churn.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 110,
   "id": "aecf3f4e-17f2-4fae-bbd6-689ba576ea3c",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import pandas as pd, numpy as np\n",
    "from timeit import default_timer as timer\n",
    "from datetime import timedelta\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "import pickle as pkl\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Input, Flatten, BatchNormalization,Normalization\n",
    "from tensorflow.python.keras.layers.recurrent import LSTM\n",
    "from keras.optimizers import SGD\n",
    "from keras.metrics import binary_accuracy\n",
    "from keras.utils import to_categorical\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from keras.optimizers import Adam\n",
    "from keras.models import Model\n",
    "# from keras.utils.vis_utils import plot_model\n",
    "import tensorflow as tf\n",
    "from keras import metrics\n",
    "import pickle as pkl\n",
    "import tensorflow.keras.backend as K\n",
    "import zipcodes\n",
    "from operator import itemgetter\n",
    "from tensorflow_addons.metrics import F1Score\n",
    "from itertools import chain\n",
    "from sklearn.metrics import ConfusionMatrixDisplay, confusion_matrix\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import precision_score, recall_score\n",
    "from sklearn.metrics import precision_recall_curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "23e4531b-8ce7-4dfb-a597-738998b84f54",
   "metadata": {
    "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>AMPERITY_ID</th>\n",
       "      <th>TOTAL_ORDERS</th>\n",
       "      <th>ECOM_TOTAL_ORDERS</th>\n",
       "      <th>STORE_TOTAL_ORDERS</th>\n",
       "      <th>CANCELLED_ORDERS</th>\n",
       "      <th>RETURNED_ORDERS</th>\n",
       "      <th>ECOM_RETURNED_ORDERS</th>\n",
       "      <th>STORE_RETURNED_ORDERS</th>\n",
       "      <th>GROSS_ORDERED_QUANTITY</th>\n",
       "      <th>ECOM_GROSS_ORDERED_QUANTITY</th>\n",
       "      <th>...</th>\n",
       "      <th>SEC_SPEND_MEN</th>\n",
       "      <th>SEC_SPEND_SUNGLASSES</th>\n",
       "      <th>SEC_SPEND_WOMEN</th>\n",
       "      <th>TOT_SEC_SPENT_R</th>\n",
       "      <th>NO_OF_SESSIONUUIDS_R</th>\n",
       "      <th>POSTAL</th>\n",
       "      <th>DISTANCE_TO_CLOSEST_STORE</th>\n",
       "      <th>PREFERRED_STORE_ID</th>\n",
       "      <th>IS_LOYAL_CUSTOMER</th>\n",
       "      <th>LOYALTY_FIRST_DATE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>581835f0-c5fc-372c-aa41-b2c63a6ddc9f</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>34982-7112</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5966b4db-0952-3ace-94aa-76c60894d6dc</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>85743</td>\n",
       "      <td>6.383023</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>b9f602ca-9662-3e1d-8cfd-ef2b1293c582</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>7936</td>\n",
       "      <td>0.803065</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3 rows × 117 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                            AMPERITY_ID  TOTAL_ORDERS  ECOM_TOTAL_ORDERS  \\\n",
       "0  581835f0-c5fc-372c-aa41-b2c63a6ddc9f             2                  2   \n",
       "1  5966b4db-0952-3ace-94aa-76c60894d6dc             1                  1   \n",
       "2  b9f602ca-9662-3e1d-8cfd-ef2b1293c582             1                  1   \n",
       "\n",
       "   STORE_TOTAL_ORDERS  CANCELLED_ORDERS  RETURNED_ORDERS  \\\n",
       "0                   0                 0                1   \n",
       "1                   0                 0                0   \n",
       "2                   0                 0                0   \n",
       "\n",
       "   ECOM_RETURNED_ORDERS  STORE_RETURNED_ORDERS  GROSS_ORDERED_QUANTITY  \\\n",
       "0                     1                      0                       3   \n",
       "1                     0                      0                       1   \n",
       "2                     0                      0                       2   \n",
       "\n",
       "   ECOM_GROSS_ORDERED_QUANTITY  ...  SEC_SPEND_MEN  SEC_SPEND_SUNGLASSES  \\\n",
       "0                            3  ...            NaN                   NaN   \n",
       "1                            1  ...            NaN                   NaN   \n",
       "2                            2  ...            NaN                   NaN   \n",
       "\n",
       "   SEC_SPEND_WOMEN  TOT_SEC_SPENT_R  NO_OF_SESSIONUUIDS_R      POSTAL  \\\n",
       "0              NaN              NaN                   NaN  34982-7112   \n",
       "1              NaN              NaN                   NaN       85743   \n",
       "2              NaN              NaN                   NaN        7936   \n",
       "\n",
       "   DISTANCE_TO_CLOSEST_STORE  PREFERRED_STORE_ID  IS_LOYAL_CUSTOMER  \\\n",
       "0                        NaN                 789                  0   \n",
       "1                   6.383023                 789                  0   \n",
       "2                   0.803065                 789                  0   \n",
       "\n",
       "   LOYALTY_FIRST_DATE  \n",
       "0                 NaN  \n",
       "1                 NaN  \n",
       "2                 NaN  \n",
       "\n",
       "[3 rows x 117 columns]"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ecom_df = pd.read_csv('sample_data.csv')\n",
    "ecom_df=ecom_df.drop('Unnamed: 0',axis=1)\n",
    "ecom_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "82f8af70-967c-421e-a296-67678ec00189",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of the dataset: (500000, 117)\n",
      "Null values in target variable: 0\n"
     ]
    }
   ],
   "source": [
    "print(\"Shape of the dataset:\",ecom_df.shape)\n",
    "ecom_df = ecom_df[~ecom_df['CHURNED'].isnull()]\n",
    "print(\"Null values in target variable:\",ecom_df['CHURNED'].isnull().sum())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "3015c582-8a49-4063-85b8-9db1d5f96af4",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TOTAL_ORDERS : int64\n",
      "ECOM_TOTAL_ORDERS : int64\n",
      "STORE_TOTAL_ORDERS : int64\n",
      "CANCELLED_ORDERS : int64\n",
      "RETURNED_ORDERS : int64\n",
      "ECOM_RETURNED_ORDERS : int64\n",
      "STORE_RETURNED_ORDERS : int64\n",
      "GROSS_ORDERED_QUANTITY : int64\n",
      "ECOM_GROSS_ORDERED_QUANTITY : int64\n",
      "STORE_GROSS_ORDERED_QUANTITY : int64\n",
      "CANCELLED_GROSS_QUANTITY : int64\n",
      "GROSS_SALES : float64\n",
      "ECOM_GROSS_SALES : float64\n",
      "STORE_GROSS_SALES : int64\n",
      "CANCELLED_GROSS_SALES : float64\n",
      "NET_SALES : float64\n",
      "ECOM_NET_SALES : float64\n",
      "STORE_NET_SALES : int64\n",
      "TOTAL_RETURNED_QUANTITY : int64\n",
      "ECOM_RETURNED_QUANTITY : int64\n",
      "STORE_RETURNED_QUANTITY : int64\n",
      "GROSS_DISCOUNT : float64\n",
      "ECOM_GROSS_DISCOUNT : float64\n",
      "STORE_GROSS_DISCOUNT : int64\n",
      "CANCELLED_RATE : float64\n",
      "AOV : float64\n",
      "ECOM_AOV : float64\n",
      "STORE_AOV : int64\n",
      "AUR : float64\n",
      "ECOM_AUR : float64\n",
      "STORE_AUR : int64\n",
      "UPT : float64\n",
      "ECOM_UPT : float64\n",
      "STORE_UPT : int64\n",
      "DEPTS_PURCHASED : int64\n",
      "CATEGORIES_PURCHSED : int64\n",
      "BRANDS_PURCHASED : int64\n",
      "CLASSES_PURCHASED : int64\n",
      "RECENCY : int64\n",
      "GROSS_SALES_3M : float64\n",
      "GROSS_SALES_6M : float64\n",
      "GROSS_SALES_9M : float64\n",
      "GROSS_SALES_12M : float64\n",
      "GROSS_SALES_15M : float64\n",
      "GROSS_SALES_18M : float64\n",
      "ORDERS_3M : int64\n",
      "ORDERS_6M : int64\n",
      "ORDERS_9M : int64\n",
      "ORDERS_12M : int64\n",
      "ORDERS_15M : int64\n",
      "ORDERS_18M : int64\n",
      "OPEN_EMAILS : float64\n",
      "CLICK_EMAILS : float64\n",
      "UNSUBSCRIBE_EMAILS : float64\n",
      "MKTG_CHANNEL_NATURAL_SEARCH : float64\n",
      "MKTG_CHANNEL_AFFILIATES : float64\n",
      "MKTG_CHANNEL_BANNER_CLICK : float64\n",
      "MKTG_CHANNEL_EMAIL : float64\n",
      "MKTG_CHANNEL_INTERNAL_SEARCH : float64\n",
      "MKTG_CHANNEL_NATURAL_PLA : float64\n",
      "MKTG_CHANNEL_PAID_SEARCH : float64\n",
      "MKTG_CHANNEL_PUSH_NOTIFICATION : float64\n",
      "MKTG_CHANNEL_REFERRAL : float64\n",
      "MKTG_CHANNEL_SEARCH : float64\n",
      "MKTG_CHANNEL_SHOPPING_COMPARISON : float64\n",
      "MKTG_CHANNEL_SMS : float64\n",
      "MKTG_CHANNEL_SOCIAL_MEDIA : float64\n",
      "MKTG_CHANNEL_SOCIAL_MEDIA_ORGANIC : float64\n",
      "MKTG_CHANNEL_UNKNOWN_PAID_CHANNEL : float64\n",
      "NO_SORT : float64\n",
      "PRICE_LOW_HIGH_SORT : float64\n",
      "PRICE_HIGH_LOW_SORT : float64\n",
      "NEW_SORT : float64\n",
      "BRAND_SORT : float64\n",
      "CLEARANCE_FIRST_SORT : float64\n",
      "BEST_SELLERS_SORT : float64\n",
      "TURN_TO_RATINGS : float64\n",
      "TOT_CART_FLAG : float64\n",
      "TOT_CHECKOUT_FLAG : float64\n",
      "TOT_ORDER_FLAG : float64\n",
      "SESSION_COUNT_3M : float64\n",
      "SESSION_COUNT_6M : float64\n",
      "SESSION_COUNT_9M : float64\n",
      "SESSION_COUNT_TOTAL : float64\n",
      "TOT_SEC_SPENT_3M : float64\n",
      "TOT_SEC_SPENT_6M : float64\n",
      "TOT_SEC_SPENT_9M : float64\n",
      "TOT_SEC_SPENT : float64\n",
      "AVG_SEC_SPENT : float64\n",
      "SEC_SPEND_ACCESSORIES : float64\n",
      "SEC_SPEND_BEAUTY_FRAGRANCE : float64\n",
      "SEC_SPEND_FOOTWEAR : float64\n",
      "SEC_SPEND_HANDBAGS : float64\n",
      "SEC_SPEND_HOME_GIFTS_LUGGAGE : float64\n",
      "SEC_SPEND_JEWELLERY : float64\n",
      "SEC_SPEND_KIDS : float64\n",
      "SEC_SPEND_MEN : float64\n",
      "SEC_SPEND_SUNGLASSES : float64\n",
      "SEC_SPEND_WOMEN : float64\n",
      "TOT_SEC_SPENT_R : float64\n",
      "NO_OF_SESSIONUUIDS_R : float64\n",
      "DISTANCE_TO_CLOSEST_STORE : float64\n",
      "PREFERRED_STORE_ID : int64\n",
      "IS_LOYAL_CUSTOMER : int64\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(500000, 117)"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ex_col = ['AMPERITY_ID','FIRST_ORDERED_DATE','LAST_ORDERED_DATE','ECOM_FIRST_ORDERED_DATE',\n",
    "          'ECOM_LAST_ORDERED_DATE','STORE_FIRST_ORDERED_DATE','STORE_LAST_ORDERED_DATE','CUST_TYPE',\n",
    "          'CHURNED','CUST_SEG','SS_EMAIL','POSTAL','LOYALTY_FIRST_DATE'\n",
    "         ]\n",
    "in_col = list(filter(lambda x: x not in ex_col, ecom_df.columns))\n",
    "\n",
    "cols_to_convert = ecom_df[in_col].select_dtypes(include='object').columns\n",
    "ecom_df[cols_to_convert] = ecom_df[cols_to_convert].apply(pd.to_numeric, errors='coerce', downcast = 'float')\n",
    "for i in in_col:\n",
    "    print(i,\":\",ecom_df[i].dtype)\n",
    "ecom_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "6dfbb837-9745-4331-b211-a2506ec771c4",
   "metadata": {
    "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>AMPERITY_ID</th>\n",
       "      <th>TOTAL_ORDERS</th>\n",
       "      <th>ECOM_TOTAL_ORDERS</th>\n",
       "      <th>STORE_TOTAL_ORDERS</th>\n",
       "      <th>CANCELLED_ORDERS</th>\n",
       "      <th>RETURNED_ORDERS</th>\n",
       "      <th>ECOM_RETURNED_ORDERS</th>\n",
       "      <th>STORE_RETURNED_ORDERS</th>\n",
       "      <th>GROSS_ORDERED_QUANTITY</th>\n",
       "      <th>ECOM_GROSS_ORDERED_QUANTITY</th>\n",
       "      <th>...</th>\n",
       "      <th>SEC_SPEND_MEN</th>\n",
       "      <th>SEC_SPEND_SUNGLASSES</th>\n",
       "      <th>SEC_SPEND_WOMEN</th>\n",
       "      <th>TOT_SEC_SPENT_R</th>\n",
       "      <th>NO_OF_SESSIONUUIDS_R</th>\n",
       "      <th>POSTAL</th>\n",
       "      <th>DISTANCE_TO_CLOSEST_STORE</th>\n",
       "      <th>PREFERRED_STORE_ID</th>\n",
       "      <th>IS_LOYAL_CUSTOMER</th>\n",
       "      <th>LOYALTY_FIRST_DATE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>35771</th>\n",
       "      <td>5f989c6a-2e73-3219-946d-7a32b1105f77</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106130</th>\n",
       "      <td>09adf44f-5497-3174-9725-4e00092cb5a6</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>241297</th>\n",
       "      <td>fc86445d-249e-3c97-8161-1235cd4ed6ba</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>257582</th>\n",
       "      <td>3ae293dc-fcf3-3730-9cd5-a1a47a6b8aae</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>308245</th>\n",
       "      <td>47d357b9-b704-398b-a442-aa9ab8bec603</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>20.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>353710</th>\n",
