TensorFlow German Credit Data on GPU.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Credit Scoring Model on GPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TensorFlow version: 2.11.0\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' \n",
    "\n",
    "import pprint as pp\n",
    "import math\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.metrics import roc_auc_score, f1_score, confusion_matrix, classification_report\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from tensorflow.keras import Sequential\n",
    "from tensorflow.keras.layers import DenseFeatures, Dense, Activation, LayerNormalization, Flatten\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.python.client import device_lib\n",
    "from IPython.display import display\n",
    "import matplotlib.pyplot as plt\n",
    "sns.set(style='darkgrid')\n",
    "print('TensorFlow version: ' + tf.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Num GPUs Available:  1\n"
     ]
    }
   ],
   "source": [
    "print(\"Num GPUs Available: \", len(tf.config.list_physical_devices('GPU')))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Default GPU Device: /device:GPU:0\n"
     ]
    }
   ],
   "source": [
    "if tf.test.gpu_device_name():\n",
    "    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))\n",
    "else:\n",
    "    print(\"Please install GPU version of TF\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[name: \"/device:CPU:0\"\n",
      "device_type: \"CPU\"\n",
      "memory_limit: 268435456\n",
      "locality {\n",
      "}\n",
      "incarnation: 4867476263508553294\n",
      "xla_global_id: -1\n",
      ", name: \"/device:GPU:0\"\n",
      "device_type: \"GPU\"\n",
      "memory_limit: 1673999156\n",
      "locality {\n",
      "  bus_id: 1\n",
      "  links {\n",
      "  }\n",
      "}\n",
      "incarnation: 8529602303512699172\n",
      "physical_device_desc: \"device: 0, name: NVIDIA GeForce RTX 3050 Laptop GPU, pci bus id: 0000:01:00.0, compute capability: 8.6\"\n",
      "xla_global_id: 416903419\n",
      "]\n"
     ]
    }
   ],
   "source": [
    "print(device_lib.list_local_devices())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "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>CustomerID</th>\n",
       "      <th>CheckingStatus</th>\n",
       "      <th>LoanDuration</th>\n",
       "      <th>CreditHistory</th>\n",
       "      <th>LoanPurpose</th>\n",
       "      <th>LoanAmount</th>\n",
       "      <th>ExistingSavings</th>\n",
       "      <th>EmploymentDuration</th>\n",
       "      <th>InstallmentPercent</th>\n",
       "      <th>Sex</th>\n",
       "      <th>...</th>\n",
       "      <th>OwnsProperty</th>\n",
       "      <th>Age</th>\n",
       "      <th>InstallmentPlans</th>\n",
       "      <th>Housing</th>\n",
       "      <th>ExistingCreditsCount</th>\n",
       "      <th>Job</th>\n",
       "      <th>Dependents</th>\n",
       "      <th>Telephone</th>\n",
       "      <th>ForeignWorker</th>\n",
       "      <th>Risk</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>713a336c-a255-4e2d-9d57-90b3e99e2f06</td>\n",
       "      <td>0_to_200</td>\n",
       "      <td>31</td>\n",
       "      <td>credits_paid_to_date</td>\n",
       "      <td>other</td>\n",
       "      <td>1889</td>\n",
       "      <td>100_to_500</td>\n",
       "      <td>less_1</td>\n",
       "      <td>3</td>\n",
       "      <td>female</td>\n",
       "      <td>...</td>\n",
       "      <td>savings_insurance</td>\n",
       "      <td>32</td>\n",
       "      <td>none</td>\n",
       "      <td>own</td>\n",
       "      <td>1</td>\n",
       "      <td>skilled</td>\n",
       "      <td>1</td>\n",
       "      <td>none</td>\n",
       "      <td>yes</td>\n",
       "      <td>No Risk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>140b363f-a3fe-4828-a33f-7284dfdb3969</td>\n",
       "      <td>less_0</td>\n",
       "      <td>18</td>\n",
       "      <td>credits_paid_to_date</td>\n",
       "      <td>car_new</td>\n",
       "      <td>462</td>\n",
       "      <td>less_100</td>\n",
       "      <td>1_to_4</td>\n",
       "      <td>2</td>\n",
       "      <td>female</td>\n",
       "      <td>...</td>\n",
       "      <td>savings_insurance</td>\n",
       "      <td>37</td>\n",
       "      <td>stores</td>\n",
       "      <td>own</td>\n",
       "      <td>2</td>\n",
       "      <td>skilled</td>\n",
       "      <td>1</td>\n",
       "      <td>none</td>\n",
       "      <td>yes</td>\n",
       "      <td>No Risk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>43b7b51d-5eda-4860-b461-ebef3d3436f4</td>\n",
       "      <td>less_0</td>\n",
       "      <td>15</td>\n",
       "      <td>prior_payments_delayed</td>\n",
       "      <td>furniture</td>\n",
       "      <td>250</td>\n",
       "      <td>less_100</td>\n",
       "      <td>1_to_4</td>\n",
       "      <td>2</td>\n",
       "      <td>male</td>\n",
       "      <td>...</td>\n",
       "      <td>real_estate</td>\n",
       "      <td>28</td>\n",
       "      <td>none</td>\n",
       "      <td>own</td>\n",
       "      <td>2</td>\n",
       "      <td>skilled</td>\n",
       "      <td>1</td>\n",
       "      <td>yes</td>\n",
       "      <td>no</td>\n",
       "      <td>No Risk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>f40eaf08-e6d1-4765-ab20-c5f7faca1635</td>\n",
       "      <td>0_to_200</td>\n",
       "      <td>28</td>\n",
       "      <td>credits_paid_to_date</td>\n",
       "      <td>retraining</td>\n",
       "      <td>3693</td>\n",
       "      <td>less_100</td>\n",
       "      <td>greater_7</td>\n",
       "      <td>3</td>\n",
       "      <td>male</td>\n",
       "      <td>...</td>\n",
       "      <td>savings_insurance</td>\n",
       "      <td>32</td>\n",
       "      <td>none</td>\n",
       "      <td>own</td>\n",
       "      <td>1</td>\n",
       "      <td>skilled</td>\n",
       "      <td>1</td>\n",
       "      <td>none</td>\n",
       "      <td>yes</td>\n",
       "      <td>No Risk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1728910a-d3ff-4799-ac50-203a3a58a3fb</td>\n",
       "      <td>no_checking</td>\n",
       "      <td>28</td>\n",
       "      <td>prior_payments_delayed</td>\n",
       "      <td>education</td>\n",
       "      <td>6235</td>\n",
       "      <td>500_to_1000</td>\n",
       "      <td>greater_7</td>\n",
       "      <td>3</td>\n",
       "      <td>male</td>\n",
       "      <td>...</td>\n",
       "      <td>unknown</td>\n",
       "      <td>57</td>\n",
       "      <td>none</td>\n",
       "      <td>own</td>\n",
       "      <td>2</td>\n",
       "      <td>skilled</td>\n",
       "      <td>1</td>\n",
