\n\ndef downsample_basic_block(x, planes, stride):\n out = F.avg_pool3d(x, kernel_size=1, stride=stride)\n zero_pads = torch.Tensor(\n out.size(0), planes - out.size(1), out.size(2), out.size(3),\n out.size(4)).zero_()\n \n if isinstance(out.data, torch.cuda.FloatTensor):\n zero_pads = zero_pads.cuda()\n\n out = Variable(torch.cat([out.data, zero_pads], dim=1))\n\n return out", "_____no_output_____" ], [ "#export\nclass BasicBlock(nn.Module):\n expansion = 1\n\n def __init__(self, in_channels, channels, stride=1, downsample=None):\n super(BasicBlock, self).__init__()\n self.conv1 = conv3x3x3(in_channels, channels, stride)\n self.bn1 = nn.BatchNorm3d(channels)\n self.relu = nn.ReLU(inplace=True)\n self.conv2 = conv3x3x3(channels, channels)\n self.bn2 = nn.BatchNorm3d(channels)\n self.downsample = downsample\n self.stride = stride\n\n def forward(self, x):\n residual = x\n\n out = self.conv1(x)\n out = self.bn1(out)\n out = self.relu(out)\n\n out = self.conv2(out)\n out = self.bn2(out)\n\n if self.downsample is not None:\n residual = self.downsample(x)\n\n out += residual\n out = self.relu(out)\n\n return out", "_____no_output_____" ], [ "#export \nclass Bottleneck(nn.Module):\n expansion = 4\n\n def __init__(self, inplanes, planes, stride=1, downsample=None):\n super(Bottleneck, self).__init__()\n self.conv1 = nn.Conv3d(inplanes, planes, kernel_size=1, bias=False)\n self.bn1 = nn.BatchNorm3d(planes)\n self.conv2 = nn.Conv3d(planes, planes, kernel_size=3, stride=stride,\n padding=1, bias=False)\n self.bn2 = nn.BatchNorm3d(planes)\n self.conv3 = nn.Conv3d(planes, planes * 4, kernel_size=1, bias=False)\n self.bn3 = nn.BatchNorm3d(planes * 4)\n self.relu = nn.ReLU(inplace=True)\n self.downsample = downsample\n self.stride = stride\n\n def forward(self, x):\n residual = x\n\n out = self.conv1(x)\n out = self.bn1(out)\n out = self.relu(out)\n\n out = self.conv2(out)\n out = self.bn2(out)\n out = self.relu(out)\n\n out = self.conv3(out)\n out = self.bn3(out)\n\n if self.downsample is not None:\n residual = self.downsample(x)\n\n out += residual\n out = self.relu(out)\n\n return out", "_____no_output_____" ], [ "#export\nclass ResNet(nn.Module):\n\n def __init__(self, block, layers, sample_size,\n sample_duration, shortcut_type='B',\n num_classes=400):\n self.inplanes = 64\n super(ResNet, self).__init__()\n \n self.conv1 = nn.Conv3d(3,\n 64,\n kernel_size=7,\n stride=(1, 2, 2),\n padding=(3, 3, 3),\n bias=False)\n self.bn1 = nn.BatchNorm3d(64)\n self.relu = nn.ReLU(inplace=True)\n self.maxpool = nn.MaxPool3d(kernel_size=(3, 3, 3), stride=2, padding=1)\n self.layer1 = self._make_layer(block, 64, layers[0], shortcut_type)\n self.layer2 = self._make_layer(block, 128, layers[1], shortcut_type, stride=2)\n self.layer3 = self._make_layer(block, 256, layers[2], shortcut_type, stride=2)\n self.layer4 = self._make_layer(block, 512, layers[3], shortcut_type, stride=2)\n last_duration = int(math.ceil(sample_duration / 16))\n last_size = int(math.ceil(sample_size / 32))\n self.avgpool = nn.AvgPool3d((last_duration, last_size, last_size), stride=1)\n self.fc = nn.Linear(512 * block.expansion, num_classes)\n\n for m in self.modules():\n if isinstance(m, nn.Conv3d):\n m.weight = nn.init.kaiming_normal_(m.weight, mode='fan_out')\n elif isinstance(m, nn.BatchNorm3d):\n m.weight.data.fill_(1)\n m.bias.data.zero_()\n\n def _make_layer(self, block, planes, blocks, shortcut_type, stride=1):\n downsample = None\n if stride != 1 or self.inplanes != planes * block.expansion:\n if shortcut_type == 'A':\n downsample = partial(\n downsample_basic_block,\n planes=planes * block.expansion,\n stride=stride)\n else:\n downsample = nn.Sequential(\n nn.Conv3d(self.inplanes,\n planes * block.expansion,\n kernel_size=1,\n stride=stride,\n bias=False), \n nn.BatchNorm3d(planes * block.expansion))\n\n layers = []\n layers.append(block(self.inplanes, planes, stride, downsample))\n self.inplanes = planes * block.expansion\n for i in range(1, blocks):\n layers.append(block(self.inplanes, planes))\n\n return nn.Sequential(*layers)\n\n def forward(self, x): \n # only when using fastai\n x = x.permute(0,2,1,3,4)\n \n with torch.no_grad():\n h = self.conv1(x)\n h = self.bn1(h)\n h = self.relu(h)\n h = self.maxpool(h)\n\n h = self.layer1(h)\n h = self.layer2(h)\n h = self.layer3(h)\n h = self.layer4[0](h)\n# h = self.layer4(h)\n\n h = self.avgpool(h)\n\n h = h.view(h.size(0), -1)\n h = self.fc(h)\n\n return h", "_____no_output_____" ], [ "#export\nclass ResNet50_3D(nn.Module):\n def __init__(self, num_classes, **kwargs):\n super(ResNet50_3D, self).__init__()\n \n if 'model_pretrained' in kwargs.keys():\n print(f\"ResNet50_3D is loading pretrained ResNet50 from {kwargs['model_pretrained']}\")\n pretrained_resnet50 = torch.load('./model-pretrained/resnet-50-kinetics.pth', map_location=torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\"))\n kwargs.pop('model_pretrained', None)\n resnet = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs) \n \n keys = [k for k,v in pretrained_resnet50['state_dict'].items()]\n pretrained_state_dict = {k[7:]: v.cpu() for k, v in pretrained_resnet50['state_dict'].items()}\n resnet.load_state_dict(pretrained_state_dict) \n \n else:\n resnet = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs)\n \n # chenage the last layer to match number of classes\n resnet.fc = nn.Linear(resnet.fc.weight.shape[1], num_classes)\n \n# self.feature_extractor = nn.Sequential(*list(resnet.children())[:-1])\n self.feature_extractor = resnet\n# self.final = nn.Sequential(\n# nn.Linear(resnet.fc.in_features, num_classes),\n# )\n \n\n def forward(self, x):\n # The input x will now be size [batch_size, c, seq_len, h, w]. \n # This is what I might get..Sequence (bs, 4, 3, 224, 224)\n #batch_size, c, h, w = x.shape\n #x = x.view(batch_size, c, h, w)\n x = self.feature_extractor(x)\n #x = x.view(batch_size, -1)\n# x = self.final(x)\n #x = x.view(batch_size, -1)\n return x\n ", "_____no_output_____" ], [ "#export\ndef resnet10(**kwargs):\n \"\"\"Constructs a ResNet-18 model.