       "      <td>edcb8967-a7a5-385e-8bb4-5d5c6e9f4b4e</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>96.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>141.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>420965</th>\n",
       "      <td>97957609-a6b1-3cf7-b3ef-fef5dfc1c545</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>1</td>\n",
       "      <td>11/9/23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>451044</th>\n",
       "      <td>ec96bcb2-5ff6-33ad-9b52-5046dc8b96cd</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>1662.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1527.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>1</td>\n",
       "      <td>9/22/23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>465695</th>\n",
       "      <td>35ca64a1-9328-3f65-8390-dfe5287bbc99</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>1</td>\n",
       "      <td>11/4/23</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9 rows × 117 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 AMPERITY_ID  TOTAL_ORDERS  ECOM_TOTAL_ORDERS  \\\n",
       "35771   5f989c6a-2e73-3219-946d-7a32b1105f77             1                  1   \n",
       "106130  09adf44f-5497-3174-9725-4e00092cb5a6             0                  0   \n",
       "241297  fc86445d-249e-3c97-8161-1235cd4ed6ba             0                  0   \n",
       "257582  3ae293dc-fcf3-3730-9cd5-a1a47a6b8aae             1                  1   \n",
       "308245  47d357b9-b704-398b-a442-aa9ab8bec603             1                  1   \n",
       "353710  edcb8967-a7a5-385e-8bb4-5d5c6e9f4b4e             1                  1   \n",
       "420965  97957609-a6b1-3cf7-b3ef-fef5dfc1c545             1                  1   \n",
       "451044  ec96bcb2-5ff6-33ad-9b52-5046dc8b96cd             1                  1   \n",
       "465695  35ca64a1-9328-3f65-8390-dfe5287bbc99             1                  1   \n",
       "\n",
       "        STORE_TOTAL_ORDERS  CANCELLED_ORDERS  RETURNED_ORDERS  \\\n",
       "35771                    0                 0                0   \n",
       "106130                   0                 1                0   \n",
       "241297                   0                 1                0   \n",
       "257582                   0                 0                0   \n",
       "308245                   0                 1                1   \n",
       "353710                   0                 0                0   \n",
       "420965                   0                 0                0   \n",
       "451044                   0                 0                0   \n",
       "465695                   0                 0                0   \n",
       "\n",
       "        ECOM_RETURNED_ORDERS  STORE_RETURNED_ORDERS  GROSS_ORDERED_QUANTITY  \\\n",
       "35771                      0                      0                       1   \n",
       "106130                     0                      0                       0   \n",
       "241297                     0                      0                       0   \n",
       "257582                     0                      0                       1   \n",
       "308245                     1                      0                       2   \n",
       "353710                     0                      0                       2   \n",
       "420965                     0                      0                       1   \n",
       "451044                     0                      0                       2   \n",
       "465695                     0                      0                       2   \n",
       "\n",
       "        ECOM_GROSS_ORDERED_QUANTITY  ...  SEC_SPEND_MEN  SEC_SPEND_SUNGLASSES  \\\n",
       "35771                             1  ...            NaN                   NaN   \n",
       "106130                            0  ...            NaN                   NaN   \n",
       "241297                            0  ...            NaN                   NaN   \n",
       "257582                            1  ...            NaN                   NaN   \n",
       "308245                            2  ...            NaN                   NaN   \n",
       "353710                            2  ...           96.0                   NaN   \n",
       "420965                            1  ...            NaN                   NaN   \n",
       "451044                            2  ...         1662.0                   NaN   \n",
       "465695                            2  ...            NaN                   NaN   \n",
       "\n",
       "        SEC_SPEND_WOMEN  TOT_SEC_SPENT_R  NO_OF_SESSIONUUIDS_R  POSTAL  \\\n",
       "35771               NaN              NaN                   NaN     NaN   \n",
       "106130              NaN              NaN                   NaN     NaN   \n",
       "241297              NaN              NaN                   NaN     NaN   \n",
       "257582              NaN              NaN                   NaN     NaN   \n",
       "308245             20.0              1.0                   1.0     NaN   \n",
       "353710            141.0              1.0                   1.0     NaN   \n",
       "420965              NaN              NaN                   NaN     NaN   \n",
       "451044           1527.0              1.0                   1.0     NaN   \n",
       "465695              NaN              NaN                   NaN     NaN   \n",
       "\n",
       "        DISTANCE_TO_CLOSEST_STORE  PREFERRED_STORE_ID  IS_LOYAL_CUSTOMER  \\\n",
       "35771                         NaN                 789                  0   \n",
       "106130                        NaN                 789                  0   \n",
       "241297                        NaN                 789                  0   \n",
       "257582                        NaN                 789                  0   \n",
       "308245                        NaN                 789                  0   \n",
       "353710                        NaN                 789                  0   \n",
       "420965                        NaN                 789                  1   \n",
       "451044                        NaN                 789                  1   \n",
       "465695                        NaN                 789                  1   \n",
       "\n",
       "        LOYALTY_FIRST_DATE  \n",
       "35771                  NaN  \n",
       "106130                 NaN  \n",
       "241297                 NaN  \n",
       "257582                 NaN  \n",
       "308245                 NaN  \n",
       "353710                 NaN  \n",
       "420965             11/9/23  \n",
       "451044             9/22/23  \n",
       "465695             11/4/23  \n",
       "\n",
       "[9 rows x 117 columns]"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ecom_df[ecom_df.POSTAL.isnull()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "03c380cb-0ab5-4acd-9e32-08c2a3a5a7ca",
   "metadata": {
    "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>AMPERITY_ID</th>\n",
       "      <th>ECOM_TOTAL_ORDERS</th>\n",
       "      <th>CANCELLED_ORDERS</th>\n",
       "      <th>ECOM_RETURNED_ORDERS</th>\n",
       "      <th>ECOM_GROSS_ORDERED_QUANTITY</th>\n",
       "      <th>CANCELLED_GROSS_QUANTITY</th>\n",
       "      <th>ECOM_GROSS_SALES</th>\n",
       "      <th>CANCELLED_GROSS_SALES</th>\n",
       "      <th>ECOM_NET_SALES</th>\n",
       "      <th>ECOM_RETURNED_QUANTITY</th>\n",
       "      <th>...</th>\n",
       "      <th>SEC_SPEND_KIDS</th>\n",
       "      <th>SEC_SPEND_MEN</th>\n",
       "      <th>SEC_SPEND_SUNGLASSES</th>\n",
       "      <th>SEC_SPEND_WOMEN</th>\n",
       "      <th>TOT_SEC_SPENT_R</th>\n",
       "      <th>NO_OF_SESSIONUUIDS_R</th>\n",
       "      <th>POSTAL</th>\n",
       "      <th>DISTANCE_TO_CLOSEST_STORE</th>\n",
       "      <th>PREFERRED_STORE_ID</th>\n",
       "      <th>IS_LOYAL_CUSTOMER</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>581835f0-c5fc-372c-aa41-b2c63a6ddc9f</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>868.97</td>\n",
       "      <td>0.0</td>\n",
       "      <td>368.98</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>34982-7112</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5966b4db-0952-3ace-94aa-76c60894d6dc</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>59.99</td>\n",
       "      <td>0.0</td>\n",
       "      <td>59.99</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>85743</td>\n",
       "      <td>6.383023</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>b9f602ca-9662-3e1d-8cfd-ef2b1293c582</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>59.96</td>\n",
       "      <td>0.0</td>\n",
       "      <td>59.96</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>7936</td>\n",
       "      <td>0.803065</td>\n",
       "      <td>789</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>20d88dae-3acf-3bfe-b3ae-11f3edbc5c7c</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>71.99</td>\n",
       "      <td>0.0</td>\n",
       "      <td>71.99</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>94.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>94127-1950</td>\n",
       "      <td>NaN</td>\n",
       "      <td>789</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>b39da4a8-f8fc-3156-bdd7-e69f7229eaa8</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>248.98</td>\n",
       "      <td>0.0</td>\n",
       "      <td>248.98</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>20837</td>\n",
       "      <td>9.022280</td>\n",
       "      <td>789</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 75 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                            AMPERITY_ID  ECOM_TOTAL_ORDERS  CANCELLED_ORDERS  \\\n",
       "0  581835f0-c5fc-372c-aa41-b2c63a6ddc9f                  2                 0   \n",
       "1  5966b4db-0952-3ace-94aa-76c60894d6dc                  1                 0   \n",
       "2  b9f602ca-9662-3e1d-8cfd-ef2b1293c582                  1                 0   \n",
       "3  20d88dae-3acf-3bfe-b3ae-11f3edbc5c7c                  1                 0   \n",
       "4  b39da4a8-f8fc-3156-bdd7-e69f7229eaa8                  1                 0   \n",
       "\n",
       "   ECOM_RETURNED_ORDERS  ECOM_GROSS_ORDERED_QUANTITY  \\\n",
       "0                     1                            3   \n",
       "1                     0                            1   \n",
       "2                     0                            2   \n",
       "3                     0                            1   \n",
       "4                     0                            2   \n",
       "\n",
       "   CANCELLED_GROSS_QUANTITY  ECOM_GROSS_SALES  CANCELLED_GROSS_SALES  \\\n",
       "0                         0            868.97                    0.0   \n",
       "1                         0             59.99                    0.0   \n",
       "2                         0             59.96                    0.0   \n",
       "3                         0             71.99                    0.0   \n",
       "4                         0            248.98                    0.0   \n",
       "\n",
       "   ECOM_NET_SALES  ECOM_RETURNED_QUANTITY  ...  SEC_SPEND_KIDS  SEC_SPEND_MEN  \\\n",
       "0          368.98                       1  ...             NaN            NaN   \n",
       "1           59.99                       0  ...             NaN            NaN   \n",
       "2           59.96                       0  ...             NaN            NaN   \n",
       "3           71.99                       0  ...             NaN            NaN   \n",
       "4          248.98                       0  ...             NaN            NaN   \n",
       "\n",
       "   SEC_SPEND_SUNGLASSES  SEC_SPEND_WOMEN  TOT_SEC_SPENT_R  \\\n",
       "0                   NaN              NaN              NaN   \n",
       "1                   NaN              NaN              NaN   \n",
       "2                   NaN              NaN              NaN   \n",
       "3                   NaN             94.0              1.0   \n",
       "4                   NaN              NaN              1.0   \n",
       "\n",
       "   NO_OF_SESSIONUUIDS_R      POSTAL  DISTANCE_TO_CLOSEST_STORE  \\\n",
       "0                   NaN  34982-7112                        NaN   \n",
       "1                   NaN       85743                   6.383023   \n",