       "      <td>none</td>\n",
       "      <td>yes</td>\n",
       "      <td>Risk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4995</th>\n",
       "      <td>e77fa77b-78da-4607-a2fa-ede36c1e968f</td>\n",
       "      <td>greater_200</td>\n",
       "      <td>27</td>\n",
       "      <td>credits_paid_to_date</td>\n",
       "      <td>furniture</td>\n",
       "      <td>4650</td>\n",
       "      <td>less_100</td>\n",
       "      <td>1_to_4</td>\n",
       "      <td>3</td>\n",
       "      <td>male</td>\n",
       "      <td>...</td>\n",
       "      <td>savings_insurance</td>\n",
       "      <td>40</td>\n",
       "      <td>none</td>\n",
       "      <td>own</td>\n",
       "      <td>1</td>\n",
       "      <td>skilled</td>\n",
       "      <td>1</td>\n",
       "      <td>none</td>\n",
       "      <td>yes</td>\n",
       "      <td>No Risk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4996</th>\n",
       "      <td>6e71db4b-375c-42e1-b4a8-3292c007967a</td>\n",
       "      <td>0_to_200</td>\n",
       "      <td>11</td>\n",
       "      <td>prior_payments_delayed</td>\n",
       "      <td>furniture</td>\n",
       "      <td>250</td>\n",
       "      <td>greater_1000</td>\n",
       "      <td>4_to_7</td>\n",
       "      <td>3</td>\n",
       "      <td>male</td>\n",
       "      <td>...</td>\n",
       "      <td>car_other</td>\n",
       "      <td>32</td>\n",
       "      <td>bank</td>\n",
       "      <td>own</td>\n",
       "      <td>1</td>\n",
       "      <td>unemployed</td>\n",
       "      <td>1</td>\n",
       "      <td>none</td>\n",
       "      <td>yes</td>\n",
       "      <td>No Risk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4997</th>\n",
       "      <td>db501d22-e0c9-4f38-bf05-7f2c1df35395</td>\n",
       "      <td>no_checking</td>\n",
       "      <td>32</td>\n",
       "      <td>outstanding_credit</td>\n",
       "      <td>appliances</td>\n",
       "      <td>6536</td>\n",
       "      <td>unknown</td>\n",
       "      <td>greater_7</td>\n",
       "      <td>5</td>\n",
       "      <td>male</td>\n",
       "      <td>...</td>\n",
       "      <td>unknown</td>\n",
       "      <td>54</td>\n",
       "      <td>stores</td>\n",
       "      <td>own</td>\n",
       "      <td>2</td>\n",
       "      <td>unskilled</td>\n",
       "      <td>2</td>\n",
       "      <td>yes</td>\n",
       "      <td>yes</td>\n",
       "      <td>Risk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4998</th>\n",
       "      <td>802055d6-6aa7-4532-bf1c-0b8b114b483d</td>\n",
       "      <td>0_to_200</td>\n",
       "      <td>38</td>\n",
       "      <td>outstanding_credit</td>\n",
       "      <td>other</td>\n",
       "      <td>1597</td>\n",
       "      <td>500_to_1000</td>\n",
       "      <td>greater_7</td>\n",
       "      <td>3</td>\n",
       "      <td>female</td>\n",
       "      <td>...</td>\n",
       "      <td>savings_insurance</td>\n",
       "      <td>27</td>\n",
       "      <td>stores</td>\n",
       "      <td>own</td>\n",
       "      <td>1</td>\n",
       "      <td>skilled</td>\n",
       "      <td>1</td>\n",
       "      <td>none</td>\n",
       "      <td>yes</td>\n",
       "      <td>Risk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4999</th>\n",
       "      <td>53094239-82f4-4b14-b2e9-7a0355a10839</td>\n",
       "      <td>less_0</td>\n",
       "      <td>12</td>\n",
       "      <td>all_credits_paid_back</td>\n",
       "      <td>car_new</td>\n",
       "      <td>4152</td>\n",
       "      <td>less_100</td>\n",
       "      <td>1_to_4</td>\n",
       "      <td>2</td>\n",
       "      <td>male</td>\n",
       "      <td>...</td>\n",
       "      <td>savings_insurance</td>\n",
       "      <td>29</td>\n",
       "      <td>none</td>\n",
       "      <td>own</td>\n",
       "      <td>2</td>\n",
       "      <td>skilled</td>\n",
       "      <td>1</td>\n",
       "      <td>none</td>\n",
       "      <td>yes</td>\n",
       "      <td>No Risk</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5000 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                CustomerID CheckingStatus  LoanDuration  \\\n",
       "0     713a336c-a255-4e2d-9d57-90b3e99e2f06       0_to_200            31   \n",
       "1     140b363f-a3fe-4828-a33f-7284dfdb3969         less_0            18   \n",
       "2     43b7b51d-5eda-4860-b461-ebef3d3436f4         less_0            15   \n",
       "3     f40eaf08-e6d1-4765-ab20-c5f7faca1635       0_to_200            28   \n",
       "4     1728910a-d3ff-4799-ac50-203a3a58a3fb    no_checking            28   \n",
       "...                                    ...            ...           ...   \n",
       "4995  e77fa77b-78da-4607-a2fa-ede36c1e968f    greater_200            27   \n",
       "4996  6e71db4b-375c-42e1-b4a8-3292c007967a       0_to_200            11   \n",
       "4997  db501d22-e0c9-4f38-bf05-7f2c1df35395    no_checking            32   \n",
       "4998  802055d6-6aa7-4532-bf1c-0b8b114b483d       0_to_200            38   \n",
       "4999  53094239-82f4-4b14-b2e9-7a0355a10839         less_0            12   \n",
       "\n",
       "               CreditHistory LoanPurpose  LoanAmount ExistingSavings  \\\n",
       "0       credits_paid_to_date       other        1889      100_to_500   \n",
       "1       credits_paid_to_date     car_new         462        less_100   \n",
       "2     prior_payments_delayed   furniture         250        less_100   \n",
       "3       credits_paid_to_date  retraining        3693        less_100   \n",
       "4     prior_payments_delayed   education        6235     500_to_1000   \n",
       "...                      ...         ...         ...             ...   \n",
       "4995    credits_paid_to_date   furniture        4650        less_100   \n",
       "4996  prior_payments_delayed   furniture         250    greater_1000   \n",
       "4997      outstanding_credit  appliances        6536         unknown   \n",
       "4998      outstanding_credit       other        1597     500_to_1000   \n",
       "4999   all_credits_paid_back     car_new        4152        less_100   \n",
       "\n",
       "     EmploymentDuration  InstallmentPercent     Sex  ...       OwnsProperty  \\\n",
       "0                less_1                   3  female  ...  savings_insurance   \n",
       "1                1_to_4                   2  female  ...  savings_insurance   \n",
       "2                1_to_4                   2    male  ...        real_estate   \n",
       "3             greater_7                   3    male  ...  savings_insurance   \n",
       "4             greater_7                   3    male  ...            unknown   \n",
