credit-part2-data-preparation-3-impute-scaler-outlier.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "7cda2f1d",
   "metadata": {
    "papermill": {
     "duration": 0.04141,
     "end_time": "2021-12-23T09:05:13.052182",
     "exception": false,
     "start_time": "2021-12-23T09:05:13.010772",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Load Lib "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b2cda463",
   "metadata": {
    "_cell_guid": "ddedf45c-cdd2-4013-973f-6ccf6c78347a",
    "_uuid": "20c9b28e-4c93-447d-8210-cad89f404354",
    "collapsed": false,
    "execution": {
     "iopub.execute_input": "2021-12-23T09:05:13.140574Z",
     "iopub.status.busy": "2021-12-23T09:05:13.138944Z",
     "iopub.status.idle": "2021-12-23T09:05:16.801020Z",
     "shell.execute_reply": "2021-12-23T09:05:16.800359Z"
    },
    "jupyter": {
     "outputs_hidden": false
    },
    "papermill": {
     "duration": 3.708695,
     "end_time": "2021-12-23T09:05:16.801190",
     "exception": false,
     "start_time": "2021-12-23T09:05:13.092495",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type='text/css'>\n",
       ".datatable table.frame { margin-bottom: 0; }\n",
       ".datatable table.frame thead { border-bottom: none; }\n",
       ".datatable table.frame tr.coltypes td {  color: #FFFFFF;  line-height: 6px;  padding: 0 0.5em;}\n",
       ".datatable .bool    { background: #DDDD99; }\n",
       ".datatable .object  { background: #565656; }\n",
       ".datatable .int     { background: #5D9E5D; }\n",
       ".datatable .float   { background: #4040CC; }\n",
       ".datatable .str     { background: #CC4040; }\n",
       ".datatable .time    { background: #40CC40; }\n",
       ".datatable .row_index {  background: var(--jp-border-color3);  border-right: 1px solid var(--jp-border-color0);  color: var(--jp-ui-font-color3);  font-size: 9px;}\n",
       ".datatable .frame tbody td { text-align: left; }\n",
       ".datatable .frame tr.coltypes .row_index {  background: var(--jp-border-color0);}\n",
       ".datatable th:nth-child(2) { padding-left: 12px; }\n",
       ".datatable .hellipsis {  color: var(--jp-cell-editor-border-color);}\n",
       ".datatable .vellipsis {  background: var(--jp-layout-color0);  color: var(--jp-cell-editor-border-color);}\n",
       ".datatable .na {  color: var(--jp-cell-editor-border-color);  font-size: 80%;}\n",
       ".datatable .sp {  opacity: 0.25;}\n",
       ".datatable .footer { font-size: 9px; }\n",
       ".datatable .frame_dimensions {  background: var(--jp-border-color3);  border-top: 1px solid var(--jp-border-color0);  color: var(--jp-ui-font-color3);  display: inline-block;  opacity: 0.6;  padding: 1px 10px 1px 5px;}\n",
       ".datatable .frame thead tr.colnames {  background-image: url('data:image/png;base64,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');  background-repeat: repeat-x;  background-size: 14px;  height: 28px;}\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Load the librarys\n",
    "import pandas as pd #To work with dataset\n",
    "import numpy as np #Math library\n",
    "import matplotlib.gridspec as gridspec\n",
    "import seaborn as sns #Graph library that use matplot in background\n",
    "import matplotlib.pyplot as plt #to plot some parameters in seaborn\n",
    "import warnings\n",
    "# Preparation  \n",
    "from sklearn.preprocessing import LabelEncoder, OrdinalEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.preprocessing import PowerTransformer, StandardScaler,Normalizer,RobustScaler,MaxAbsScaler,MinMaxScaler,QuantileTransformer\n",
    "from sklearn.preprocessing import FunctionTransformer\n",
    "from sklearn.preprocessing import PolynomialFeatures\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.preprocessing import KBinsDiscretizer\n",
    "# Import StandardScaler from scikit-learn\n",
    "\n",
    "from sklearn.experimental import enable_iterative_imputer\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.impute import KNNImputer,IterativeImputer\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.compose import make_column_transformer,ColumnTransformer\n",
    "from sklearn.pipeline import make_pipeline, Pipeline,FeatureUnion\n",
    "from sklearn.manifold import TSNE\n",
    "# Import train_test_split()\n",
    "# Metrics\n",
    "from sklearn.metrics import roc_auc_score, average_precision_score\n",
    "from sklearn.metrics import make_scorer,mean_absolute_error\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.metrics import roc_curve,confusion_matrix\n",
    "from datetime import datetime, date\n",
    "from sklearn.linear_model import ElasticNet, Lasso,  BayesianRidge, LassoLarsIC\n",
    "from sklearn.linear_model import LinearRegression, RidgeCV\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "#import tensorflow as tf \n",
    "#from tensorflow.keras import layers\n",
    "#from tensorflow.keras.callbacks import EarlyStopping\n",
    "#from tensorflow.keras.callbacks import LearningRateScheduler\n",
    "#import smogn\n",
    "from sklearn.experimental import enable_hist_gradient_boosting\n",
    "from sklearn.base import BaseEstimator, TransformerMixin, RegressorMixin, clone\n",
    "from sklearn.kernel_ridge import KernelRidge\n",
    "from sklearn.ensemble import GradientBoostingRegressor,RandomForestRegressor\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.ensemble import HistGradientBoostingClassifier\n",
    "# For training random forest model\n",
    "import lightgbm as lgb\n",
    "from scipy import sparse\n",
    "from sklearn.neighbors import KNeighborsRegressor \n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.cluster import KMeans \n",
    "# Model selection\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.feature_selection import SelectKBest\n",
    "from sklearn.feature_selection import f_regression,f_classif,chi2\n",
    "from sklearn.feature_selection import mutual_info_regression\n",
    "from sklearn.feature_selection import SelectPercentile\n",
    "from sklearn.feature_selection import mutual_info_classif,VarianceThreshold\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from lightgbm import LGBMClassifier\n",
    "import lightgbm as lgbm\n",
    "#from catboost import CatBoostRegressor, CatBoostClassifier\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn import svm\n",
    "from xgboost import XGBClassifier,XGBRegressor\n",
    "from sklearn import set_config\n",
    "from itertools import combinations\n",
    "# Cluster :\n",
    "from sklearn.cluster import MiniBatchKMeans\n",
    "#from yellowbrick.cluster import KElbowVisualizer\n",
    "#import smong \n",
    "import category_encoders as ce\n",
    "import warnings\n",
    "#import optuna \n",
    "from joblib import Parallel, delayed\n",
    "import joblib \n",
    "from sklearn import set_config\n",
    "from typing import List, Optional, Union\n",
    "set_config(display='diagram')\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "73cdc857",
   "metadata": {
    "_cell_guid": "38c6befb-892d-4ef3-9c63-13be5959b671",
    "_uuid": "2953f9f3-02f8-4fa9-a7fc-cb93565c2183",
    "papermill": {
     "duration": 0.040454,
     "end_time": "2021-12-23T09:05:16.883863",
     "exception": false,
     "start_time": "2021-12-23T09:05:16.843409",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "\n",
    "# Load the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "6d2472cc",
   "metadata": {
    "_cell_guid": "98a5b4a6-90e9-449d-96cc-0d85c9ff1c77",
    "_uuid": "e086592f-9f6a-4592-80b5-c1e7ad227e02",
    "collapsed": false,
    "execution": {
     "iopub.execute_input": "2021-12-23T09:05:16.970520Z",
     "iopub.status.busy": "2021-12-23T09:05:16.969981Z",
     "iopub.status.idle": "2021-12-23T09:05:17.390392Z",
     "shell.execute_reply": "2021-12-23T09:05:17.390935Z"
    },
    "jupyter": {
     "outputs_hidden": false
    },
    "papermill": {
     "duration": 0.466595,
     "end_time": "2021-12-23T09:05:17.391091",
     "exception": false,
     "start_time": "2021-12-23T09:05:16.924496",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 220 ms, sys: 48.3 ms, total: 269 ms\n",
      "Wall time: 401 ms\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>SeriousDlqin2yrs</th>\n",
       "      <th>RevolvingUtilizationOfUnsecuredLines</th>\n",
       "      <th>age</th>\n",
       "      <th>NumberOfTime30-59DaysPastDueNotWorse</th>\n",
       "      <th>DebtRatio</th>\n",
       "      <th>MonthlyIncome</th>\n",
       "      <th>NumberOfOpenCreditLinesAndLoans</th>\n",
       "      <th>NumberOfTimes90DaysLate</th>\n",
       "      <th>NumberRealEstateLoansOrLines</th>\n",
       "      <th>NumberOfTime60-89DaysPastDueNotWorse</th>\n",
       "      <th>NumberOfDependents</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.766127</td>\n",
       "      <td>45</td>\n",
       "      <td>2</td>\n",
       "      <td>0.802982</td>\n",
       "      <td>9120.0</td>\n",
       "      <td>13</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0.957151</td>\n",
       "      <td>40</td>\n",
       "      <td>0</td>\n",
       "      <td>0.121876</td>\n",
       "      <td>2600.0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0.658180</td>\n",
       "      <td>38</td>\n",
       "      <td>1</td>\n",
       "      <td>0.085113</td>\n",
       "      <td>3042.0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0  SeriousDlqin2yrs  RevolvingUtilizationOfUnsecuredLines  age  \\\n",
       "0           1                 1                              0.766127   45   \n",
       "1           2                 0                              0.957151   40   \n",
       "2           3                 0                              0.658180   38   \n",
       "\n",
       "   NumberOfTime30-59DaysPastDueNotWorse  DebtRatio  MonthlyIncome  \\\n",
       "0                                     2   0.802982         9120.0   \n",
       "1                                     0   0.121876         2600.0   \n",
       "2                                     1   0.085113         3042.0   \n",
       "\n",
       "   NumberOfOpenCreditLinesAndLoans  NumberOfTimes90DaysLate  \\\n",
       "0                               13                        0   \n",
       "1                                4                        0   \n",
       "2                                2                        1   \n",
       "\n",
       "   NumberRealEstateLoansOrLines  NumberOfTime60-89DaysPastDueNotWorse  \\\n",
       "0                             6                                     0   \n",
       "1                             0                                     0   \n",
       "2                             0                                     0   \n",
       "\n",
       "   NumberOfDependents  \n",
       "0                 2.0  \n",
       "1                 1.0  \n",
       "2                 0.0  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time \n",
    "train = pd.read_csv('../input/GiveMeSomeCredit/cs-training.csv')\n",
    "test = pd.read_csv('../input/GiveMeSomeCredit/cs-test.csv')\n",
    "train.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "53b8ac2f",
   "metadata": {
    "papermill": {
     "duration": 0.041888,
     "end_time": "2021-12-23T09:05:17.477875",
     "exception": false,
     "start_time": "2021-12-23T09:05:17.435987",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Convert Dtypes :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "7762757a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:05:17.589910Z",
     "iopub.status.busy": "2021-12-23T09:05:17.588335Z",
     "iopub.status.idle": "2021-12-23T09:05:17.643782Z",
     "shell.execute_reply": "2021-12-23T09:05:17.643102Z"
    },
    "papermill": {
     "duration": 0.123543,
     "end_time": "2021-12-23T09:05:17.643920",
     "exception": false,
     "start_time": "2021-12-23T09:05:17.520377",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Convert Dtypes :\n",
    "train[train.select_dtypes(['int64','int16','float16','float32','float64','int8']).columns] = train[train.select_dtypes(['int64','int16','float16','float32','float64','int8']).columns].apply(pd.to_numeric)\n",
    "train[train.select_dtypes(['object','category']).columns] = train.select_dtypes(['object','category']).apply(lambda x: x.astype('category'))\n",
    "# Convert Dtypes :\n",
    "test[test.select_dtypes(['int64','int16','float16','float32','float64','int8']).columns] = test[test.select_dtypes(['int64','int16','float16','float32','float64','int8']).columns].apply(pd.to_numeric)\n",
    "test[test.select_dtypes(['object','category']).columns] = test.select_dtypes(['object','category']).apply(lambda x: x.astype('category'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bae21e98",
   "metadata": {
    "papermill": {
     "duration": 0.044503,
     "end_time": "2021-12-23T09:05:17.732228",
     "exception": false,
     "start_time": "2021-12-23T09:05:17.687725",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Reduce Memory"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "62705c75",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:05:17.838026Z",
     "iopub.status.busy": "2021-12-23T09:05:17.836299Z",
     "iopub.status.idle": "2021-12-23T09:05:17.879973Z",
     "shell.execute_reply": "2021-12-23T09:05:17.880660Z"
    },
    "papermill": {
     "duration": 0.105626,
     "end_time": "2021-12-23T09:05:17.880862",
     "exception": false,
     "start_time": "2021-12-23T09:05:17.775236",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Memory usage of dataframe is 13.73 MB\n",
      "Memory usage after optimization is: 3.29 MB\n",
      "Decreased by 76.0%\n",
      "Memory usage of dataframe is 9.29 MB\n",
      "Memory usage after optimization is: 2.90 MB\n",
      "Decreased by 68.7%\n"
     ]
    }
   ],
   "source": [
    "# Author : https://www.kaggle.com/gemartin/load-data-reduce-memory-usage\n",
    "def reduce_mem_usage(df):\n",
    "    \"\"\" iterate through all the columns of a dataframe and modify the data type\n",
    "        to reduce memory usage.        \n",
    "    \"\"\"\n",
    "    start_mem = df.memory_usage().sum() / 1024**2\n",
    "    print('Memory usage of dataframe is {:.2f} MB'.format(start_mem))\n",
    "    \n",
    "    for col in df.columns:\n",
    "        col_type = df[col].dtype\n",
    "        name =df[col].dtype.name \n",
    "        \n",
    "        if col_type != object and col_type.name != 'category':\n",
    "        #if name != \"category\":    \n",
    "            c_min = df[col].min()\n",
    "            c_max = df[col].max()\n",
    "            if str(col_type)[:3] == 'int':\n",
    "                if c_min > np.iinfo(np.int8).min and c_max < np.iinfo(np.int8).max:\n",
    "                    df[col] = df[col].astype(np.int8)\n",
    "                elif c_min > np.iinfo(np.int16).min and c_max < np.iinfo(np.int16).max:\n",
    "                    df[col] = df[col].astype(np.int16)\n",
    "                elif c_min > np.iinfo(np.int32).min and c_max < np.iinfo(np.int32).max:\n",
    "                    df[col] = df[col].astype(np.int32)\n",
    "                elif c_min > np.iinfo(np.int64).min and c_max < np.iinfo(np.int64).max:\n",
    "                    df[col] = df[col].astype(np.int64)  \n",
    "            else:\n",
    "                if c_min > np.finfo(np.float16).min and c_max < np.finfo(np.float16).max:\n",
    "                    df[col] = df[col].astype(np.float16)\n",
    "                elif c_min > np.finfo(np.float32).min and c_max < np.finfo(np.float32).max:\n",
    "                    df[col] = df[col].astype(np.float32)\n",
    "                else:\n",
    "                    df[col] = df[col].astype(np.float64)\n",
    "        else:\n",
    "            df[col] = df[col].astype('category')\n",
    "\n",
    "    end_mem = df.memory_usage().sum() / 1024**2\n",
    "    print('Memory usage after optimization is: {:.2f} MB'.format(end_mem))\n",
    "    print('Decreased by {:.1f}%'.format(100 * (start_mem - end_mem) / start_mem))\n",
    "    \n",
    "    return df\n",
    "train= reduce_mem_usage(train)\n",
    "test= reduce_mem_usage(test)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "61f15a9f",
   "metadata": {
    "papermill": {
     "duration": 0.042529,
     "end_time": "2021-12-23T09:05:17.969022",
     "exception": false,
     "start_time": "2021-12-23T09:05:17.926493",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# X and Y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "7c7a6132",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:05:18.063038Z",
     "iopub.status.busy": "2021-12-23T09:05:18.062102Z",
     "iopub.status.idle": "2021-12-23T09:05:18.065456Z",
     "shell.execute_reply": "2021-12-23T09:05:18.064881Z"
    },
    "papermill": {
     "duration": 0.053188,
     "end_time": "2021-12-23T09:05:18.065634",
     "exception": false,
     "start_time": "2021-12-23T09:05:18.012446",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Cardinality : \n",
    "# - RevolvingUtilizationOfUnsecuredLines :125728, high Outlier\n",
    "# - DebtRatio :114194 , high Outlier \n",
    "# deal with outlier + bin \n",
    "PERCENTAGE = [\"RevolvingUtilizationOfUnsecuredLines\", \"DebtRatio\"]\n",
    "# MonthlyIncome:13594 , high outlier +bin \n",
    "REAL= [\"MonthlyIncome\"]\n",
    "# Can be considred as cat \n",
    "NUMERIC_DISCRET_low = [\"NumberOfDependents\",\n",
    "                       \"NumberOfTime60-89DaysPastDueNotWorse\",\n",
    "                       \"NumberRealEstateLoansOrLines\",\n",
    "                       \"NumberOfTimes90DaysLate\",\n",
    "                       \"NumberOfOpenCreditLinesAndLoans\",\n",
    "                       \"NumberOfTime30-59DaysPastDueNotWorse\",\n",
    "                       \"age\"]\n",
    "Late_Pay_Cols = ['NumberOfTime30-59DaysPastDueNotWorse',\n",
    "                 'NumberOfTimes90DaysLate',\n",
    "                 'NumberOfTime60-89DaysPastDueNotWorse']\n",
    "TARGET = [\"SeriousDlqin2yrs\"]\n",
    "\n",
    "#also change the type for TARGET to categorical\n",
    "#df[TARGET] = df[TARGET].astype('category')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9a536875",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:05:18.159043Z",
     "iopub.status.busy": "2021-12-23T09:05:18.158102Z",
     "iopub.status.idle": "2021-12-23T09:05:18.159925Z",
     "shell.execute_reply": "2021-12-23T09:05:18.160353Z"
    },
    "papermill": {
     "duration": 0.050469,
     "end_time": "2021-12-23T09:05:18.160488",
     "exception": false,
     "start_time": "2021-12-23T09:05:18.110019",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "y = train['SeriousDlqin2yrs']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "fa2f55b1",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:05:18.252219Z",
     "iopub.status.busy": "2021-12-23T09:05:18.251444Z",
     "iopub.status.idle": "2021-12-23T09:05:18.308503Z",
     "shell.execute_reply": "2021-12-23T09:05:18.308948Z"
    },
    "papermill": {