       "2                   NaN        7936                   0.803065   \n",
       "3                   1.0  94127-1950                        NaN   \n",
       "4                   1.0       20837                   9.022280   \n",
       "\n",
       "   PREFERRED_STORE_ID IS_LOYAL_CUSTOMER  \n",
       "0                 789                 0  \n",
       "1                 789                 0  \n",
       "2                 789                 0  \n",
       "3                 789                 1  \n",
       "4                 789                 1  \n",
       "\n",
       "[5 rows x 75 columns]"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "col_l = ['FIRST_ORDERED_DATE','LAST_ORDERED_DATE','ECOM_FIRST_ORDERED_DATE',\n",
    "        'ECOM_LAST_ORDERED_DATE','STORE_FIRST_ORDERED_DATE','STORE_LAST_ORDERED_DATE','LOYALTY_FIRST_DATE',\n",
    "        'Unnamed: 0', 'SS_EMAIL', 'CUST_TYPE',\n",
    "        'TOTAL_ORDERS',\n",
    "        'STORE_TOTAL_ORDERS',\n",
    "        'RETURNED_ORDERS',\n",
    "        'STORE_RETURNED_ORDERS',\n",
    "        'GROSS_ORDERED_QUANTITY',\n",
    "        'STORE_GROSS_ORDERED_QUANTITY',\n",
    "        'GROSS_SALES',\n",
    "        'STORE_GROSS_SALES',\n",
    "        'NET_SALES',\n",
    "        'STORE_NET_SALES',\n",
    "        'TOTAL_RETURNED_QUANTITY',\n",
    "        'STORE_RETURNED_QUANTITY',\n",
    "        'GROSS_DISCOUNT',\n",
    "        'STORE_GROSS_DISCOUNT',\n",
    "        'AOV',\n",
    "        'STORE_AOV',\n",
    "        'AUR',\n",
    "        'STORE_AUR',\n",
    "        'UPT',\n",
    "        'STORE_UPT',\n",
    "        'RECENCY',\n",
    "        'GROSS_SALES_3M',\n",
    "        'GROSS_SALES_6M',\n",
    "        'GROSS_SALES_9M',\n",
    "        'GROSS_SALES_12M',\n",
    "        'GROSS_SALES_15M',\n",
    "        'GROSS_SALES_18M',\n",
    "        'ORDERS_3M',\n",
    "        'ORDERS_6M',\n",
    "        'ORDERS_9M',\n",
    "        'ORDERS_12M',\n",
    "        'ORDERS_15M',\n",
    "        'ORDERS_18M'\n",
    "        ]\n",
    "\n",
    "df1 = ecom_df[ecom_df.columns[~ecom_df.columns.isin(col_l)]]\n",
    "df1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "5209283e-db91-4e09-b9fa-f4e310e86da9",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape aftert removing store cols: (500000, 75)\n"
     ]
    }
   ],
   "source": [
    "print(\"Shape aftert removing store cols:\",df1.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "1c786e85-ae35-4506-9b14-2198ced01387",
   "metadata": {
    "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>POSTAL</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>34982-7112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>85743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7936</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>94127-1950</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>20837</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499985</th>\n",
       "      <td>84096-2091</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499986</th>\n",
       "      <td>33432-6366</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499992</th>\n",
       "      <td>12189-1711</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499997</th>\n",
       "      <td>21075-5240</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499998</th>\n",
       "      <td>99163-5741</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>231488 rows × 1 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            POSTAL\n",
       "0       34982-7112\n",
       "1            85743\n",
       "2             7936\n",
       "3       94127-1950\n",
       "4            20837\n",
       "...            ...\n",
       "499985  84096-2091\n",
       "499986  33432-6366\n",
       "499992  12189-1711\n",
       "499997  21075-5240\n",
       "499998  99163-5741\n",
       "\n",
       "[231488 rows x 1 columns]"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1['POSTAL'].drop_duplicates().to_frame()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "53f10a20-e05e-4dd5-9ca6-24e07711c3d9",
   "metadata": {
    "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>POSTAL</th>\n",
       "      <th>POSTAL_NEW</th>\n",
       "      <th>zip</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>34982-7112</td>\n",
       "      <td>34982</td>\n",
       "      <td>34982</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>85743</td>\n",
       "      <td>85743</td>\n",
       "      <td>85743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7936</td>\n",
       "      <td>07936</td>\n",
       "      <td>07936</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>94127-1950</td>\n",
       "      <td>94127</td>\n",
       "      <td>94127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>20837</td>\n",
       "      <td>20837</td>\n",
       "      <td>20837</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499995</th>\n",
       "      <td>28207</td>\n",
       "      <td>28207</td>\n",
       "      <td>28207</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499996</th>\n",
       "      <td>33155</td>\n",
       "      <td>33155</td>\n",
       "      <td>33155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499997</th>\n",
       "      <td>21075-5240</td>\n",
       "      <td>21075</td>\n",
       "      <td>21075</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499998</th>\n",
       "      <td>99163-5741</td>\n",
       "      <td>99163</td>\n",
       "      <td>99163</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499999</th>\n",
       "      <td>63043</td>\n",
       "      <td>63043</td>\n",
       "      <td>63043</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>500000 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            POSTAL POSTAL_NEW    zip\n",
       "0       34982-7112      34982  34982\n",
       "1            85743      85743  85743\n",
       "2             7936      07936  07936\n",
       "3       94127-1950      94127  94127\n",
       "4            20837      20837  20837\n",
       "...            ...        ...    ...\n",
       "499995       28207      28207  28207\n",
       "499996       33155      33155  33155\n",
       "499997  21075-5240      21075  21075\n",
       "499998  99163-5741      99163  99163\n",
       "499999       63043      63043  63043\n",
       "\n",
       "[500000 rows x 3 columns]"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# FILL IN comments for this cell\n",
    "\n",
    "empty_zips = list(df1[df1.POSTAL.isnull()].index)\n",
    "df1['POSTAL_NEW'] = df1['POSTAL'].apply(lambda x: '00'+str(x) if len(str(x))==3 and x is not None and x is not np.nan\n",
    "                                  else ('0'+str(x) if len(str(x))==4 and x is not None and x is not np.nan \n",
    "                                        else x))\n",
    "df1['POSTAL_NEW'].replace({pd.NA: None}, inplace=True)\n",
    "df1.POSTAL_NEW[df1.POSTAL_NEW.apply(lambda x: len(str(x)) in [12,13,14,15,16])]\n",
    "df1['POSTAL_NEW'] = df1['POSTAL_NEW'].apply(lambda x:str(x).strip()[:5])\n",
    "\n",
    "df1['POSTAL_NEW'].replace({'None': None}, inplace=True)\n",
    "\n",
    "all_zips = list(map(itemgetter('zip_code'), zipcodes.filter_by(country='US')))\n",
    "\n",
    "df1 = pd.merge(left = df1,\n",
    "        right = pd.DataFrame({'zip':all_zips}),\n",
    "        left_on = 'POSTAL_NEW', right_on='zip',\n",
    "        how = 'left')\n",
    "\n",
    "df1[['POSTAL','POSTAL_NEW','zip']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "527ad225-4132-47c3-a40f-ed0c4e46797a",
   "metadata": {
    "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>zip</th>\n",
       "      <th>postal_group</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>34982</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>85743</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>07936</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>94127</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>20837</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499995</th>\n",
       "      <td>28207</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499996</th>\n",
       "      <td>33155</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499997</th>\n",
       "      <td>21075</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499998</th>\n",
       "      <td>99163</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>499999</th>\n",
       "      <td>63043</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>500000 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          zip  postal_group\n",
       "0       34982             3\n",
       "1       85743             8\n",
       "2       07936             0\n",
       "3       94127             9\n",
       "4       20837             2\n",
       "...       ...           ...\n",
       "499995  28207             2\n",
       "499996  33155             3\n",
       "499997  21075             2\n",
       "499998  99163             9\n",
       "499999  63043             6\n",
       "\n",
       "[500000 rows x 2 columns]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1['postal_group'] = df1.zip.apply(lambda x: int(str(x)[0]) if x is not np.nan and x is not None else 999)\n",
    "df1[['zip','postal_group']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "932ca8aa-db62-41b2-912c-37a556e9441f",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1579, 0)"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1.zip.isnull().sum(), df1.postal_group.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "6800e39a-6636-4447-a755-3c54c4a9dfe5",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "postal_group\n",
       "1      105254\n",
       "9       70701\n",
       "3       69166\n",
       "0       67750\n",
       "2       48619\n",
       "7       47079\n",
       "4       31046\n",
       "6       29123\n",
       "8       18804\n",
       "5       10879\n",
       "999      1579\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2 = df1.drop(columns=['POSTAL','POSTAL_NEW','zip'], errors='ignore')\n",
    "df2.postal_group.value_counts()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a9cac15-46bc-436a-b4af-594b7c32990b",
   "metadata": {},
   "source": [
    "### Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "95d0c445-2760-40b1-ace1-273df4af4b9d",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 1] ['reactive' 'lapsed' 'new' 'existing']\n"
     ]
    }
   ],
   "source": [
    "df2.CHURNED.replace({'YES': 1, 'NO': 0}, inplace = True)\n",
    "print(df2.CHURNED.unique(), df2.CUST_SEG.unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "0dd38b2d-f1fa-4dbd-8f07-7e91dc7e3efc",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(334084, 74)"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df = df2[df2['CUST_SEG'] != 'new']\n",
    "train_df.drop('CUST_SEG', axis=1, inplace=True, errors='ignore')\n",
    "train_df = train_df.fillna(0)\n",
    "train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "702dce0d-f2f1-4d68-8402-36eae1911962",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(267267, 73) (66817, 73)\n",
      "(267267,) (66817,)\n",
      "(165916, 74)\n"
     ]
    }
   ],
   "source": [
    "df_x = train_df.drop('CHURNED', axis=1) \n",
    "df_y = train_df['CHURNED']\n",
    "xtrain, xval, ytrain, yval = train_test_split(df_x, df_y, test_size=0.2,random_state=10, shuffle=True, stratify=df_y)\n",
    "\n",
    "test_df = df2[df2['CUST_SEG']=='new']\n",
    "test_df.drop('CUST_SEG', axis=1, inplace=True, errors='ignore') \n",
    "\n",
    "print(xtrain.shape, xval.shape)\n",
    "print(ytrain.shape, yval.shape)\n",
    "print(test_df.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "a323ce68-a712-4b67-846f-aafbf812ed68",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "xtrain_amp_df = xtrain['AMPERITY_ID']\n",
    "xval_amp_df = xval['AMPERITY_ID']\n",
    "\n",
    "xtrain.drop('AMPERITY_ID', axis=1, inplace=True)\n",
    "xval.drop('AMPERITY_ID', axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "0ae0c89c-3668-413b-a852-43c0c3db6d2f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "\n",
    "scaler = MinMaxScaler(feature_range=(0,1))\n",
    "xtrain = scaler.fit_transform(xtrain)\n",
    "xval = scaler.transform(xval)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "b2fe9585-72ea-4efc-8a23-39ee5df3b512",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def recall_m(y_true, y_pred):\n",
    "    true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))\n",