       "...                 ...                 ...     ...  ...                ...   \n",
       "4995             1_to_4                   3    male  ...  savings_insurance   \n",
       "4996             4_to_7                   3    male  ...          car_other   \n",
       "4997          greater_7                   5    male  ...            unknown   \n",
       "4998          greater_7                   3  female  ...  savings_insurance   \n",
       "4999             1_to_4                   2    male  ...  savings_insurance   \n",
       "\n",
       "      Age InstallmentPlans  Housing ExistingCreditsCount         Job  \\\n",
       "0      32             none      own                    1     skilled   \n",
       "1      37           stores      own                    2     skilled   \n",
       "2      28             none      own                    2     skilled   \n",
       "3      32             none      own                    1     skilled   \n",
       "4      57             none      own                    2     skilled   \n",
       "...   ...              ...      ...                  ...         ...   \n",
       "4995   40             none      own                    1     skilled   \n",
       "4996   32             bank      own                    1  unemployed   \n",
       "4997   54           stores      own                    2   unskilled   \n",
       "4998   27           stores      own                    1     skilled   \n",
       "4999   29             none      own                    2     skilled   \n",
       "\n",
       "      Dependents Telephone  ForeignWorker     Risk  \n",
       "0              1      none            yes  No Risk  \n",
       "1              1      none            yes  No Risk  \n",
       "2              1       yes             no  No Risk  \n",
       "3              1      none            yes  No Risk  \n",
       "4              1      none            yes     Risk  \n",
       "...          ...       ...            ...      ...  \n",
       "4995           1      none            yes  No Risk  \n",
       "4996           1      none            yes  No Risk  \n",
       "4997           2       yes            yes     Risk  \n",
       "4998           1      none            yes     Risk  \n",
       "4999           1      none            yes  No Risk  \n",
       "\n",
       "[5000 rows x 22 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# credit_data = pd.read_csv('german_credit_data.csv')\n",
    "# credit_data.to_hdf('credit_data.h5', key='df', mode='w')\n",
    "credit_data = pd.read_hdf('credit_data.h5', 'df')  \n",
    "display(credit_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = credit_data.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "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>No Risk</th>\n",
       "      <th>Risk</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Risk</th>\n",
       "      <td>3330</td>\n",
       "      <td>1670</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      No Risk  Risk\n",
       "Risk     3330  1670"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Name of the label column.\n",
    "label = 'Risk'\n",
    "id_column = 'CustomerID'\n",
    "data[label].value_counts().to_frame().T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 5000 entries, 0 to 4999\n",
      "Data columns (total 22 columns):\n",
      " #   Column                    Non-Null Count  Dtype \n",
      "---  ------                    --------------  ----- \n",
      " 0   CustomerID                5000 non-null   object\n",
      " 1   CheckingStatus            5000 non-null   object\n",
      " 2   LoanDuration              5000 non-null   int64 \n",
      " 3   CreditHistory             5000 non-null   object\n",
      " 4   LoanPurpose               5000 non-null   object\n",
      " 5   LoanAmount                5000 non-null   int64 \n",
      " 6   ExistingSavings           5000 non-null   object\n",
      " 7   EmploymentDuration        5000 non-null   object\n",
      " 8   InstallmentPercent        5000 non-null   int64 \n",
      " 9   Sex                       5000 non-null   object\n",
      " 10  OthersOnLoan              5000 non-null   object\n",
      " 11  CurrentResidenceDuration  5000 non-null   int64 \n",
      " 12  OwnsProperty              5000 non-null   object\n",
      " 13  Age                       5000 non-null   int64 \n",
      " 14  InstallmentPlans          5000 non-null   object\n",
      " 15  Housing                   5000 non-null   object\n",
      " 16  ExistingCreditsCount      5000 non-null   int64 \n",
      " 17  Job                       5000 non-null   object\n",
      " 18  Dependents                5000 non-null   int64 \n",
      " 19  Telephone                 5000 non-null   object\n",
      " 20  ForeignWorker             5000 non-null   object\n",
      " 21  Risk                      5000 non-null   object\n",
      "dtypes: int64(7), object(15)\n",
      "memory usage: 859.5+ KB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "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>No Risk</th>\n",
       "      <th>Risk</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Risk</th>\n",
       "      <td>3330</td>\n",
       "      <td>1670</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      No Risk  Risk\n",
       "Risk     3330  1670"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Name of the label column.\n",
    "label = 'Risk'\n",
    "id_column = 'CustomerID'\n",
    "data[label].value_counts().to_frame().T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "label_new = label + '_target'\n",
    "data[label_new] = data[label].apply(lambda x: 1 if x == 'Risk' else 0)\n",
    "data.drop(label, axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train data shape: (3333, 22)\n",
      "Test data shape: (1667, 22)\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>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Risk_target</th>\n",
       "      <td>2220</td>\n",
       "      <td>1113</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                0     1\n",
       "Risk_target  2220  1113"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train = 2/3\n",
    "test = 1 - train\n",
    "label_lst = [label_new]\n",
    "train_data, test_data = train_test_split(data, test_size=test, random_state=0, stratify=data[label_lst])\n",
    "print(f\"Train data shape: {train_data.shape}\")\n",
    "print(f\"Test data shape: {test_data.shape}\")\n",
    "train_data[label_new].value_counts().to_frame().T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training dataset\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>Risk_target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>sum</th>\n",
