Upload 2 files

P2 - Secom Notebook2 - Mercury.ipynb

@@ -26,7 +26,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 431,
+   "execution_count": 1,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -34,13 +34,6 @@
    },
    "outputs": [],
    "source": [
-    "# import pandas for data manipulation\n",
-    "# import numpy for numerical computation\n",
-    "# import seaborn for data visualization\n",
-    "# import matplotlib for data visualization\n",
-    "# import stats for statistical analysis\n",
-    "# import train_test_split for splitting data into training and testing sets\n",
-    "\n",
     "import mercury as mr\n",
     "import pandas as pd\n",
     "import numpy as np\n",
@@ -48,8 +41,7 @@
     "import matplotlib.pyplot as plt\n",
     "from scipy import stats\n",
     "from sklearn.model_selection import train_test_split\n",
-    "import warnings\n",
-    "warnings.filterwarnings('ignore')\n",
+    "\n",
     "from mlxtend.plotting import plot_confusion_matrix\n",
     "from sklearn.metrics import confusion_matrix, accuracy_score, precision_score, recall_score, f1_score\n",
     "from mlxtend.plotting import plot_confusion_matrix"
@@ -57,7 +49,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 432,
+   "execution_count": 2,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -68,7 +60,7 @@
      "data": {
       "application/mercury+json": {
        "allow_download": true,
-       "code_uid": "App.0.40.24.1-randf68a3764",
+       "code_uid": "App.0.40.24.1-rand53016c34",
        "continuous_update": false,
        "description": "Recumpute everything dynamically",
        "full_screen": true,
@@ -100,7 +92,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 433,
+   "execution_count": 3,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -141,7 +133,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 434,
+   "execution_count": 4,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -151,18 +143,18 @@
     {
      "data": {
       "application/mercury+json": {
-       "code_uid": "Text.0.40.15.11-randa5faa9c1",
+       "code_uid": "Text.0.40.15.11-rand81961de4",
        "disabled": false,
        "hidden": false,
        "label": "Test Size Ratio",
-       "model_id": "a2eb64736c1146fc835a6b2afa84c9c8",
+       "model_id": "cddcc5c10139484dbc19e59ce26f012c",
        "rows": 1,
        "url_key": "",
        "value": "0.25",
        "widget": "Text"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "a2eb64736c1146fc835a6b2afa84c9c8",
+       "model_id": "cddcc5c10139484dbc19e59ce26f012c",
        "version_major": 2,
        "version_minor": 0
       },
@@ -176,18 +168,18 @@
     {
      "data": {
       "application/mercury+json": {
-       "code_uid": "Text.0.40.15.14-rand83abdf01",
+       "code_uid": "Text.0.40.15.14-rand72283006",
        "disabled": false,
        "hidden": false,
        "label": "Random State Integer",
-       "model_id": "7c9d97ed67cb4252a11f2802fc495482",
+       "model_id": "dcac3f415e624b61aac8a3578e285bca",
        "rows": 1,
        "url_key": "",
        "value": "13",
        "widget": "Text"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "7c9d97ed67cb4252a11f2802fc495482",
+       "model_id": "dcac3f415e624b61aac8a3578e285bca",
        "version_major": 2,
        "version_minor": 0
       },
@@ -236,7 +228,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 435,
+   "execution_count": 5,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -339,7 +331,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 436,
+   "execution_count": 6,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -347,26 +339,27 @@
    },
    "outputs": [],
    "source": [
-    "def outlier_removal(z_df, z_threshold=4):\n",
+    "def outlier_removal(z_df, action = 'ignore', z_threshold=3):\n",
     "    \n",
     "    global outlier_var\n",
     "    global outlier_removal_report0\n",
     "    global outlier_removal_report1\n",
     "\n",
-    "\n",
-    "    if z_threshold == 'none':\n",
-    "        outlier_removal_report0 = 'No outliers were removed'\n",
+    "    if action == 'ignore':\n",
+    "        outlier_removal_report0 = 'No z-score threshold was selected'\n",