     "duration": 0.105701,
     "end_time": "2021-12-23T09:05:18.309112",
     "exception": false,
     "start_time": "2021-12-23T09:05:18.203411",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DebtRatio                               30457\n",
       "NumberOfTime30-59DaysPastDueNotWorse    23982\n",
       "SeriousDlqin2yrs                        10026\n",
       "NumberOfTimes90DaysLate                  8338\n",
       "NumberOfTime60-89DaysPastDueNotWorse     7604\n",
       "MonthlyIncome                            1557\n",
       "NumberOfDependents                        991\n",
       "NumberOfOpenCreditLinesAndLoans           442\n",
       "RevolvingUtilizationOfUnsecuredLines      339\n",
       "NumberRealEstateLoansOrLines              209\n",
       "dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def get_outlier_counts(df, outlier_threshold=1.5):\n",
    "    Q1 = df.quantile(0.25)\n",
    "    Q3 = df.quantile(0.75)\n",
    "    IQR = Q3 - Q1\n",
    "\n",
    "    outlier_counts = ((df < (Q1 - outlier_threshold * IQR)) | (df > (Q3 + outlier_threshold * IQR))).sum()\n",
    "    return outlier_counts[outlier_counts>0].sort_values(ascending=False)\n",
    "\n",
    "get_outlier_counts(train, outlier_threshold=3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "10d5144c",
   "metadata": {
    "papermill": {
     "duration": 0.043339,
     "end_time": "2021-12-23T09:05:18.395036",
     "exception": false,
     "start_time": "2021-12-23T09:05:18.351697",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Imput before Descritaziation : \n",
    "\n",
    "## IterativeImputer\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "548b8e41",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:05:18.490995Z",
     "iopub.status.busy": "2021-12-23T09:05:18.490371Z",
     "iopub.status.idle": "2021-12-23T09:05:18.495311Z",
     "shell.execute_reply": "2021-12-23T09:05:18.494904Z"
    },
    "papermill": {
     "duration": 0.057677,
     "end_time": "2021-12-23T09:05:18.495430",
     "exception": false,
     "start_time": "2021-12-23T09:05:18.437753",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 338 µs, sys: 0 ns, total: 338 µs\n",
      "Wall time: 343 µs\n"
     ]
    }
   ],
   "source": [
    "%%time \n",
    "param =  {   \"verbosity\": 0,\n",
    "            #\"objective\": \"binary:logistic\",\n",
    "            #\"eval_metric\": \"auc\",\n",
    "            'random_state': 42,\n",
    "            # regression\n",
    "            'objective':'reg:squarederror', \n",
    "             'eval_metric': 'mae',\n",
    "            #early_stopping_rounds=100 ,\n",
    "            'gpu_id':0, \n",
    "            'predictor':\"gpu_predictor\",\n",
    "            # use exact for small dataset.\n",
    "            \"tree_method\": \"exact\",\n",
    "            # big data :\n",
    "            # 'tree_method': 'gpu_hist',\n",
    "            # defines booster, gblinear for linear functions.\n",
    "             'booster': 'gbtree', \n",
    "            'lambda': 8.544792472633987e-07,\n",
    "            'alpha': 0.31141671752487043,\n",
    "            'subsample': 0.8779467596981366, \n",
    "            'colsample_bytree': 0.9759532762677546,\n",
    "            'learning_rate': 0.008686087328805853, \n",
    "            'n_estimators': 6988,\n",
    "            'max_depth': 9,\n",
    "            'min_child_weight': 2, \n",
    "            'eta': 3.7603213457541647e-06,\n",
    "            'gamma': 2.1478058456847449e-07,\n",
    "            'grow_policy': 'lossguide'}\n",
    "                \n",
    "\n",
    "#model_xgb = XGBRegressor(\n",
    "       #objective=\"mae\",\n",
    " #   **xgb_params2)\n",
    "\n",
    "numeric_transformer1 = Pipeline(\n",
    "                            steps=[\n",
    "                            ('imputer', SimpleImputer(strategy='median'\n",
    "                                                      ,add_indicator=True)),\n",
    "                            ('scaler', PowerTransformer()),#(Numerical Input, Numerical Output)\n",
    "                            # Create an SelectKBest object to select features with two best ANOVA F-Values\n",
    "                            #The F-value scores examine if, when we group the numerical feature by the target vector, the means for each group are significantly different\n",
    "                           # ('reducedim',  SelectPercentile(f_classif,percentile=90))\n",
    "                            ]\n",
    "                            )\n",
    "numeric_transformer2 = Pipeline(\n",
    "                            steps=[\n",
    "                            #('imputer', SimpleImputer(strategy='median'\n",
    "                             #                         ,add_indicator=True)),\n",
    "                            ('scaler', PowerTransformer()),#(Numerical Input, Numerical Output)\n",
    "                            # Create an SelectKBest object to select features with two best ANOVA F-Values\n",
    "                            #The F-value scores examine if, when we group the numerical feature by the target vector, the means for each group are significantly different\n",
    "                           # ('reducedim',  SelectPercentile(f_classif,percentile=90))\n",
    "                            ]\n",
    "                            )\n",
    "\n",
    "pipe_xgbr1 = Pipeline(\n",
    "                    steps=[\n",
    "                        ('preprocessor', numeric_transformer1),\n",
    "                        ('classifier', XGBRegressor(\n",
    "                      #objective=\"mae\",\n",
    "                       **param))\n",
    "                    ]\n",
    "                )\n",
    "pipe_xgbr2 = Pipeline(\n",
    "                    steps=[\n",
    "                        ('preprocessor', numeric_transformer2),\n",
    "                        ('classifier', XGBRegressor(\n",
    "                      #objective=\"mae\",\n",
    "                       **param))\n",
    "                    ]\n",
    "                )\n",
    "\n",
    "#imputer=IterativeImputer(estimator=XGBRegressor(\n",
    " #                     #objective=\"mae\",\n",
    "                        #add_indicator=True,\n",
    "  #                     **param),max_iter=2,random_state=42)\n",
    "#train=train.drop(['Unnamed: 0','SeriousDlqin2yrs'], axis=1)\n",
    "#test=test.drop(['Unnamed: 0','SeriousDlqin2yrs'], axis=1)\n",
    "#imputer.fit(train)\n",
    "#train = imputer.transform(train)\n",
    "#test = imputer.transform(test)\n",
    "#features=['RevolvingUtilizationOfUnsecuredLines', 'age',\n",
    "#      'NumberOfTime30-59DaysPastDueNotWorse', 'DebtRatio', 'MonthlyIncome',\n",
    " #      'NumberOfOpenCreditLinesAndLoans', 'NumberOfTimes90DaysLate',\n",
    " #     'NumberRealEstateLoansOrLines', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
    " #     'NumberOfDependents']\n",
    "#train=pd.DataFrame(train,columns=features)\n",
    "#test=pd.DataFrame(test,columns=features)\n",
    "#train= reduce_mem_usage(train)\n",
    "#test= reduce_mem_usage(test)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8e1cf22d",
   "metadata": {
    "papermill": {
     "duration": 0.042852,
     "end_time": "2021-12-23T09:05:18.582170",
     "exception": false,
     "start_time": "2021-12-23T09:05:18.539318",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Rgression Imputer "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "c105aaa5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:05:18.677469Z",
     "iopub.status.busy": "2021-12-23T09:05:18.676699Z",
     "iopub.status.idle": "2021-12-23T09:29:20.418249Z",
     "shell.execute_reply": "2021-12-23T09:29:20.418819Z"
    },
    "papermill": {
     "duration": 1441.793333,
     "end_time": "2021-12-23T09:29:20.419024",
     "exception": false,
     "start_time": "2021-12-23T09:05:18.625691",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 45min 35s, sys: 6.39 s, total: 45min 41s\n",
      "Wall time: 24min 1s\n"
     ]
    }
   ],
   "source": [
    "%%time \n",
    "train=train.drop(['Unnamed: 0','SeriousDlqin2yrs'], axis=1)\n",
    "test=test.drop(['Unnamed: 0','SeriousDlqin2yrs'], axis=1)\n",
    "train_final= pd.concat( [train, test], axis=0) \n",
    "#testdf_income= train_final[train_final['MonthlyIncome'].isnull()==True]\n",
    "traindf_income = train_final[train_final['MonthlyIncome'].isnull()==False]\n",
    "y_income = traindf_income['MonthlyIncome']\n",
    "X_income=traindf_income.drop([\"MonthlyIncome\"],axis=1)\n",
    "pipe_xgbr1.fit(X_income, y_income)\n",
    "train_income_missing=train[train['MonthlyIncome'].isnull()==True].drop([\"MonthlyIncome\"],axis=1)\n",
    "test_income_missing=test[test['MonthlyIncome'].isnull()==True].drop([\"MonthlyIncome\"],axis=1)\n",
    "train_predicted = pipe_xgbr1.predict(train_income_missing)\n",
    "test_predicted = pipe_xgbr1.predict(test_income_missing)\n",
    "train.loc[(train.MonthlyIncome.isnull()), 'MonthlyIncome'] = train_predicted\n",
    "test.loc[(test.MonthlyIncome.isnull()), 'MonthlyIncome'] = test_predicted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "e1b0a0fa",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:29:20.511696Z",
     "iopub.status.busy": "2021-12-23T09:29:20.510757Z",
     "iopub.status.idle": "2021-12-23T09:29:20.523993Z",
     "shell.execute_reply": "2021-12-23T09:29:20.524437Z"
    },
    "papermill": {
     "duration": 0.061827,
     "end_time": "2021-12-23T09:29:20.524599",
     "exception": false,
     "start_time": "2021-12-23T09:29:20.462772",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 150000 entries, 0 to 149999\n",
      "Data columns (total 10 columns):\n",
      " #   Column                                Non-Null Count   Dtype  \n",
      "---  ------                                --------------   -----  \n",
      " 0   RevolvingUtilizationOfUnsecuredLines  150000 non-null  float16\n",
      " 1   age                                   150000 non-null  int8   \n",
      " 2   NumberOfTime30-59DaysPastDueNotWorse  150000 non-null  int8   \n",
      " 3   DebtRatio                             150000 non-null  float32\n",
      " 4   MonthlyIncome                         150000 non-null  float32\n",
      " 5   NumberOfOpenCreditLinesAndLoans       150000 non-null  int8   \n",
      " 6   NumberOfTimes90DaysLate               150000 non-null  int8   \n",
      " 7   NumberRealEstateLoansOrLines          150000 non-null  int8   \n",
      " 8   NumberOfTime60-89DaysPastDueNotWorse  150000 non-null  int8   \n",
      " 9   NumberOfDependents                    146076 non-null  float16\n",
      "dtypes: float16(2), float32(2), int8(6)\n",
      "memory usage: 2.6 MB\n"
     ]
    }
   ],
   "source": [
    "train.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "e8822be5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:29:20.618211Z",
     "iopub.status.busy": "2021-12-23T09:29:20.617347Z",
     "iopub.status.idle": "2021-12-23T09:29:20.628917Z",
     "shell.execute_reply": "2021-12-23T09:29:20.629422Z"
    },
    "papermill": {
     "duration": 0.060464,
     "end_time": "2021-12-23T09:29:20.629595",
     "exception": false,
     "start_time": "2021-12-23T09:29:20.569131",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 101503 entries, 0 to 101502\n",
      "Data columns (total 10 columns):\n",
      " #   Column                                Non-Null Count   Dtype  \n",
      "---  ------                                --------------   -----  \n",
      " 0   RevolvingUtilizationOfUnsecuredLines  101503 non-null  float16\n",
      " 1   age                                   101503 non-null  int8   \n",
      " 2   NumberOfTime30-59DaysPastDueNotWorse  101503 non-null  int8   \n",
      " 3   DebtRatio                             101503 non-null  float32\n",
      " 4   MonthlyIncome                         101503 non-null  float32\n",
      " 5   NumberOfOpenCreditLinesAndLoans       101503 non-null  int8   \n",
      " 6   NumberOfTimes90DaysLate               101503 non-null  int8   \n",
      " 7   NumberRealEstateLoansOrLines          101503 non-null  int8   \n",
      " 8   NumberOfTime60-89DaysPastDueNotWorse  101503 non-null  int8   \n",
      " 9   NumberOfDependents                    98877 non-null   float16\n",
      "dtypes: float16(2), float32(2), int8(6)\n",
      "memory usage: 1.7 MB\n"
     ]
    }
   ],
   "source": [
    "test.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "8e5bf9a1",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T09:29:20.731250Z",
     "iopub.status.busy": "2021-12-23T09:29:20.730422Z",
     "iopub.status.idle": "2021-12-23T10:01:12.047106Z",
     "shell.execute_reply": "2021-12-23T10:01:12.046166Z"
    },
    "papermill": {
     "duration": 1911.370073,
     "end_time": "2021-12-23T10:01:12.047270",
     "exception": false,
     "start_time": "2021-12-23T09:29:20.677197",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "traindf_NumberOfDependents = train_final[train_final['NumberOfDependents'].isnull()==False]\n",
    "y_NumberOfDependents = traindf_NumberOfDependents['NumberOfDependents']\n",
    "X_NumberOfDependents=traindf_NumberOfDependents.drop([\"NumberOfDependents\"],axis=1)\n",
    "pipe_xgbr2.fit(X_NumberOfDependents, y_NumberOfDependents)\n",
    "train_NumberOfDependents_missing=train[train['NumberOfDependents'].isnull()==True].drop([\"NumberOfDependents\"],axis=1)\n",
    "test_NumberOfDependents_missing=test[test['NumberOfDependents'].isnull()==True].drop([\"NumberOfDependents\"],axis=1)\n",
    "train_predicted = pipe_xgbr2.predict(train_NumberOfDependents_missing)\n",
    "test_predicted = pipe_xgbr2.predict(test_NumberOfDependents_missing)\n",
    "train.loc[(train.NumberOfDependents.isnull()), 'NumberOfDependents'] = train_predicted\n",
    "test.loc[(test.NumberOfDependents.isnull()), 'NumberOfDependents'] = test_predicted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "c7ce542e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:12.142139Z",
     "iopub.status.busy": "2021-12-23T10:01:12.141169Z",
     "iopub.status.idle": "2021-12-23T10:01:12.153610Z",
     "shell.execute_reply": "2021-12-23T10:01:12.154349Z"
    },
    "papermill": {
     "duration": 0.062126,
     "end_time": "2021-12-23T10:01:12.154543",
     "exception": false,
     "start_time": "2021-12-23T10:01:12.092417",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 150000 entries, 0 to 149999\n",
      "Data columns (total 10 columns):\n",
      " #   Column                                Non-Null Count   Dtype  \n",
      "---  ------                                --------------   -----  \n",
      " 0   RevolvingUtilizationOfUnsecuredLines  150000 non-null  float16\n",
      " 1   age                                   150000 non-null  int8   \n",
      " 2   NumberOfTime30-59DaysPastDueNotWorse  150000 non-null  int8   \n",
      " 3   DebtRatio                             150000 non-null  float32\n",
      " 4   MonthlyIncome                         150000 non-null  float32\n",
      " 5   NumberOfOpenCreditLinesAndLoans       150000 non-null  int8   \n",
      " 6   NumberOfTimes90DaysLate               150000 non-null  int8   \n",
      " 7   NumberRealEstateLoansOrLines          150000 non-null  int8   \n",
      " 8   NumberOfTime60-89DaysPastDueNotWorse  150000 non-null  int8   \n",
      " 9   NumberOfDependents                    150000 non-null  float16\n",
      "dtypes: float16(2), float32(2), int8(6)\n",
      "memory usage: 2.6 MB\n"
     ]
    }
   ],
   "source": [
    "train.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "0e3936c2",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:12.251435Z",
     "iopub.status.busy": "2021-12-23T10:01:12.250629Z",
     "iopub.status.idle": "2021-12-23T10:01:12.261956Z",
     "shell.execute_reply": "2021-12-23T10:01:12.262346Z"
    },
    "papermill": {
     "duration": 0.06095,
     "end_time": "2021-12-23T10:01:12.262498",
     "exception": false,
     "start_time": "2021-12-23T10:01:12.201548",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 101503 entries, 0 to 101502\n",
      "Data columns (total 10 columns):\n",
      " #   Column                                Non-Null Count   Dtype  \n",
      "---  ------                                --------------   -----  \n",
      " 0   RevolvingUtilizationOfUnsecuredLines  101503 non-null  float16\n",
      " 1   age                                   101503 non-null  int8   \n",
      " 2   NumberOfTime30-59DaysPastDueNotWorse  101503 non-null  int8   \n",
      " 3   DebtRatio                             101503 non-null  float32\n",
      " 4   MonthlyIncome                         101503 non-null  float32\n",
      " 5   NumberOfOpenCreditLinesAndLoans       101503 non-null  int8   \n",
      " 6   NumberOfTimes90DaysLate               101503 non-null  int8   \n",
      " 7   NumberRealEstateLoansOrLines          101503 non-null  int8   \n",
      " 8   NumberOfTime60-89DaysPastDueNotWorse  101503 non-null  int8   \n",
      " 9   NumberOfDependents                    101503 non-null  float16\n",
      "dtypes: float16(2), float32(2), int8(6)\n",
      "memory usage: 1.7 MB\n"
     ]
    }
   ],
   "source": [
    "test.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6ff719d1",
   "metadata": {
    "papermill": {
     "duration": 0.045363,
     "end_time": "2021-12-23T10:01:12.359343",
     "exception": false,
     "start_time": "2021-12-23T10:01:12.313980",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Add Bin "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "d4b85bbb",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:12.475331Z",
     "iopub.status.busy": "2021-12-23T10:01:12.474239Z",
     "iopub.status.idle": "2021-12-23T10:01:13.052725Z",
     "shell.execute_reply": "2021-12-23T10:01:13.052172Z"
    },
    "papermill": {
     "duration": 0.646534,
     "end_time": "2021-12-23T10:01:13.052862",
     "exception": false,
     "start_time": "2021-12-23T10:01:12.406328",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Add bin data \n",
    "# initializing append_str\n",
    "append_str = 'cat_'\n",
    "# Append suffix / prefix to strings in list\n",
    "num_features1=[\"RevolvingUtilizationOfUnsecuredLines\", \"DebtRatio\",\"MonthlyIncome\"]\n",
    "num_features2=[\"NumberOfDependents\",\n",
    "                       \"NumberOfTime60-89DaysPastDueNotWorse\",\n",
    "                       \"NumberRealEstateLoansOrLines\",\n",
    "                       \"NumberOfTimes90DaysLate\",\n",
    "                       \"NumberOfOpenCreditLinesAndLoans\",\n",
    "                       \"NumberOfTime30-59DaysPastDueNotWorse\",\n",
    "                       \"age\"]\n",
    "cat_features1 = [append_str + sub for sub in num_features1]\n",
    "cat_features2 = [append_str + sub for sub in num_features2]\n",
    "\n",
    "# create the discretizer object with strategy quantile and 1000 bins\n",