\n \"\"\"\n model = ResNet(BasicBlock, [1, 1, 1, 1], **kwargs)\n return model\n\n\ndef resnet18(**kwargs):\n \"\"\"Constructs a ResNet-18 model.\n \"\"\"\n model = ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)\n return model\n\n\ndef resnet34(**kwargs):\n \"\"\"Constructs a ResNet-34 model.\n \"\"\"\n model = ResNet(BasicBlock, [3, 4, 6, 3], **kwargs)\n return model\n\n\ndef resnet50(**kwargs):\n \"\"\"Constructs a ResNet-50 model.\n \"\"\"\n print('function resnet50')\n model = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs)\n return model\n\ndef resnet101(**kwargs):\n \"\"\"Constructs a ResNet-101 model.\n \"\"\"\n model = ResNet(Bottleneck, [3, 4, 23, 3], **kwargs)\n return model\n\n\ndef resnet152(**kwargs):\n \"\"\"Constructs a ResNet-101 model.\n \"\"\"\n model = ResNet(Bottleneck, [3, 8, 36, 3], **kwargs)\n return model\n\n\ndef resnet200(**kwargs):\n \"\"\"Constructs a ResNet-101 model.\n \"\"\"\n model = ResNet(Bottleneck, [3, 24, 36, 3], **kwargs)\n return model", "_____no_output_____" ], [ "model = ResNet50_3D(num_classes=101, sample_size=224, sample_duration=16, model_pretrained='./model-pretrained/resnet-50-kinetics.pth')", "ResNet50_3D is loading pretrained ResNet50 from ./model-pretrained/resnet-50-kinetics.pth\n" ], [ "model", "_____no_output_____" ], [ "model = resnet10(sample_size=224, sample_duration=16)", "_____no_output_____" ], [ "model", "_____no_output_____" ], [ "#hide\nfrom nbdev.export import *\nnotebook2script()", "Converted 01_dataset_ucf101.ipynb.\nConverted 02_avi.ipynb.\nConverted 04_data_augmentation.ipynb.\nConverted 05_models.ipynb.\nConverted 06_models-resnet_3d.ipynb.\nConverted 07_utils.ipynb.\nConverted 10_run-baseline.ipynb.\nConverted 11_run-sequence-convlstm.ipynb.\nConverted 12_run-sequence-3d.ipynb.\nConverted 14_fastai_sequence.ipynb.\nConverted index.ipynb.\n" ] ] ]

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

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# OUTDATED, the examples moved to the gallery\n## See https://empymod.github.io/emg3d-gallery\n\n----\n\n# 3D with tri-axial anisotropy comparison between `emg3d` and `SimPEG`\n\n`SimPEG` is an open source python package for simulation and gradient based parameter estimation in geophysical applications, see https://simpeg.xyz. We can use `emg3d` as a solver for `SimPEG`, and compare it with the forward solver `Pardiso`.\n\n#### Requires\n- **emg3d >= 0.9.0**\n- ``discretize``, ``SimPEG``, ``pymatsolver``\n- ``numpy``, ``scipy``, ``numba``, ``matplotlib``\n\nNote, in order to use the `Pardiso`-solver `pymatsolver` has to be installed via `conda`, not via `pip`!", "_____no_output_____" ] ], [ [ "import time\nimport emg3d\nimport discretize\nimport numpy as np\nimport SimPEG, pymatsolver\nfrom SimPEG.EM import FDEM\nfrom SimPEG import Mesh, Maps\nfrom SimPEG.Survey import Data\nimport matplotlib.pyplot as plt\nfrom timeit import default_timer\nfrom contextlib import contextmanager\nfrom datetime import datetime, timedelta\nfrom pymatsolver import Pardiso as Solver\nfrom matplotlib.colors import LogNorm, SymLogNorm\n\n%load_ext