     "        outlier_var = 'none'\n",
-    "        return z_df\n",
-    "        \n",
-    "    else:\n",
+    "        z_df_copy = z_df.copy()\n",
+    "        outlier_removal_report1 = 'No outliers were removed from the dataset'\n",
+    "    \n",
+    "    if action == 'remove':\n",
+    "                \n",
     "        outlier_removal_report0 = 'The z-score threshold is:', z_threshold\n",
     "\n",
     "        z_df_copy = z_df.copy()\n",
     "\n",
     "        z_scores = np.abs(stats.zscore(z_df_copy))\n",
     "\n",
-    "     # Identify the outliers in the dataset using the z-score method\n",
+    "        # Identify the outliers in the dataset using the z-score method\n",
     "        outliers_mask = z_scores > z_threshold\n",
     "        z_df_copy[outliers_mask] = np.nan\n",
     "\n",
@@ -375,6 +368,25 @@
     "\n",
     "        outlier_var = z_threshold\n",
     "\n",
+    "    if action == 'push':\n",
+    "\n",
+    "        # push the outliers to the threshold value\n",
+    "        outlier_removal_report0 = 'The z-score threshold is:', z_threshold\n",
+    "\n",
+    "        z_df_copy = z_df.copy()\n",
+    "\n",
+    "        z_scores = np.abs(stats.zscore(z_df_copy))\n",
+    "\n",
+    "        # Identify the outliers in the dataset using the z-score method\n",
+    "        outliers_mask = z_scores > z_threshold\n",
+    "        z_df_copy[outliers_mask] = np.sign(z_df_copy[outliers_mask]) * (3 * np.std(z_df_copy)) + np.mean(z_df_copy)\n",
+    "\n",
+    "        outliers_count = np.count_nonzero(outliers_mask)\n",
+    "        outlier_removal_report1 = 'The number of outliers pushed to the boundaries is:', outliers_count\n",
+    "\n",
+    "        outlier_var = str(action) + '-' + str(z_threshold) + 's'\n",
+    "\n",
+    "   \n",
     "    return z_df_copy"
    ]
   },
@@ -392,7 +404,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 437,
+   "execution_count": 7,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -471,7 +483,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 438,
+   "execution_count": 8,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -487,7 +499,7 @@
     "    global imputation_report0\n",
     "    global imputation_report1\n",
     "    global imputation_report2\n",
-    "    global imputation_report3\n",
+    "  \n",
     "\n",
     "    imputation_report0 = 'Number of missing values before imputation: ', df_transform.isnull().sum().sum()\n",
     "\n",
@@ -560,7 +572,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 439,
+   "execution_count": 9,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -593,9 +605,9 @@
     "        return X_train_filtered, selected_columns\n",
     "    \n",
     "    if method == 'none':\n",
-    "        feature_selection_report = 'No feature selection has been applied'\n",
+    "        feature_selection_report0 = 'No feature selection has been applied'\n",
     "        X_train_filtered = X_train\n",
-    "        feature_selection_report = 'Shape of the training set after no feature selection: ', X_train_filtered.shape\n",
+    "        feature_selection_report1 = 'Shape of the training set after no feature selection: ', X_train_filtered.shape\n",
     "        feature_selection_var = 'none'\n",
     "        selected_features = X_train_filtered.columns\n",
     "        return X_train_filtered, selected_features        \n",
@@ -650,7 +662,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 440,
+   "execution_count": 10,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -665,13 +677,17 @@
     "    global imbalance_var\n",
     "    global imbalance_report0\n",
     "    global imbalance_report1\n",
+    "    global imbalance_report2\n",
+    "    \n",
+    "\n",
     "\n",
     "    if method == 'smote':        \n",
     "        from imblearn.over_sampling import SMOTE\n",
     "        sm = SMOTE(random_state=42)\n",
     "        X_train_res, y_train_res = sm.fit_resample(X_train, y_train)\n",
     "        imbalance_report0 = 'Shape of the training set after oversampling with SMOTE: ', X_train_res.shape\n",
-    "        imbalance_report1 = 'Value counts of the target variable after oversampling with SMOTE: ', y_train_res.value_counts()\n",