    "discretizer1 = KBinsDiscretizer(n_bins=40, encode='ordinal',strategy='quantile')\n",
    "discretizer2 = KBinsDiscretizer(n_bins=4, encode='ordinal',strategy='quantile')\n",
    "\n",
    "pipeline1 = Pipeline([\n",
    "        ('imputer', SimpleImputer( strategy='median')),\n",
    "        ('bin', discretizer1)\n",
    "    ])\n",
    "# fit the discretizer to the train set\n",
    "pipeline1.fit(train.loc[:,num_features1])\n",
    "# apply the discretisation\n",
    "train_cat1 = pipeline1.transform(train.loc[:,num_features1])\n",
    "test_cat1 = pipeline1.transform(test.loc[:,num_features1])\n",
    "train_df1=pd.DataFrame(train_cat1,columns=cat_features1).astype('category')\n",
    "test_df1=pd.DataFrame(test_cat1,columns=cat_features1).astype('category')\n",
    "train_final1= pd.concat( [train.loc[:,num_features1], train_df1], axis=1) \n",
    "test_final1= pd.concat( [test.loc[:,num_features1], test_df1], axis=1) \n",
    "\n",
    "pipeline2 = Pipeline([\n",
    "        ('imputer', SimpleImputer( strategy='median')),\n",
    "        ('bin', discretizer2)\n",
    "    ])\n",
    "# fit the discretizer to the train set\n",
    "pipeline2.fit(train.loc[:,num_features2])\n",
    "# apply the discretisation\n",
    "train_cat2 = pipeline2.transform(train.loc[:,num_features2])\n",
    "test_cat2 = pipeline2.transform(test.loc[:,num_features2])\n",
    "train_df2=pd.DataFrame(train_cat2,columns=cat_features2).astype('category')\n",
    "test_df2=pd.DataFrame(test_cat2,columns=cat_features2).astype('category')\n",
    "train_final2= pd.concat( [train.loc[:,num_features2], train_df2], axis=1) \n",
    "test_final2= pd.concat( [test.loc[:,num_features2], test_df2], axis=1) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "39e63db2",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:13.149215Z",
     "iopub.status.busy": "2021-12-23T10:01:13.148696Z",
     "iopub.status.idle": "2021-12-23T10:01:13.155133Z",
     "shell.execute_reply": "2021-12-23T10:01:13.154619Z"
    },
    "papermill": {
     "duration": 0.056585,
     "end_time": "2021-12-23T10:01:13.155257",
     "exception": false,
     "start_time": "2021-12-23T10:01:13.098672",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "train_final= pd.concat( [train_final1, train_final2], axis=1) \n",
    "test_final= pd.concat( [test_final1, test_final2], axis=1) "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bfd3e4a4",
   "metadata": {
    "papermill": {
     "duration": 0.045055,
     "end_time": "2021-12-23T10:01:13.245663",
     "exception": false,
     "start_time": "2021-12-23T10:01:13.200608",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Final X and y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "522d8a48",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:13.341505Z",
     "iopub.status.busy": "2021-12-23T10:01:13.340647Z",
     "iopub.status.idle": "2021-12-23T10:01:13.342412Z",
     "shell.execute_reply": "2021-12-23T10:01:13.342880Z"
    },
    "papermill": {
     "duration": 0.051873,
     "end_time": "2021-12-23T10:01:13.343015",
     "exception": false,
     "start_time": "2021-12-23T10:01:13.291142",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Pour le train test\n",
    "target= \"SeriousDlqin2yrs\"\n",
    "X = train_final# axis=1\n",
    "X_test_final =test_final# axis=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "b6d55a9a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:13.441007Z",
     "iopub.status.busy": "2021-12-23T10:01:13.438264Z",
     "iopub.status.idle": "2021-12-23T10:01:13.458607Z",
     "shell.execute_reply": "2021-12-23T10:01:13.459093Z"
    },
    "papermill": {
     "duration": 0.070478,
     "end_time": "2021-12-23T10:01:13.459246",
     "exception": false,
     "start_time": "2021-12-23T10:01:13.388768",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 101503 entries, 0 to 101502\n",
      "Data columns (total 20 columns):\n",
      " #   Column                                    Non-Null Count   Dtype   \n",
      "---  ------                                    --------------   -----   \n",
      " 0   RevolvingUtilizationOfUnsecuredLines      101503 non-null  float16 \n",
      " 1   DebtRatio                                 101503 non-null  float32 \n",
      " 2   MonthlyIncome                             101503 non-null  float32 \n",
      " 3   cat_RevolvingUtilizationOfUnsecuredLines  101503 non-null  category\n",
      " 4   cat_DebtRatio                             101503 non-null  category\n",
      " 5   cat_MonthlyIncome                         101503 non-null  category\n",
      " 6   NumberOfDependents                        101503 non-null  float16 \n",
      " 7   NumberOfTime60-89DaysPastDueNotWorse      101503 non-null  int8    \n",
      " 8   NumberRealEstateLoansOrLines              101503 non-null  int8    \n",
      " 9   NumberOfTimes90DaysLate                   101503 non-null  int8    \n",
      " 10  NumberOfOpenCreditLinesAndLoans           101503 non-null  int8    \n",
      " 11  NumberOfTime30-59DaysPastDueNotWorse      101503 non-null  int8    \n",
      " 12  age                                       101503 non-null  int8    \n",
      " 13  cat_NumberOfDependents                    101503 non-null  category\n",
      " 14  cat_NumberOfTime60-89DaysPastDueNotWorse  101503 non-null  category\n",
      " 15  cat_NumberRealEstateLoansOrLines          101503 non-null  category\n",
      " 16  cat_NumberOfTimes90DaysLate               101503 non-null  category\n",
      " 17  cat_NumberOfOpenCreditLinesAndLoans       101503 non-null  category\n",
      " 18  cat_NumberOfTime30-59DaysPastDueNotWorse  101503 non-null  category\n",
      " 19  cat_age                                   101503 non-null  category\n",
      "dtypes: category(10), float16(2), float32(2), int8(6)\n",
      "memory usage: 2.7 MB\n"
     ]
    }
   ],
   "source": [
    "X_test_final.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "df15bf91",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:13.558505Z",
     "iopub.status.busy": "2021-12-23T10:01:13.557545Z",
     "iopub.status.idle": "2021-12-23T10:01:13.579005Z",
     "shell.execute_reply": "2021-12-23T10:01:13.578464Z"
    },
    "papermill": {
     "duration": 0.071868,
     "end_time": "2021-12-23T10:01:13.579139",
     "exception": false,
     "start_time": "2021-12-23T10:01:13.507271",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 150000 entries, 0 to 149999\n",
      "Data columns (total 20 columns):\n",
      " #   Column                                    Non-Null Count   Dtype   \n",
      "---  ------                                    --------------   -----   \n",
      " 0   RevolvingUtilizationOfUnsecuredLines      150000 non-null  float16 \n",
      " 1   DebtRatio                                 150000 non-null  float32 \n",
      " 2   MonthlyIncome                             150000 non-null  float32 \n",
      " 3   cat_RevolvingUtilizationOfUnsecuredLines  150000 non-null  category\n",
      " 4   cat_DebtRatio                             150000 non-null  category\n",
      " 5   cat_MonthlyIncome                         150000 non-null  category\n",
      " 6   NumberOfDependents                        150000 non-null  float16 \n",
      " 7   NumberOfTime60-89DaysPastDueNotWorse      150000 non-null  int8    \n",
      " 8   NumberRealEstateLoansOrLines              150000 non-null  int8    \n",
      " 9   NumberOfTimes90DaysLate                   150000 non-null  int8    \n",
      " 10  NumberOfOpenCreditLinesAndLoans           150000 non-null  int8    \n",
      " 11  NumberOfTime30-59DaysPastDueNotWorse      150000 non-null  int8    \n",
      " 12  age                                       150000 non-null  int8    \n",
      " 13  cat_NumberOfDependents                    150000 non-null  category\n",
      " 14  cat_NumberOfTime60-89DaysPastDueNotWorse  150000 non-null  category\n",
      " 15  cat_NumberRealEstateLoansOrLines          150000 non-null  category\n",
      " 16  cat_NumberOfTimes90DaysLate               150000 non-null  category\n",
      " 17  cat_NumberOfOpenCreditLinesAndLoans       150000 non-null  category\n",
      " 18  cat_NumberOfTime30-59DaysPastDueNotWorse  150000 non-null  category\n",
      " 19  cat_age                                   150000 non-null  category\n",
      "dtypes: category(10), float16(2), float32(2), int8(6)\n",
      "memory usage: 4.0 MB\n"
     ]
    }
   ],
   "source": [
    "X.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d2bc5b46",
   "metadata": {
    "papermill": {
     "duration": 0.04595,
     "end_time": "2021-12-23T10:01:13.696724",
     "exception": false,
     "start_time": "2021-12-23T10:01:13.650774",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Features Engineer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "54533c55",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:13.799397Z",
     "iopub.status.busy": "2021-12-23T10:01:13.798522Z",
     "iopub.status.idle": "2021-12-23T10:01:13.800184Z",
     "shell.execute_reply": "2021-12-23T10:01:13.800704Z"
    },
    "papermill": {
     "duration": 0.057948,
     "end_time": "2021-12-23T10:01:13.800848",
     "exception": false,
     "start_time": "2021-12-23T10:01:13.742900",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def feature_eng(df):\n",
    "    \n",
    "    df['Weighted_Sum_PastDue'] = 2 * df['NumberOfTime30-59DaysPastDueNotWorse'] + 3 * df['NumberOfTime60-89DaysPastDueNotWorse'] + 6 * df['NumberOfTimes90DaysLate']\n",
    "    \n",
    "    df['90days_out_of_TotalPastDue'] = df['NumberOfTimes90DaysLate'] / (df['Weighted_Sum_PastDue'])\n",
    "    df.loc[df['Weighted_Sum_PastDue']==0, '90days_out_of_TotalPastDue'] = 0\n",
    "   \n",
    "    df['RemainingLines'] = df['NumberOfOpenCreditLinesAndLoans'] - df['NumberRealEstateLoansOrLines']\n",
    "\n",
    "    df['Loans_vs_Other_Lines'] = df['RemainingLines'] / (1 + df['NumberRealEstateLoansOrLines'])\n",
    "\n",
    "    df['Debt_per_Real_Estate_Loan'] = df['DebtRatio'] / df['NumberRealEstateLoansOrLines']\n",
    "    df.loc[df['NumberRealEstateLoansOrLines']==0, 'Debt_per_Real_Estate_Loan'] = 0\n",
    "\n",
    "    df['Disposable_Income_per_person'] = (df['MonthlyIncome'] - df['DebtRatio']) / (df['NumberOfDependents'] + 1)\n",
    "    df.loc[df['NumberOfDependents']==0, 'Disposable_Income_per_person'] = 0\n",
    "\n",
    "    df['RemainingLines_per_person'] = df['RemainingLines'] / (df['NumberOfDependents'] + 1)\n",
    "\n",
    "    df['NumberRE_X_DebtRatio_X_age'] = df['NumberRealEstateLoansOrLines'] * df['DebtRatio'] / df['age']\n",
    "\n",
    "    df['RevolvingUtilization_per_age'] = df['RevolvingUtilizationOfUnsecuredLines'] / df['age']\n",
    "\n",
    "    #df.drop(['MonthlyIncome', 'NumberRealEstateLoansOrLines'], axis=1, inplace=True)\n",
    "    df[df==np.inf]=np.nan\n",
    "    #df.fillna(df.median(), inplace=True)\n",
    "    #credit:https://www.kaggle.com/jamesngoa/credit-scoring-feature-eng-and-prediction-0-8674\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "9a95cd01",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:13.907797Z",
     "iopub.status.busy": "2021-12-23T10:01:13.906820Z",
     "iopub.status.idle": "2021-12-23T10:01:14.059662Z",
     "shell.execute_reply": "2021-12-23T10:01:14.060095Z"
    },
    "papermill": {
     "duration": 0.20981,
     "end_time": "2021-12-23T10:01:14.060255",
     "exception": false,
     "start_time": "2021-12-23T10:01:13.850445",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Memory usage of dataframe is 8.87 MB\n",
      "Memory usage after optimization is: 6.87 MB\n",
      "Decreased by 22.6%\n",
      "Memory usage of dataframe is 6.01 MB\n",
      "Memory usage after optimization is: 4.65 MB\n",
      "Decreased by 22.6%\n"
     ]
    }
   ],
   "source": [
    "X_new=X.copy()\n",
    "X_new=feature_eng(X_new)\n",
    "X_test_final_new=X_test_final.copy()\n",
    "X_test_final_new=feature_eng(X_test_final_new)\n",
    "X_new= reduce_mem_usage(X_new)\n",
    "X_test_final_new= reduce_mem_usage(X_test_final_new)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "40f79570",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:14.160783Z",
     "iopub.status.busy": "2021-12-23T10:01:14.159900Z",
     "iopub.status.idle": "2021-12-23T10:01:14.162490Z",
     "shell.execute_reply": "2021-12-23T10:01:14.161918Z"
    },
    "papermill": {
     "duration": 0.0547,
     "end_time": "2021-12-23T10:01:14.162624",
     "exception": false,
     "start_time": "2021-12-23T10:01:14.107924",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "del train\n",
    "del test \n",
    "del train_final\n",
    "del test_final"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bdf67cc1",
   "metadata": {
    "papermill": {
     "duration": 0.048422,
     "end_time": "2021-12-23T10:01:14.261631",
     "exception": false,
     "start_time": "2021-12-23T10:01:14.213209",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "\n",
    "## Drop NAN "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "80772c71",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:14.368321Z",
     "iopub.status.busy": "2021-12-23T10:01:14.367435Z",
     "iopub.status.idle": "2021-12-23T10:01:14.475438Z",
     "shell.execute_reply": "2021-12-23T10:01:14.474789Z"
    },
    "papermill": {
     "duration": 0.166231,
     "end_time": "2021-12-23T10:01:14.475576",
     "exception": false,
     "start_time": "2021-12-23T10:01:14.309345",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "X_new=X.copy()\n",
    "X_new=feature_eng(X_new)\n",
    "X_test_final_new=X_test_final.copy()\n",
    "X_test_final_new=feature_eng(X_test_final)\n",
    "X_without_missing=X.drop(['MonthlyIncome', 'NumberOfDependents'], axis=1)\n",
    "X_new_without_missing=X_new.drop(['MonthlyIncome', 'NumberOfDependents'], axis=1)\n",
    "X_test_final_without_missing=X_test_final.drop(['MonthlyIncome', 'NumberOfDependents'], axis=1)\n",
    "X_test_final_new_without_missing=X_test_final_new.drop(['MonthlyIncome', 'NumberOfDependents'], axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4b2ae09f",
   "metadata": {
    "papermill": {
     "duration": 0.047381,
     "end_time": "2021-12-23T10:01:14.571298",
     "exception": false,
     "start_time": "2021-12-23T10:01:14.523917",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Extract final cat  and num Features \n",
    "## Cat Features "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "0a844a53",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:14.679946Z",
     "iopub.status.busy": "2021-12-23T10:01:14.679116Z",
     "iopub.status.idle": "2021-12-23T10:01:14.686748Z",
     "shell.execute_reply": "2021-12-23T10:01:14.685794Z"
    },
    "papermill": {
     "duration": 0.068425,
     "end_time": "2021-12-23T10:01:14.686874",
     "exception": false,
     "start_time": "2021-12-23T10:01:14.618449",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# select non-numeric columns\n",
    "cat_columns = X.select_dtypes(exclude=['int64','int16','float16','float32','float64','int8']).columns\n",
    "cat_columns_new = X_new.select_dtypes(exclude=['int64','int16','float16','float32','float64','int8']).columns\n",
    "cat_columns_without_missing = X_without_missing.select_dtypes(exclude=['int64','int16','float16','float32','float64','int8']).columns\n",
    "cat_columns_new_without_missing  = X_new_without_missing.select_dtypes(exclude=['int64','int16','float16','float32','float64','int8']).columns\n",
    "cat_columns_new_without_missing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b2ed9d1",
   "metadata": {
    "papermill": {
     "duration": 0.0474,
     "end_time": "2021-12-23T10:01:14.782088",
     "exception": false,
     "start_time": "2021-12-23T10:01:14.734688",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "**Num Features**\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "86987d6d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:14.900152Z",
     "iopub.status.busy": "2021-12-23T10:01:14.899284Z",
     "iopub.status.idle": "2021-12-23T10:01:14.921160Z",
     "shell.execute_reply": "2021-12-23T10:01:14.920674Z"
    },
    "papermill": {
     "duration": 0.086207,
     "end_time": "2021-12-23T10:01:14.921284",
     "exception": false,
     "start_time": "2021-12-23T10:01:14.835077",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90DaysLate', 'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age', 'Weighted_Sum_PastDue',\n",
       "       '90days_out_of_TotalPastDue', 'RemainingLines', 'Loans_vs_Other_Lines',\n",
       "       'Debt_per_Real_Estate_Loan', 'Disposable_Income_per_person',\n",
       "       'RemainingLines_per_person', 'NumberRE_X_DebtRatio_X_age',\n",
       "       'RevolvingUtilization_per_age'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# select the float columns\n",
    "num_columns = X.select_dtypes(include=['int64','int16','float16','float32','float64','int8']).columns\n",
    "num_columns_new = X_new.select_dtypes(include=['int64','int16','float16','float32','float64','int8']).columns\n",
    "num_columns_without_missing = X_without_missing.select_dtypes(include=['int64','int16','float16','float32','float64','int8']).columns\n",
    "num_columns_new_without_missing = X_new_without_missing.select_dtypes(include=['int64','int16','float16','float32','float64','int8']).columns\n",
    "num_columns_new_without_missing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a82649f",
   "metadata": {
    "papermill": {
     "duration": 0.053315,
     "end_time": "2021-12-23T10:01:15.028734",
     "exception": false,
     "start_time": "2021-12-23T10:01:14.975419",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Outlier handle:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "12651f67",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:15.133307Z",
     "iopub.status.busy": "2021-12-23T10:01:15.132416Z",
     "iopub.status.idle": "2021-12-23T10:01:15.134280Z",
     "shell.execute_reply": "2021-12-23T10:01:15.134713Z"
    },
    "papermill": {
     "duration": 0.057968,
     "end_time": "2021-12-23T10:01:15.134859",
     "exception": false,
     "start_time": "2021-12-23T10:01:15.076891",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Outlier Handle \n",
    "class OutlierReplace(BaseEstimator,TransformerMixin):\n",
    "    def __init__(self,factor=1.5):\n",
    "        self.factor = factor\n",
    "\n",
    "    def outlier_removal(self,X,y=None):\n",
    "        X = pd.Series(X).copy()\n",
    "        #qmin=X.quantile(0.01)\n",
    "        qmax=X.quantile(0.995)\n",
    "        #q1 = X.quantile(0.25)\n",
    "        #q3 = X.quantile(0.75)\n",
    "        #iqr = q3 - q1\n",
    "        #lower_bound = q1 - (self.factor * iqr)\n",
    "        #upper_bound = q3 + (self.factor * iqr)\n",
    "        upper_bound = X.quantile(0.995)\n",
    "        #X.loc[((X < lower_bound) | (X > upper_bound))] = np.nan \n",
    "        #X.loc[X < lower_bound] = qmin\n",
    "        X.loc[X > upper_bound] = qmax\n",