    "    possible_positives = K.sum(K.round(K.clip(y_true, 0, 1)))\n",
    "    recall = true_positives / (possible_positives + K.epsilon())\n",
    "    return recall\n",
    "\n",
    "def precision_m(y_true, y_pred):\n",
    "    true_positives = K.sum(K.round(K.clip(float(y_true) * float(y_pred), 0, 1)))\n",
    "    predicted_positives = K.sum(K.round(K.clip(y_pred, 0, 1)))\n",
    "    precision = true_positives / (predicted_positives + K.epsilon())\n",
    "    return precision\n",
    "\n",
    "def f1_m(y_true, y_pred):\n",
    "    print(y_pred.shape)\n",
    "    print(y_true.shape)\n",
    "    print('-------------------------')\n",
    "    precision = precision_m(y_true, y_pred)\n",
    "    recall = recall_m(y_true, y_pred)\n",
    "    return 2*((precision*recall)/(precision+recall+K.epsilon()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "a1224d67-491f-487c-8968-9d520a8765c1",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_4\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " dense_16 (Dense)            (None, 72)                5256      \n",
      "                                                                 \n",
      " dense_17 (Dense)            (None, 64)                4672      \n",
      "                                                                 \n",
      " dense_18 (Dense)            (None, 10)                650       \n",
      "                                                                 \n",
      " dense_19 (Dense)            (None, 1)                 11        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 10589 (41.36 KB)\n",
      "Trainable params: 10589 (41.36 KB)\n",
      "Non-trainable params: 0 (0.00 Byte)\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = Sequential([\n",
    "    Dense(72, input_shape=(72,), activation='relu'),\n",
    "    Dense(64, activation='relu'),\n",
    "    Dense(10,activation = 'relu'),\n",
    "    # we use sigmoid for binary output\n",
    "    # output layer\n",
    "    Dense(1, activation='sigmoid')\n",
    "    ])\n",
    "\n",
    "model.compile(optimizer = 'adam',\n",
    "             loss = 'binary_crossentropy',\n",
    "             metrics = ['acc', f1_m, precision_m, recall_m]\n",
    "             )\n",
    "# model.build()\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "86e80d9b-20de-4bd8-8bcf-d619327331e5",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "(None, 1)\n",
      "(None, 1)\n",
      "-------------------------\n",
      "(None, 1)\n",
      "(None, 1)\n",
      "-------------------------\n",
      "2051/2089 [============================>.] - ETA: 0s - loss: 0.2361 - acc: 0.9038 - f1_m: 0.9414 - precision_m: 0.9026 - recall_m: 0.9844(None, 1)\n",
      "(None, 1)\n",
      "-------------------------\n",
      "2089/2089 [==============================] - 1s 619us/step - loss: 0.2357 - acc: 0.9039 - f1_m: 0.9414 - precision_m: 0.9027 - recall_m: 0.9844 - val_loss: 0.2190 - val_acc: 0.9045 - val_f1_m: 0.9417 - val_precision_m: 0.9048 - val_recall_m: 0.9825\n",
      "Epoch 2/5\n",
      "2089/2089 [==============================] - 1s 561us/step - loss: 0.2148 - acc: 0.9058 - f1_m: 0.9425 - precision_m: 0.9057 - recall_m: 0.9830 - val_loss: 0.2121 - val_acc: 0.9048 - val_f1_m: 0.9423 - val_precision_m: 0.9006 - val_recall_m: 0.9886\n",
      "Epoch 3/5\n",
      "2089/2089 [==============================] - 1s 576us/step - loss: 0.2105 - acc: 0.9062 - f1_m: 0.9427 - precision_m: 0.9058 - recall_m: 0.9832 - val_loss: 0.2327 - val_acc: 0.9037 - val_f1_m: 0.9419 - val_precision_m: 0.8956 - val_recall_m: 0.9940\n",
      "Epoch 4/5\n",
      "2089/2089 [==============================] - 1s 565us/step - loss: 0.2070 - acc: 0.9066 - f1_m: 0.9428 - precision_m: 0.9060 - recall_m: 0.9833 - val_loss: 0.2055 - val_acc: 0.9054 - val_f1_m: 0.9425 - val_precision_m: 0.9028 - val_recall_m: 0.9864\n",
      "Epoch 5/5\n",
      "2089/2089 [==============================] - 1s 560us/step - loss: 0.2049 - acc: 0.9065 - f1_m: 0.9429 - precision_m: 0.9064 - recall_m: 0.9830 - val_loss: 0.2033 - val_acc: 0.9056 - val_f1_m: 0.9423 - val_precision_m: 0.9063 - val_recall_m: 0.9820\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.src.callbacks.History at 0x372f8cf90>"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(xtrain, \n",
    "          ytrain, \n",
    "          epochs=5,\n",
    "          batch_size = 128,\n",
    "          # validation_split=0.3\n",
    "          validation_data=(xval, yval)\n",
    "         )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "6ddd5500-f6d5-429a-8d4f-c31fd93f414d",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8353/8353 [==============================] - 2s 203us/step\n",
      "2089/2089 [==============================] - 0s 201us/step\n"
     ]
    }
   ],
   "source": [
    "train_preds = model.predict(xtrain)\n",
    "val_preds = model.predict(xval)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "b165d982-ca0c-4fab-88a0-5f88001020b6",
   "metadata": {
    "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>actual</th>\n",
       "      <th>pred_prob</th>\n",
       "      <th>preds</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>72575</th>\n",
       "      <td>1</td>\n",
       "      <td>0.960875</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>318731</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000028</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>317673</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999676</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>423991</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999471</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>120136</th>\n",
       "      <td>1</td>\n",
       "      <td>0.998934</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>84492</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999422</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>58084</th>\n",
       "      <td>1</td>\n",
       "      <td>0.997148</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>261152</th>\n",
       "      <td>1</td>\n",
       "      <td>0.997029</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164570</th>\n",
       "      <td>1</td>\n",
       "      <td>0.712596</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>128511</th>\n",
       "      <td>1</td>\n",
       "      <td>0.998876</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>267267 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        actual  pred_prob  preds\n",
       "72575        1   0.960875      1\n",
       "318731       0   0.000028      0\n",
       "317673       1   0.999676      1\n",
       "423991       1   0.999471      1\n",
       "120136       1   0.998934      1\n",
       "...        ...        ...    ...\n",
       "84492        1   0.999422      1\n",
       "58084        1   0.997148      1\n",
       "261152       1   0.997029      1\n",
       "164570       1   0.712596      1\n",
       "128511       1   0.998876      1\n",
       "\n",
       "[267267 rows x 3 columns]"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_pred_df = pd.DataFrame({'actual' : ytrain,\n",
    "             'pred_prob' : list(map(itemgetter(0), train_preds))\n",
    "             })\n",
    "\n",
    "train_pred_df['preds'] = train_pred_df.pred_prob.apply(lambda x : 1 if x>0.5 else 0)\n",
    "train_pred_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "98e8de04-7236-43c9-b6dc-746fbe7fb918",
   "metadata": {
    "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>actual</th>\n",
       "      <th>pred_prob</th>\n",
       "      <th>preds</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>199620</th>\n",
       "      <td>1</td>\n",
       "      <td>0.844078</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129557</th>\n",
       "      <td>1</td>\n",
       "      <td>0.800608</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>371238</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999533</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424416</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164771</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999345</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19307</th>\n",
       "      <td>1</td>\n",
       "      <td>0.851092</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>250804</th>\n",
       "      <td>1</td>\n",
       "      <td>0.997194</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>422463</th>\n",
       "      <td>0</td>\n",
       "      <td>0.612055</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>337836</th>\n",
       "      <td>1</td>\n",
       "      <td>0.997053</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31698</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999358</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>66817 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        actual  pred_prob  preds\n",
       "199620       1   0.844078      1\n",
       "129557       1   0.800608      1\n",
       "371238       1   0.999533      1\n",
       "424416       0   0.000001      0\n",
       "164771       1   0.999345      1\n",
       "...        ...        ...    ...\n",
       "19307        1   0.851092      1\n",
       "250804       1   0.997194      1\n",
       "422463       0   0.612055      1\n",
       "337836       1   0.997053      1\n",
       "31698        1   0.999358      1\n",
       "\n",
       "[66817 rows x 3 columns]"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val_pred_df = pd.DataFrame({'actual' : yval,\n",
    "             'pred_prob' : list(map(itemgetter(0), val_preds))\n",
    "             })\n",
    "\n",
    "val_pred_df['preds'] = val_pred_df.pred_prob.apply(lambda x : 1 if x>0.5 else 0)\n",
    "val_pred_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "669ec170-6697-4c42-8a9e-2ae3b6fd3cb3",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x2c9a4b1d0>"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "classes = np.unique(np.concatenate([val_pred_df['actual'], val_pred_df['preds']]))\n",
    "\n",
    "train_cm = confusion_matrix(train_pred_df['actual'], train_pred_df['preds'], labels=classes)\n",
    "ConfusionMatrixDisplay(train_cm, display_labels=classes).plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "61ab74f3-c0cd-4a5a-b1be-d02b7fe01d83",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x2c9ccc050>"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "val_cm = confusion_matrix(val_pred_df['actual'], val_pred_df['preds'], labels=classes)\n",
    "ConfusionMatrixDisplay(val_cm, display_labels=classes).plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "b5eae1ef-8588-43e9-b5d1-59eae9552890",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 35561  21122]\n",
      " [  3741 206843]]\n",
      "\n",
      " [[ 8813  5358]\n",
      " [  948 51698]]\n"
     ]
    }
   ],
   "source": [
    "print(train_cm)\n",
    "\n",
    "train_tp = train_cm[1,1]\n",
    "train_tn = train_cm[0,0]\n",
    "train_fp = train_cm[0,1]\n",
    "train_fn = train_cm[1,0]\n",
    "\n",
    "print(\"\\n\",val_cm)\n",
    "\n",
    "val_tp = val_cm[1,1]\n",
    "val_tn = val_cm[0,0]\n",
    "val_fp = val_cm[0,1]\n",
    "val_fn = val_cm[1,0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "4523d452-0a05-451c-83ed-ae65397208c7",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train data\n",
      "\n",
      "        actual  pred_prob  preds                           AMPERITY_ID\n",
      "72575        1   0.960875      1  734bdaa4-c88b-3d61-a2f1-fc1fe41d90c6\n",
      "318731       0   0.000028      0  59b96990-0442-375f-9885-944b4576b1ed\n",
      "317673       1   0.999676      1  db81ec32-381d-3cac-8e85-2551b02ff041 \n",
      "\n",
      "validation data\n",
      "\n",
      "        actual  pred_prob  preds                           AMPERITY_ID\n",
      "199620       1   0.844078      1  ce1838dc-9fe5-320d-8e74-7382e9d040aa\n",
      "129557       1   0.800608      1  de707d89-12c0-3393-8ac0-c4c73baad9f3\n",
      "371238       1   0.999533      1  63fc387b-0105-389e-aa5c-9ea8c764fbbb\n"
     ]
    }
   ],
   "source": [
    "train_pred_df['AMPERITY_ID'] = df_x.loc[train_pred_df.index]['AMPERITY_ID']\n",
    "val_pred_df['AMPERITY_ID'] = df_x.loc[val_pred_df.index]['AMPERITY_ID']\n",
    "print(\"train data\\n\")\n",
    "print(train_pred_df.head(3),\"\\n\")\n",
    "print(\"validation data\\n\")\n",