+    "        imbalance_report1 = 'Value counts of the target variable after oversampling with SMOTE: '\n",
+    "        imbalance_report2 = y_train_res.value_counts()\n",
     "        imbalance_var = 'smote'\n",
     "        return X_train_res, y_train_res\n",
     "    \n",
@@ -680,7 +696,8 @@
     "        rus = RandomUnderSampler(random_state=42)\n",
     "        X_train_res, y_train_res = rus.fit_resample(X_train, y_train)\n",
     "        imbalance_report0 = 'Shape of the training set after undersampling with RandomUnderSampler: ', X_train_res.shape\n",
-    "        imbalance_report1 = 'Value counts of the target variable after undersampling with RandomUnderSampler: ', y_train_res.value_counts()\n",
+    "        imbalance_report1 = 'Value counts of the target variable after undersampling with RandomUnderSampler: '\n",
+    "        imbalance_report2 = y_train_res.value_counts()\n",
     "        imbalance_var = 'undersampling'\n",
     "        return X_train_res, y_train_res\n",
     "    \n",
@@ -689,7 +706,8 @@
     "        ros = RandomOverSampler(random_state=42)\n",
     "        X_train_res, y_train_res = ros.fit_resample(X_train, y_train)\n",
     "        imbalance_report0 = 'Shape of the training set after oversampling with RandomOverSampler: ', X_train_res.shape\n",
-    "        imbalance_report1 = 'Value counts of the target variable after oversampling with RandomOverSampler: ', y_train_res.value_counts()\n",
+    "        imbalance_report1 = 'Value counts of the target variable after oversampling with RandomOverSampler: '\n",
+    "        imbalance_report2 = y_train_res.value_counts()\n",
     "        imbalance_var = 'rose'\n",
     "        return X_train_res, y_train_res\n",
     "    \n",
@@ -698,7 +716,8 @@
     "        X_train_res = X_train\n",
     "        y_train_res = y_train\n",
     "        imbalance_report0 = 'Shape of the training set after no resampling: ', X_train_res.shape\n",
-    "        imbalance_report1 = 'Value counts of the target variable after no resampling: ', y_train_res.value_counts()\n",
+    "        imbalance_report1 = 'Value counts of the target variable after no resampling: '\n",
+    "        imbalance_report2 = y_train_res.value_counts()\n",
     "        imbalance_var = 'none'\n",
     "        return X_train_res, y_train_res\n",
     "    \n",
@@ -706,7 +725,9 @@
     "        print('Please choose a valid resampling method: smote, rose, undersampling or none')\n",
     "        X_train_res = X_train\n",
     "        y_train_res = y_train\n",
-    "        return X_train_res, y_train_res"
+    "        return X_train_res, y_train_res\n",
+    "    \n",
+    "    "
    ]
   },
   {
@@ -723,7 +744,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 441,
+   "execution_count": 11,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -800,7 +821,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 442,
+   "execution_count": 12,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -826,7 +847,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 443,
+   "execution_count": 13,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -931,7 +952,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 444,
+   "execution_count": 14,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -941,18 +962,18 @@
     {
      "data": {
       "application/mercury+json": {
-       "code_uid": "Text.0.40.15.8-rand27c6053f",
+       "code_uid": "Text.0.40.15.8-randd265d777",
        "disabled": false,
        "hidden": false,
        "label": "Missing Value Threeshold",
-       "model_id": "9bf214b16a4342099c9edd6fdda6cca9",
+       "model_id": "aec705cdd896483f9d28d01d5e488a64",
        "rows": 1,
        "url_key": "",
        "value": "50",
        "widget": "Text"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "9bf214b16a4342099c9edd6fdda6cca9",
+       "model_id": "aec705cdd896483f9d28d01d5e488a64",
        "version_major": 2,
        "version_minor": 0
       },
@@ -966,18 +987,18 @@
     {
      "data": {
       "application/mercury+json": {
-       "code_uid": "Text.0.40.15.11-rand5d52d01b",
+       "code_uid": "Text.0.40.15.11-rand7024cef0",