    "        return pd.Series(X)\n",
    "\n",
    "    def fit(self,X,y=None):\n",
    "        return self\n",
    "\n",
    "    def transform(self,X,y=None):\n",
    "        return X.apply(self.outlier_removal) "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "50a202ae",
   "metadata": {
    "papermill": {
     "duration": 0.048173,
     "end_time": "2021-12-23T10:01:15.231213",
     "exception": false,
     "start_time": "2021-12-23T10:01:15.183040",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Baseline  Pipe0:OWE+Powertransformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "3369954c",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:15.342102Z",
     "iopub.status.busy": "2021-12-23T10:01:15.337651Z",
     "iopub.status.idle": "2021-12-23T10:01:15.789367Z",
     "shell.execute_reply": "2021-12-23T10:01:15.789858Z"
    },
    "papermill": {
     "duration": 0.509546,
     "end_time": "2021-12-23T10:01:15.790019",
     "exception": false,
     "start_time": "2021-12-23T10:01:15.280473",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"29f909a6-3683-45ce-be6f-97da0a1922b7\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"29f909a6-3683-45ce-be6f-97da0a1922b7\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89Da...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"75f0465a-8ff3-469e-bcc4-a75dcb84ee11\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"75f0465a-8ff3-469e-bcc4-a75dcb84ee11\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEs...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"71ea1b87-fa93-4229-9bec-19fca60d56f8\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"71ea1b87-fa93-4229-9bec-19fca60d56f8\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"9cfd2432-4cae-48e7-9c7a-e4c2dadbd1d0\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"9cfd2432-4cae-48e7-9c7a-e4c2dadbd1d0\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"c270e598-8296-4f06-891e-88bf87f77b77\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"c270e598-8296-4f06-891e-88bf87f77b77\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"670a8a4a-413f-4674-a515-a4a000a689d2\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"670a8a4a-413f-4674-a515-a4a000a689d2\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"e6cd963f-6b94-450b-8d7b-bbbc1b4e7762\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"e6cd963f-6b94-450b-8d7b-bbbc1b4e7762\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"d0e8426d-2e4c-427d-b84f-f66dba04a2b7\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"d0e8426d-2e4c-427d-b84f-f66dba04a2b7\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"535990cd-f6a4-4f8d-b672-386b06538ba0\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"535990cd-f6a4-4f8d-b672-386b06538ba0\">WOEEncoder</label><div class=\"sk-toggleable__content\"><pre>WOEEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"4441fc4a-8722-4fe2-82c0-5269f59d1bf6\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"4441fc4a-8722-4fe2-82c0-5269f59d1bf6\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"fffa6ab5-cd8d-4d7d-a645-69412fbabe13\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"fffa6ab5-cd8d-4d7d-a645-69412fbabe13\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89Da...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Cat pipeline\n",
    "categorical_transformer = Pipeline(\n",
    "                    steps=[\n",
    "                        ('imputer', SimpleImputer(strategy='most_frequent',\n",
    "                                                  fill_value='missing',\n",
    "                                                  add_indicator=True)),\n",
    "                        ('encoder',  ce.woe.WOEEncoder()),#(Numerical Input, Categorical Output)\n",
    "                         ('reducedim',  SelectPercentile( mutual_info_classif, percentile=90))\n",
    "\n",
    "                    ]\n",
    "                    ) \n",
    "#Define vnum pipeline\n",
    "numeric_transformer = Pipeline(\n",
    "                            steps=[\n",
    "                            ('imputer', SimpleImputer(strategy='median'\n",
    "                                                      ,add_indicator=True)),\n",
    "                            ('scaler', PowerTransformer()),#(Numerical Input, Numerical Output)\n",
    "                            # Create an SelectKBest object to select features with two best ANOVA F-Values\n",
    "                            #The F-value scores examine if, when we group the numerical feature by the target vector, the means for each group are significantly different\n",
    "                            ('reducedim',  SelectPercentile(f_classif,percentile=90))\n",
    "                            ]\n",
    "                            )\n",
    "# Features union cat + num \n",
    "# WOE+PowerTransformer\n",
    "preprocessor_woe_powertransformer = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer, num_columns),\n",
    "                ('categorical', categorical_transformer, cat_columns)\n",
    "            ]\n",
    "            )\n",
    "lgbm_param={'learning_rate': 0.0018069834369607075,\n",
    "                             'max_depth': 8,\n",
    "                             #'max_features': 4,\n",
    "                             'min_samples_leaf': 47,\n",
    "                             #'min_samples_split': 389,\n",
    "                             'subsample': 0.8573598985000007,\n",
    "                             #'n_iter_no_change': 300,\n",
    "                             'n_estimators': 5000,\n",
    "                            # 'verbose': 0,\n",
    "                             'random_state': 42,\n",
    "                             'metric': 'auc',\n",
    "                             \"device_type\" : \"gpu\",\n",
    "                            'boosting_type': 'gbdt',\n",
    "                           # 'tree_method': \"gpu_hist\"\n",
    "                           }\n",
    "pipe_lgbm_woe_powertransformer = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_woe_powertransformer),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_woe_powertransformer"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "31e641ea",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-19T13:04:51.69428Z",
     "iopub.status.busy": "2021-12-19T13:04:51.693953Z",
     "iopub.status.idle": "2021-12-19T13:04:51.697766Z",
     "shell.execute_reply": "2021-12-19T13:04:51.697057Z",
     "shell.execute_reply.started": "2021-12-19T13:04:51.694242Z"
    },
    "papermill": {
     "duration": 0.049253,
     "end_time": "2021-12-23T10:01:15.888669",
     "exception": false,
     "start_time": "2021-12-23T10:01:15.839416",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Add outlier Handle:outlier+woe+powertransormer "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "1f4c48d7",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:16.026344Z",
     "iopub.status.busy": "2021-12-23T10:01:16.016496Z",
     "iopub.status.idle": "2021-12-23T10:01:16.475117Z",
     "shell.execute_reply": "2021-12-23T10:01:16.475537Z"
    },
    "papermill": {
     "duration": 0.538007,
     "end_time": "2021-12-23T10:01:16.475704",
     "exception": false,
     "start_time": "2021-12-23T10:01:15.937697",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"bb9d13b7-4cfc-4132-a0e0-4e4f0ad76aeb\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"bb9d13b7-4cfc-4132-a0e0-4e4f0ad76aeb\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'Numbe...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"bcecb8f5-30e4-465b-9157-ee0c160286f8\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"bcecb8f5-30e4-465b-9157-ee0c160286f8\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('outlierhandle',\n",
       "                                                  OutlierReplace()),\n",
       "                                                 ('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-8...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"6fd9a607-d556-483c-9463-e8a5569c6daf\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"6fd9a607-d556-483c-9463-e8a5569c6daf\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"8ff2a3cb-10e8-4e7d-b68d-20133c265248\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"8ff2a3cb-10e8-4e7d-b68d-20133c265248\">OutlierReplace</label><div class=\"sk-toggleable__content\"><pre>OutlierReplace()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"59629f7b-8c40-4fa2-9920-bfe0996c3080\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"59629f7b-8c40-4fa2-9920-bfe0996c3080\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f3e525e3-360d-44f8-9d30-dfe1a60069c3\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f3e525e3-360d-44f8-9d30-dfe1a60069c3\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"0aac3352-4429-410c-a7ae-55a3ceb21fe7\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"0aac3352-4429-410c-a7ae-55a3ceb21fe7\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"4692b056-66a6-495b-8e47-f75f68e2999a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"4692b056-66a6-495b-8e47-f75f68e2999a\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"1089415f-5db8-4811-8f87-c2689892ac0a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"1089415f-5db8-4811-8f87-c2689892ac0a\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"66fd85d8-8999-4ecc-a55a-44d718726f04\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"66fd85d8-8999-4ecc-a55a-44d718726f04\">WOEEncoder</label><div class=\"sk-toggleable__content\"><pre>WOEEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b291e69d-b61d-4c4f-9d06-40468be881c2\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b291e69d-b61d-4c4f-9d06-40468be881c2\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"2382de8c-ac28-4aed-8128-c1e0804b3a1a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"2382de8c-ac28-4aed-8128-c1e0804b3a1a\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'Numbe...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Define vnum pipeline\n",
    "numeric_transformer_outlier = Pipeline(\n",
    "                            steps=[('outlierhandle', OutlierReplace()),\n",
    "                            ('imputer', SimpleImputer(strategy='median'\n",
    "                                                      ,add_indicator=True)),\n",
    "                            ('scaler', PowerTransformer()),#(Numerical Input, Numerical Output)\n",
    "                            # Create an SelectKBest object to select features with two best ANOVA F-Values\n",
    "                            #The F-value scores examine if, when we group the numerical feature by the target vector, the means for each group are significantly different\n",
    "                            ('reducedim',  SelectPercentile(f_classif,percentile=90))\n",
    "                            ]\n",
    "                            )\n",
    "# Features union cat + num \n",
    "# WOE+PowerTransformer\n",
    "preprocessor_woe_powertransformer_outlier = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_outlier, num_columns),\n",
    "                ('categorical', categorical_transformer, cat_columns)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_woe_powertransformer_outlier = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_woe_powertransformer_outlier),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_woe_powertransformer_outlier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b550e08c",
   "metadata": {
    "papermill": {
     "duration": 0.050086,
     "end_time": "2021-12-23T10:01:16.577142",
     "exception": false,
     "start_time": "2021-12-23T10:01:16.527056",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "### Delete missing:woe+powertransormer for not null data \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "665f5eca",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:16.679926Z",
     "iopub.status.busy": "2021-12-23T10:01:16.679075Z",
     "iopub.status.idle": "2021-12-23T10:01:17.122319Z",
     "shell.execute_reply": "2021-12-23T10:01:17.122792Z"
    },
    "papermill": {
     "duration": 0.4963,
     "end_time": "2021-12-23T10:01:17.122955",
     "exception": false,
     "start_time": "2021-12-23T10:01:16.626655",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"162a5770-f5a4-4986-9a80-d9032f07225a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"162a5770-f5a4-4986-9a80-d9032f07225a\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateL...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"99be7d6e-b22f-47fd-b750-e2187a919368\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"99be7d6e-b22f-47fd-b750-e2187a919368\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90Day...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"01254121-78dc-4ae0-910c-e0fc9e7de4a9\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"01254121-78dc-4ae0-910c-e0fc9e7de4a9\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90DaysLate', 'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"939fe4f5-8d73-475b-8599-6d6f64a3262e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"939fe4f5-8d73-475b-8599-6d6f64a3262e\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"bf7037e1-d4d4-4432-a999-f797e56e1f5e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"bf7037e1-d4d4-4432-a999-f797e56e1f5e\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"cad2a20a-d7b0-49ac-b1a4-a2864bab9932\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"cad2a20a-d7b0-49ac-b1a4-a2864bab9932\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"99d7a634-50a5-42fe-a6ff-a7a774ca6316\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"99d7a634-50a5-42fe-a6ff-a7a774ca6316\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"df22957d-9493-4a85-89b7-9fef9113cd87\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"df22957d-9493-4a85-89b7-9fef9113cd87\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"cc17c481-c17a-4094-a96e-5cf06f4f01bd\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"cc17c481-c17a-4094-a96e-5cf06f4f01bd\">WOEEncoder</label><div class=\"sk-toggleable__content\"><pre>WOEEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"39f3ca64-6ad2-4ed6-bd92-59461fcf183e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"39f3ca64-6ad2-4ed6-bd92-59461fcf183e\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"d39a2513-0da9-4ce4-9a1d-0eb3fd159263\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"d39a2513-0da9-4ce4-9a1d-0eb3fd159263\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateL...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Features union cat + num \n",
    "# WOE+PowerTransformer\n",
    "preprocessor_woe_powertransformer_without_missing = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer, num_columns_without_missing),\n",
    "                ('categorical', categorical_transformer, cat_columns_without_missing)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_woe_powertransformer_without_missing = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_woe_powertransformer_without_missing),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_woe_powertransformer_without_missing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c342d43e",
   "metadata": {
    "papermill": {
     "duration": 0.053756,
     "end_time": "2021-12-23T10:01:17.229129",
     "exception": false,
     "start_time": "2021-12-23T10:01:17.175373",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Pipe2: CatBoostEncoder + Power transformer\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "56b76b6c",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:17.463193Z",
     "iopub.status.busy": "2021-12-23T10:01:17.452789Z",
     "iopub.status.idle": "2021-12-23T10:01:17.886058Z",
     "shell.execute_reply": "2021-12-23T10:01:17.886509Z"
    },
    "papermill": {
     "duration": 0.605462,
     "end_time": "2021-12-23T10:01:17.886680",
     "exception": false,
     "start_time": "2021-12-23T10:01:17.281218",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"4e321dde-acce-4a76-b674-18a48d99ee00\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"4e321dde-acce-4a76-b674-18a48d99ee00\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89Da...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"6a525e45-eece-4bc8-a5f2-9af0c92bbbba\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"6a525e45-eece-4bc8-a5f2-9af0c92bbbba\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEs...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"e9ec031b-4e01-4239-be39-a01ba374743a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"e9ec031b-4e01-4239-be39-a01ba374743a\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f8db3663-2c42-45a5-ab06-1e15edbc171c\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f8db3663-2c42-45a5-ab06-1e15edbc171c\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"53b04c1b-de3a-4602-bade-af498cb5f6fb\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"53b04c1b-de3a-4602-bade-af498cb5f6fb\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"4a80f90d-a850-4b0b-8d2c-964b37233015\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"4a80f90d-a850-4b0b-8d2c-964b37233015\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f00c5865-6436-4627-800d-ffb88b69fc06\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f00c5865-6436-4627-800d-ffb88b69fc06\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"15593abe-607d-4450-a671-3d9ef737a031\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"15593abe-607d-4450-a671-3d9ef737a031\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"3668fbb1-79b3-4b12-af8f-8c4564ad82b1\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"3668fbb1-79b3-4b12-af8f-8c4564ad82b1\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"a33d97c0-b3dc-4e14-a5ba-471eaf96d110\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"a33d97c0-b3dc-4e14-a5ba-471eaf96d110\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f137f600-91bb-4f63-9432-d02952996ac6\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f137f600-91bb-4f63-9432-d02952996ac6\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89Da...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Define cat pipeline\n",
    "# Cat pipeline\n",
    "categorical_transformer_catboost = Pipeline(\n",
    "                    steps=[\n",
    "                        ('imputer', SimpleImputer(strategy='most_frequent',\n",
    "                                                  fill_value='missing',\n",
    "                                                  add_indicator=True)),\n",
    "                        ('encoder',  ce.cat_boost.CatBoostEncoder()),#(Numerical Input, Categorical Output)\n",
    "                         ('reducedim',  SelectPercentile( mutual_info_classif,\n",
    "                                                         percentile=90))\n",
    "                    ]\n",
    "                    ) \n",
    "#Define vnum pipeline\n",
    "numeric_transformer = Pipeline(\n",