       "      <td>1113</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>3333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Risk_target\n",
       "sum           1113\n",
       "count         3333"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test dataset\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>Risk_target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>sum</th>\n",
       "      <td>557</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1667</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Risk_target\n",
       "sum            557\n",
       "count         1667"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Training dataset\")\n",
    "display(train_data.agg({label_new : ['sum', 'count']}))\n",
    "print(\"Test dataset\")\n",
    "display(test_data.agg({label_new : ['sum', 'count']}))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def nominal_to_ratio(df, column, target):\n",
    "    agr = df.groupby(column, as_index=False, dropna=False)[target].aggregate(['sum','count'])\n",
    "    agr['ratio'] = agr['sum'] / agr['count']\n",
    "    agr.reset_index(inplace=True)\n",
    "    agr = agr[[column, 'ratio']]\n",
    "    return dict(agr.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['Ratio_CheckingStatus', 'Ratio_CreditHistory', 'Ratio_LoanPurpose', 'Ratio_ExistingSavings', 'Ratio_EmploymentDuration', 'Ratio_Sex', 'Ratio_OthersOnLoan', 'Ratio_OwnsProperty', 'Ratio_InstallmentPlans', 'Ratio_Housing', 'Ratio_Job', 'Ratio_Telephone', 'Ratio_ForeignWorker']\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>CustomerID</th>\n",
       "      <th>LoanDuration</th>\n",
       "      <th>LoanAmount</th>\n",
       "      <th>InstallmentPercent</th>\n",
       "      <th>CurrentResidenceDuration</th>\n",
       "      <th>Age</th>\n",
       "      <th>ExistingCreditsCount</th>\n",
       "      <th>Dependents</th>\n",
       "      <th>Risk_target</th>\n",
       "      <th>Ratio_CheckingStatus</th>\n",
       "      <th>...</th>\n",
       "      <th>Ratio_ExistingSavings</th>\n",
       "      <th>Ratio_EmploymentDuration</th>\n",
       "      <th>Ratio_Sex</th>\n",
       "      <th>Ratio_OthersOnLoan</th>\n",
       "      <th>Ratio_OwnsProperty</th>\n",
       "      <th>Ratio_InstallmentPlans</th>\n",
       "      <th>Ratio_Housing</th>\n",
       "      <th>Ratio_Job</th>\n",
       "      <th>Ratio_Telephone</th>\n",
       "      <th>Ratio_ForeignWorker</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3985</th>\n",
       "      <td>1569bb6b-f83f-45fb-9284-864289d8b5db</td>\n",
       "      <td>20</td>\n",
       "      <td>3822</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>35</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.095488</td>\n",
       "      <td>...</td>\n",
       "      <td>0.143533</td>\n",
       "      <td>0.730706</td>\n",
       "      <td>0.356211</td>\n",
       "      <td>0.267500</td>\n",
       "      <td>0.439681</td>\n",
       "      <td>0.174917</td>\n",
       "      <td>0.324096</td>\n",
       "      <td>0.257403</td>\n",
       "      <td>0.533036</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2467</th>\n",
       "      <td>46420c4e-fe41-4517-a6a6-33b2104b25c8</td>\n",
       "      <td>36</td>\n",
       "      <td>6040</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>55</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.569337</td>\n",
       "      <td>...</td>\n",
       "      <td>0.518919</td>\n",
       "      <td>0.730706</td>\n",
       "      <td>0.356211</td>\n",
       "      <td>0.683084</td>\n",
       "      <td>0.705263</td>\n",
       "      <td>0.223373</td>\n",
       "      <td>0.324096</td>\n",
       "      <td>0.257403</td>\n",
       "      <td>0.533036</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3290</th>\n",
       "      <td>ae2546dd-5ce1-4198-86d2-c971f88f9391</td>\n",
       "      <td>36</td>\n",
       "      <td>6203</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>43</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.569337</td>\n",
       "      <td>...</td>\n",
       "      <td>0.625000</td>\n",
       "      <td>0.730706</td>\n",
       "      <td>0.356211</td>\n",
       "      <td>0.683084</td>\n",
       "      <td>0.439681</td>\n",
       "      <td>0.223373</td>\n",
       "      <td>0.324096</td>\n",
       "      <td>0.257403</td>\n",
       "      <td>0.533036</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2523</th>\n",
       "      <td>78c7bf4e-f4bd-490c-abda-88c0b77e730b</td>\n",
       "      <td>8</td>\n",
       "      <td>1312</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>24</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.095488</td>\n",
       "      <td>...</td>\n",
       "      <td>0.143533</td>\n",
       "      <td>0.237374</td>\n",
       "      <td>0.297468</td>\n",
       "      <td>0.267500</td>\n",
       "      <td>0.265471</td>\n",
       "      <td>0.174917</td>\n",
       "      <td>0.118949</td>\n",
       "      <td>0.210811</td>\n",
       "      <td>0.198892</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1084</th>\n",
       "      <td>bd697901-dbe2-40a1-aa82-89db5c8c1005</td>\n",
       "      <td>8</td>\n",
       "      <td>1638</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>38</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.569337</td>\n",
       "      <td>...</td>\n",
       "      <td>0.338501</td>\n",
       "      <td>0.080906</td>\n",
       "      <td>0.356211</td>\n",
       "      <td>0.267500</td>\n",
       "      <td>0.055405</td>\n",
       "      <td>0.386151</td>\n",
       "      <td>0.118949</td>\n",
       "      <td>0.392941</td>\n",
       "      <td>0.533036</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3882</th>\n",
       "      <td>3a07fe8e-d341-4112-a09a-827e1073b704</td>\n",
       "      <td>19</td>\n",
       "      <td>4349</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>23</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0.224771</td>\n",
       "      <td>...</td>\n",
       "      <td>0.483501</td>\n",
       "      <td>0.080906</td>\n",
       "      <td>0.356211</td>\n",
       "      <td>0.267500</td>\n",
       "      <td>0.439681</td>\n",
       "      <td>0.174917</td>\n",
       "      <td>0.324096</td>\n",
       "      <td>0.257403</td>\n",
       "      <td>0.533036</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2474</th>\n",
       "      <td>77d08099-3a56-4140-996e-cd0a18a41155</td>\n",
       "      <td>12</td>\n",
       "      <td>3223</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>35</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.569337</td>\n",