        "disabled": false,
        "hidden": false,
        "label": "Variance Threshold",
-       "model_id": "98b6b9bb59ec43f1bc6c824e38f4eddd",
+       "model_id": "19c44ea2727948d09fa75e29c62844d8",
        "rows": 1,
        "url_key": "",
        "value": "0.05",
        "widget": "Text"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "98b6b9bb59ec43f1bc6c824e38f4eddd",
+       "model_id": "19c44ea2727948d09fa75e29c62844d8",
        "version_major": 2,
        "version_minor": 0
       },
@@ -991,18 +1012,18 @@
     {
      "data": {
       "application/mercury+json": {
-       "code_uid": "Text.0.40.15.14-randd7d692a8",
+       "code_uid": "Text.0.40.15.14-randa98919fc",
        "disabled": false,
        "hidden": false,
        "label": "Correlation Threshold",
-       "model_id": "b4e4bb3cc6414fcaa12c01b283081d96",
+       "model_id": "dd6d45837f4b45d3a4d58e9821188473",
        "rows": 1,
        "url_key": "",
        "value": "0.95",
        "widget": "Text"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "b4e4bb3cc6414fcaa12c01b283081d96",
+       "model_id": "dd6d45837f4b45d3a4d58e9821188473",
        "version_major": 2,
        "version_minor": 0
       },
@@ -1013,6 +1034,35 @@
      "metadata": {},
      "output_type": "display_data"
     },
+    {
+     "data": {
+      "application/mercury+json": {
+       "choices": [
+        "ignore",
+        "remove",
+        "push"
+       ],
+       "code_uid": "Select.0.40.16.19-rand82f0f224",
+       "disabled": false,
+       "hidden": false,
+       "label": "Outlier Action",
+       "model_id": "021d7973917845d4842f11c6d9f2fc67",
+       "url_key": "",
+       "value": "ignore",
+       "widget": "Select"
+      },
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "021d7973917845d4842f11c6d9f2fc67",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "mercury.Select"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
     {
      "data": {
       "application/mercury+json": {
@@ -1022,17 +1072,17 @@
         4,
         5
        ],
-       "code_uid": "Select.0.40.16.18-rand6188731c",
+       "code_uid": "Select.0.40.16.22-rand043a8dbf",
        "disabled": false,
        "hidden": false,
-       "label": "Outlier Removal Threshold",
-       "model_id": "48828625c53c4fe9ae8ad3abdab7bca6",
+       "label": "Outlier Action Threshold",
+       "model_id": "e0e42c0aef7845838da1aa1c70828482",
        "url_key": "",
-       "value": 5,
+       "value": "none",
        "widget": "Select"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "48828625c53c4fe9ae8ad3abdab7bca6",
+       "model_id": "e0e42c0aef7845838da1aa1c70828482",
        "version_major": 2,
        "version_minor": 0
       },
@@ -1052,17 +1102,17 @@
         "minmax",
         "robust"
        ],
-       "code_uid": "Select.0.40.16.25-rand4ff0ac92",
+       "code_uid": "Select.0.40.16.29-rand05477265",
        "disabled": false,
        "hidden": false,
        "label": "Scaling Variables",
-       "model_id": "4268185d86f34c559e1444de3c1739d9",
+       "model_id": "3938af66d22649bfb1cdaa77c4fb9684",
        "url_key": "",
-       "value": "standard",
+       "value": "none",
        "widget": "Select"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "4268185d86f34c559e1444de3c1739d9",
+       "model_id": "3938af66d22649bfb1cdaa77c4fb9684",
        "version_major": 2,
        "version_minor": 0
       },
@@ -1082,17 +1132,17 @@
         "knn",
         "most_frequent"
        ],
-       "code_uid": "Select.0.40.16.29-rand9bb317f9",
+       "code_uid": "Select.0.40.16.33-randade919ed",
        "disabled": false,
        "hidden": false,
        "label": "Imputation Methods",
-       "model_id": "a147c118c8f14de28b280232786f146a",
+       "model_id": "9367b36a66f148d2acd012c4f1c4fd71",
        "url_key": "",
        "value": "median",
        "widget": "Select"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "a147c118c8f14de28b280232786f146a",
+       "model_id": "9367b36a66f148d2acd012c4f1c4fd71",
        "version_major": 2,
        "version_minor": 0
       },
@@ -1113,17 +1163,17 @@
         "pca",
         "boruta"
        ],
-       "code_uid": "Select.0.40.16.34-rand7cda1892",