    "                            steps=[\n",
    "                            ('imputer', SimpleImputer(strategy='median'\n",
    "                                                      ,add_indicator=True)),\n",
    "                            ('scaler', PowerTransformer()),#(Numerical Input, Numerical Output)\n",
    "                            # Create an SelectKBest object to select features with two best ANOVA F-Values\n",
    "                            #The F-value scores examine if, when we group the numerical feature by the target vector, the means for each group are significantly different\n",
    "                            ('reducedim',  SelectPercentile(f_classif,percentile=90))\n",
    "                            ]\n",
    "                            )\n",
    "# Features union cat + num \n",
    "# Catboost+PowerTransformer\n",
    "preprocessor_catboost_powertransformer = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer, num_columns),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_powertransformer = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_powertransformer),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_powertransformer"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c77c0d6",
   "metadata": {
    "papermill": {
     "duration": 0.055873,
     "end_time": "2021-12-23T10:01:17.994742",
     "exception": false,
     "start_time": "2021-12-23T10:01:17.938869",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Handle Outlier:CatBoostEncoder + Power transformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "fab281ee",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:18.114105Z",
     "iopub.status.busy": "2021-12-23T10:01:18.113234Z",
     "iopub.status.idle": "2021-12-23T10:01:18.562446Z",
     "shell.execute_reply": "2021-12-23T10:01:18.562867Z"
    },
    "papermill": {
     "duration": 0.515148,
     "end_time": "2021-12-23T10:01:18.563024",
     "exception": false,
     "start_time": "2021-12-23T10:01:18.047876",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"5c69cdac-c311-4a9b-b03e-345693d7353e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"5c69cdac-c311-4a9b-b03e-345693d7353e\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'Numbe...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f3b72e3f-9337-42fb-b690-cc77c328758f\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f3b72e3f-9337-42fb-b690-cc77c328758f\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('outlierhandle',\n",
       "                                                  OutlierReplace()),\n",
       "                                                 ('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-8...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"feaf6dbc-feb7-4fd0-9ab5-1c718cc0913a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"feaf6dbc-feb7-4fd0-9ab5-1c718cc0913a\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"0784b5f6-ff2b-496c-b85f-70f7e2d374ff\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"0784b5f6-ff2b-496c-b85f-70f7e2d374ff\">OutlierReplace</label><div class=\"sk-toggleable__content\"><pre>OutlierReplace()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"52877e28-ce12-474b-a858-0751ff8e4c09\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"52877e28-ce12-474b-a858-0751ff8e4c09\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"99ea07ad-7418-4e71-96c0-6c2759682069\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"99ea07ad-7418-4e71-96c0-6c2759682069\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"3a0e1cc7-e37d-4e4c-964b-22242cd4e50e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"3a0e1cc7-e37d-4e4c-964b-22242cd4e50e\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ec52c143-c075-4fdb-9dbe-dafdfe257eeb\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ec52c143-c075-4fdb-9dbe-dafdfe257eeb\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"00767168-bcb8-47dc-b0fb-7e0a61734cf2\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"00767168-bcb8-47dc-b0fb-7e0a61734cf2\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"3b7dadcc-7a30-4bef-ab09-69430eb421a6\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"3b7dadcc-7a30-4bef-ab09-69430eb421a6\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"e7f43683-f60d-4b61-ae76-38b9b8f6bc8a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"e7f43683-f60d-4b61-ae76-38b9b8f6bc8a\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"529ee74a-89fd-4da6-abe6-0ee0e0400790\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"529ee74a-89fd-4da6-abe6-0ee0e0400790\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'Numbe...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Define vnum pipeline\n",
    "numeric_transformer_outlier = Pipeline(\n",
    "                            steps=[('outlierhandle', OutlierReplace()),\n",
    "                            ('imputer', SimpleImputer(strategy='median'\n",
    "                                                      ,add_indicator=True)),\n",
    "                            ('scaler', PowerTransformer()),#(Numerical Input, Numerical Output)\n",
    "                            # Create an SelectKBest object to select features with two best ANOVA F-Values\n",
    "                            #The F-value scores examine if, when we group the numerical feature by the target vector, the means for each group are significantly different\n",
    "                            ('reducedim',  SelectPercentile(f_classif,percentile=90))\n",
    "                            ]\n",
    "                            )\n",
    "# Features union cat + num \n",
    "# WOE+PowerTransformer\n",
    "preprocessor_catboost_powertransformer_outlier = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_outlier, num_columns),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_powertransformer_outlier = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_powertransformer_outlier),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_powertransformer_outlier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "28b8936c",
   "metadata": {
    "papermill": {
     "duration": 0.051453,
     "end_time": "2021-12-23T10:01:18.666249",
     "exception": false,
     "start_time": "2021-12-23T10:01:18.614796",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Drop misssing :CatBoostEncoder + Power transformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "688340bb",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:18.817741Z",
     "iopub.status.busy": "2021-12-23T10:01:18.802219Z",
     "iopub.status.idle": "2021-12-23T10:01:19.206295Z",
     "shell.execute_reply": "2021-12-23T10:01:19.206805Z"
    },
    "papermill": {
     "duration": 0.489377,
     "end_time": "2021-12-23T10:01:19.206963",
     "exception": false,
     "start_time": "2021-12-23T10:01:18.717586",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"2b0a7faf-0bbe-430c-9653-a2e0f504d36a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"2b0a7faf-0bbe-430c-9653-a2e0f504d36a\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateL...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"83fea9cd-3377-4bc1-9886-8894e91986c9\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"83fea9cd-3377-4bc1-9886-8894e91986c9\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90Day...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"6af57d8f-fea5-4fd6-ac79-0343fd3c06d6\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"6af57d8f-fea5-4fd6-ac79-0343fd3c06d6\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90DaysLate', 'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"91fe66ba-99cc-4a7b-a0f1-5cb001a4058c\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"91fe66ba-99cc-4a7b-a0f1-5cb001a4058c\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ace5ec03-5a0c-4cb3-802c-73f70cab1eaa\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ace5ec03-5a0c-4cb3-802c-73f70cab1eaa\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"30d4e064-9047-466b-a828-ac5401678024\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"30d4e064-9047-466b-a828-ac5401678024\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"84319d5e-9360-4a90-accd-e327aa9686bd\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"84319d5e-9360-4a90-accd-e327aa9686bd\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f7f2d1fb-497d-483f-a6a6-5280c869eb88\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f7f2d1fb-497d-483f-a6a6-5280c869eb88\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"dcf40a62-28b9-49e5-9b78-5e9e29b8b890\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"dcf40a62-28b9-49e5-9b78-5e9e29b8b890\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b8045869-d263-4cbc-bca4-5d1eb75431d8\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b8045869-d263-4cbc-bca4-5d1eb75431d8\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"06861596-ee99-4b9e-b538-4d19b956dbb3\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"06861596-ee99-4b9e-b538-4d19b956dbb3\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateL...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Features union cat + num \n",
    "# WOE+PowerTransformer\n",
    "\n",
    "pipe_lgbm_catboost_powertransformer_without_missing = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer, num_columns_without_missing),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns_without_missing)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_powertransformer_without_missing = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', pipe_lgbm_catboost_powertransformer_without_missing),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_powertransformer_without_missing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "653d2366",
   "metadata": {
    "papermill": {
     "duration": 0.052113,
     "end_time": "2021-12-23T10:01:19.311636",
     "exception": false,
     "start_time": "2021-12-23T10:01:19.259523",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Messing +Outlier handle :CatBoostEncoder + Power transformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "0d29c72a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:19.428675Z",
     "iopub.status.busy": "2021-12-23T10:01:19.427825Z",
     "iopub.status.idle": "2021-12-23T10:01:19.878798Z",
     "shell.execute_reply": "2021-12-23T10:01:19.878329Z"
    },
    "papermill": {
     "duration": 0.5143,
     "end_time": "2021-12-23T10:01:19.878931",
     "exception": false,
     "start_time": "2021-12-23T10:01:19.364631",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"bd9778e9-9636-4eff-a202-01ab8ca7d286\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"bd9778e9-9636-4eff-a202-01ab8ca7d286\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89Days...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"23ea8bac-9b96-4548-9249-4f222aa2a69f\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"23ea8bac-9b96-4548-9249-4f222aa2a69f\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('outlierhandle',\n",
       "                                                  OutlierReplace()),\n",
       "                                                 ('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEsta...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"6310be62-7173-4bda-b6c8-0d78ee439ee8\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"6310be62-7173-4bda-b6c8-0d78ee439ee8\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90DaysLate', 'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b97d0be7-e9c9-4cd4-aa24-d99faa7e3e76\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b97d0be7-e9c9-4cd4-aa24-d99faa7e3e76\">OutlierReplace</label><div class=\"sk-toggleable__content\"><pre>OutlierReplace()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"81fd005e-f20f-4a64-8909-051f1f027d27\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"81fd005e-f20f-4a64-8909-051f1f027d27\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"451caf76-0ae7-4a08-b59e-e9c647434cf4\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"451caf76-0ae7-4a08-b59e-e9c647434cf4\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"55b47e4b-94bf-4085-89b1-413a9fbfc63a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"55b47e4b-94bf-4085-89b1-413a9fbfc63a\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"746a8b09-16aa-4dce-b29c-363b2237f100\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"746a8b09-16aa-4dce-b29c-363b2237f100\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"3f8c8946-0c22-46c3-ad35-b25ce36d20cf\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"3f8c8946-0c22-46c3-ad35-b25ce36d20cf\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"fbb8d3ce-798e-40b5-be09-a98878f0716f\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"fbb8d3ce-798e-40b5-be09-a98878f0716f\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"6d4ba71d-5820-42ec-a020-e06f6a4ece54\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"6d4ba71d-5820-42ec-a020-e06f6a4ece54\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"993797de-b19c-4851-b8c6-3e0e4af88bb3\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"993797de-b19c-4851-b8c6-3e0e4af88bb3\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89Days...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Features union cat + num \n",
    "# WOE+PowerTransformer\n",
    "preprocessor_catboost_powertransformer_outlier_without_missing = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_outlier, num_columns_without_missing),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns_without_missing)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_powertransformer_outlier_without_missing = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_powertransformer_outlier_without_missing),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_powertransformer_outlier_without_missing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "98ce8c2c",
   "metadata": {
    "papermill": {
     "duration": 0.053713,
     "end_time": "2021-12-23T10:01:19.986358",
     "exception": false,
     "start_time": "2021-12-23T10:01:19.932645",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Pipe 3: catboost +qauntile transformer  \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "b4236865",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:20.147543Z",
     "iopub.status.busy": "2021-12-23T10:01:20.106520Z",
     "iopub.status.idle": "2021-12-23T10:01:20.663068Z",
     "shell.execute_reply": "2021-12-23T10:01:20.664046Z"
    },
    "papermill": {
     "duration": 0.623897,
     "end_time": "2021-12-23T10:01:20.664265",
     "exception": false,
     "start_time": "2021-12-23T10:01:20.040368",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"fefea3c2-1ffd-4d18-92c3-b49e511ffb8a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"fefea3c2-1ffd-4d18-92c3-b49e511ffb8a\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-8...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"0a5cbad3-0d35-433b-aa86-9ef22e998bf3\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"0a5cbad3-0d35-433b-aa86-9ef22e998bf3\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler',\n",
       "                                                  QuantileTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRea...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"a64990aa-ba77-4e36-baa3-11a9347e229d\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"a64990aa-ba77-4e36-baa3-11a9347e229d\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"50d22bc9-2ec3-44e3-8d59-d82d71b1f485\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"50d22bc9-2ec3-44e3-8d59-d82d71b1f485\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b82525ad-3497-4378-b3ee-92bcc97ca37b\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b82525ad-3497-4378-b3ee-92bcc97ca37b\">QuantileTransformer</label><div class=\"sk-toggleable__content\"><pre>QuantileTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"84fca0d1-c11c-4464-a4c2-3d8b5798fc5e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"84fca0d1-c11c-4464-a4c2-3d8b5798fc5e\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"0d5171aa-653a-469c-b71b-979a69976157\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"0d5171aa-653a-469c-b71b-979a69976157\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"0caffb76-d5ce-43b7-acf6-7d9c621a0f8f\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"0caffb76-d5ce-43b7-acf6-7d9c621a0f8f\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"432572b2-121d-490b-a1f7-c625f821e0b2\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"432572b2-121d-490b-a1f7-c625f821e0b2\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"fc91371a-d551-4ceb-af83-93941adc8477\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"fc91371a-d551-4ceb-af83-93941adc8477\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"fdc1b355-c2dc-4bd2-bf95-b7c66789ec0c\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"fdc1b355-c2dc-4bd2-bf95-b7c66789ec0c\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-8...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# complete pipe :\n",
    "# select the float/cat columns\n",
    "#cat_feautres = X.select_dtypes(include=['object','category']).columns\n",
    "#num_features = X.select_dtypes(exclude=['object','category']).columns\n",
    "categorical_transformer_catboost = Pipeline(\n",
    "                    steps=[\n",
    "                        ('imputer', SimpleImputer(strategy='most_frequent',\n",
    "                                                  fill_value='missing',\n",
    "                                                  add_indicator=True)),\n",
    "                        ('encoder',  ce.cat_boost.CatBoostEncoder()),#(Numerical Input, Categorical Output)\n",
    "                         ('reducedim',  SelectPercentile( mutual_info_classif,\n",
    "                                                        percentile=90))\n",
    "\n",
    "                    ]\n",
    "                    ) \n",
    "numeric_transformer_quantile = Pipeline(\n",
    "                            steps=[\n",
    "                            ('imputer', SimpleImputer(strategy='median'\n",
    "                                                      ,add_indicator=True)),\n",
    "                            ('scaler', QuantileTransformer()),#(Numerical Input, Numerical Output)\n",
    "                            # Create an SelectKBest object to select features with two best ANOVA F-Values\n",
    "                            #The F-value scores examine if, when we group the numerical feature by the target vector, the means for each group are significantly different\n",
    "                            ('reducedim',  SelectPercentile(f_classif,\n",
    "                                                            percentile=90))\n",
    "                                # i used 90 before\n",
    "\n",
    "                            ]\n",
    "                            )\n",
    "# Features union cat + num \n",