memory_profiler", "_____no_output_____" ], [ "# Style adjustments\n%matplotlib notebook\nplt.style.use('ggplot')", "_____no_output_____" ] ], [ [ "## Model and survey parameters", "_____no_output_____" ] ], [ [ "# Depths (0 is sea-surface)\nwater_depth = 1000\ntarget_x = np.r_[-500, 500]\ntarget_y = target_x\ntarget_z = -water_depth + np.r_[-400, -100]\n\n# Resistivities\nres_air = 2e8\nres_sea = 0.33\nres_back = [1., 2., 3.] # Background in x-, y-, and z-directions\nres_target = 100.\n\nfreq = 1.0\n\nsrc = [-100, 100, 0, 0, -900, -900]", "_____no_output_____" ] ], [ [ "## Mesh and source-field", "_____no_output_____" ] ], [ [ "# skin depth\nskin_depth = 503/np.sqrt(res_back[0]/freq)\nprint(f\"\\nThe skin_depth is {skin_depth} m.\\n\")\n\ncs = 100 # 100 m min_width of cells\n\npf = 1.15 # Padding factor x- and y-directions\npfz = 1.35 # z-direction\nnpadx = 12 # Nr of padding in x- and y-directions\nnpadz = 9 # z-direction\n\ndomain_x = 4000 # x- and y-domain\ndomain_z = - target_z[0] # z-domain\n\n# Create mesh\nmesh = Mesh.TensorMesh(\n [[(cs, npadx, -pf), (cs, int(domain_x/cs)), (cs, npadx, pf)], \n [(cs, npadx, -pf), (cs, int(domain_x/cs)), (cs, npadx, pf)], \n [(cs, npadz, -pfz), (cs, int(domain_z/cs)), (cs, npadz, pfz)]]\n)\n\n# Center mesh\nmesh.x0 = np.r_[-mesh.hx.sum()/2, -mesh.hy.sum()/2, -mesh.hz[:-npadz].sum()]\n\n# Create the source field for this mesh and given frequency\nsfield = emg3d.utils.get_source_field(mesh, src, freq, strength=0)\n\n# We take the receiver locations at the actual CCx-locations\nrec_x = mesh.vectorCCx[12:-12]\nprint(f\"Receiver locations:\\n{rec_x}\\n\")\n\nmesh", "\nThe skin_depth is 503.0 m.\n\nReceiver locations:\n[-1950. -1850. -1750. -1650. -1550. -1450. -1350. -1250. -1150. -1050.\n -950. -850. -750. -650. -550. -450. -350. -250. -150. -50.\n 50. 150. 250. 350. 450. 550. 650. 750. 850. 950.\n 1050. 1150. 1250. 1350. 1450. 1550. 1650. 1750. 1850. 1950.]\n\n" ] ], [ [ "## Create model", "_____no_output_____" ] ], [ [ "# Layered_background\nres_x = res_air*np.ones(mesh.nC)\nres_x[mesh.gridCC[:, 2] <= 0] = res_sea\n\nres_y = res_x.copy()\nres_z = res_x.copy()\n\nres_x[mesh.gridCC[:, 2] <= -water_depth] = res_back[0]\nres_y[mesh.gridCC[:, 2] <= -water_depth] = res_back[1]\nres_z[mesh.gridCC[:, 2] <= -water_depth] = res_back[2]\n\nres_x_bg = res_x.copy()\nres_y_bg = res_y.copy()\nres_z_bg = res_z.copy()\n\n# Include the target\ntarget_inds = (\n (mesh.gridCC[:, 0] >= target_x[0]) & (mesh.gridCC[:, 0] <= target_x[1]) &\n (mesh.gridCC[:, 1] >= target_y[0]) & (mesh.gridCC[:, 1] <= target_y[1]) &\n (mesh.gridCC[:, 2] >= target_z[0]) & (mesh.gridCC[:, 2] <= target_z[1])\n)\nres_x[target_inds] = res_target\nres_y[target_inds] = res_target\nres_z[target_inds] = res_target\n\n# Create emg3d-models for given frequency\npmodel = emg3d.utils.Model(mesh, res_x, res_y, res_z)\npmodel_bg = emg3d.utils.Model(mesh, res_x_bg, res_y_bg, res_z_bg)\n\n# Plot a slice\nmesh.plot_3d_slicer(pmodel.res_x, zslice=-1100, clim=[0, 2],\n xlim=(-4000, 4000), ylim=(-4000, 4000), zlim=(-2000, 