+       "code_uid": "Select.0.40.16.38-randff2c0505",
        "disabled": false,
        "hidden": false,
        "label": "Feature Selection",
-       "model_id": "ed31020a12d842a9b6e77a88344adfd6",
+       "model_id": "d9e985281de24f87a5e3d20d79348999",
        "url_key": "",
-       "value": "lasso",
+       "value": "none",
        "widget": "Select"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "ed31020a12d842a9b6e77a88344adfd6",
+       "model_id": "d9e985281de24f87a5e3d20d79348999",
        "version_major": 2,
        "version_minor": 0
       },
@@ -1143,17 +1193,17 @@
         "undersampling",
         "rose"
        ],
-       "code_uid": "Select.0.40.16.38-randc6301b14",
+       "code_uid": "Select.0.40.16.42-rand81ad2a30",
        "disabled": false,
        "hidden": false,
        "label": "Imbalance Treatment",
-       "model_id": "ef37d1810f974d2081c0cd9bed1d4384",
+       "model_id": "368fdc5c1dfa46029723962befc161a1",
        "url_key": "",
-       "value": "smote",
+       "value": "none",
        "widget": "Select"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "ef37d1810f974d2081c0cd9bed1d4384",
+       "model_id": "368fdc5c1dfa46029723962befc161a1",
        "version_major": 2,
        "version_minor": 0
       },
@@ -1176,17 +1226,17 @@
         "decision_tree",
         "xgboost"
        ],
-       "code_uid": "Select.0.40.16.42-randce0898a7",
+       "code_uid": "Select.0.40.16.46-rand5302564f",
        "disabled": false,
        "hidden": false,
        "label": "Model Selection",
-       "model_id": "02c163a5f04e4dde8adda8eb149814d0",
+       "model_id": "ad9082261f9e49b387ff963824152a05",
        "url_key": "",
        "value": "random_forest",
        "widget": "Select"
       },
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "02c163a5f04e4dde8adda8eb149814d0",
+       "model_id": "ad9082261f9e49b387ff963824152a05",
        "version_major": 2,
        "version_minor": 0
       },
@@ -1216,14 +1266,18 @@
     "input_correlation_threshold = float(input_correlation_threshold.value)\n",
     "\n",
     "# input outlier removal variables\n",
-    "input_outlier_removal_threshold = mr.Select(label=\"Outlier Removal Threshold\", value=5, choices=['none', 3, 4, 5])      # 'none' or zscore from 0 to 100\n",
+    "\n",
+    "input_outlier_action = mr.Select(label=\"Outlier Action\", value='ignore', choices=['ignore', 'remove', 'push'])      # 'ignore', 'remove', 'push'\n",
+    "input_outlier_action = str(input_outlier_action.value)\n",
+    "\n",
+    "input_outlier_removal_threshold = mr.Select(label=\"Outlier Action Threshold\", value='none', choices=['none', 3, 4, 5])      # 'none' or zscore from 0 to 100\n",
     "if input_outlier_removal_threshold.value != 'none':\n",
     "    input_outlier_removal_threshold = int(input_outlier_removal_threshold.value)\n",
     "elif input_outlier_removal_threshold.value == 'none':\n",
     "    input_outlier_removal_threshold = str(input_outlier_removal_threshold.value)\n",
     "\n",
     "# input scaling variables\n",
-    "input_scale_model = mr.Select(label=\"Scaling Variables\", value=\"standard\", choices=['none', 'standard', 'minmax', 'robust'])              # 'none', 'normal', 'standard', 'minmax', 'robust'\n",
+    "input_scale_model = mr.Select(label=\"Scaling Variables\", value=\"none\", choices=['none', 'standard', 'minmax', 'robust'])              # 'none', 'normal', 'standard', 'minmax', 'robust'\n",
     "input_scale_model = str(input_scale_model.value)\n",
     "\n",
     "# input imputation variables\n",
@@ -1232,11 +1286,11 @@
     "input_imputation_method = str(input_imputation_method.value)\n",
     "\n",
     "# import feature selection variables\n",
-    "input_feature_selection = mr.Select(label=\"Feature Selection\", value=\"lasso\", choices=['none', 'lasso', 'rfe', 'pca', 'boruta'])          # 'none', 'lasso', 'rfe', 'pca', 'boruta'\n",
+    "input_feature_selection = mr.Select(label=\"Feature Selection\", value=\"none\", choices=['none', 'lasso', 'rfe', 'pca', 'boruta'])          # 'none', 'lasso', 'rfe', 'pca', 'boruta'\n",
     "input_feature_selection = str(input_feature_selection.value)\n",
     "\n",