    "# Catboost+QuantileTransformer\n",
    "preprocessor_catboost_qauntile = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_quantile, num_columns),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_qauntile = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_qauntile),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_qauntile"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fca0885d",
   "metadata": {
    "papermill": {
     "duration": 0.087664,
     "end_time": "2021-12-23T10:01:20.840312",
     "exception": false,
     "start_time": "2021-12-23T10:01:20.752648",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Add Outlier Handle:catboost +qauntile transformer  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "84566022",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:21.047638Z",
     "iopub.status.busy": "2021-12-23T10:01:21.031976Z",
     "iopub.status.idle": "2021-12-23T10:01:21.509306Z",
     "shell.execute_reply": "2021-12-23T10:01:21.509742Z"
    },
    "papermill": {
     "duration": 0.580535,
     "end_time": "2021-12-23T10:01:21.509893",
     "exception": false,
     "start_time": "2021-12-23T10:01:20.929358",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"80e215e6-de51-4348-b907-cb63ad0832b0\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"80e215e6-de51-4348-b907-cb63ad0832b0\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'Nu...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ee7fe5bd-8837-4d6f-83c3-cdd585263bd6\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ee7fe5bd-8837-4d6f-83c3-cdd585263bd6\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('outlierhandle',\n",
       "                                                  OutlierReplace()),\n",
       "                                                 ('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler',\n",
       "                                                  QuantileTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime6...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"d82c49a6-2e7a-41b7-9097-9428c49c3e24\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"d82c49a6-2e7a-41b7-9097-9428c49c3e24\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"beee8ea9-9f63-496e-8bd0-614af752d0e3\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"beee8ea9-9f63-496e-8bd0-614af752d0e3\">OutlierReplace</label><div class=\"sk-toggleable__content\"><pre>OutlierReplace()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ff2d2781-8609-4102-b7cb-ef9fe83a06aa\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ff2d2781-8609-4102-b7cb-ef9fe83a06aa\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"7cf06d79-df91-47ed-8dd3-0ab038c28b75\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"7cf06d79-df91-47ed-8dd3-0ab038c28b75\">QuantileTransformer</label><div class=\"sk-toggleable__content\"><pre>QuantileTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"089ff22f-f82c-49cb-8577-ef97c7d506e2\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"089ff22f-f82c-49cb-8577-ef97c7d506e2\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"9ec9f4e3-9ce5-48c0-a1b6-5da46d941fde\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"9ec9f4e3-9ce5-48c0-a1b6-5da46d941fde\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"fab4da5d-cf4e-43af-bb73-324b3039111e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"fab4da5d-cf4e-43af-bb73-324b3039111e\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"92249ca6-da71-4d7a-875a-9fb6ae2e7054\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"92249ca6-da71-4d7a-875a-9fb6ae2e7054\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"eb09a224-8664-4383-9487-d58ed4f223dd\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"eb09a224-8664-4383-9487-d58ed4f223dd\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"cce09113-9cf2-4cfd-9294-dd9dd6c974e3\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"cce09113-9cf2-4cfd-9294-dd9dd6c974e3\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'Nu...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# complete pipe :\n",
    "# select the float/cat columns\n",
    "#cat_feautres = X.select_dtypes(include=['object','category']).columns\n",
    "#num_features = X.select_dtypes(exclude=['object','category']).columns\n",
    "categorical_transformer_catboost = Pipeline(\n",
    "                    steps=[\n",
    "                        ('imputer', SimpleImputer(strategy='most_frequent',\n",
    "                                                  fill_value='missing',\n",
    "                                                  add_indicator=True)),\n",
    "                        ('encoder',  ce.cat_boost.CatBoostEncoder()),#(Numerical Input, Categorical Output)\n",
    "                         ('reducedim',  SelectPercentile( mutual_info_classif,\n",
    "                                                        percentile=90))\n",
    "\n",
    "                    ]\n",
    "                    ) \n",
    "numeric_transformer_quantile_outlier = Pipeline(\n",
    "                            steps=[('outlierhandle', OutlierReplace()),\n",
    "                            ('imputer', SimpleImputer(strategy='median'\n",
    "                                                      ,add_indicator=True)),\n",
    "                            ('scaler', QuantileTransformer()),#(Numerical Input, Numerical Output)\n",
    "                            # Create an SelectKBest object to select features with two best ANOVA F-Values\n",
    "                            #The F-value scores examine if, when we group the numerical feature by the target vector, the means for each group are significantly different\n",
    "                            ('reducedim',  SelectPercentile(f_classif,\n",
    "                                                            percentile=90))\n",
    "                                # i used 90 before\n",
    "\n",
    "                            ]\n",
    "                            )\n",
    "# Features union cat + num \n",
    "# Catboost+QuantileTransformer\n",
    "preprocessor_catboost_qauntile_outlier = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_quantile_outlier, num_columns),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_qauntile_outlier = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_qauntile_outlier),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "\n",
    "pipe_lgbm_catboost_qauntile_outlier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c44b0191",
   "metadata": {
    "papermill": {
     "duration": 0.054758,
     "end_time": "2021-12-23T10:01:21.620535",
     "exception": false,
     "start_time": "2021-12-23T10:01:21.565777",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Drop Missing:catboost +qauntile transformer  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "d215e06c",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:21.789186Z",
     "iopub.status.busy": "2021-12-23T10:01:21.763367Z",
     "iopub.status.idle": "2021-12-23T10:01:22.181070Z",
     "shell.execute_reply": "2021-12-23T10:01:22.181494Z"
    },
    "papermill": {
     "duration": 0.506787,
     "end_time": "2021-12-23T10:01:22.181676",
     "exception": false,
     "start_time": "2021-12-23T10:01:21.674889",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ca3b94ae-13c0-42c5-a96f-663fff07a1db\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ca3b94ae-13c0-42c5-a96f-663fff07a1db\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEsta...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"e0d9630b-86f2-4e55-98b3-e9718a9c2db0\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"e0d9630b-86f2-4e55-98b3-e9718a9c2db0\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler',\n",
       "                                                  QuantileTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ff9beb3d-800a-4af7-aaf5-6e6d13e02219\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ff9beb3d-800a-4af7-aaf5-6e6d13e02219\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90DaysLate', 'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f29b3dba-c5b3-4d0b-8ad2-b00dc5d00e45\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f29b3dba-c5b3-4d0b-8ad2-b00dc5d00e45\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"d40af509-4840-4c47-9c5d-31f3f379c963\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"d40af509-4840-4c47-9c5d-31f3f379c963\">QuantileTransformer</label><div class=\"sk-toggleable__content\"><pre>QuantileTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"84fd2933-93bc-4e6d-8ca3-5871c5347300\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"84fd2933-93bc-4e6d-8ca3-5871c5347300\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f890bcf5-af81-4987-95ed-7f0f36f97fb3\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f890bcf5-af81-4987-95ed-7f0f36f97fb3\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b81affe4-a29b-4132-8579-b56463445705\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b81affe4-a29b-4132-8579-b56463445705\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"90c96088-c8ed-41a0-bba1-ae62caabdac2\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"90c96088-c8ed-41a0-bba1-ae62caabdac2\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"71451fa1-efff-4728-a51c-43f29b4f2d0e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"71451fa1-efff-4728-a51c-43f29b4f2d0e\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"d3fbd96a-50a6-4610-b174-d53176ed21d8\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"d3fbd96a-50a6-4610-b174-d53176ed21d8\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEsta...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Features union cat + num \n",
    "# Catboost+QuantileTransformer\n",
    "preprocessor_catboost_qauntile_without_missing = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_quantile, num_columns_without_missing ),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns_without_missing )\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_qauntile_without_missing  = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_qauntile_without_missing ),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_qauntile_without_missing "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ad08ff8a",
   "metadata": {
    "papermill": {
     "duration": 0.057796,
     "end_time": "2021-12-23T10:01:22.295452",
     "exception": false,
     "start_time": "2021-12-23T10:01:22.237656",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Drop Null+handle outlier : catboost +qauntile transformer  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "13d133c7",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:22.415513Z",
     "iopub.status.busy": "2021-12-23T10:01:22.414688Z",
     "iopub.status.idle": "2021-12-23T10:01:22.896088Z",
     "shell.execute_reply": "2021-12-23T10:01:22.896484Z"
    },
    "papermill": {
     "duration": 0.545561,
     "end_time": "2021-12-23T10:01:22.896651",
     "exception": false,
     "start_time": "2021-12-23T10:01:22.351090",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"76d9a375-c330-4987-a9dc-a1603b35177e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"76d9a375-c330-4987-a9dc-a1603b35177e\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89D...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"5567e650-6857-4e30-95bd-9cdc9844c4b0\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"5567e650-6857-4e30-95bd-9cdc9844c4b0\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('outlierhandle',\n",
       "                                                  OutlierReplace()),\n",
       "                                                 ('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler',\n",
       "                                                  QuantileTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealE...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ac1a42c4-7004-46f1-a16d-cf5e3af2f8ab\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ac1a42c4-7004-46f1-a16d-cf5e3af2f8ab\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90DaysLate', 'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"c9c16885-6714-4966-bcd1-9994876d8f10\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"c9c16885-6714-4966-bcd1-9994876d8f10\">OutlierReplace</label><div class=\"sk-toggleable__content\"><pre>OutlierReplace()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"de58b3f9-490a-49fd-aefc-c1501f1b0265\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"de58b3f9-490a-49fd-aefc-c1501f1b0265\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"29caab90-d181-4913-9ffd-0508a2a81fda\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"29caab90-d181-4913-9ffd-0508a2a81fda\">QuantileTransformer</label><div class=\"sk-toggleable__content\"><pre>QuantileTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"a59db223-bad0-4693-8f9d-9ba05d5765bc\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"a59db223-bad0-4693-8f9d-9ba05d5765bc\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ac51c41e-9576-45cf-862f-2d018431a901\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ac51c41e-9576-45cf-862f-2d018431a901\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"120c0745-3851-4115-b30d-e4e3c874fc59\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"120c0745-3851-4115-b30d-e4e3c874fc59\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"677eed4c-89c5-4593-a416-0d325dca2e73\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"677eed4c-89c5-4593-a416-0d325dca2e73\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b831fef5-015f-41b6-897c-55776199af82\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b831fef5-015f-41b6-897c-55776199af82\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"4469cd31-98e8-42a1-8601-edfd0fe7cfec\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"4469cd31-98e8-42a1-8601-edfd0fe7cfec\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89D...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Features union cat + num \n",
    "# Catboost+QuantileTransformer\n",
    "preprocessor_catboost_qauntile_outlier_without_missing = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_quantile_outlier, num_columns_without_missing),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns_without_missing)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_qauntile_outlier_without_missing = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_qauntile_outlier_without_missing),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_qauntile_outlier_without_missing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "317a803f",
   "metadata": {
    "papermill": {
     "duration": 0.056545,
     "end_time": "2021-12-23T10:01:23.008966",
     "exception": false,
     "start_time": "2021-12-23T10:01:22.952421",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Pipe 4 Catoost + RobustScaler : \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "3ca8fcb3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:23.128889Z",
     "iopub.status.busy": "2021-12-23T10:01:23.128066Z",
     "iopub.status.idle": "2021-12-23T10:01:23.573783Z",
     "shell.execute_reply": "2021-12-23T10:01:23.574213Z"
    },
    "papermill": {
     "duration": 0.509643,
     "end_time": "2021-12-23T10:01:23.574357",
     "exception": false,
     "start_time": "2021-12-23T10:01:23.064714",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"1667fd92-ab77-4fbd-9f28-4bc9433f45a6\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"1667fd92-ab77-4fbd-9f28-4bc9433f45a6\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   RobustScaler()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPa...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"3b22d052-280f-47d4-a191-87f108742d03\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"3b22d052-280f-47d4-a191-87f108742d03\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', RobustScaler()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstate...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ff615ff5-e5f4-443f-9f9d-2bbb8dd4d17c\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ff615ff5-e5f4-443f-9f9d-2bbb8dd4d17c\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"114d3787-0e60-4d36-bebb-8c72936f9851\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"114d3787-0e60-4d36-bebb-8c72936f9851\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"5f18515a-32dd-4408-a8f3-a90ce7552a9c\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"5f18515a-32dd-4408-a8f3-a90ce7552a9c\">RobustScaler</label><div class=\"sk-toggleable__content\"><pre>RobustScaler()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"c7068b68-8727-437e-82c7-b3bb8a51f130\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"c7068b68-8727-437e-82c7-b3bb8a51f130\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"6cb6dcf9-8dc0-473b-a417-94507ec2e916\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"6cb6dcf9-8dc0-473b-a417-94507ec2e916\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f5325516-8ee2-46f2-97be-78847f4158d6\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f5325516-8ee2-46f2-97be-78847f4158d6\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"8e980dd1-aca6-4977-9884-7c2eaf1471ce\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"8e980dd1-aca6-4977-9884-7c2eaf1471ce\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"7e075e52-9fb5-407f-9337-2a68de1a782a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"7e075e52-9fb5-407f-9337-2a68de1a782a\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"5660b03c-44d6-4e80-aa9f-47886d3f3a2b\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"5660b03c-44d6-4e80-aa9f-47886d3f3a2b\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   RobustScaler()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPa...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "categorical_transformer_catboost = Pipeline(\n",
    "                    steps=[\n",
    "                        ('imputer', SimpleImputer(strategy='most_frequent',\n",
    "                                                  fill_value='missing',\n",
    "                                                  add_indicator=True)),\n",
    "                        ('encoder',  ce.cat_boost.CatBoostEncoder()),#(Numerical Input, Categorical Output)\n",
    "                         ('reducedim',  SelectPercentile( mutual_info_classif,\n",
    "                                                        percentile=90))\n",
    "                    ]\n",
    "                    ) \n",
    "\n",
    "\n",
    "numeric_transformer_robustscaler = Pipeline(\n",
    "                            steps=[\n",
    "                            ('imputer', SimpleImputer(strategy='median'\n",
    "                                                      ,add_indicator=True)),\n",
    "                            ('scaler', RobustScaler()),#(Numerical Input, Numerical Output)\n",
    "                            # Create an SelectKBest object to select features with two best ANOVA F-Values\n",