500))", "_____no_output_____" ] ], [ [ "## Calculate `emg3d`", "_____no_output_____" ] ], [ [ "%memit em3_tg = emg3d.solver.solver(mesh, pmodel, sfield, verb=3, nu_pre=0, semicoarsening=True)", "\n:: emg3d START :: 21:03:39 ::\n\n MG-cycle : 'F' sslsolver : False\n semicoarsening : True [1 2 3] tol : 1e-06\n linerelaxation : False [0] maxit : 50\n nu_{i,1,c,2} : 0, 0, 1, 2 verb : 3\n Original grid : 64 x 64 x 32 => 131,072 cells\n Coarsest grid : 2 x 2 x 2 => 8 cells\n Coarsest level : 5 ; 5 ; 4 \n\n [hh:mm:ss] rel. error [abs. error, last/prev] l s\n\n h_\n 2h_ \\ /\n 4h_ \\ /\\ / \n 8h_ \\ /\\ / \\ / \n 16h_ \\ /\\ / \\ / \\ / \n 32h_ \\/\\/ \\/ \\/ \\/ \n\n [21:03:40] 5.227e-02 after 1 F-cycles [2.918e-07, 0.052] 0 1\n [21:03:40] 6.108e-03 after 2 F-cycles [3.410e-08, 0.117] 0 2\n [21:03:41] 7.462e-04 after 3 F-cycles [4.166e-09, 0.122] 0 3\n [21:03:41] 1.133e-04 after 4 F-cycles [6.328e-10, 0.152] 0 1\n [21:03:42] 3.089e-05 after 5 F-cycles [1.724e-10, 0.272] 0 2\n [21:03:42] 4.369e-06 after 6 F-cycles [2.439e-11, 0.141] 0 3\n [21:03:43] 1.529e-06 after 7 F-cycles [8.538e-12, 0.350] 0 1\n [21:03:43] 9.542e-07 after 8 F-cycles [5.327e-12, 0.624] 0 2\n\n > CONVERGED\n > MG cycles : 8\n > Final rel. error : 9.542e-07\n\n:: emg3d END :: 21:03:43 :: runtime = 0:00:04\n\npeak memory: 329.74 MiB, increment: 62.88 MiB\n" ], [ "%memit em3_bg = emg3d.solver.solver(mesh, pmodel_bg, sfield, verb=3, nu_pre=0, semicoarsening=True)", "\n:: emg3d START :: 21:03:43 ::\n\n MG-cycle : 'F' sslsolver : False\n semicoarsening : True [1 2 3] tol : 1e-06\n linerelaxation : False [0] maxit : 50\n nu_{i,1,c,2} : 0, 0, 1, 2 verb : 3\n Original grid : 64 x 64 x 32 => 131,072 cells\n Coarsest grid : 2 x 2 x 2 => 8 cells\n Coarsest level : 5 ; 5 ; 4 \n\n [hh:mm:ss] rel. error [abs. error, last/prev] l s\n\n h_\n 2h_ \\ /\n 4h_ \\ /\\ / \n 8h_ \\ /\\ / \\ / \n 16h_ \\ /\\ / \\ / \\ / \n 32h_ \\/\\/ \\/ \\/ \\/ \n\n [21:03:44] 5.250e-02 after 1 F-cycles [2.931e-07, 0.052] 0 1\n [21:03:45] 6.468e-03 after 2 F-cycles [3.611e-08, 0.123] 0 2\n [21:03:45] 8.049e-04 after 3 F-cycles [4.494e-09, 0.124] 0 3\n [21:03:45] 1.435e-04 after 4 F-cycles [8.012e-10, 0.178] 0 1\n [21:03:46] 4.756e-05 after 5 F-cycles [2.655e-10, 0.331] 0 2\n [21:03:46] 5.863e-06 after 6 F-cycles [3.274e-11, 0.123] 0 3\n [21:03:47] 1.947e-06 after 7 F-cycles [1.087e-11, 0.332] 0 1\n [21:03:47] 1.068e-06 after 8 F-cycles [5.965e-12, 0.549] 0 2\n [21:03:48] 3.441e-07 after 9 F-cycles [1.921e-12, 0.322] 0 3\n\n > CONVERGED\n > MG cycles : 9\n > Final rel. error : 3.441e-07\n\n:: emg3d END :: 21:03:48 :: runtime = 0:00:04\n\npeak memory: 337.88 MiB, increment: 26.70 MiB\n" ] ], [ [ "## Calculate `SimPEG`", "_____no_output_____" ] ], [ [ "# Set up the PDE\nprob = FDEM.Problem3D_e(mesh, sigmaMap=Maps.IdentityMap(mesh), Solver=Solver)\n\n# Set up the receivers\nrx_locs = Mesh.utils.ndgrid([rec_x, np.r_[0], np.r_[-water_depth]])\nrx_list = [\n FDEM.Rx.Point_e(orientation='x', component=\"real\", locs=rx_locs), \n FDEM.Rx.Point_e(orientation='x', component=\"imag\", locs=rx_locs)\n]\n\n# We use the emg3d-source-vector, to ensure we use the same in both cases\nsrc_sp = FDEM.Src.RawVec_e(rx_list, s_e=sfield.vector, freq=freq)\nsrc_list = [src_sp]\nsurvey = FDEM.Survey(src_list)\n\n# Create the simulation\nprob.pair(survey)", "_____no_output_____" ], [ "@contextmanager\ndef ctimeit(before=''):\n \"\"\"Print time used by commands run within the context manager.