    "print(val_pred_df.head(3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "id": "0aa2b837-d28b-45a2-ab0d-35ee204efe43",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train acc: 0.906973176636098  |  validation acc: 0.905622820539683\n",
      "train f1: 0.9433062211976313  |  validation f1: 0.9425170006016299\n",
      "\n",
      "Train Sensitivity: 0.98\n",
      "Train Specificity: 0.91\n",
      "\n",
      "Validation Sensitivity: 0.98\n",
      "Validation Specificity: 0.91\n"
     ]
    }
   ],
   "source": [
    "train_acc = metrics.accuracy_score(train_pred_df['actual'], train_pred_df['preds'])\n",
    "val_acc = metrics.accuracy_score(val_pred_df['actual'], val_pred_df['preds'])\n",
    "\n",
    "train_f1 = metrics.f1_score(train_pred_df['actual'], train_pred_df['preds'])\n",
    "val_f1 = metrics.f1_score(val_pred_df['actual'], val_pred_df['preds'])\n",
    "\n",
    "print('train acc: {}  |  validation acc: {}'.format(train_acc, val_acc))\n",
    "print('train f1: {}  |  validation f1: {}'.format(train_f1, val_f1))\n",
    "\n",
    "train_sensitivity = train_tp/float(train_tp+train_fn)\n",
    "print('\\nTrain Sensitivity: {}'.format(round(train_sensitivity,2)))\n",
    "\n",
    "train_specificity = train_tp/float(train_tp+train_fp)\n",
    "print('Train Specificity: {}\\n'.format(round(train_specificity,2)))\n",
    "\n",
    "val_sensitivity = val_tp/float(val_tp+val_fn)\n",
    "print('Validation Sensitivity: {}'.format(round(val_sensitivity,2)))\n",
    "\n",
    "val_specificity = val_tp/float(val_tp+val_fp)\n",
    "print('Validation Specificity: {}'.format(round(val_specificity,2)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "id": "c8ac1134-9e5a-48be-94cf-68b2c095c6e1",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 500x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def draw_roc( actual, probs ):\n",
    "    fpr, tpr, thresholds = metrics.roc_curve( actual, probs,\n",
    "                                              drop_intermediate = False )\n",
    "    auc_score = metrics.roc_auc_score( actual, probs )\n",
    "    plt.figure(figsize=(5, 5))\n",
    "    plt.plot( fpr, tpr, label='ROC curve (area = %0.2f)' % auc_score )\n",
    "    plt.plot([0, 1], [0, 1], 'k--')\n",
    "    plt.xlim([0.0, 1.0])\n",
    "    plt.ylim([0.0, 1.05])\n",
    "    plt.xlabel('False Positive Rate or [1 - True Negative Rate]')\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.title('Receiver operating characteristic example')\n",
    "    plt.legend(loc=\"lower right\")\n",
    "    plt.show()\n",
    "\n",
    "    return None\n",
    "\n",
    "fpr, tpr, thresholds = metrics.roc_curve(val_pred_df['actual'], val_pred_df['pred_prob'], drop_intermediate = False )\n",
    "draw_roc(train_pred_df['actual'], train_pred_df['pred_prob'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "id": "b750a59d-c176-4e6c-8886-c5dc9b800629",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 500x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_roc(val_pred_df['actual'], val_pred_df['pred_prob'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "id": "489bf05f-330a-4b8a-a69d-9b3b0135d8c9",
   "metadata": {
    "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>actual</th>\n",
       "      <th>pred_prob</th>\n",
       "      <th>preds</th>\n",
       "      <th>AMPERITY_ID</th>\n",
       "      <th>0.0</th>\n",
       "      <th>0.05</th>\n",
       "      <th>0.1</th>\n",
       "      <th>0.15</th>\n",
       "      <th>0.2</th>\n",
       "      <th>0.25</th>\n",
       "      <th>...</th>\n",
       "      <th>0.5</th>\n",
       "      <th>0.55</th>\n",
       "      <th>0.6</th>\n",
       "      <th>0.65</th>\n",
       "      <th>0.7</th>\n",
       "      <th>0.75</th>\n",
       "      <th>0.8</th>\n",
       "      <th>0.85</th>\n",
       "      <th>0.9</th>\n",
       "      <th>0.95</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>72575</th>\n",
       "      <td>1</td>\n",
       "      <td>0.960875</td>\n",
       "      <td>1</td>\n",
       "      <td>734bdaa4-c88b-3d61-a2f1-fc1fe41d90c6</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>318731</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000028</td>\n",
       "      <td>0</td>\n",
       "      <td>59b96990-0442-375f-9885-944b4576b1ed</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>317673</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999676</td>\n",
       "      <td>1</td>\n",
       "      <td>db81ec32-381d-3cac-8e85-2551b02ff041</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>423991</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999471</td>\n",
       "      <td>1</td>\n",
       "      <td>12ed5047-f90a-3678-8e4d-441270d1c301</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>120136</th>\n",
       "      <td>1</td>\n",
       "      <td>0.998934</td>\n",
       "      <td>1</td>\n",
       "      <td>75a508e1-6b3f-37f3-aba0-040b52df8889</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122568</th>\n",
       "      <td>1</td>\n",
       "      <td>0.974941</td>\n",
       "      <td>1</td>\n",
       "      <td>e1ca557b-8c22-3b73-8270-da731fee59eb</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>282023</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999533</td>\n",
       "      <td>1</td>\n",
       "      <td>4033ecc5-8816-38c0-9f6f-88783a3cbfbb</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>238916</th>\n",
       "      <td>1</td>\n",
       "      <td>0.989985</td>\n",
       "      <td>1</td>\n",
       "      <td>644172fb-bbc6-3421-a4ee-8990f15b2f1d</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>296285</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999146</td>\n",
       "      <td>1</td>\n",
       "      <td>29f55456-ff68-3fd7-88b9-926dcaff9977</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>341108</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999623</td>\n",
       "      <td>1</td>\n",
       "      <td>025ba356-1d0c-3a94-a269-813fc830a44f</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>401856</th>\n",
       "      <td>1</td>\n",
       "      <td>0.998961</td>\n",
       "      <td>1</td>\n",
       "      <td>85f30b23-02ce-3fbc-8ad5-6345593e8628</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>408481</th>\n",
       "      <td>1</td>\n",
       "      <td>0.998758</td>\n",
       "      <td>1</td>\n",
       "      <td>a3d9aaef-7d71-338a-990c-58bfc51fb507</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>476630</th>\n",
       "      <td>1</td>\n",
       "      <td>0.917632</td>\n",
       "      <td>1</td>\n",
       "      <td>c39b0781-3f48-329a-91b3-d3a7ba8f206f</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>142987</th>\n",
       "      <td>1</td>\n",
       "      <td>0.988625</td>\n",
       "      <td>1</td>\n",
       "      <td>c73f4502-3793-381c-ac0f-b3274f191a49</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90324</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999313</td>\n",
       "      <td>1</td>\n",
       "      <td>8ca35d7d-c1b1-32e7-bf8f-6ab24661c99b</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111115</th>\n",
       "      <td>1</td>\n",
       "      <td>0.750193</td>\n",
       "      <td>1</td>\n",
       "      <td>85db7517-4422-3be8-890e-d72149ee3d68</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>323032</th>\n",
       "      <td>1</td>\n",
       "      <td>0.762379</td>\n",
       "      <td>1</td>\n",
       "      <td>3ac861ac-b243-3917-ad8d-8437a23f1894</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>430457</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999177</td>\n",
       "      <td>1</td>\n",
       "      <td>faa627ea-92a1-3c09-89d5-e2f2ec2e13f7</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45842</th>\n",
       "      <td>1</td>\n",
       "      <td>0.986779</td>\n",
       "      <td>1</td>\n",
       "      <td>3b741a25-160e-3e9e-976c-07b019f18a5c</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>390217</th>\n",
       "      <td>1</td>\n",
       "      <td>0.720321</td>\n",
       "      <td>1</td>\n",
       "      <td>4ae56161-5a5c-3597-8300-f4a15c094bc8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>291623</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999199</td>\n",
       "      <td>1</td>\n",
       "      <td>e7349bc2-0ad2-384b-bc65-d3d8c10443bd</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105590</th>\n",
       "      <td>1</td>\n",
       "      <td>0.665250</td>\n",
       "      <td>1</td>\n",
       "      <td>abaae21e-089b-3402-b9fe-3918b4b633b6</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>401685</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000242</td>\n",
       "      <td>0</td>\n",
       "      <td>1e3fd98c-151e-32b5-a5a1-3ee90233e1c6</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>144972</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999111</td>\n",
       "      <td>1</td>\n",
       "      <td>f90c7748-8348-300e-a25e-9111235ed941</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>176436</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000747</td>\n",
       "      <td>0</td>\n",
       "      <td>a27032cb-417e-389e-a6a6-b686024ca96f</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>384087</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999658</td>\n",
       "      <td>1</td>\n",
       "      <td>51069de1-959e-36ea-9115-bb2a114e0c3b</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>379527</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999571</td>\n",
       "      <td>1</td>\n",
       "      <td>739ab892-28a0-332a-9af3-3e2669ab610b</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>361906</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999333</td>\n",
       "      <td>1</td>\n",
       "      <td>9224cf2f-5aa7-3741-934c-6dd544512a01</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>188344</th>\n",
       "      <td>1</td>\n",
       "      <td>0.638904</td>\n",
       "      <td>1</td>\n",
       "      <td>86c8c111-888d-31f5-9897-b865c2b3c4fc</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>372675</th>\n",
       "      <td>1</td>\n",
       "      <td>0.759084</td>\n",
       "      <td>1</td>\n",
       "      <td>a357b419-5bd0-389d-85a4-e68cb7ef727b</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>292571</th>\n",
       "      <td>0</td>\n",
       "      <td>0.634827</td>\n",
       "      <td>1</td>\n",
       "      <td>c729b062-b3f0-3b30-a216-9ccea8ee006b</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>272189</th>\n",
       "      <td>1</td>\n",
       "      <td>0.525587</td>\n",
       "      <td>1</td>\n",
       "      <td>8f4eebca-f36d-3b8b-a63c-c32cacced1f3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61849</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999148</td>\n",
       "      <td>1</td>\n",
       "      <td>08509629-1d40-3829-9f5a-225215ada661</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33042</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999783</td>\n",
       "      <td>1</td>\n",
       "      <td>d4ec1d41-db86-30ae-ab92-69351e258af9</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>498827</th>\n",
       "      <td>1</td>\n",
       "      <td>0.766704</td>\n",
       "      <td>1</td>\n",
       "      <td>6defac97-322e-3183-84f6-c04e9a1e6b35</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>477005</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999339</td>\n",
       "      <td>1</td>\n",
       "      <td>a33bafc3-c677-3e1f-b855-6a3323452239</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>228751</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999072</td>\n",
       "      <td>1</td>\n",
       "      <td>f8690193-384f-365d-b619-3c3a9c93ceab</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>378889</th>\n",
       "      <td>1</td>\n",
       "      <td>0.611081</td>\n",
       "      <td>1</td>\n",
       "      <td>de5faf81-ec5a-39cc-98b8-e53947d3b0fd</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>365821</th>\n",
       "      <td>1</td>\n",
       "      <td>0.993939</td>\n",
       "      <td>1</td>\n",
       "      <td>b557121a-38c0-3afb-9358-16750518e5fe</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92622</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999153</td>\n",
       "      <td>1</td>\n",
       "      <td>85d1ccfa-a88f-370a-9072-7c3e08ca507c</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>304179</th>\n",
       "      <td>1</td>\n",
       "      <td>0.993523</td>\n",
       "      <td>1</td>\n",