       "      <td>...</td>\n",
       "      <td>0.625000</td>\n",
       "      <td>0.237374</td>\n",
       "      <td>0.356211</td>\n",
       "      <td>0.267500</td>\n",
       "      <td>0.439681</td>\n",
       "      <td>0.386151</td>\n",
       "      <td>0.118949</td>\n",
       "      <td>0.392941</td>\n",
       "      <td>0.198892</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3191</th>\n",
       "      <td>8a9dc768-9b0c-4f0e-9ac7-72393810f754</td>\n",
       "      <td>33</td>\n",
       "      <td>6712</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>43</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.414286</td>\n",
       "      <td>...</td>\n",
       "      <td>0.518919</td>\n",
       "      <td>0.413187</td>\n",
       "      <td>0.297468</td>\n",
       "      <td>0.267500</td>\n",
       "      <td>0.705263</td>\n",
       "      <td>0.386151</td>\n",
       "      <td>0.324096</td>\n",
       "      <td>0.347636</td>\n",
       "      <td>0.533036</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3664</th>\n",
       "      <td>d83f91fe-d8fb-43d6-affe-c2a00e0836e5</td>\n",
       "      <td>33</td>\n",
       "      <td>4539</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>51</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.569337</td>\n",
       "      <td>...</td>\n",
       "      <td>0.518919</td>\n",
       "      <td>0.730706</td>\n",
       "      <td>0.356211</td>\n",
       "      <td>0.683084</td>\n",
       "      <td>0.265471</td>\n",
       "      <td>0.223373</td>\n",
       "      <td>0.324096</td>\n",
       "      <td>0.257403</td>\n",
       "      <td>0.533036</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3668</th>\n",
       "      <td>ad350fa7-ca99-4d7f-865a-0ec62525e6b1</td>\n",
       "      <td>20</td>\n",
       "      <td>4183</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>21</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0.224771</td>\n",
       "      <td>...</td>\n",
       "      <td>0.483501</td>\n",
       "      <td>0.237374</td>\n",
       "      <td>0.356211</td>\n",
       "      <td>0.267500</td>\n",
       "      <td>0.439681</td>\n",
       "      <td>0.386151</td>\n",
       "      <td>0.324096</td>\n",
       "      <td>0.347636</td>\n",
       "      <td>0.533036</td>\n",
       "      <td>0.328427</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3333 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                CustomerID  LoanDuration  LoanAmount  \\\n",
       "3985  1569bb6b-f83f-45fb-9284-864289d8b5db            20        3822   \n",
       "2467  46420c4e-fe41-4517-a6a6-33b2104b25c8            36        6040   \n",
       "3290  ae2546dd-5ce1-4198-86d2-c971f88f9391            36        6203   \n",
       "2523  78c7bf4e-f4bd-490c-abda-88c0b77e730b             8        1312   \n",
       "1084  bd697901-dbe2-40a1-aa82-89db5c8c1005             8        1638   \n",
       "...                                    ...           ...         ...   \n",
       "3882  3a07fe8e-d341-4112-a09a-827e1073b704            19        4349   \n",
       "2474  77d08099-3a56-4140-996e-cd0a18a41155            12        3223   \n",
       "3191  8a9dc768-9b0c-4f0e-9ac7-72393810f754            33        6712   \n",
       "3664  d83f91fe-d8fb-43d6-affe-c2a00e0836e5            33        4539   \n",
       "3668  ad350fa7-ca99-4d7f-865a-0ec62525e6b1            20        4183   \n",
       "\n",
       "      InstallmentPercent  CurrentResidenceDuration  Age  ExistingCreditsCount  \\\n",
       "3985                   3                         3   35                     1   \n",
       "2467                   5                         4   55                     2   \n",
       "3290                   4                         5   43                     2   \n",
       "2523                   1                         1   24                     1   \n",
       "1084                   3                         4   38                     1   \n",
       "...                  ...                       ...  ...                   ...   \n",
       "3882                   3                         2   23                     2   \n",
       "2474                   4                         3   35                     1   \n",
       "3191                   3                         4   43                     3   \n",
       "3664                   4                         5   51                     2   \n",
       "3668                   2                         3   21                     1   \n",
       "\n",
       "      Dependents  Risk_target  Ratio_CheckingStatus  ...  \\\n",
       "3985           1            1              0.095488  ...   \n",
       "2467           2            1              0.569337  ...   \n",
       "3290           2            1              0.569337  ...   \n",
       "2523           1            0              0.095488  ...   \n",
       "1084           1            0              0.569337  ...   \n",
       "...          ...          ...                   ...  ...   \n",
       "3882           2            0              0.224771  ...   \n",
       "2474           1            0              0.569337  ...   \n",
       "3191           2            1              0.414286  ...   \n",
       "3664           2            1              0.569337  ...   \n",
       "3668           2            0              0.224771  ...   \n",
       "\n",
       "      Ratio_ExistingSavings  Ratio_EmploymentDuration  Ratio_Sex  \\\n",
       "3985               0.143533                  0.730706   0.356211   \n",
       "2467               0.518919                  0.730706   0.356211   \n",
       "3290               0.625000                  0.730706   0.356211   \n",
       "2523               0.143533                  0.237374   0.297468   \n",
       "1084               0.338501                  0.080906   0.356211   \n",
       "...                     ...                       ...        ...   \n",
       "3882               0.483501                  0.080906   0.356211   \n",
       "2474               0.625000                  0.237374   0.356211   \n",
       "3191               0.518919                  0.413187   0.297468   \n",
       "3664               0.518919                  0.730706   0.356211   \n",
       "3668               0.483501                  0.237374   0.356211   \n",
       "\n",
       "      Ratio_OthersOnLoan  Ratio_OwnsProperty  Ratio_InstallmentPlans  \\\n",
       "3985            0.267500            0.439681                0.174917   \n",
       "2467            0.683084            0.705263                0.223373   \n",