\"\"\"\n t0 = default_timer()\n yield\n t1 = default_timer() - t0\n print(f\"{before}{timedelta(seconds=np.round(t1))}\")", "_____no_output_____" ], [ "with ctimeit(\"SimPEG runtime: \"):\n %memit spg_tg_dobs = survey.dpred(np.vstack([1./res_x, 1./res_y, 1./res_z]).T)\nspg_tg = Data(survey, dobs=spg_tg_dobs)", "peak memory: 10468.39 MiB, increment: 10138.32 MiB\nSimPEG runtime: 0:03:52\n" ], [ "with ctimeit(\"SimPEG runtime: \"):\n %memit spg_bg_dobs = survey.dpred(np.vstack([1./res_x_bg, 1./res_y_bg, 1./res_z_bg]).T)\nspg_bg = Data(survey, dobs=spg_bg_dobs)", "peak memory: 10460.16 MiB, increment: 9731.63 MiB\nSimPEG runtime: 0:03:53\n" ] ], [ [ "## Plot result", "_____no_output_____" ] ], [ [ "ix1, ix2 = 12, 12\niy = 32\niz = 13\n\nmesh.vectorCCx[ix1], mesh.vectorCCx[-ix2-1], mesh.vectorNy[iy], mesh.vectorNz[iz]", "_____no_output_____" ], [ "plt.figure(figsize=(9, 6))\n\nplt.subplot(221)\nplt.title('|Real(response)|')\nplt.semilogy(rec_x/1e3, np.abs(em3_bg.fx[ix1:-ix2, iy, iz].real))\nplt.semilogy(rec_x/1e3, np.abs(em3_tg.fx[ix1:-ix2, iy, iz].real))\nplt.semilogy(rec_x/1e3, np.abs(spg_bg[src_sp, rx_list[0]]), 'C4--')\nplt.semilogy(rec_x/1e3, np.abs(spg_tg[src_sp, rx_list[0]]), 'C5--')\nplt.xlabel('Offset (km)')\nplt.ylabel('$E_x$ (V/m)')\n\nplt.subplot(223)\nplt.title('|Imag(response)|')\nplt.semilogy(rec_x/1e3, np.abs(em3_bg.fx[ix1:-ix2, iy, iz].imag), label='emg3d BG')\nplt.semilogy(rec_x/1e3, np.abs(em3_tg.fx[ix1:-ix2, iy, iz].imag), label='emg3d target')\nplt.semilogy(rec_x/1e3, np.abs(spg_bg[src_sp, rx_list[1]]), 'C4--', label='SimPEG BG')\nplt.semilogy(rec_x/1e3, np.abs(spg_tg[src_sp, rx_list[1]]), 'C5--', label='SimPEG target')\nplt.xlabel('Offset (km)')\nplt.ylabel('$E_x$ (V/m)')\nplt.legend()\n\nplt.subplot(222)\nplt.title('Relative error Real')\nplt.semilogy(rec_x/1e3, 100*np.abs((spg_bg[src_sp, rx_list[0]]-em3_bg.fx[ix1:-ix2, iy, iz].real)/\n em3_bg.fx[ix1:-ix2, iy, iz].real), label='BG')\nplt.semilogy(rec_x/1e3, 100*np.abs((spg_tg[src_sp, rx_list[0]]-em3_tg.fx[ix1:-ix2, iy, iz].real)/\n em3_tg.fx[ix1:-ix2, iy, iz].real), label='target')\n\nplt.xlabel('Offset (km)')\nplt.ylabel('Rel. Error (%)')\nplt.legend()\n\nplt.subplot(224)\nplt.title('Relative error (%) Imag')\nplt.semilogy(rec_x/1e3, 100*np.abs((spg_bg[src_sp, rx_list[1]]-em3_bg.fx[ix1:-ix2, iy, iz].imag)/\n em3_bg.fx[ix1:-ix2, iy, iz].imag), label='BG')\nplt.semilogy(rec_x/1e3, 100*np.abs((spg_tg[src_sp, rx_list[1]]-em3_tg.fx[ix1:-ix2, iy, iz].imag)/\n em3_tg.fx[ix1:-ix2, iy, iz].imag), label='target')\n\nplt.xlabel('Offset (km)')\nplt.ylabel('Rel. Error (%)')\nplt.legend()\n\nplt.tight_layout()\nplt.show()", "_____no_output_____" ], [ "emg3d.Report([discretize, SimPEG, pymatsolver])", "_____no_output_____" ] ] ]