       "      <td>e054bd9f-6bfb-338b-aaea-88fcf724856b</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>316724</th>\n",
       "      <td>1</td>\n",
       "      <td>0.896932</td>\n",
       "      <td>1</td>\n",
       "      <td>9cae98e2-d097-3e09-9600-6fe8be1dfaae</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>480748</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999167</td>\n",
       "      <td>1</td>\n",
       "      <td>dbc60caa-72ab-3457-acc2-4ce5a2bad44f</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>447099</th>\n",
       "      <td>1</td>\n",
       "      <td>0.770461</td>\n",
       "      <td>1</td>\n",
       "      <td>d60f17a7-4f34-3325-a9fc-3a3be846d746</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>323891</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999148</td>\n",
       "      <td>1</td>\n",
       "      <td>281cbbab-bb67-324e-be34-36a0ba7911c7</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>237813</th>\n",
       "      <td>1</td>\n",
       "      <td>0.990606</td>\n",
       "      <td>1</td>\n",
       "      <td>5fa9464b-801f-37c7-bca2-c07a254a6601</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>237526</th>\n",
       "      <td>1</td>\n",
       "      <td>0.710253</td>\n",
       "      <td>1</td>\n",
       "      <td>fc37ec83-a242-3825-9ca4-3062376b9b29</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>273032</th>\n",
       "      <td>1</td>\n",
       "      <td>0.998550</td>\n",
       "      <td>1</td>\n",
       "      <td>80761dc1-b31a-3a02-8d3a-b48e5ad1cd1f</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91010</th>\n",
       "      <td>0</td>\n",
       "      <td>0.036472</td>\n",
       "      <td>0</td>\n",
       "      <td>e6831bc1-aabf-3519-9e7e-a1199de903f7</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79644</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999530</td>\n",
       "      <td>1</td>\n",
       "      <td>373be62f-836f-3377-9dde-9b51d4762287</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>50 rows × 24 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        actual  pred_prob  preds                           AMPERITY_ID  0.0  \\\n",
       "72575        1   0.960875      1  734bdaa4-c88b-3d61-a2f1-fc1fe41d90c6    1   \n",
       "318731       0   0.000028      0  59b96990-0442-375f-9885-944b4576b1ed    1   \n",
       "317673       1   0.999676      1  db81ec32-381d-3cac-8e85-2551b02ff041    1   \n",
       "423991       1   0.999471      1  12ed5047-f90a-3678-8e4d-441270d1c301    1   \n",
       "120136       1   0.998934      1  75a508e1-6b3f-37f3-aba0-040b52df8889    1   \n",
       "122568       1   0.974941      1  e1ca557b-8c22-3b73-8270-da731fee59eb    1   \n",
       "282023       1   0.999533      1  4033ecc5-8816-38c0-9f6f-88783a3cbfbb    1   \n",
       "238916       1   0.989985      1  644172fb-bbc6-3421-a4ee-8990f15b2f1d    1   \n",
       "296285       1   0.999146      1  29f55456-ff68-3fd7-88b9-926dcaff9977    1   \n",
       "341108       1   0.999623      1  025ba356-1d0c-3a94-a269-813fc830a44f    1   \n",
       "401856       1   0.998961      1  85f30b23-02ce-3fbc-8ad5-6345593e8628    1   \n",
       "408481       1   0.998758      1  a3d9aaef-7d71-338a-990c-58bfc51fb507    1   \n",
       "476630       1   0.917632      1  c39b0781-3f48-329a-91b3-d3a7ba8f206f    1   \n",
       "142987       1   0.988625      1  c73f4502-3793-381c-ac0f-b3274f191a49    1   \n",
       "90324        1   0.999313      1  8ca35d7d-c1b1-32e7-bf8f-6ab24661c99b    1   \n",
       "111115       1   0.750193      1  85db7517-4422-3be8-890e-d72149ee3d68    1   \n",
       "323032       1   0.762379      1  3ac861ac-b243-3917-ad8d-8437a23f1894    1   \n",
       "430457       1   0.999177      1  faa627ea-92a1-3c09-89d5-e2f2ec2e13f7    1   \n",
       "45842        1   0.986779      1  3b741a25-160e-3e9e-976c-07b019f18a5c    1   \n",
       "390217       1   0.720321      1  4ae56161-5a5c-3597-8300-f4a15c094bc8    1   \n",
       "291623       1   0.999199      1  e7349bc2-0ad2-384b-bc65-d3d8c10443bd    1   \n",
       "105590       1   0.665250      1  abaae21e-089b-3402-b9fe-3918b4b633b6    1   \n",
       "401685       0   0.000242      0  1e3fd98c-151e-32b5-a5a1-3ee90233e1c6    1   \n",
       "144972       1   0.999111      1  f90c7748-8348-300e-a25e-9111235ed941    1   \n",
       "176436       0   0.000747      0  a27032cb-417e-389e-a6a6-b686024ca96f    1   \n",
       "384087       1   0.999658      1  51069de1-959e-36ea-9115-bb2a114e0c3b    1   \n",
       "379527       1   0.999571      1  739ab892-28a0-332a-9af3-3e2669ab610b    1   \n",
       "361906       1   0.999333      1  9224cf2f-5aa7-3741-934c-6dd544512a01    1   \n",
       "188344       1   0.638904      1  86c8c111-888d-31f5-9897-b865c2b3c4fc    1   \n",
       "372675       1   0.759084      1  a357b419-5bd0-389d-85a4-e68cb7ef727b    1   \n",
       "292571       0   0.634827      1  c729b062-b3f0-3b30-a216-9ccea8ee006b    1   \n",
       "272189       1   0.525587      1  8f4eebca-f36d-3b8b-a63c-c32cacced1f3    1   \n",
       "61849        1   0.999148      1  08509629-1d40-3829-9f5a-225215ada661    1   \n",
       "33042        1   0.999783      1  d4ec1d41-db86-30ae-ab92-69351e258af9    1   \n",
       "498827       1   0.766704      1  6defac97-322e-3183-84f6-c04e9a1e6b35    1   \n",
       "477005       1   0.999339      1  a33bafc3-c677-3e1f-b855-6a3323452239    1   \n",
       "228751       1   0.999072      1  f8690193-384f-365d-b619-3c3a9c93ceab    1   \n",
       "378889       1   0.611081      1  de5faf81-ec5a-39cc-98b8-e53947d3b0fd    1   \n",
       "365821       1   0.993939      1  b557121a-38c0-3afb-9358-16750518e5fe    1   \n",
       "92622        1   0.999153      1  85d1ccfa-a88f-370a-9072-7c3e08ca507c    1   \n",
       "304179       1   0.993523      1  e054bd9f-6bfb-338b-aaea-88fcf724856b    1   \n",
       "316724       1   0.896932      1  9cae98e2-d097-3e09-9600-6fe8be1dfaae    1   \n",
       "480748       1   0.999167      1  dbc60caa-72ab-3457-acc2-4ce5a2bad44f    1   \n",
       "447099       1   0.770461      1  d60f17a7-4f34-3325-a9fc-3a3be846d746    1   \n",
       "323891       1   0.999148      1  281cbbab-bb67-324e-be34-36a0ba7911c7    1   \n",
       "237813       1   0.990606      1  5fa9464b-801f-37c7-bca2-c07a254a6601    1   \n",
       "237526       1   0.710253      1  fc37ec83-a242-3825-9ca4-3062376b9b29    1   \n",
       "273032       1   0.998550      1  80761dc1-b31a-3a02-8d3a-b48e5ad1cd1f    1   \n",
       "91010        0   0.036472      0  e6831bc1-aabf-3519-9e7e-a1199de903f7    1   \n",
       "79644        1   0.999530      1  373be62f-836f-3377-9dde-9b51d4762287    1   \n",
       "\n",
       "        0.05  0.1  0.15  0.2  0.25  ...  0.5  0.55  0.6  0.65  0.7  0.75  0.8  \\\n",
       "72575      1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "318731     0    0     0    0     0  ...    0     0    0     0    0     0    0   \n",
       "317673     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "423991     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "120136     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "122568     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "282023     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "238916     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "296285     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "341108     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "401856     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "408481     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "476630     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "142987     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "90324      1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "111115     1    1     1    1     1  ...    1     1    1     1    1     1    0   \n",
       "323032     1    1     1    1     1  ...    1     1    1     1    1     1    0   \n",
       "430457     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "45842      1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "390217     1    1     1    1     1  ...    1     1    1     1    1     0    0   \n",
       "291623     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "105590     1    1     1    1     1  ...    1     1    1     1    0     0    0   \n",
       "401685     0    0     0    0     0  ...    0     0    0     0    0     0    0   \n",
       "144972     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "176436     0    0     0    0     0  ...    0     0    0     0    0     0    0   \n",
       "384087     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "379527     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "361906     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "188344     1    1     1    1     1  ...    1     1    1     0    0     0    0   \n",
       "372675     1    1     1    1     1  ...    1     1    1     1    1     1    0   \n",
       "292571     1    1     1    1     1  ...    1     1    1     0    0     0    0   \n",
       "272189     1    1     1    1     1  ...    1     0    0     0    0     0    0   \n",
       "61849      1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "33042      1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "498827     1    1     1    1     1  ...    1     1    1     1    1     1    0   \n",
       "477005     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "228751     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "378889     1    1     1    1     1  ...    1     1    1     0    0     0    0   \n",
       "365821     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "92622      1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "304179     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "316724     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "480748     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "447099     1    1     1    1     1  ...    1     1    1     1    1     1    0   \n",
       "323891     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "237813     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "237526     1    1     1    1     1  ...    1     1    1     1    1     0    0   \n",
       "273032     1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "91010      0    0     0    0     0  ...    0     0    0     0    0     0    0   \n",
       "79644      1    1     1    1     1  ...    1     1    1     1    1     1    1   \n",
       "\n",
       "        0.85  0.9  0.95  \n",
       "72575      1    1     1  \n",
       "318731     0    0     0  \n",
       "317673     1    1     1  \n",
       "423991     1    1     1  \n",
       "120136     1    1     1  \n",
       "122568     1    1     1  \n",
       "282023     1    1     1  \n",
       "238916     1    1     1  \n",
       "296285     1    1     1  \n",
       "341108     1    1     1  \n",
       "401856     1    1     1  \n",
       "408481     1    1     1  \n",
       "476630     1    1     0  \n",
       "142987     1    1     1  \n",
       "90324      1    1     1  \n",
       "111115     0    0     0  \n",
       "323032     0    0     0  \n",
       "430457     1    1     1  \n",
       "45842      1    1     1  \n",
       "390217     0    0     0  \n",
       "291623     1    1     1  \n",
       "105590     0    0     0  \n",
       "401685     0    0     0  \n",
       "144972     1    1     1  \n",
       "176436     0    0     0  \n",
       "384087     1    1     1  \n",
       "379527     1    1     1  \n",
       "361906     1    1     1  \n",
       "188344     0    0     0  \n",
       "372675     0    0     0  \n",
       "292571     0    0     0  \n",
       "272189     0    0     0  \n",
       "61849      1    1     1  \n",
       "33042      1    1     1  \n",
       "498827     0    0     0  \n",
       "477005     1    1     1  \n",
       "228751     1    1     1  \n",
       "378889     0    0     0  \n",
       "365821     1    1     1  \n",
       "92622      1    1     1  \n",
       "304179     1    1     1  \n",
       "316724     1    0     0  \n",
       "480748     1    1     1  \n",
       "447099     0    0     0  \n",
       "323891     1    1     1  \n",
       "237813     1    1     1  \n",
       "237526     0    0     0  \n",
       "273032     1    1     1  \n",
       "91010      0    0     0  \n",
       "79644      1    1     1  \n",
       "\n",
       "[50 rows x 24 columns]"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "numbers = np.arange(0,1,0.05).round(2)   #[float(x)/10 for x in range(1)]\n",