       "3290            0.683084            0.439681                0.223373   \n",
       "2523            0.267500            0.265471                0.174917   \n",
       "1084            0.267500            0.055405                0.386151   \n",
       "...                  ...                 ...                     ...   \n",
       "3882            0.267500            0.439681                0.174917   \n",
       "2474            0.267500            0.439681                0.386151   \n",
       "3191            0.267500            0.705263                0.386151   \n",
       "3664            0.683084            0.265471                0.223373   \n",
       "3668            0.267500            0.439681                0.386151   \n",
       "\n",
       "      Ratio_Housing  Ratio_Job  Ratio_Telephone  Ratio_ForeignWorker  \n",
       "3985       0.324096   0.257403         0.533036             0.328427  \n",
       "2467       0.324096   0.257403         0.533036             0.328427  \n",
       "3290       0.324096   0.257403         0.533036             0.328427  \n",
       "2523       0.118949   0.210811         0.198892             0.328427  \n",
       "1084       0.118949   0.392941         0.533036             0.328427  \n",
       "...             ...        ...              ...                  ...  \n",
       "3882       0.324096   0.257403         0.533036             0.328427  \n",
       "2474       0.118949   0.392941         0.198892             0.328427  \n",
       "3191       0.324096   0.347636         0.533036             0.328427  \n",
       "3664       0.324096   0.257403         0.533036             0.328427  \n",
       "3668       0.324096   0.347636         0.533036             0.328427  \n",
       "\n",
       "[3333 rows x 22 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "nominalFeatures = ['CheckingStatus','CreditHistory','LoanPurpose','ExistingSavings','EmploymentDuration','Sex',\n",
    "'OthersOnLoan','OwnsProperty','InstallmentPlans','Housing','Job','Telephone','ForeignWorker']\n",
    "\n",
    "train_data_ratio = train_data.copy()\n",
    "test_data_ratio = test_data.copy()\n",
    "\n",
    "ratioFeatures = []\n",
    "\n",
    "for name in nominalFeatures:\n",
    "    ratio_map = nominal_to_ratio(train_data_ratio, name, label_new)\n",
    "    ratio_name = 'Ratio_' + name\n",
    "    train_data_ratio[ratio_name] = train_data_ratio[name].map(ratio_map)\n",
    "    train_data_ratio.drop(name, axis=1, inplace=True)\n",
    "    \n",
    "    test_data_ratio[ratio_name] = test_data_ratio[name].map(ratio_map)\n",
    "    test_data_ratio.drop(name, axis=1, inplace=True)\n",
    "    ratioFeatures.append(ratio_name)\n",
    "\n",
    "print(ratioFeatures)\n",
    "display(train_data_ratio)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Risk_target\n",
      "['LoanDuration', 'LoanAmount', 'InstallmentPercent', 'CurrentResidenceDuration', 'Age', 'ExistingCreditsCount', 'Dependents', 'Ratio_CheckingStatus', 'Ratio_CreditHistory', 'Ratio_LoanPurpose', 'Ratio_ExistingSavings', 'Ratio_EmploymentDuration', 'Ratio_Sex', 'Ratio_OthersOnLoan', 'Ratio_OwnsProperty', 'Ratio_InstallmentPlans', 'Ratio_Housing', 'Ratio_Job', 'Ratio_Telephone', 'Ratio_ForeignWorker']\n",
      "[]\n"
     ]
    }
   ],
   "source": [
    "label = label_new\n",
    "id_column = 'CustomerID'\n",
    "\n",
    "### Define continuous list\n",
    "numericFeatures  = ['LoanDuration','LoanAmount','InstallmentPercent','CurrentResidenceDuration','Age','ExistingCreditsCount','Dependents']  \n",
    "numericFeatures = numericFeatures + ratioFeatures\n",
    "### Define the categorical list\n",
    "objectFeatures = []\n",
    "\n",
    "# Target column name.\n",
    "TARGET_COLUMN_NAME = label\n",
    "# Numeric feature names.\n",
    "NUMERIC_FEATURE_NAMES = numericFeatures\n",
    "# Categorical features and their vocabulary lists.\n",
    "CATEGORICAL_FEATURE_NAMES = objectFeatures\n",
    "\n",
    "print(TARGET_COLUMN_NAME)\n",
    "print(NUMERIC_FEATURE_NAMES)\n",
    "print(CATEGORICAL_FEATURE_NAMES)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def specify_feature_usages(df, label):\n",
    "    feature_usages = list()\n",
    "    feature_names = list()\n",
    "\n",
    "    for feature_name in NUMERIC_FEATURE_NAMES:\n",
    "        \n",
    "        mean = df[feature_name].mean()\n",
    "        std = df[feature_name].std()\n",
    "        \n",
    "        def zscore(x):\n",
    "            x = tf.dtypes.cast(x, tf.float32)\n",
    "            return (x - mean)/std\n",
    "        \n",
    "        feature_usage = tf.feature_column.numeric_column(key=feature_name, normalizer_fn=zscore)\n",
    "        feature_usages.append(feature_usage)\n",
    "        feature_names.append(feature_name)\n",
    "\n",
    "    for feature_name in CATEGORICAL_FEATURE_NAMES:\n",
    "        \n",
    "        aggregate = df.groupby(feature_name)[label].agg(['sum','count'])\n",
    "        aggregate['share'] = aggregate['sum'] / aggregate['count']\n",
    "        aggregate.sort_values('share', ascending=False, inplace=True)\n",
    "        vocabulary = aggregate.index.values.tolist()\n",
    "        \n",
    "        feature_usage = tf.feature_column.indicator_column(\n",
    "            tf.feature_column.categorical_column_with_vocabulary_list(\n",
    "            key=feature_name, vocabulary_list=vocabulary, default_value=0\n",
    "        ))\n",
    "        feature_usages.append(feature_usage)\n",
    "        feature_names.append(feature_name)\n",
    "\n",
    "    return feature_usages, feature_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "inputs=20, output=1\n",
      "[NumericColumn(key='LoanDuration', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8ba008bb80>),\n",
      " NumericColumn(key='LoanAmount', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8ba008bee0>),\n",
      " NumericColumn(key='InstallmentPercent', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8ba008b310>),\n",
      " NumericColumn(key='CurrentResidenceDuration', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8ba008b280>),\n",
      " NumericColumn(key='Age', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def0d0>),\n",
      " NumericColumn(key='ExistingCreditsCount', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def160>),\n",
      " NumericColumn(key='Dependents', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def1f0>),\n",
      " NumericColumn(key='Ratio_CheckingStatus', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def280>),\n",
      " NumericColumn(key='Ratio_CreditHistory', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def310>),\n",