    "                            #The F-value scores examine if, when we group the numerical feature by the target vector, the means for each group are significantly different\n",
    "                            ('reducedim',  SelectPercentile(f_classif,\n",
    "                                                           percentile=90))\n",
    "                            ]\n",
    "                            )\n",
    "# Features union cat + num \n",
    "# Catboost+QuantileTransformer\n",
    "preprocessor_catboost_robust = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_robustscaler, num_columns),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_robust = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_robust),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_robust"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "684bbdbc",
   "metadata": {
    "papermill": {
     "duration": 0.055754,
     "end_time": "2021-12-23T10:01:23.687237",
     "exception": false,
     "start_time": "2021-12-23T10:01:23.631483",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Pipe6 :cat+power+fe \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "975b33ed",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:23.825465Z",
     "iopub.status.busy": "2021-12-23T10:01:23.824565Z",
     "iopub.status.idle": "2021-12-23T10:01:24.258451Z",
     "shell.execute_reply": "2021-12-23T10:01:24.258917Z"
    },
    "papermill": {
     "duration": 0.515365,
     "end_time": "2021-12-23T10:01:24.259071",
     "exception": false,
     "start_time": "2021-12-23T10:01:23.743706",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"d804a631-fbe4-44cf-ad9c-285b7d205fb8\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"d804a631-fbe4-44cf-ad9c-285b7d205fb8\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89Da...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b25dde58-ef06-4137-91d8-0cd30ac5042f\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b25dde58-ef06-4137-91d8-0cd30ac5042f\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEs...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"3e0abf94-f958-4b4f-8fb7-e5b1dd0deae4\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"3e0abf94-f958-4b4f-8fb7-e5b1dd0deae4\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age', 'Weighted_Sum_PastDue',\n",
       "       '90days_out_of_TotalPastDue', 'RemainingLines', 'Loans_vs_Other_Lines',\n",
       "       'Debt_per_Real_Estate_Loan', 'Disposable_Income_per_person',\n",
       "       'RemainingLines_per_person', 'NumberRE_X_DebtRatio_X_age',\n",
       "       'RevolvingUtilization_per_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f2b943ad-fd90-4cdc-b46d-5cafa4cd02c7\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f2b943ad-fd90-4cdc-b46d-5cafa4cd02c7\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"5c97d998-a5ed-40d8-bb56-4fcfa5a7a1ae\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"5c97d998-a5ed-40d8-bb56-4fcfa5a7a1ae\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"1949be6d-b3b5-4666-a7d1-00708656ced0\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"1949be6d-b3b5-4666-a7d1-00708656ced0\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f5ef56e3-929c-417b-a11e-e9a48391a991\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f5ef56e3-929c-417b-a11e-e9a48391a991\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"190a84fb-2bad-443f-b258-bb8a4cef4f6a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"190a84fb-2bad-443f-b258-bb8a4cef4f6a\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"c62237a1-c218-4113-93ae-308af6609ae9\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"c62237a1-c218-4113-93ae-308af6609ae9\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"548918da-8787-4170-840f-d92141ca898a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"548918da-8787-4170-840f-d92141ca898a\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"c3b2521e-ab26-40d9-97ff-dd89a36873db\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"c3b2521e-ab26-40d9-97ff-dd89a36873db\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89Da...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocessor_catboost_powertransformer_new_fe = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer, num_columns_new),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns_new)\n",
    "            ]\n",
    "            )\n",
    "\n",
    "pipe_lgbm_catboost_powertransformer_new_fe = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_powertransformer_new_fe),\n",
    "        ('classifier', lgbm.LGBMClassifier( n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_powertransformer_new_fe"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a0624b9d",
   "metadata": {
    "papermill": {
     "duration": 0.056991,
     "end_time": "2021-12-23T10:01:24.373577",
     "exception": false,
     "start_time": "2021-12-23T10:01:24.316586",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Handle Outlier :cat+power+fe "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "0a767c92",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:24.538222Z",
     "iopub.status.busy": "2021-12-23T10:01:24.517629Z",
     "iopub.status.idle": "2021-12-23T10:01:24.962575Z",
     "shell.execute_reply": "2021-12-23T10:01:24.962950Z"
    },
    "papermill": {
     "duration": 0.530862,
     "end_time": "2021-12-23T10:01:24.963103",
     "exception": false,
     "start_time": "2021-12-23T10:01:24.432241",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"72f4dba0-e6e3-4d2c-b435-9b7b1ffb8c2d\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"72f4dba0-e6e3-4d2c-b435-9b7b1ffb8c2d\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'Numbe...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"96094f75-4708-4ecb-bd72-1d58c6ab063d\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"96094f75-4708-4ecb-bd72-1d58c6ab063d\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('outlierhandle',\n",
       "                                                  OutlierReplace()),\n",
       "                                                 ('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-8...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"a6433d6f-fbea-4920-9a8e-7459d49e2888\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"a6433d6f-fbea-4920-9a8e-7459d49e2888\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age', 'Weighted_Sum_PastDue',\n",
       "       '90days_out_of_TotalPastDue', 'RemainingLines', 'Loans_vs_Other_Lines',\n",
       "       'Debt_per_Real_Estate_Loan', 'Disposable_Income_per_person',\n",
       "       'RemainingLines_per_person', 'NumberRE_X_DebtRatio_X_age',\n",
       "       'RevolvingUtilization_per_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"79cc91a2-1733-4c5a-8983-e3096594ee89\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"79cc91a2-1733-4c5a-8983-e3096594ee89\">OutlierReplace</label><div class=\"sk-toggleable__content\"><pre>OutlierReplace()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"25c0786e-89f7-4013-a4d2-e9761d6fb0d1\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"25c0786e-89f7-4013-a4d2-e9761d6fb0d1\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"c5ffd315-e8b5-4eb0-801a-ef291a0429e1\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"c5ffd315-e8b5-4eb0-801a-ef291a0429e1\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"3aea587e-d212-4c49-8a8b-94577512a4dc\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"3aea587e-d212-4c49-8a8b-94577512a4dc\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"8038ee85-ea03-4658-8758-21f61d55ccd2\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"8038ee85-ea03-4658-8758-21f61d55ccd2\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"e20f2b54-ed63-48c5-a1b3-692a923a3b26\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"e20f2b54-ed63-48c5-a1b3-692a923a3b26\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"46c73904-b087-4543-8837-0a522d1dfa37\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"46c73904-b087-4543-8837-0a522d1dfa37\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"5f8414f3-c103-48e9-87b5-5554b87257f7\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"5f8414f3-c103-48e9-87b5-5554b87257f7\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"a9f61878-e29c-481b-9481-471f3c151ee4\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"a9f61878-e29c-481b-9481-471f3c151ee4\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('outlierhandle',\n",
       "                                                                   OutlierReplace()),\n",
       "                                                                  ('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'Numbe...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocessor_catboost_powertransformer_new_fe_outlier = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_outlier, num_columns_new),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns_new)\n",
    "            ]\n",
    "            )\n",
    "\n",
    "pipe_lgbm_catboost_powertransformer_new_fe_outlier= Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_powertransformer_new_fe_outlier),\n",
    "        ('classifier', lgbm.LGBMClassifier( n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_powertransformer_new_fe_outlier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a89c1e3",
   "metadata": {
    "papermill": {
     "duration": 0.059782,
     "end_time": "2021-12-23T10:01:25.084088",
     "exception": false,
     "start_time": "2021-12-23T10:01:25.024306",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Drop missing:cat+power+fe "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "317ac3e5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:25.256922Z",
     "iopub.status.busy": "2021-12-23T10:01:25.236389Z",
     "iopub.status.idle": "2021-12-23T10:01:25.665023Z",
     "shell.execute_reply": "2021-12-23T10:01:25.665421Z"
    },
    "papermill": {
     "duration": 0.522169,
     "end_time": "2021-12-23T10:01:25.665586",
     "exception": false,
     "start_time": "2021-12-23T10:01:25.143417",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"c3d6faba-1ba9-4eed-9dbd-9d1360e49e92\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"c3d6faba-1ba9-4eed-9dbd-9d1360e49e92\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateL...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"d1c20049-7e55-4aad-aaac-63147fe5406e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"d1c20049-7e55-4aad-aaac-63147fe5406e\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler', PowerTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90Day...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f37809df-4e51-4a4c-a91e-99b64c006437\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f37809df-4e51-4a4c-a91e-99b64c006437\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90DaysLate', 'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age', 'Weighted_Sum_PastDue',\n",
       "       '90days_out_of_TotalPastDue', 'RemainingLines', 'Loans_vs_Other_Lines',\n",
       "       'Debt_per_Real_Estate_Loan', 'Disposable_Income_per_person',\n",
       "       'RemainingLines_per_person', 'NumberRE_X_DebtRatio_X_age',\n",
       "       'RevolvingUtilization_per_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"5dcce3d8-293d-4b89-9d63-e19274e50e0c\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"5dcce3d8-293d-4b89-9d63-e19274e50e0c\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"6e57bc1b-8246-4008-a170-fd79cd080408\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"6e57bc1b-8246-4008-a170-fd79cd080408\">PowerTransformer</label><div class=\"sk-toggleable__content\"><pre>PowerTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b413c5a0-e962-450a-aa59-f3d1b9855187\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b413c5a0-e962-450a-aa59-f3d1b9855187\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"cfd41ee3-4392-4322-b699-d9dcece5a143\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"cfd41ee3-4392-4322-b699-d9dcece5a143\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"7a3761d6-3840-423d-a7a4-54202b5342a3\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"7a3761d6-3840-423d-a7a4-54202b5342a3\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b6aad11c-5a8d-4ddc-92e3-87230e767549\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b6aad11c-5a8d-4ddc-92e3-87230e767549\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"6c133125-c9e7-4790-95a9-1dc6f7263c2b\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"6c133125-c9e7-4790-95a9-1dc6f7263c2b\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"dc9b5389-b6e1-4776-b69c-fed82b3e2a88\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"dc9b5389-b6e1-4776-b69c-fed82b3e2a88\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   PowerTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateL...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocessor_catboost_powertransformer_new_fe_without_missing= ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer, num_columns_new_without_missing),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns_new_without_missing)\n",
    "            ]\n",
    "            )\n",
    "pipe_lgbm_catboost_powertransformer_new_fe_without_missing = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_powertransformer_new_fe_without_missing),\n",
    "        ('classifier', lgbm.LGBMClassifier( n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_powertransformer_new_fe_without_missing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da559c92",
   "metadata": {
    "papermill": {
     "duration": 0.059343,
     "end_time": "2021-12-23T10:01:25.783965",
     "exception": false,
     "start_time": "2021-12-23T10:01:25.724622",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Pipe 7: cat +qauntile +new feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "e94d90cd",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:25.920455Z",
     "iopub.status.busy": "2021-12-23T10:01:25.910550Z",
     "iopub.status.idle": "2021-12-23T10:01:26.366708Z",
     "shell.execute_reply": "2021-12-23T10:01:26.366260Z"
    },
    "papermill": {
     "duration": 0.523818,
     "end_time": "2021-12-23T10:01:26.366835",
     "exception": false,
     "start_time": "2021-12-23T10:01:25.843017",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"90b269d2-32a3-4027-a6a7-bf18b015d875\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"90b269d2-32a3-4027-a6a7-bf18b015d875\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-8...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b3b0d76c-a1fe-4c1f-b616-a51509a20a3a\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b3b0d76c-a1fe-4c1f-b616-a51509a20a3a\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler',\n",
       "                                                  QuantileTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRea...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"64767200-23b5-4124-a9e3-9ca2c168f08b\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"64767200-23b5-4124-a9e3-9ca2c168f08b\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'NumberRealEstateLoansOrLines', 'NumberOfTimes90DaysLate',\n",
       "       'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age', 'Weighted_Sum_PastDue',\n",
       "       '90days_out_of_TotalPastDue', 'RemainingLines', 'Loans_vs_Other_Lines',\n",
       "       'Debt_per_Real_Estate_Loan', 'Disposable_Income_per_person',\n",
       "       'RemainingLines_per_person', 'NumberRE_X_DebtRatio_X_age',\n",
       "       'RevolvingUtilization_per_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"b6b92ddf-f823-45be-b739-5736972c1def\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"b6b92ddf-f823-45be-b739-5736972c1def\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"9f61a05e-9b23-4792-b273-0f98544a2dc0\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"9f61a05e-9b23-4792-b273-0f98544a2dc0\">QuantileTransformer</label><div class=\"sk-toggleable__content\"><pre>QuantileTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"e54097a2-8bb3-4823-ae31-205737965dcb\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"e54097a2-8bb3-4823-ae31-205737965dcb\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f0c8508f-e4fa-4828-9e34-c2b4dbcab2c0\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f0c8508f-e4fa-4828-9e34-c2b4dbcab2c0\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"3e2836ef-8f1a-4472-ac15-1de835d801a2\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"3e2836ef-8f1a-4472-ac15-1de835d801a2\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"085fe090-6e7c-48a0-80d8-4495122e91cc\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"085fe090-6e7c-48a0-80d8-4495122e91cc\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"aadfb54c-5872-41cd-8506-0917570cf785\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"aadfb54c-5872-41cd-8506-0917570cf785\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"a48502b8-b767-48b4-98cc-79592fae1ce9\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"a48502b8-b767-48b4-98cc-79592fae1ce9\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio', 'MonthlyIncome',\n",
       "       'NumberOfDependents', 'NumberOfTime60-8...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocessor_catboost_qauntile_new = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_quantile, num_columns_new),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns_new)\n",
    "            ]\n",
    "            )\n",
    "\n",
    "pipe_lgbm_catboost_qauntile_new = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_qauntile_new),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_qauntile_new"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f66249e8",
   "metadata": {
    "papermill": {
     "duration": 0.060003,
     "end_time": "2021-12-23T10:01:26.486896",
     "exception": false,
     "start_time": "2021-12-23T10:01:26.426893",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## drop missing:cat +qauntile +new feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "b237913b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:26.610892Z",
     "iopub.status.busy": "2021-12-23T10:01:26.610089Z",
     "iopub.status.idle": "2021-12-23T10:01:27.063676Z",
     "shell.execute_reply": "2021-12-23T10:01:27.063230Z"
    },
    "papermill": {
     "duration": 0.516595,
     "end_time": "2021-12-23T10:01:27.063805",
     "exception": false,