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

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

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Portfolio Optimization\nThe portfolio optimization problem is a combinatorial optimization problem that seeks the optimal combination of assets based on the balance between risk and return.", "_____no_output_____" ], [ "## Cost Function\nThe cost function for solving the portfolio optimization problem is", "_____no_output_____" ], [ "$$\nE = -\\sum \\mu_i q_i + \\gamma \\sum \\delta_{i,j}q_i q_j\n$$", "_____no_output_____" ], [ "The 1st term shows the return of the assets and the 2nd as risk we estimate.", "_____no_output_____" ], [ "## Example\nNow, let's choose two of the six assets and find the optimal combination.", "_____no_output_____" ] ], [ [ "import numpy as np\nfrom blueqat import vqe\nfrom blueqat.pauli import I, X, Y, Z\nfrom blueqat.pauli import from_qubo\nfrom blueqat.pauli import qubo_bit as q\nfrom blueqat import Circuit", "_____no_output_____" ] ], [ [ "Use the following as return data", "_____no_output_____" ] ], [ [ "asset_return = np.diag([-0.026,-0.031,-0.007,-0.022,-0.010,-0.055])\nprint(asset_return)", "[[-0.026 0. 0. 0. 0. 0. ]\n [ 0. -0.031 0. 0. 0. 0. ]\n [ 0. 0. -0.007 0. 0. 0. ]\n [ 0. 0. 0. -0.022 0. 0. ]\n [ 0. 0. 0. 0. -0.01 0. ]\n [ 0. 0. 0. 0. 0. -0.055]]\n" ] ], [ [ "Use the following as risk data", "_____no_output_____" ] ], [ [ "asset_risk = [[0,0.0015,0.0012,0.0018,0.0022,0.0012],[0,0,0.0017,0.0022,0.0005,0.0019],[0,0,0,0.0040,0.0032,0.0024],[0,0,0,0,0.0012,0.0076],[0,0,0,0,0,0.0021],[0,0,0,0,0,0]]\nnp.asarray(asset_risk)", "_____no_output_____" ] ], [ [ "It is then converted to Hamiltonian and calculated. In addition, this time there is the constraint of selecting two out of six assets, which is implemented using XYmixer.", "_____no_output_____" ] ], [ [ "#convert qubo to pauli\nqubo = asset_return + np.asarray(asset_risk)*0.5\nhamiltonian = from_qubo(qubo)\n\ninit = Circuit(6).x[0,1]\nmixer = I()*0\nfor i in range(5):\n for j in range(i+1, 6):\n mixer += (X[i]*X[j] + Y[i]*Y[j])*0.5\n\nstep = 1\n\nresult = vqe.Vqe(vqe.QaoaAnsatz(hamiltonian, step, init, mixer)).run()\nprint(result.most_common(12))", "(((0, 0, 0, 0, 1, 1), 0.9999994909753838), ((0, 0, 1, 0, 0, 1), 5.082984897154482e-07), ((0, 0, 0, 1, 0, 1), 4.5366922587264707e-10), ((0, 1, 0, 0, 0, 1), 1.8161653052486483e-10), ((0, 0, 1, 0, 1, 0), 9.063023975859213e-11), ((0, 0, 0, 1, 1, 0), 1.618082643262117e-13), ((1, 0, 0, 0, 0, 1), 1.6228830532381364e-14), ((0, 1, 0, 0, 1, 0), 1.618262335727315e-14), ((0, 1, 0, 1, 0, 0), 9.246381676299576e-17), ((0, 0, 1, 1, 0, 0), 8.260881354525854e-20), ((1, 0, 0, 1, 0, 0), 1.650702183251039e-20), ((0, 1, 1, 0, 0, 0), 8.25793442702573e-21))\n" ], [ "result.circuit.run(backend=\"draw\")", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Доверительные интервалы на основе bootstrap", "_____no_output_____" ] ], [ [ "import numpy as