      " NumericColumn(key='Ratio_LoanPurpose', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def3a0>),\n",
      " NumericColumn(key='Ratio_ExistingSavings', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def430>),\n",
      " NumericColumn(key='Ratio_EmploymentDuration', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def4c0>),\n",
      " NumericColumn(key='Ratio_Sex', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def550>),\n",
      " NumericColumn(key='Ratio_OthersOnLoan', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def5e0>),\n",
      " NumericColumn(key='Ratio_OwnsProperty', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def670>),\n",
      " NumericColumn(key='Ratio_InstallmentPlans', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def700>),\n",
      " NumericColumn(key='Ratio_Housing', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def790>),\n",
      " NumericColumn(key='Ratio_Job', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def820>),\n",
      " NumericColumn(key='Ratio_Telephone', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def8b0>),\n",
      " NumericColumn(key='Ratio_ForeignWorker', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=<function specify_feature_usages.<locals>.zscore at 0x7f8b38def940>)]\n"
     ]
    }
   ],
   "source": [
    "train_data = train_data_ratio.copy()\n",
    "test_data = test_data_ratio.copy()\n",
    "\n",
    "feature_columns, feature_names = specify_feature_usages(train_data, label)\n",
    "l_inputs = len(feature_columns)\n",
    "l_outputs = len([label])\n",
    "print('inputs={inputs}, output={output}'.format(inputs=l_inputs,output=l_outputs))\n",
    "pp.pprint(feature_columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# A utility method to create a tf.data dataset from a Pandas Dataframe\n",
    "def df_to_dataset(dataframe, target_cols, shuffle=True):\n",
    "    dataframe = dataframe.copy()\n",
    "    total_rows = dataframe.shape[0]\n",
    "    batch_size = int(total_rows/10)\n",
    "    labels = dataframe[target_cols]\n",
    "    features = dataframe.drop(target_cols, axis=1)\n",
    "    ds = tf.data.Dataset.from_tensor_slices((dict(features), labels))\n",
    "    if shuffle:\n",
    "        ds = ds.shuffle(buffer_size=len(features))\n",
    "        ds = ds.batch(batch_size)\n",
    "    return ds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert the dataset into a TensorFlow dataset.\n",
    "train_dataset = df_to_dataset(train_data, label, True)\n",
    "test_dataset = df_to_dataset(test_data, label, True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(DenseFeatures(feature_columns))\n",
    "model.add(LayerNormalization(axis=1))\n",
    "model.add(Dense(l_inputs, activation=tf.nn.sigmoid))\n",
    "model.add(Dense(int(l_inputs/2)+1, activation=tf.nn.sigmoid))\n",
    "model.add(Dense(l_outputs, activation=tf.nn.softmax))\n",
    "\n",
    "METRICS = [\n",
    "    keras.metrics.BinaryAccuracy(name='accuracy'),\n",
    "    keras.metrics.Precision(name='precision'),\n",
    "    keras.metrics.Recall(name='recall'),\n",
    "    keras.metrics.AUC(name='auc')\n",
    "]\n",
    "\n",
    "model.compile(optimizer=keras.optimizers.Adam(),\n",
    "                loss=keras.losses.BinaryCrossentropy(),\n",
    "                metrics=METRICS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Layers in a Sequential model should only have a single input tensor. Received: inputs={'CustomerID': <tf.Tensor 'IteratorGetNext:2' shape=(None,) dtype=string>, 'LoanDuration': <tf.Tensor 'IteratorGetNext:7' shape=(None,) dtype=int64>, 'LoanAmount': <tf.Tensor 'IteratorGetNext:6' shape=(None,) dtype=int64>, 'InstallmentPercent': <tf.Tensor 'IteratorGetNext:5' shape=(None,) dtype=int64>, 'CurrentResidenceDuration': <tf.Tensor 'IteratorGetNext:1' shape=(None,) dtype=int64>, 'Age': <tf.Tensor 'IteratorGetNext:0' shape=(None,) dtype=int64>, 'ExistingCreditsCount': <tf.Tensor 'IteratorGetNext:4' shape=(None,) dtype=int64>, 'Dependents': <tf.Tensor 'IteratorGetNext:3' shape=(None,) dtype=int64>, 'Ratio_CheckingStatus': <tf.Tensor 'IteratorGetNext:8' shape=(None,) dtype=float64>, 'Ratio_CreditHistory': <tf.Tensor 'IteratorGetNext:9' shape=(None,) dtype=float64>, 'Ratio_LoanPurpose': <tf.Tensor 'IteratorGetNext:16' shape=(None,) dtype=float64>, 'Ratio_ExistingSavings': <tf.Tensor 'IteratorGetNext:11' shape=(None,) dtype=float64>, 'Ratio_EmploymentDuration': <tf.Tensor 'IteratorGetNext:10' shape=(None,) dtype=float64>, 'Ratio_Sex': <tf.Tensor 'IteratorGetNext:19' shape=(None,) dtype=float64>, 'Ratio_OthersOnLoan': <tf.Tensor 'IteratorGetNext:17' shape=(None,) dtype=float64>, 'Ratio_OwnsProperty': <tf.Tensor 'IteratorGetNext:18' shape=(None,) dtype=float64>, 'Ratio_InstallmentPlans': <tf.Tensor 'IteratorGetNext:14' shape=(None,) dtype=float64>, 'Ratio_Housing': <tf.Tensor 'IteratorGetNext:13' shape=(None,) dtype=float64>, 'Ratio_Job': <tf.Tensor 'IteratorGetNext:15' shape=(None,) dtype=float64>, 'Ratio_Telephone': <tf.Tensor 'IteratorGetNext:20' shape=(None,) dtype=float64>, 'Ratio_ForeignWorker': <tf.Tensor 'IteratorGetNext:12' shape=(None,) dtype=float64>}. Consider rewriting this model with the Functional API.\n",
      "WARNING:tensorflow:Layers in a Sequential model should only have a single input tensor. Received: inputs={'CustomerID': <tf.Tensor 'IteratorGetNext:2' shape=(None,) dtype=string>, 'LoanDuration': <tf.Tensor 'IteratorGetNext:7' shape=(None,) dtype=int64>, 'LoanAmount': <tf.Tensor 'IteratorGetNext:6' shape=(None,) dtype=int64>, 'InstallmentPercent': <tf.Tensor 'IteratorGetNext:5' shape=(None,) dtype=int64>, 'CurrentResidenceDuration': <tf.Tensor 'IteratorGetNext:1' shape=(None,) dtype=int64>, 'Age': <tf.Tensor 'IteratorGetNext:0' shape=(None,) dtype=int64>, 'ExistingCreditsCount': <tf.Tensor 'IteratorGetNext:4' shape=(None,) dtype=int64>, 'Dependents': <tf.Tensor 'IteratorGetNext:3' shape=(None,) dtype=int64>, 'Ratio_CheckingStatus': <tf.Tensor 'IteratorGetNext:8' shape=(None,) dtype=float64>, 'Ratio_CreditHistory': <tf.Tensor 'IteratorGetNext:9' shape=(None,) dtype=float64>, 'Ratio_LoanPurpose': <tf.Tensor 'IteratorGetNext:16' shape=(None,) dtype=float64>, 'Ratio_ExistingSavings': <tf.Tensor 'IteratorGetNext:11' shape=(None,) dtype=float64>, 'Ratio_EmploymentDuration': <tf.Tensor 'IteratorGetNext:10' shape=(None,) dtype=float64>, 'Ratio_Sex': <tf.Tensor 'IteratorGetNext:19' shape=(None,) dtype=float64>, 'Ratio_OthersOnLoan': <tf.Tensor 'IteratorGetNext:17' shape=(None,) dtype=float64>, 'Ratio_OwnsProperty': <tf.Tensor 'IteratorGetNext:18' shape=(None,) dtype=float64>, 'Ratio_InstallmentPlans': <tf.Tensor 'IteratorGetNext:14' shape=(None,) dtype=float64>, 'Ratio_Housing': <tf.Tensor 'IteratorGetNext:13' shape=(None,) dtype=float64>, 'Ratio_Job': <tf.Tensor 'IteratorGetNext:15' shape=(None,) dtype=float64>, 'Ratio_Telephone': <tf.Tensor 'IteratorGetNext:20' shape=(None,) dtype=float64>, 'Ratio_ForeignWorker': <tf.Tensor 'IteratorGetNext:12' shape=(None,) dtype=float64>}. Consider rewriting this model with the Functional API.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f8ba002e1f0>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(train_dataset, verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training dataset\n",