     "start_time": "2021-12-23T10:01:26.547210",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>div.sk-top-container {color: black;background-color: white;}div.sk-toggleable {background-color: white;}label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.2em 0.3em;box-sizing: border-box;text-align: center;}div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}div.sk-estimator {font-family: monospace;background-color: #f0f8ff;margin: 0.25em 0.25em;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;}div.sk-estimator:hover {background-color: #d4ebff;}div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 2em;bottom: 0;left: 50%;}div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;}div.sk-item {z-index: 1;}div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;}div.sk-parallel-item {display: flex;flex-direction: column;position: relative;background-color: white;}div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}div.sk-parallel-item:only-child::after {width: 0;}div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0.2em;box-sizing: border-box;padding-bottom: 0.1em;background-color: white;position: relative;}div.sk-label label {font-family: monospace;font-weight: bold;background-color: white;display: inline-block;line-height: 1.2em;}div.sk-label-container {position: relative;z-index: 2;text-align: center;}div.sk-container {display: inline-block;position: relative;}</style><div class=\"sk-top-container\"><div class=\"sk-container\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"ed615ea9-2d2c-4668-8529-f8b73c4f45c1\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"ed615ea9-2d2c-4668-8529-f8b73c4f45c1\">Pipeline</label><div class=\"sk-toggleable__content\"><pre>Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEsta...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"942e58f6-40f9-45db-b107-61c25cb7704e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"942e58f6-40f9-45db-b107-61c25cb7704e\">preprocessor: ColumnTransformer</label><div class=\"sk-toggleable__content\"><pre>ColumnTransformer(transformers=[('numerical',\n",
       "                                 Pipeline(steps=[('imputer',\n",
       "                                                  SimpleImputer(add_indicator=True,\n",
       "                                                                strategy='median')),\n",
       "                                                 ('scaler',\n",
       "                                                  QuantileTransformer()),\n",
       "                                                 ('reducedim',\n",
       "                                                  SelectPercentile(percentile=90))]),\n",
       "                                 Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90...\n",
       "                                                                   score_func=<function mutual_info_classif at 0x7fab4fedf950>))]),\n",
       "                                 Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"54d56166-510c-4923-8555-a484c60f44f1\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"54d56166-510c-4923-8555-a484c60f44f1\">numerical</label><div class=\"sk-toggleable__content\"><pre>Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEstateLoansOrLines',\n",
       "       'NumberOfTimes90DaysLate', 'NumberOfOpenCreditLinesAndLoans',\n",
       "       'NumberOfTime30-59DaysPastDueNotWorse', 'age', 'Weighted_Sum_PastDue',\n",
       "       '90days_out_of_TotalPastDue', 'RemainingLines', 'Loans_vs_Other_Lines',\n",
       "       'Debt_per_Real_Estate_Loan', 'Disposable_Income_per_person',\n",
       "       'RemainingLines_per_person', 'NumberRE_X_DebtRatio_X_age',\n",
       "       'RevolvingUtilization_per_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"3022f426-7f01-42f3-96a4-01b913db053e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"3022f426-7f01-42f3-96a4-01b913db053e\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, strategy='median')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"0c4acf7c-e5fc-4aca-a90b-fad78192a5ca\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"0c4acf7c-e5fc-4aca-a90b-fad78192a5ca\">QuantileTransformer</label><div class=\"sk-toggleable__content\"><pre>QuantileTransformer()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"9b0ced64-cdcf-4b4e-84dc-c06b29518359\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"9b0ced64-cdcf-4b4e-84dc-c06b29518359\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90)</pre></div></div></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"0413ca91-fd97-4341-8907-f0327dda0342\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"0413ca91-fd97-4341-8907-f0327dda0342\">categorical</label><div class=\"sk-toggleable__content\"><pre>Index(['cat_RevolvingUtilizationOfUnsecuredLines', 'cat_DebtRatio',\n",
       "       'cat_MonthlyIncome', 'cat_NumberOfDependents',\n",
       "       'cat_NumberOfTime60-89DaysPastDueNotWorse',\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object')</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"f123cfae-e779-44ca-9a03-a88d724901ad\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"f123cfae-e779-44ca-9a03-a88d724901ad\">SimpleImputer</label><div class=\"sk-toggleable__content\"><pre>SimpleImputer(add_indicator=True, fill_value='missing',\n",
       "              strategy='most_frequent')</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"cf94ee4e-e4af-4326-aaf8-207ccd41fca0\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"cf94ee4e-e4af-4326-aaf8-207ccd41fca0\">CatBoostEncoder</label><div class=\"sk-toggleable__content\"><pre>CatBoostEncoder()</pre></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"c688f713-dd2c-46d8-a7a7-7309e122402e\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"c688f713-dd2c-46d8-a7a7-7309e122402e\">SelectPercentile</label><div class=\"sk-toggleable__content\"><pre>SelectPercentile(percentile=90,\n",
       "                 score_func=<function mutual_info_classif at 0x7fab4fedf950>)</pre></div></div></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"c5dee0bc-9639-4cc8-967b-c1694bf171af\" type=\"checkbox\" ><label class=\"sk-toggleable__label\" for=\"c5dee0bc-9639-4cc8-967b-c1694bf171af\">LGBMClassifier</label><div class=\"sk-toggleable__content\"><pre>LGBMClassifier(device_type='gpu', learning_rate=0.0018069834369607075,\n",
       "               max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "               n_estimators=5000, random_state=42, subsample=0.8573598985000007,\n",
       "               verbose=-1)</pre></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "Pipeline(steps=[('preprocessor',\n",
       "                 ColumnTransformer(transformers=[('numerical',\n",
       "                                                  Pipeline(steps=[('imputer',\n",
       "                                                                   SimpleImputer(add_indicator=True,\n",
       "                                                                                 strategy='median')),\n",
       "                                                                  ('scaler',\n",
       "                                                                   QuantileTransformer()),\n",
       "                                                                  ('reducedim',\n",
       "                                                                   SelectPercentile(percentile=90))]),\n",
       "                                                  Index(['RevolvingUtilizationOfUnsecuredLines', 'DebtRatio',\n",
       "       'NumberOfTime60-89DaysPastDueNotWorse', 'NumberRealEsta...\n",
       "       'cat_NumberRealEstateLoansOrLines', 'cat_NumberOfTimes90DaysLate',\n",
       "       'cat_NumberOfOpenCreditLinesAndLoans',\n",
       "       'cat_NumberOfTime30-59DaysPastDueNotWorse', 'cat_age'],\n",
       "      dtype='object'))])),\n",
       "                ('classifier',\n",
       "                 LGBMClassifier(device_type='gpu',\n",
       "                                learning_rate=0.0018069834369607075,\n",
       "                                max_depth=8, metric='auc', min_samples_leaf=47,\n",
       "                                n_estimators=5000, random_state=42,\n",
       "                                subsample=0.8573598985000007, verbose=-1))])"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocessor_catboost_qauntile_new_without_missing = ColumnTransformer(\n",
    "            transformers=[\n",
    "                ('numerical', numeric_transformer_quantile, num_columns_new_without_missing),\n",
    "                ('categorical', categorical_transformer_catboost, cat_columns_new_without_missing)\n",
    "            ]\n",
    "            )\n",
    "\n",
    "pipe_lgbm_catboost_qauntile_new_without_missing = Pipeline(\n",
    "    steps=[\n",
    "        ('preprocessor', preprocessor_catboost_qauntile_new_without_missing),\n",
    "        ('classifier', lgbm.LGBMClassifier( \n",
    "            #n_jobs=-1,\n",
    "                                           verbose=-1,\n",
    "                                           **lgbm_param))\n",
    "    ]\n",
    ")\n",
    "pipe_lgbm_catboost_qauntile_new_without_missing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "3c00c539",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:27.189622Z",
     "iopub.status.busy": "2021-12-23T10:01:27.188838Z",
     "iopub.status.idle": "2021-12-23T10:01:27.195421Z",
     "shell.execute_reply": "2021-12-23T10:01:27.194907Z"
    },
    "papermill": {
     "duration": 0.070856,
     "end_time": "2021-12-23T10:01:27.195532",
     "exception": false,
     "start_time": "2021-12-23T10:01:27.124676",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def plot_roc(y,y_pred):\n",
    "    from sklearn.metrics import auc\n",
    "    fpr, tpr, thresholds = roc_curve(y_true=y, y_score=y_pred)\n",
    "    auc = auc(fpr, tpr)\n",
    "    plt.figure(1)\n",
    "    plt.plot([0, 1], [0, 1], 'k--')\n",
    "    plt.plot(fpr, tpr, label='Model(area = {:.3f})'.format(auc))\n",
    "    plt.xlabel('False positive rate')\n",
    "    plt.ylabel('True positive rate')\n",
    "    plt.title('ROC curve')\n",
    "    plt.legend(loc='best')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f569bf12",
   "metadata": {
    "papermill": {
     "duration": 0.064611,
     "end_time": "2021-12-23T10:01:27.322374",
     "exception": false,
     "start_time": "2021-12-23T10:01:27.257763",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Test Pipes with CV "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "070246c4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:27.452744Z",
     "iopub.status.busy": "2021-12-23T10:01:27.451839Z",
     "iopub.status.idle": "2021-12-23T10:01:27.453639Z",
     "shell.execute_reply": "2021-12-23T10:01:27.454071Z"
    },
    "papermill": {
     "duration": 0.069352,
     "end_time": "2021-12-23T10:01:27.454275",
     "exception": false,
     "start_time": "2021-12-23T10:01:27.384923",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def plot_roc(y,y_pred):\n",
    "    from sklearn.metrics import auc\n",
    "    fpr, tpr, thresholds = roc_curve(y_true=y, y_score=y_pred)\n",
    "    auc = auc(fpr, tpr)\n",
    "    plt.figure(1)\n",
    "    plt.plot([0, 1], [0, 1], 'k--')\n",
    "    plt.plot(fpr, tpr, label='Model(area = {:.3f})'.format(auc))\n",
    "    plt.xlabel('False positive rate')\n",
    "    plt.ylabel('True positive rate')\n",
    "    plt.title('ROC curve')\n",
    "    plt.legend(loc='best')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "14690b0e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:01:27.584769Z",
     "iopub.status.busy": "2021-12-23T10:01:27.584175Z",
     "iopub.status.idle": "2021-12-23T10:19:18.107934Z",
     "shell.execute_reply": "2021-12-23T10:19:18.108662Z"
    },
    "papermill": {
     "duration": 1070.593722,
     "end_time": "2021-12-23T10:19:18.108875",
     "exception": false,
     "start_time": "2021-12-23T10:01:27.515153",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 0 | AUC: 0.8669130495204391\n",
      "Fold 0 | AP: 0.407665684789223\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 1 | AUC: 0.8626155540665279\n",
      "Fold 1 | AP: 0.3849630973055488\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 2 | AUC: 0.8669757677029638\n",
      "Fold 2 | AP: 0.40533600002146797\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 3 | AUC: 0.8645943643670688\n",
      "Fold 3 | AP: 0.3961029196417055\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 4 | AUC: 0.8664786878202287\n",
      "Fold 4 | AP: 0.40800623246057943\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 5 | AUC: 0.8733771892866091\n",
      "Fold 5 | AP: 0.43988118488798555\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 6 | AUC: 0.8598904134865003\n",
      "Fold 6 | AP: 0.37859680901603604\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 7 | AUC: 0.8591809808016486\n",
      "Fold 7 | AP: 0.3984755991549263\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 8 | AUC: 0.874316126997327\n",
      "Fold 8 | AP: 0.4434923753324774\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set with min_child_samples=20, will be overridden by min_samples_leaf=47. Current value: min_data_in_leaf=47\n",
      "Fold 9 | AUC: 0.8643415768833814\n",
      "Fold 9 | AP: 0.38182803234569784\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Overall mean auc : 0.8658683710932694\n",
      "\n",
      "Overall mean mean average : 0.40443479349556477\n",
      "CPU times: user 31min 3s, sys: 28.3 s, total: 31min 31s\n",
      "Wall time: 17min 50s\n"
     ]
    }
   ],
   "source": [
    "%%time \n",
    "# evaluate each strategy on the dataset\n",
    "results = list()\n",
    "# Setting a 10-fold stratified cross-validation (note: shuffle=True)\n",
    "SEED = 42\n",
    "FOLDS = 10\n",
    "skf = StratifiedKFold(n_splits=FOLDS, shuffle=True, random_state=SEED)\n",
    "# CV interations\n",
    "# Create arrays for the features and the response variable\n",
    "roc_auc = list()\n",
    "average_precision = list()\n",
    "#X1=X.iloc[0:2000,:].copy()\n",
    "#y1=y[0:2000].copy()\n",
    "oof = np.empty((X.shape[0],))\n",
    "#oof_bin = np.empty((X.shape[0],))\n",
    "predictions=[]\n",
    "mean_auc = 0\n",
    "mean_ap=0\n",
    "best_iteration = list()\n",
    "for fold, (train_idx, test_idx) in enumerate(skf.split(X, y)):    \n",
    "    X_train, y_train = X_new.iloc[list(train_idx), :], y.iloc[list(train_idx)]\n",
    "    X_test, y_test = X_new.iloc[list(test_idx), :],y.iloc[list(test_idx)]\n",
    "    pipe_lgbm_catboost_qauntile.fit(X_train,y_train)\n",
    "    preds = pipe_lgbm_catboost_qauntile.predict_proba(X_test)[:,1]\n",
    "    oof[test_idx] = preds\n",
    "    auc_score= roc_auc_score(y_true=y_test, y_score=preds)\n",
    "    roc_auc.append(auc_score)\n",
    "    print(f\"Fold {fold} | AUC: {auc_score}\")\n",
    "    ap=average_precision_score(y_true=y_test, y_score=preds)\n",
    "    average_precision.append(ap)\n",
    "    print(f\"Fold {fold} | AP: {ap}\")\n",
    "    plot_roc(y_test,preds)\n",
    "    mean_auc += auc_score / FOLDS\n",
    "    mean_ap+=ap/FOLDS\n",
    "    predictions.append(pipe_lgbm_catboost_qauntile.predict_proba(X_test_final)[:,1]) \n",
    "    #Mean of the predictions\n",
    "print(f\"\\nOverall mean auc : {mean_auc}\")\n",
    "print(f\"\\nOverall mean mean average : {mean_ap}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "1025fdc5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:19:18.264661Z",
     "iopub.status.busy": "2021-12-23T10:19:18.263820Z",
     "iopub.status.idle": "2021-12-23T10:19:18.266116Z",
     "shell.execute_reply": "2021-12-23T10:19:18.265614Z"
    },
    "papermill": {
     "duration": 0.08064,
     "end_time": "2021-12-23T10:19:18.266227",
     "exception": false,
     "start_time": "2021-12-23T10:19:18.185587",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "#pipe_lgbm_woe_powertransformer.fit(X,y)\n",
    "#predictions =pipe_lgbm_woe_powertransformer.predict_proba(X_test_final)[:,1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "6998891f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:19:18.429873Z",
     "iopub.status.busy": "2021-12-23T10:19:18.428976Z",
     "iopub.status.idle": "2021-12-23T10:19:18.436588Z",
     "shell.execute_reply": "2021-12-23T10:19:18.437281Z"
    },
    "papermill": {
     "duration": 0.096709,
     "end_time": "2021-12-23T10:19:18.437507",
     "exception": false,
     "start_time": "2021-12-23T10:19:18.340798",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "predictions = np.mean(np.column_stack(predictions), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "8a896c31",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-23T10:19:18.716395Z",
     "iopub.status.busy": "2021-12-23T10:19:18.715583Z",
     "iopub.status.idle": "2021-12-23T10:19:19.230152Z",
     "shell.execute_reply": "2021-12-23T10:19:19.230666Z"
    },
    "papermill": {
     "duration": 0.6526,
     "end_time": "2021-12-23T10:19:19.230892",
     "exception": false,
     "start_time": "2021-12-23T10:19:18.578292",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Id</th>\n",
       "      <th>Probability</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0.066887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>0.047580</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>0.012182</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>0.075306</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>0.109521</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101498</th>\n",
       "      <td>101499</td>\n",
       "      <td>0.034113</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101499</th>\n",
       "      <td>101500</td>\n",
       "      <td>0.419753</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101500</th>\n",
       "      <td>101501</td>\n",
       "      <td>0.005690</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101501</th>\n",
       "      <td>101502</td>\n",
       "      <td>0.062747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101502</th>\n",
       "      <td>101503</td>\n",
       "      <td>0.035845</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>101503 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            Id  Probability\n",
       "0            1     0.066887\n",
       "1            2     0.047580\n",
       "2            3     0.012182\n",
       "3            4     0.075306\n",
       "4            5     0.109521\n",
       "...        ...          ...\n",
       "101498  101499     0.034113\n",
       "101499  101500     0.419753\n",
       "101500  101501     0.005690\n",
       "101501  101502     0.062747\n",
       "101502  101503     0.035845\n",
       "\n",
       "[101503 rows x 2 columns]"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Save the predictions to a CSV file\n",
    "sub = pd.read_csv('../input/GiveMeSomeCredit/sampleEntry.csv')\n",
    "sub['Probability']=predictions\n",
    "sub.to_csv('pipe_lgbm_catboost_qauntile_estimator_XGB_CV.csv', index=False)\n",
    "sub"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.12"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 4456.802832,
   "end_time": "2021-12-23T10:19:20.739972",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2021-12-23T09:05:03.937140",
   "version": "2.3.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}