     "# input imbalance treatment variables\n",
-    "input_imbalance_treatment = mr.Select(label=\"Imbalance Treatment\", value=\"smote\", choices=['none', 'smote', 'undersampling', 'rose'])       # 'none', 'smote', 'undersampling', 'rose'\n",
+    "input_imbalance_treatment = mr.Select(label=\"Imbalance Treatment\", value=\"none\", choices=['none', 'smote', 'undersampling', 'rose'])       # 'none', 'smote', 'undersampling', 'rose'\n",
     "input_imbalance_treatment = str(input_imbalance_treatment.value)\n",
     "\n",
     "# input model\n",
@@ -1255,7 +1309,7 @@
     "\n",
     "# remove outliers from train dataset\n",
     "\n",
-    "X_train_dropped_outliers = outlier_removal(X_train2, input_outlier_removal_threshold)\n",
+    "X_train_dropped_outliers = outlier_removal(X_train2, input_outlier_action, input_outlier_removal_threshold)\n",
     "\n",
     "# scale the training and testing sets\n",
     "\n",
@@ -1291,7 +1345,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 445,
+   "execution_count": 15,
    "metadata": {
     "slideshow": {
      "slide_type": "skip"
@@ -1316,7 +1370,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 446,
+   "execution_count": 16,
    "metadata": {
     "slideshow": {
      "slide_type": "slide"
@@ -1328,14 +1382,14 @@
      "output_type": "stream",
      "text": [
       "   Accuracy  Precision  Recall  F1-score\n",
-      "0      0.89       0.15    0.15      0.15\n"
+      "0      0.93        0.0     0.0       0.0\n"
      ]
     },
     {
      "data": {
-      "image/png": "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",
+      "image/png": "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",
       "text/plain": [
-       "<Figure size 500x500 with 1 Axes>"
+       "<Figure size 640x480 with 1 Axes>"
       ]
      },
      "metadata": {},
@@ -1372,7 +1426,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 447,
+   "execution_count": 17,
    "metadata": {
     "slideshow": {
      "slide_type": "slide"
@@ -1392,11 +1446,11 @@
       "('New shape of the dataframe is: ', (1175, 179))\n",
       "------------------------------------------\n",
       "OUTLIER REMOVAL\n",
-      "('The z-score threshold is:', 5)\n",
-      "('The number of outliers removed from the dataset is:', 163)\n",
+      "No z-score threshold was selected\n",
+      "No outliers were removed from the dataset\n",
       "------------------------------------------\n",
       "SCALING\n",
-      "The dataframe has been scaled using the standard scaling model\n",
+      "The dataframe has not been scaled\n",
       "------------------------------------------\n",
       "IMPUTATION\n",
       "('Number of missing values before imputation: ', 1196)\n",
@@ -1404,15 +1458,16 @@
       "('Number of missing values after imputation: ', 0)\n",
       "------------------------------------------\n",
       "FEATURE SELECTION\n",
-      "('Selected method is: ', 'lasso')\n",
-      "('Shape of the training set after feature selection with LassoCV: ', (1175, 14))\n",
+      "No feature selection has been applied\n",
+      "('Shape of the training set after no feature selection: ', (1175, 179))\n",
       "------------------------------------------\n",
       "IMBALANCE TREATMENT\n",
-      "('Shape of the training set after oversampling with SMOTE: ', (2194, 14))\n",
-      "('Value counts of the target variable after oversampling with SMOTE: ', pass/fail\n",
+      "('Shape of the training set after no resampling: ', (1175, 179))\n",
+      "Value counts of the target variable after no resampling: \n",
+      "pass/fail\n",
       "0            1097\n",
-      "1            1097\n",
-      "dtype: int64)\n"
+      "1              78\n",
+      "dtype: int64\n"
      ]
     }
    ],
@@ -1443,7 +1498,8 @@
     "print('------------------------------------------')\n",
     "print('IMBALANCE TREATMENT')\n",
     "print(imbalance_report0)\n",
-    "print(imbalance_report1)"
+    "print(imbalance_report1)\n",
+    "print(imbalance_report2)"
    ]
   }
  ],

requirements.txt

@@ -3,11 +3,9 @@ pandas
 numpy
 seaborn
 matplotlib
+scipy
 sklearn
-imblearn
-xgboost
 mlxtend
-boruta
 imblearn
 xgboost
-scipy
+boruta