      "WARNING:tensorflow:Layers in a Sequential model should only have a single input tensor. Received: inputs={'CustomerID': <tf.Tensor 'IteratorGetNext:2' shape=(None,) dtype=string>, 'LoanDuration': <tf.Tensor 'IteratorGetNext:7' shape=(None,) dtype=int64>, 'LoanAmount': <tf.Tensor 'IteratorGetNext:6' shape=(None,) dtype=int64>, 'InstallmentPercent': <tf.Tensor 'IteratorGetNext:5' shape=(None,) dtype=int64>, 'CurrentResidenceDuration': <tf.Tensor 'IteratorGetNext:1' shape=(None,) dtype=int64>, 'Age': <tf.Tensor 'IteratorGetNext:0' shape=(None,) dtype=int64>, 'ExistingCreditsCount': <tf.Tensor 'IteratorGetNext:4' shape=(None,) dtype=int64>, 'Dependents': <tf.Tensor 'IteratorGetNext:3' shape=(None,) dtype=int64>, 'Ratio_CheckingStatus': <tf.Tensor 'IteratorGetNext:8' shape=(None,) dtype=float64>, 'Ratio_CreditHistory': <tf.Tensor 'IteratorGetNext:9' shape=(None,) dtype=float64>, 'Ratio_LoanPurpose': <tf.Tensor 'IteratorGetNext:16' shape=(None,) dtype=float64>, 'Ratio_ExistingSavings': <tf.Tensor 'IteratorGetNext:11' shape=(None,) dtype=float64>, 'Ratio_EmploymentDuration': <tf.Tensor 'IteratorGetNext:10' shape=(None,) dtype=float64>, 'Ratio_Sex': <tf.Tensor 'IteratorGetNext:19' shape=(None,) dtype=float64>, 'Ratio_OthersOnLoan': <tf.Tensor 'IteratorGetNext:17' shape=(None,) dtype=float64>, 'Ratio_OwnsProperty': <tf.Tensor 'IteratorGetNext:18' shape=(None,) dtype=float64>, 'Ratio_InstallmentPlans': <tf.Tensor 'IteratorGetNext:14' shape=(None,) dtype=float64>, 'Ratio_Housing': <tf.Tensor 'IteratorGetNext:13' shape=(None,) dtype=float64>, 'Ratio_Job': <tf.Tensor 'IteratorGetNext:15' shape=(None,) dtype=float64>, 'Ratio_Telephone': <tf.Tensor 'IteratorGetNext:20' shape=(None,) dtype=float64>, 'Ratio_ForeignWorker': <tf.Tensor 'IteratorGetNext:12' shape=(None,) dtype=float64>}. Consider rewriting this model with the Functional API.\n",
      "loss 10.157008171081543\n",
      "accuracy 0.33393338322639465\n",
      "precision 0.33393338322639465\n",
      "recall 1.0\n",
      "auc 0.5\n",
      "Test dataset\n",
      "loss 10.153963088989258\n",
      "accuracy 0.33413317799568176\n",
      "precision 0.33413317799568176\n",
      "recall 1.0\n",
      "auc 0.5\n"
     ]
    }
   ],
   "source": [
    "# evaluate the model\n",
    "print(\"Training dataset\")\n",
    "scores = model.evaluate(train_dataset, verbose=0)\n",
    "scores_metrics = zip(model.metrics_names, scores)\n",
    "for m, s in list(scores_metrics):\n",
    "    print(m, s)\n",
    "    \n",
    "print(\"Test dataset\")\n",
    "scores = model.evaluate(test_dataset, verbose=0)\n",
    "scores_metrics = zip(model.metrics_names, scores)\n",
    "for m, s in list(scores_metrics):\n",
    "    print(m, s)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Layers in a Sequential model should only have a single input tensor. Received: inputs={'CustomerID': <tf.Tensor 'IteratorGetNext:2' shape=(None,) dtype=string>, 'LoanDuration': <tf.Tensor 'IteratorGetNext:7' shape=(None,) dtype=int64>, 'LoanAmount': <tf.Tensor 'IteratorGetNext:6' shape=(None,) dtype=int64>, 'InstallmentPercent': <tf.Tensor 'IteratorGetNext:5' shape=(None,) dtype=int64>, 'CurrentResidenceDuration': <tf.Tensor 'IteratorGetNext:1' shape=(None,) dtype=int64>, 'Age': <tf.Tensor 'IteratorGetNext:0' shape=(None,) dtype=int64>, 'ExistingCreditsCount': <tf.Tensor 'IteratorGetNext:4' shape=(None,) dtype=int64>, 'Dependents': <tf.Tensor 'IteratorGetNext:3' shape=(None,) dtype=int64>, 'Ratio_CheckingStatus': <tf.Tensor 'IteratorGetNext:8' shape=(None,) dtype=float64>, 'Ratio_CreditHistory': <tf.Tensor 'IteratorGetNext:9' shape=(None,) dtype=float64>, 'Ratio_LoanPurpose': <tf.Tensor 'IteratorGetNext:16' shape=(None,) dtype=float64>, 'Ratio_ExistingSavings': <tf.Tensor 'IteratorGetNext:11' shape=(None,) dtype=float64>, 'Ratio_EmploymentDuration': <tf.Tensor 'IteratorGetNext:10' shape=(None,) dtype=float64>, 'Ratio_Sex': <tf.Tensor 'IteratorGetNext:19' shape=(None,) dtype=float64>, 'Ratio_OthersOnLoan': <tf.Tensor 'IteratorGetNext:17' shape=(None,) dtype=float64>, 'Ratio_OwnsProperty': <tf.Tensor 'IteratorGetNext:18' shape=(None,) dtype=float64>, 'Ratio_InstallmentPlans': <tf.Tensor 'IteratorGetNext:14' shape=(None,) dtype=float64>, 'Ratio_Housing': <tf.Tensor 'IteratorGetNext:13' shape=(None,) dtype=float64>, 'Ratio_Job': <tf.Tensor 'IteratorGetNext:15' shape=(None,) dtype=float64>, 'Ratio_Telephone': <tf.Tensor 'IteratorGetNext:20' shape=(None,) dtype=float64>, 'Ratio_ForeignWorker': <tf.Tensor 'IteratorGetNext:12' shape=(None,) dtype=float64>}. Consider rewriting this model with the Functional API.\n",
      "11/11 [==============================] - 0s 6ms/step\n"
     ]
    }
   ],
   "source": [
    "y_predicted = model.predict(test_dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_hist = pd.DataFrame(y_predicted)\n",
    "df_hist.columns = [\"Risk_prediction\"]\n",
    "sns.histplot(data=df_hist, x=\"Risk_prediction\", bins=10)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 557    0]\n",
      " [1110    0]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.00      0.00      0.00      1110\n",
      "           1       0.33      1.00      0.50       557\n",
      "\n",
      "    accuracy                           0.33      1667\n",
      "   macro avg       0.17      0.50      0.25      1667\n",
      "weighted avg       0.11      0.33      0.17      1667\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.8/dist-packages/sklearn/metrics/_classification.py:1344: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/usr/local/lib/python3.8/dist-packages/sklearn/metrics/_classification.py:1344: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/usr/local/lib/python3.8/dist-packages/sklearn/metrics/_classification.py:1344: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "y_predicted = np.where(y_predicted > 0.5, 1, 0)\n",
    "y_actual = test_data[label]\n",
    "matrix = confusion_matrix(y_actual, y_predicted, labels=[1,0])\n",
    "report = classification_report(y_actual, y_predicted)\n",
    "print(matrix)\n",
    "print(report)"
   ]
  }
 ],
 "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": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}