np\nimport pandas as pd", "_____no_output_____" ], [ "%pylab inline", "Populating the interactive namespace from numpy and matplotlib\n" ] ], [ [ "## Загрузка данных", "_____no_output_____" ], [ "### Время ремонта телекоммуникаций", "_____no_output_____" ], [ "Verizon — основная региональная телекоммуникационная компания (Incumbent Local Exchange Carrier, ILEC) в западной \nчасти США. В связи с этим данная компания обязана предоставлять сервис ремонта телекоммуникационного оборудования \nне только для своих клиентов, но и для клиентов других локальных телекоммуникационых компаний (Competing Local Exchange Carriers, CLEC). При этом в случаях, когда время ремонта оборудования для клиентов других компаний существенно выше, чем для собственных, Verizon может быть оштрафована. ", "_____no_output_____" ] ], [ [ "data = pd.read_csv('verizon.txt', sep='\\t')\ndata.shape", "_____no_output_____" ], [ "data.head()", "_____no_output_____" ], [ "data.Group.value_counts()", "_____no_output_____" ], [ "pylab.figure(figsize(12, 5))\npylab.subplot(1,2,1)\npylab.hist(data[data.Group == 'ILEC'].Time, bins = 20, color = 'b', range = (0, 100), label = 'ILEC')\npylab.legend()\n\npylab.subplot(1,2,2)\npylab.hist(data[data.Group == 'CLEC'].Time, bins = 20, color = 'r', range = (0, 100), label = 'CLEC')\npylab.legend()\n\npylab.show()", "_____no_output_____" ] ], [ [ "## Bootstrap", "_____no_output_____" ] ], [ [ "def get_bootstrap_samples(data, n_samples):\n indices = np.random.randint(0, len(data), (n_samples, len(data)))\n samples = data[indices]\n return samples", "_____no_output_____" ], [ "def stat_intervals(stat, alpha):\n boundaries = np.percentile(stat, [100 * alpha / 2., 100 * (1 - alpha / 2.)])\n return boundaries", "_____no_output_____" ] ], [ [ "### Интервальная оценка медианы", "_____no_output_____" ] ], [ [ "ilec_time = data[data.Group == 'ILEC'].Time.values\nclec_time = data[data.Group == 'CLEC'].Time.values", "_____no_output_____" ], [ "np.random.seed(0)\n\nilec_median_scores = map(np.median, get_bootstrap_samples(ilec_time, 1000))\nclec_median_scores = map(np.median, get_bootstrap_samples(clec_time, 1000))\n\nprint \"95% confidence interval for the ILEC median repair time:\", stat_intervals(ilec_median_scores, 0.05)\nprint \"95% confidence interval for the CLEC median repair time:\", stat_intervals(clec_median_scores, 0.05)", "_____no_output_____" ] ], [ [ "### Точечная оценка разности медиан", "_____no_output_____" ] ], [ [ "print \"difference between medians:\", np.median(clec_time) - np.median(ilec_time)", "_____no_output_____" ] ], [ [ "### Интервальная оценка разности медиан", "_____no_output_____" ] ], [ [ "delta_median_scores = map(lambda x: x[1] - x[0], zip(ilec_median_scores, clec_median_scores))", "_____no_output_____" ], [ "print \"95% confidence interval for the difference between medians\", stat_intervals(delta_median_scores, 0.05)", "_____no_output_____" ] ] ]

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

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

[ "MIT" ]

[ "MIT" ]