    "for i in numbers:\n",
    "    train_pred_df[i]= train_pred_df.pred_prob.map(lambda x: 1 if x > i else 0)\n",
    "train_pred_df.head(50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "id": "b33dbbfa-0207-4cf1-858c-9a363e0b25ae",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "      prob  accuracy     sensi     speci\n",
      "0.00  0.00  0.787965  1.000000  0.000229\n",
      "0.05  0.05  0.899973  0.999658  0.529630\n",
      "0.10  0.10  0.900657  0.999601  0.533070\n",
      "0.15  0.15  0.901043  0.999563  0.535028\n",
      "0.20  0.20  0.901424  0.999478  0.537145\n",
      "0.25  0.25  0.901877  0.999273  0.540038\n",
      "0.30  0.30  0.902741  0.998647  0.546442\n",
      "0.35  0.35  0.904141  0.996647  0.560468\n",
      "0.40  0.40  0.905791  0.993452  0.580121\n",
      "0.45  0.45  0.906722  0.988736  0.602032\n",
      "0.50  0.50  0.906973  0.982235  0.627366\n",
      "0.55  0.55  0.905701  0.972173  0.658751\n",
      "0.60  0.60  0.902263  0.957024  0.698816\n",
      "0.65  0.65  0.894439  0.934226  0.746626\n",
      "0.70  0.70  0.879166  0.898098  0.808832\n",
      "0.75  0.75  0.850902  0.842338  0.882716\n",
      "0.80  0.80  0.822956  0.792629  0.935624\n",
      "0.85  0.85  0.803339  0.761563  0.958541\n",
      "0.90  0.90  0.785349  0.735265  0.971420\n",
      "0.95  0.95  0.763274  0.705077  0.979482\n"
     ]
    }
   ],
   "source": [
    "# Now let's calculate accuracy sensitivity and specificity for various probability cutoffs.\n",
    "cutoff_df = pd.DataFrame( columns = ['prob','accuracy','sensi','speci'])\n",
    "for i in numbers:\n",
    "    cm1 = metrics.confusion_matrix(train_pred_df.actual, train_pred_df[i] )\n",
    "    total1=sum(sum(cm1))\n",
    "    accuracy = (cm1[0,0]+cm1[1,1])/total1\n",
    "    \n",
    "    speci = cm1[0,0]/(cm1[0,0]+cm1[0,1])\n",
    "    sensi = cm1[1,1]/(cm1[1,0]+cm1[1,1])\n",
    "    cutoff_df.loc[i] =[ i ,accuracy,sensi,speci]\n",
    "print(cutoff_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "id": "88ff7d29-d9ec-4700-89e2-9f6caa521e75",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Let's plot accuracy sensitivity and specificity for various probabilities.\n",
    "cutoff_df.plot.line(x='prob', y=['accuracy','sensi','speci'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "590cc8d9-7cf3-4717-a80b-4f2c5f0ac9b8",
   "metadata": {
    "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>actual</th>\n",
       "      <th>pred_prob</th>\n",
       "      <th>preds</th>\n",
       "      <th>AMPERITY_ID</th>\n",
       "      <th>0.0</th>\n",
       "      <th>0.05</th>\n",
       "      <th>0.1</th>\n",
       "      <th>0.15</th>\n",
       "      <th>0.2</th>\n",
       "      <th>0.25</th>\n",
       "      <th>...</th>\n",
       "      <th>0.55</th>\n",
       "      <th>0.6</th>\n",
       "      <th>0.65</th>\n",
       "      <th>0.7</th>\n",
       "      <th>0.75</th>\n",
       "      <th>0.8</th>\n",
       "      <th>0.85</th>\n",
       "      <th>0.9</th>\n",
       "      <th>0.95</th>\n",
       "      <th>final_preds</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>72575</th>\n",
       "      <td>1</td>\n",
       "      <td>0.960875</td>\n",
       "      <td>1</td>\n",
       "      <td>734bdaa4-c88b-3d61-a2f1-fc1fe41d90c6</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>318731</th>\n",
       "      <td>0</td>\n",
       "      <td>0.000028</td>\n",
       "      <td>0</td>\n",
       "      <td>59b96990-0442-375f-9885-944b4576b1ed</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>317673</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999676</td>\n",
       "      <td>1</td>\n",
       "      <td>db81ec32-381d-3cac-8e85-2551b02ff041</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>423991</th>\n",
       "      <td>1</td>\n",
       "      <td>0.999471</td>\n",
       "      <td>1</td>\n",
       "      <td>12ed5047-f90a-3678-8e4d-441270d1c301</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>120136</th>\n",
       "      <td>1</td>\n",
       "      <td>0.998934</td>\n",
       "      <td>1</td>\n",
       "      <td>75a508e1-6b3f-37f3-aba0-040b52df8889</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 25 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        actual  pred_prob  preds                           AMPERITY_ID  0.0  \\\n",
       "72575        1   0.960875      1  734bdaa4-c88b-3d61-a2f1-fc1fe41d90c6    1   \n",
       "318731       0   0.000028      0  59b96990-0442-375f-9885-944b4576b1ed    1   \n",
       "317673       1   0.999676      1  db81ec32-381d-3cac-8e85-2551b02ff041    1   \n",
       "423991       1   0.999471      1  12ed5047-f90a-3678-8e4d-441270d1c301    1   \n",
       "120136       1   0.998934      1  75a508e1-6b3f-37f3-aba0-040b52df8889    1   \n",
       "\n",
       "        0.05  0.1  0.15  0.2  0.25  ...  0.55  0.6  0.65  0.7  0.75  0.8  \\\n",
       "72575      1    1     1    1     1  ...     1    1     1    1     1    1   \n",
       "318731     0    0     0    0     0  ...     0    0     0    0     0    0   \n",
       "317673     1    1     1    1     1  ...     1    1     1    1     1    1   \n",
       "423991     1    1     1    1     1  ...     1    1     1    1     1    1   \n",
       "120136     1    1     1    1     1  ...     1    1     1    1     1    1   \n",
       "\n",
       "        0.85  0.9  0.95  final_preds  \n",
       "72575      1    1     1            1  \n",
       "318731     0    0     0            0  \n",
       "317673     1    1     1            1  \n",
       "423991     1    1     1            1  \n",
       "120136     1    1     1            1  \n",
       "\n",
       "[5 rows x 25 columns]"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_pred_df['final_preds'] = train_pred_df.pred_prob.apply(lambda x : 1 if x>0.75 else 0)\n",
    "train_pred_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "id": "f40c8596-21bd-412b-b25c-e548aa5620fc",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train acc: 0.85\n",
      "train f1: 0.90\n",
      "\n",
      "Train ==> Sensitivity/Recall: 0.84 | Specificity: 0.96\n",
      "Train ==> Precision: 0.96 | Recall: 0.84\n",
      "\n",
      "[[ 50035   6648]\n",
      " [ 33201 177383]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Lets re-evaluate once again\n",
    "\n",
    "\n",
    "train_acc = metrics.accuracy_score(train_pred_df['actual'], train_pred_df['final_preds'])\n",
    "\n",
    "train_f1 = metrics.f1_score(train_pred_df['actual'], train_pred_df['final_preds'])\n",
    "\n",
    "train_cm = confusion_matrix(train_pred_df['actual'], train_pred_df['final_preds'], labels=classes)\n",
    "ConfusionMatrixDisplay(train_cm, display_labels=classes).plot()\n",
    "\n",
    "\n",
    "train_tp = train_cm[1,1]\n",
    "train_tn = train_cm[0,0]\n",
    "train_fp = train_cm[0,1]\n",
    "train_fn = train_cm[1,0]\n",
    "\n",
    "train_sensitivity = train_tp/float(train_tp+train_fn)\n",
    "train_specificity = train_tp/float(train_tp+train_fp)\n",
    "train_precision = precision_score(train_pred_df['actual'], train_pred_df['final_preds'])\n",
    "\n",
    "print('train acc: {:.2f}'.format(train_acc))\n",
    "print('train f1: {:.2f}'.format(train_f1))\n",
    "\n",
    "print('\\nTrain ==> Sensitivity/Recall: {:.2f} | Specificity: {:.2f}'.format(train_sensitivity, train_specificity))\n",
    "\n",
    "print('Train ==> Precision: {:.2f} | Recall: {:.2f}\\n'.format(train_precision,train_sensitivity))\n",
    "\n",
    "print(train_cm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "id": "845b638d-c104-45f3-94a2-eb07a4e70c15",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "p, r, thresholds = precision_recall_curve(train_pred_df['actual'], train_pred_df['pred_prob'])\n",
    "plt.plot(thresholds, p[:-1], \"g-\", label = 'precision')\n",
    "plt.plot(thresholds, r[:-1], \"r-\", label = 'recall')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "id": "3c6e8e13-44a7-4938-b451-ce78eb6dba8e",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best cut-off/threshold point for test data -> 0.65\n",
      "Precision at Max -> 0.93\n",
      "Recall at Max -> 0.93\n"
     ]
    }
   ],
   "source": [
    "for i,j,k in zip(thresholds, p[:-1],r[:-1]):\n",
    "    if j == k:\n",
    "        best_cutoff = i\n",
    "        print('Best cut-off/threshold point for test data -> {:.2f}\\nPrecision at Max -> {:.2f}\\nRecall at Max -> {:.2f}'.format(i, j, k))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "id": "e4edfb15-af6a-44c9-9a9b-fe9607b900f4",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train acc: 0.89\n",
      "train f1: 0.93\n",
      "\n",
      "Train ==> Sensitivity/Recall: 0.93 | Specificity: 0.93\n",
      "Train ==> Precision: 0.93 | Recall: 0.93\n",
      "\n",
      "[[ 42486  14197]\n",
      " [ 14198 196386]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_pred_df['final_preds'] = train_pred_df.pred_prob.apply(lambda x : 1 if x>best_cutoff else 0)\n",
    "train_pred_df.head()\n",
    "\n",
    "train_acc = metrics.accuracy_score(train_pred_df['actual'], train_pred_df['final_preds'])\n",
    "\n",
    "train_f1 = metrics.f1_score(train_pred_df['actual'], train_pred_df['final_preds'])\n",
    "\n",
    "train_cm = confusion_matrix(train_pred_df['actual'], train_pred_df['final_preds'], labels=classes)\n",
    "ConfusionMatrixDisplay(train_cm, display_labels=classes).plot()\n",
    "\n",
    "\n",
    "train_tp = train_cm[1,1]\n",
    "train_tn = train_cm[0,0]\n",
    "train_fp = train_cm[0,1]\n",
    "train_fn = train_cm[1,0]\n",
    "\n",
    "train_sensitivity = train_tp/float(train_tp+train_fn)\n",
    "train_specificity = train_tp/float(train_tp+train_fp)\n",
    "train_precision = precision_score(train_pred_df['actual'], train_pred_df['final_preds'])\n",
    "\n",
    "print('train acc: {:.2f}'.format(train_acc))\n",
    "print('train f1: {:.2f}'.format(train_f1))\n",
    "\n",
    "print('\\nTrain ==> Sensitivity/Recall: {:.2f} | Specificity: {:.2f}'.format(train_sensitivity, train_specificity))\n",
    "\n",
    "print('Train ==> Precision: {:.2f} | Recall: {:.2f}\\n'.format(train_precision,train_sensitivity))\n",
    "\n",
    "print(train_cm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "id": "191faad0-f9f1-44c8-ae72-83572cbfc884",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "val acc: 0.89\n",
      "val f1: 0.93\n",
      "\n",
      "Validation ==> Sensitivity/Recall: 0.93 | Specificity: 0.93\n",
      "Validation ==> Precision: 0.93 | Recall: 0.93\n",
      "\n",
      "[[10562  3609]\n",
      " [ 3534 49112]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAg0AAAGwCAYAAAAqpFaiAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABAzUlEQVR4nO3deXhU5fn/8c9kmywkQwIkIRI2WQQDisFCpAqUTb6s+qtiQ1NUwAUV8xXEWqqiLURcAJWvSLEVqli0RVwxgqIosgeCLBFFtiAJCRKykXXm/P6IjB2CMsNJCHDer+s619U55znnPDPSzD33/TzPsRmGYQgAAOAM/Bq6AwAA4MJA0AAAALxC0AAAALxC0AAAALxC0AAAALxC0AAAALxC0AAAALwS0NAdMMPlcunw4cMKDw+XzWZr6O4AAHxkGIaKi4sVFxcnP7/6+x1bXl6uyspK09cJCgpScHBwHfTownRBBw2HDx9WfHx8Q3cDAGBSdna2WrRoUS/XLi8vV5tWjZSb5zR9rdjYWO3bt8+ygcMFHTSEh4dLkq694n8V4G9v4N4A9cPYktXQXQDqTbWqtEbL3X/P60NlZaVy85w6kNFaEeFnn80oKnapVeJ+VVZWEjRciE6WJAL87Qrwt+Z/QFz8DFtgQ3cBqD8/PsjgXJSYG4Xb1Cj87O/jEmXwCzpoAADAW07DJaeJpy05DVfddeYCRdAAALAElwy5dPZRg5lzLxZMuQQAAF4h0wAAsASXXDJTYDB39sWBoAEAYAlOw5DTOPsSg5lzLxaUJwAAgFfINAAALIGBkOYRNAAALMElQ06CBlMoTwAAAK+QaQAAWALlCfMIGgAAlsDsCfMoTwAAAK+QaQAAWILrx83M+VZH0AAAsASnydkTZs69WBA0AAAswWnI5FMu664vFyrGNAAAAK+QaQAAWAJjGswjaAAAWIJLNjllM3W+1VGeAAAAXiHTAACwBJdRs5k53+oIGgAAluA0WZ4wc+7FgvIEAADwCpkGAIAlkGkwj6ABAGAJLsMml2Fi9oSJcy8WlCcAAIBXyDQAACyB8oR5BA0AAEtwyk9OEwl2Zx325UJF0AAAsATD5JgGgzENjGkAAADeIdMAALAExjSYR9AAALAEp+Enp2FiTAPLSFOeAAAA3iHTAACwBJdscpn4rewSqQaCBgCAJTCmwTzKEwAAwCtkGgAAlmB+ICTlCYIGAIAl1IxpMPHAKsoTlCcAAIB3yDQAACzBZfLZE8yeIGgAAFgEYxrMI2gAAFiCS36s02ASYxoAAIBXyDQAACzBadjkNPF4azPnXiwIGgAAluA0ORDSSXmC8gQAAPAOmQYAgCW4DD+5TMyecDF7gqABAGANlCfMozwBAAC8QqYBAGAJLpmbAeGqu65csAgaAACWYH5xJ5LzfAIAAMArZBoAAJZg/tkT/M4maAAAWIJLNrlkZkwDK0ISNAAALIFMg3l8AgAAwCtkGgAAlmB+cSd+ZxM0AAAswWXY5DKzTgNPuSRsAgAA3iHTAACwBJfJ8gSLOxE0AAAswvxTLgka+AQAAIBXyDQAACzBKZucJhZoMnPuxYKgAQBgCZQnzOMTAAAAXiHTAACwBKfMlRicddeVCxZBAwDAEihPmEfQAACwBB5YZR6fAAAA9SwtLU02m02pqanufYZhaNq0aYqLi1NISIj69OmjnTt3epxXUVGh++67T02bNlVYWJiGDx+uQ4cOebQpKChQSkqKHA6HHA6HUlJSdPz4cY82Bw8e1LBhwxQWFqamTZtq4sSJqqys9Pl9EDQAACzBkE0uE5txluMhNm3apL/97W/q2rWrx/6nnnpKs2bN0ty5c7Vp0ybFxsZqwIABKi4udrdJTU3VsmXLtGTJEq1Zs0YlJSUaOnSonM6fRlgkJycrMzNT6enpSk9PV2ZmplJSUtzHnU6nhgwZotLSUq1Zs0ZLlizR0qVLNWnSJJ/fC0EDAMASTpYnzGy+Kikp0ejRo7VgwQJFRka69xuGoTlz5mjq1Km68cYblZCQoEWLFunEiRN6/fXXJUmFhYX6+9//rmeffVb9+/dXt27d9Nprr2n79u36+OOPJUlZWVlKT0/Xyy+/rKSkJCUlJWnBggV6//33tXv3bknSihUrtGvXLr322mvq1q2b+vfvr2effVYLFixQUVGRT++HoAEAAB8UFRV5bBUVFT/b9p577tGQIUPUv39/j/379u1Tbm6uBg4c6N5nt9vVu3dvrV27VpKUkZGhqqoqjzZxcXFKSEhwt1m3bp0cDod69OjhbtOzZ085HA6PNgkJCYqLi3O3GTRokCoqKpSRkeHTeydoAABYwslHY5vZJCk+Pt49fsDhcCgtLe2091uyZIm2bNly2uO5ubmSpJiYGI/9MTEx7mO5ubkKCgryyFCcrk10dHSt60dHR3u0OfU+kZGRCgoKcrfxFrMnAACW4DT5lMuT52ZnZysiIsK9326312qbnZ2t+++/XytWrFBwcPDPXtNm8xwnYRhGrX2nOrXN6dqfTRtvkGkAAMAHERERHtvpgoaMjAzl5eUpMTFRAQEBCggI0OrVq/X8888rICDA/cv/1F/6eXl57mOxsbGqrKxUQUHBL7Y5cuRIrfvn5+d7tDn1PgUFBaqqqqqVgTgTggYAgCXUVXnCG/369dP27duVmZnp3rp3767Ro0crMzNTbdu2VWxsrFauXOk+p7KyUqtXr9Y111wjSUpMTFRgYKBHm5ycHO3YscPdJikpSYWFhdq4caO7zYYNG1RYWOjRZseOHcrJyXG3WbFihex2uxITE336DClPAAAswSU/uUz8Vvbl3PDwcCUkJHjsCwsLU5MmTdz7U1NTNWPGDLVv317t27fXjBkzFBoaquTkZEmSw+HQ2LFjNWnSJDVp0kRRUVGaPHmyunTp4h5Y2alTJ11//fUaP3685s+fL0m64447NHToUHXs2FGSNHDgQHXu3FkpKSl6+umndezYMU2ePFnjx4/3KLN4g6ABAIAGMGXKFJWVlWnChAkqKChQjx49tGLFCoWHh7vbzJ49WwEBAbr55ptVVlamfv36aeHChfL393e3Wbx4sSZOnOieZTF8+HDNnTvXfdzf318ffPCBJkyYoF69eikkJETJycl65plnfO6zzTAMw8R7blBFRUVyOBzqe9UfFeD/8wNNgAuZsXlHQ3cBqDfVRpU+0zsqLCz0+Vevt05+V9z9xY2yNwo86+tUlFRp3rVv1Wtfz3dkGgAAluDruITTnW91BA0AAEswTD7l0uCBVcyeAAAA3iHTAACwBKdscp7lQ6dOnm91BA0AAEtwGebGJbgu2GkDdYfyBAAA8AqZhotcwuVH9NsbstT+0mNq0qRMj0+/Tus2xP9XC0O//912DR64R40aVWr3N030fy9drQPZjd0tnpq+Ul275Hlc97PPW+nJZ37tse9X3b9X8qjtatP6uMorArRjZ7T+knadJKlN6wKN+u1OXd4pXxERFTqSF6YP0tvrnfcuq6+3Dosa+oejGvKHHxQTXylJOrA7WItnx2jzpz9NkYtvV66xf85R154lsvnVtJl+Vyvlfx8kSQoMcmn8o4fVZ+Rx2YMNbV3TSHMfvkRHc4Lc12jX5YTGTs1RhytOyOW0ac1yh+ZPi1P5CX/h/OQyORDSzLkXC4KGi1ywvVr79jXWyk/a6pGHv6h1/KYbd+mGEVma9VySDn0fod/dvEMznlilcROGqazsp/nMyz9qp1cXd3W/rqj0/MPYK+mgUu/doFdevULbvoqVzSa1bnXcfbx9u2MqLAzWU7OuUf7RUHXudFQT79kgl8um9z7oWPdvHJaVnxOof8xorsP7a54HMOCmY5r2yn7dM7CDDnwTrOatKjTr7T1KXxKlV5+JUWmRv1q2r1Bl+U9p67seP6weA4qUdncrFRX4645Hc/TEP/fp3kEd5HLZFBVTpSeX7NXqdxvr/6ZeotBGLt31xPeaPCdbf72jdQO9c5yJSza5TIxLMHPuxaLBg4YXX3xRTz/9tHJycnT55Zdrzpw5uvbaaxu6WxeNzVsu0eYtl/zMUUM3DP9aS95M0JfrWkqSnp2TpH/9c6n6Xrdfyz9q725ZUeGvguMhp72Kn59Ld43frJcXdtNHK9u59x/6/qdfdis+vtTjnNwj4erUMV+9krIJGlCnNqx0eLxeOLO5hv7hB12WWKoD3wTr1j/mauOqCP39r3HuNrkHf3rgUGi4U4N+d0xPT4zX1i9qVuabeV9LvbZ5l7pdW6yM1RHq0b9I1dU2zf3TJTJ+rJHP/VMLzVv5jeJaV7gDFuBi06C5ljfeeEOpqamaOnWqtm7dqmuvvVaDBw/WwYMHG7JblhEbU6KoqHJtyWzu3ldV7a/tO2PUqVO+R9u+vffrjdf+o/lz39e427YoJKTKfazdpcfUrGmZXC6b5s5ZrtcXLtVfHlulVvHHf/H+YWFVKi4O+sU2gBl+foZ6jyiQPdSlrM1hstkM/apfkb7fa9f017/TG1/t1HPvf6uk6wvd57TvekKBQYYyVv+0lO+xI4E68HWwOl99QpIUaHepusrmDhgkuTMVl/+q9By9O/jKadhMb1bXoEHDrFmzNHbsWI0bN06dOnXSnDlzFB8fr3nz5jVktywjMrJcklRw3HMJ7oLjwYpqXO5+vWp1G818ppem/Km/Xn8jQb++5qAeefhz9/HmsSWSpN//7iv9640EPfqXPiopCdJTaR+rUaOK0967U8d8XdvroEc2A6grrS8r09vfbtf7+7/SxCcP6YmxrXXw22A1blqt0EYujbo3T5s/jdDDv2urL9Mj9OjL+9WlZ82/46joalVW2FRS6JmILTgaoMhmNcHytjXhimxWpd/enaeAQJcaOap12x9zfzy/Sjg/nRzTYGazugb7BCorK5WRkeF+wMZJAwcO1Nq1a097TkVFhYqKijw21IFTomebDP33zKL0Fe20dVtzHTjYWKu/aK2/PnmtrroyV+3aHqtpb6tpveTfNWWOPd810aznkmQY0nW9ameNWsUf12NTV+v1NxK09b+yHEBdOfSdXRMGdND9Q9vr/X821eTnDqpl+3LZfvyLt+6jCC1b0Ex7d4bozbkx2vBxhIb84YdfvKbNJvf/Vw58E6xnUlvq/92Zr3e/265/Ze5SzsEgHcsLkMvFr1FcvBosaDh69KicTqdiYmI89sfExCg3N/e056SlpcnhcLi3+Pj407aDdwoKajIMkZFlHvsbN66olX34b3u+i1JVlZ/i4mqCtmMFNWMdDh78qZZcVe2v3NxGatbMM1XbMr5QT07/RB+uaKd/vdmlTt4HcKrqKj8d3m/Xt1+F6pW05tq3K0Qjx+Wr6Ji/qqtqvvT/W/a3dkVfUjPb4lhegILshho5qj3aNG5SrYKjP2UfPl0Wqd9debmSr+qsmy6/XK8+EyNHk2rlHqTkdr5yyeZ+/sRZbQyEbPh1Gmw2z/8IhmHU2nfSww8/rMLCQveWnZ19Lrp40co90kjHjgWr25U57n0BAU51ufyIsrKa/ex5rVoWKjDQ5Q4W9uxpospKP7Vo8VPmx9/fpZiYUuXlh/10XvxxzZz+sT5e1UaLXruy7t8Q8AsCgwxVV/npm22hanGpZ9nskrYVyjtU82X/7Vehqqq06arrStzHo6Kr1Oqycu3aFFrrusePBqr8hL96jziuqgo/bfk8vFYbnB+MH2dPnO1mEDQ03OyJpk2byt/fv1ZWIS8vr1b24SS73S67nVHJvggOrlJc82L369iYErVtc0zFxXblHw3Tsncv0y2/3anDhyP0/eFw3XLTDlVUBOjTz1tLkprHFqtv7/3alBGnoiK7WsYXavztW7Tnu0jt+jGwOFEWqA/S2+v3v/tK+flhyssP029v2CVJ+mJNzayMkwHDlszmeuvtTopsXJPdcLlsKiziseaoO7f9MUebVoUr/3CQQho51WfEcXW9pkR/Ht1WkvTvF6P1p5cOaMf6MG1b20jd+xar54AiPfjbmhk+J4r99dG/onTHY4dVVOCv4uP+Gv9IjvZ/HeyeTSFJw287ql2bQ1VW6q+rrivWuEcO6x8zmqu0iHUazlc85dK8BgsagoKClJiYqJUrV+qGG25w71+5cqVGjBjRUN266HRod0xPzfjY/frOcVskSSs/aatnn0vSv9/qLLvdqXvv2qhGjSr19TdN9afHfuNeo6Gq2k9XXpGrkcO+VnBItY4eDdXGTZfotSVd5HL9lKh6+ZWr5HT66cEH1iooqFq7v2mqP07tp5LSmiDv2l8fVOPGFfpNn/36TZ/97vOOHAnTmPEj6/+DgGU0blatB184qKjoap0o9te+rGD9eXRbdwZgbbpDz//xEt1yb57u/sv3OrTXrr+Mb62dGxu5r/HStDg5ndLUlw4oKMSlzDXhemxMG4/xCh2vPKGUSbkKDnPp0B67np/SQp8sjTrn7xc4l2yGYTTYatpvvPGGUlJS9NJLLykpKUl/+9vftGDBAu3cuVOtWrU64/lFRUVyOBzqe9UfFeDPr1VcnIzNOxq6C0C9qTaq9JneUWFhoSIiIs58wlk4+V1xw8rbFBh29mNOqkortWzAK/Xa1/Ndgy7uNGrUKP3www964oknlJOTo4SEBC1fvtyrgAEAAF9QnjCvwVeEnDBhgiZMmNDQ3QAAAGfQ4EEDAADnAs+eMI+gAQBgCZQnzGvwdRoAAMCFgUwDAMASyDSYR9AAALAEggbzKE8AAACvkGkAAFgCmQbzCBoAAJZgyNy0yQZbPvk8QtAAALAEMg3mMaYBAAB4hUwDAMASyDSYR9AAALAEggbzKE8AAACvkGkAAFgCmQbzCBoAAJZgGDYZJr74zZx7saA8AQAAvEKmAQBgCS7ZTC3uZObciwVBAwDAEhjTYB7lCQAA4BUyDQAAS2AgpHkEDQAAS6A8YR5BAwDAEsg0mMeYBgAA4BUyDQAASzBMlifINBA0AAAswpBkGObOtzrKEwAAwCtkGgAAluCSTTZWhDSFoAEAYAnMnjCP8gQAAPAKmQYAgCW4DJtsLO5kCkEDAMASDMPk7AmmT1CeAAAA3iHTAACwBAZCmkfQAACwBIIG8wgaAACWwEBI8xjTAAAAvEKmAQBgCcyeMI+gAQBgCTVBg5kxDXXYmQsU5QkAAOAVMg0AAEtg9oR5BA0AAEswftzMnG91lCcAAIBXyDQAACyB8oR5BA0AAGugPmEaQQMAwBpMZhpEpoExDQAAwDtkGgAAlsCKkOYRNAAALIGBkOZRngAAAF4h0wAAsAbDZm4wI5kGggYAgDUwpsE8yhMAANSDefPmqWvXroqIiFBERISSkpL04Ycfuo8bhqFp06YpLi5OISEh6tOnj3bu3OlxjYqKCt13331q2rSpwsLCNHz4cB06dMijTUFBgVJSUuRwOORwOJSSkqLjx497tDl48KCGDRumsLAwNW3aVBMnTlRlZaXP74mgAQBgDUYdbD5o0aKFnnzySW3evFmbN2/Wb37zG40YMcIdGDz11FOaNWuW5s6dq02bNik2NlYDBgxQcXGx+xqpqalatmyZlixZojVr1qikpERDhw6V0+l0t0lOTlZmZqbS09OVnp6uzMxMpaSkuI87nU4NGTJEpaWlWrNmjZYsWaKlS5dq0qRJvr0hSTbDuHATLkVFRXI4HOp71R8V4B/c0N0B6oWxeUdDdwGoN9VGlT7TOyosLFRERES93OPkd0XLvz0qv9Cz/65wnSjXwTueMNXXqKgoPf3007r99tsVFxen1NRUPfTQQ5JqsgoxMTGaOXOm7rzzThUWFqpZs2Z69dVXNWrUKEnS4cOHFR8fr+XLl2vQoEHKyspS586dtX79evXo0UOStH79eiUlJenrr79Wx44d9eGHH2ro0KHKzs5WXFycJGnJkiW69dZblZeX59N78WpMw/PPP+/1BSdOnOh1WwAALjRFRUUer+12u+x2+y+e43Q69e9//1ulpaVKSkrSvn37lJubq4EDB3pcp3fv3lq7dq3uvPNOZWRkqKqqyqNNXFycEhIStHbtWg0aNEjr1q2Tw+FwBwyS1LNnTzkcDq1du1YdO3bUunXrlJCQ4A4YJGnQoEGqqKhQRkaG+vbt6/V79ypomD17tlcXs9lsBA0AgPNXHeTW4+PjPV4/9thjmjZt2mnbbt++XUlJSSovL1ejRo20bNkyde7cWWvXrpUkxcTEeLSPiYnRgQMHJEm5ubkKCgpSZGRkrTa5ubnuNtHR0bXuGx0d7dHm1PtERkYqKCjI3cZbXgUN+/bt8+miAACcb+pqcafs7GyPlP4vZRk6duyozMxMHT9+XEuXLtWYMWO0evVq93GbzbM/hmHU2le7H55tTtf+bNp446wHQlZWVmr37t2qrq4+20sAAHDu1NFAyJOzIU5uvxQ0BAUFqV27durevbvS0tJ0xRVX6LnnnlNsbKwk1fqln5eX584KxMbGqrKyUgUFBb/Y5siRI7Xum5+f79Hm1PsUFBSoqqqqVgbiTHwOGk6cOKGxY8cqNDRUl19+uQ4ePCipZizDk08+6evlAACwDMMwVFFRoTZt2ig2NlYrV650H6usrNTq1at1zTXXSJISExMVGBjo0SYnJ0c7duxwt0lKSlJhYaE2btzobrNhwwYVFhZ6tNmxY4dycnLcbVasWCG73a7ExESf+u9z0PDwww9r27Zt+uyzzxQc/NMo1P79++uNN97w9XIAAJwjtjrYvPenP/1JX3zxhfbv36/t27dr6tSp+uyzzzR69GjZbDalpqZqxowZWrZsmXbs2KFbb71VoaGhSk5OliQ5HA6NHTtWkyZN0ieffKKtW7fq97//vbp06aL+/ftLkjp16qTrr79e48eP1/r167V+/XqNHz9eQ4cOVceOHSVJAwcOVOfOnZWSkqKtW7fqk08+0eTJkzV+/HifZ4H4vCLk22+/rTfeeEM9e/b0qIV07txZ3333na+XAwDg3DiLtRZqne+DI0eOKCUlRTk5OXI4HOratavS09M1YMAASdKUKVNUVlamCRMmqKCgQD169NCKFSsUHh7uvsbs2bMVEBCgm2++WWVlZerXr58WLlwof39/d5vFixdr4sSJ7lkWw4cP19y5c93H/f399cEHH2jChAnq1auXQkJClJycrGeeecbnj8DndRpCQ0O1Y8cOtW3bVuHh4dq2bZvatm2rbdu26brrrlNhYaHPnThbrNMAK2CdBlzMzuU6DfHzpskvxMQ6DWXlyr57Wr329Xznc3ni6quv1gcffOB+fTLbsGDBAiUlJdVdzwAAqEvneEXIi5HP5Ym0tDRdf/312rVrl6qrq/Xcc89p586dWrduncc0EgAAzis85dI0nzMN11xzjb788kudOHFCl156qVasWKGYmBitW7fO51GYAADgwnFWj8bu0qWLFi1aVNd9AQCg3vBobPPOKmhwOp1atmyZsrKyZLPZ1KlTJ40YMUIBAWd1OQAA6t85nj1xMfL5W37Hjh0aMWKEcnNz3XNAv/nmGzVr1kzvvvuuunTpUuedBAAADc/nMQ3jxo3T5ZdfrkOHDmnLli3asmWLsrOz1bVrV91xxx310UcAAMw7ORDSzGZxPmcatm3bps2bN3s8dSsyMlLTp0/X1VdfXaedAwCgrtiMms3M+Vbnc6ahY8eOp304Rl5entq1a1cnnQIAoM6xToNpXgUNRUVF7m3GjBmaOHGi/vOf/+jQoUM6dOiQ/vOf/yg1NVUzZ86s7/4CAIAG4lV5onHjxh7PmTAMQzfffLN738mVqIcNGyan01kP3QQAwCQWdzLNq6Dh008/re9+AABQv5hyaZpXQUPv3r3rux8AAOA8d9arMZ04cUIHDx5UZWWlx/6uXbua7hQAAHWOTINpPgcN+fn5uu222/Thhx+e9jhjGgAA5yWCBtN8nnKZmpqqgoICrV+/XiEhIUpPT9eiRYvUvn17vfvuu/XRRwAAcB7wOdOwatUqvfPOO7r66qvl5+enVq1aacCAAYqIiFBaWpqGDBlSH/0EAMAcZk+Y5nOmobS0VNHR0ZKkqKgo5efnS6p58uWWLVvqtncAANSRkytCmtms7qxWhNy9e7ck6corr9T8+fP1/fff66WXXlLz5s3rvIMAAOD84HN5IjU1VTk5OZKkxx57TIMGDdLixYsVFBSkhQsX1nX/AACoGwyENM3noGH06NHu/92tWzft379fX3/9tVq2bKmmTZvWaecAAMD546zXaTgpNDRUV111VV30BQCAemOTyadc1llPLlxeBQ0PPPCA1xecNWvWWXcGAACcv7wKGrZu3erVxf77oVbnkrElS4YtsEHuDdS3jw5nNnQXgHpTVOxSZIdzdDOmXJrGA6sAANbAQEjTfJ5yCQAArMn0QEgAAC4IZBpMI2gAAFiC2VUdWRGS8gQAAPASmQYAgDVQnjDtrDINr776qnr16qW4uDgdOHBAkjRnzhy98847ddo5AADqjFEHm8X5HDTMmzdPDzzwgP7nf/5Hx48fl9PplCQ1btxYc+bMqev+AQCA84TPQcMLL7ygBQsWaOrUqfL393fv7969u7Zv316nnQMAoK7waGzzfB7TsG/fPnXr1q3WfrvdrtLS0jrpFAAAdY4VIU3zOdPQpk0bZWZm1tr/4YcfqnPnznXRJwAA6h5jGkzzOdPw4IMP6p577lF5ebkMw9DGjRv1r3/9S2lpaXr55Zfro48AAOA84HPQcNttt6m6ulpTpkzRiRMnlJycrEsuuUTPPfecbrnllvroIwAAprG4k3lntU7D+PHjNX78eB09elQul0vR0dF13S8AAOoW6zSYZmpxp6ZNm9ZVPwAAwHnO56ChTZs2stl+fgTp3r17TXUIAIB6YXbaJJkG34OG1NRUj9dVVVXaunWr0tPT9eCDD9ZVvwAAqFuUJ0zzOWi4//77T7v///7v/7R582bTHQIAAOenOnvK5eDBg7V06dK6uhwAAHWLdRpMq7OnXP7nP/9RVFRUXV0OAIA6xZRL83wOGrp16+YxENIwDOXm5io/P18vvvhinXYOAACcP3wOGkaOHOnx2s/PT82aNVOfPn102WWX1VW/AADAecanoKG6ulqtW7fWoEGDFBsbW199AgCg7jF7wjSfBkIGBATo7rvvVkVFRX31BwCAesGjsc3zefZEjx49tHXr1vroCwAAOI/5PKZhwoQJmjRpkg4dOqTExESFhYV5HO/atWuddQ4AgDpFtsAUr4OG22+/XXPmzNGoUaMkSRMnTnQfs9lsMgxDNptNTqez7nsJAIBZjGkwzeugYdGiRXryySe1b9+++uwPAAA4T3kdNBhGTYjVqlWreusMAAD1hcWdzPNpTMMvPd0SAIDzGuUJ03wKGjp06HDGwOHYsWOmOgQAAM5PPgUNjz/+uBwOR331BQCAekN5wjyfgoZbbrlF0dHR9dUXAADqD+UJ07xe3InxDAAAWJvPsycAALggkWkwzeugweVy1Wc/AACoV4xpMM/nZaQBALggkWkwzecHVgEAAGsi0wAAsAYyDaYRNAAALIExDeZRngAAAF4h0wAAsAbKE6YRNAAALIHyhHmUJwAAgFfINAAArIHyhGkEDQAAayBoMI3yBAAA8ApBAwDAEmx1sPkiLS1NV199tcLDwxUdHa2RI0dq9+7dHm0Mw9C0adMUFxenkJAQ9enTRzt37vRoU1FRofvuu09NmzZVWFiYhg8frkOHDnm0KSgoUEpKihwOhxwOh1JSUnT8+HGPNgcPHtSwYcMUFhampk2bauLEiaqsrPTpPRE0AACswaiDzQerV6/WPffco/Xr12vlypWqrq7WwIEDVVpa6m7z1FNPadasWZo7d642bdqk2NhYDRgwQMXFxe42qampWrZsmZYsWaI1a9aopKREQ4cOldPpdLdJTk5WZmam0tPTlZ6erszMTKWkpLiPO51ODRkyRKWlpVqzZo2WLFmipUuXatKkST69J5txAT/zuqioSA6HQ300QgG2wIbuDlAvPjqc2dBdAOpNUbFLkR32qrCwUBEREfVzjx+/Ky6/a4b87cFnfR1nRbl2vvQnZWdne/TVbrfLbref8fz8/HxFR0dr9erVuu6662QYhuLi4pSamqqHHnpIUk1WISYmRjNnztSdd96pwsJCNWvWTK+++qpGjRolSTp8+LDi4+O1fPlyDRo0SFlZWercubPWr1+vHj16SJLWr1+vpKQkff311+rYsaM+/PBDDR06VNnZ2YqLi5MkLVmyRLfeeqvy8vK8/uzJNAAA4IP4+Hh3GcDhcCgtLc2r8woLCyVJUVFRkqR9+/YpNzdXAwcOdLex2+3q3bu31q5dK0nKyMhQVVWVR5u4uDglJCS426xbt04Oh8MdMEhSz5495XA4PNokJCS4AwZJGjRokCoqKpSRkeH1e2f2BADAGupo9sTpMg1nPNUw9MADD+jXv/61EhISJEm5ubmSpJiYGI+2MTExOnDggLtNUFCQIiMja7U5eX5ubq6io6Nr3TM6Otqjzan3iYyMVFBQkLuNNwgaAADWUQcF+YiICJ9LKffee6+++uorrVmzptYxm81ziKVhGLX2nerUNqdrfzZtzoTyBAAA9ei+++7Tu+++q08//VQtWrRw74+NjZWkWr/08/Ly3FmB2NhYVVZWqqCg4BfbHDlypNZ98/PzPdqcep+CggJVVVXVykD8EoIGAIAlnHz2hJnNF4Zh6N5779Vbb72lVatWqU2bNh7H27Rpo9jYWK1cudK9r7KyUqtXr9Y111wjSUpMTFRgYKBHm5ycHO3YscPdJikpSYWFhdq4caO7zYYNG1RYWOjRZseOHcrJyXG3WbFihex2uxITE71+T5QnAADWcI5XhLznnnv0+uuv65133lF4eLj7l77D4VBISIhsNptSU1M1Y8YMtW/fXu3bt9eMGTMUGhqq5ORkd9uxY8dq0qRJatKkiaKiojR58mR16dJF/fv3lyR16tRJ119/vcaPH6/58+dLku644w4NHTpUHTt2lCQNHDhQnTt3VkpKip5++mkdO3ZMkydP1vjx430qtRA0AABQD+bNmydJ6tOnj8f+V155RbfeeqskacqUKSorK9OECRNUUFCgHj16aMWKFQoPD3e3nz17tgICAnTzzTerrKxM/fr108KFC+Xv7+9us3jxYk2cONE9y2L48OGaO3eu+7i/v78++OADTZgwQb169VJISIiSk5P1zDPP+PSeWKcBOM+xTgMuZudynYYu42bIP8jEOg2V5dr+8p/qta/nOzINAABr4IFVpjEQEgAAeIVMAwDAEs5mBsSp51sdQQMAwBooT5hG0AAAsAaCBtMY0wAAALxCpgEAYAmMaTCPoAEAYA2UJ0yjPAEAALxCpgEAYAk2w5DNxCLIZs69WBA0AACsgfKEaZQnAACAV8g0AAAsgdkT5hE0AACsgfKEaZQnAACAV8g0AAAsgfKEeQQNAABroDxhGkEDAMASyDSYx5gGAADgFTINAABroDxhGkEDAMAyKDGYQ3kCAAB4hUwDAMAaDKNmM3O+xRE0AAAsgdkT5lGeAAAAXiHTAACwBmZPmEbQAACwBJurZjNzvtVRngAAAF4h02AxQ/9wVEP+8INi4islSQd2B2vx7Bht/jRCkjRp9kENHFXgcU5WRqhSh7V3v544M1vdri1Rk5gqlZ3wU9bmMP19enNl7wmudb/AIJee++BbXXp5ue4e0EF7d4bU47uD1S15IVqvpMVp5Lh83f3E95KkgvwA/X16nDJWh6u00F8JPUt0z18P6ZK2le7zlr/WRJ8ui9Se7SE6UeKvpVnb1cjh9Lj268/FaOPHEdq7M0QBQYbe+nq7x/Hvdgbrzbkx2rExTEUFAYppUakhfziqG8Ydrf83Du9QnjCNoMFi8nMC9Y8ZzXV4v12SNOCmY5r2yn7dM7CDDnxT86W/aVW4nv3fePc51VU2j2t8+1WoVr0VqfzvgxQeWa3fTzqiGf/aqzE9Osnl8mw79s85+iE3UJdeXl7P7wxWtzszRMtfa6I2ncvc+wxDevz2NvIPMDTtlb0KbeTSW39rpj+OaqcFq79WcGhNvrm8zE/d+xSpe58i/SMt7rTXr6606bphx9Wpe6k++leTWsf3fBUqR5NqPTT3gJrFVWnX5jA992C8/PykEbcTOJwPmD1hXoOWJz7//HMNGzZMcXFxstlsevvttxuyO5awYaVDm1ZF6Pu9dn2/166FM5urvNRPlyWWuttUVdpUkB/o3oqPe8aWHy5uoh0bGunIoSDt2R6qRTNjFX1JlTt7cVL3vkVK7F2sBU+c/o8wUFfKSv00895WSn06W+H/lSH4fq9dWRlhuu/JQ+p4ZZni21Xo3rRDKjvhp0+XNXa3u3F8vkbdl6fLEk/87D3+8GCubrwjX20uO30APOh3xzThr9+ra1KpmreqVL//V6CBo37Qlx866ux9wqST6zSY2SyuQYOG0tJSXXHFFZo7d25DdsOy/PwM9R5RIHuoS1mbw9z7uyaV6I2vdurvX2Qp9elsOZpU/ew17CFODRx1TDkHgpR/ONC9v3HTKqU+fUhP3ddSFWUMnUH9mvunFvpVvyJddV2Jx/6qyprMV5D9pxFs/v5SYKChnZsa1Xu/Sov9Fd7YeeaGwAWiQcsTgwcP1uDBg71uX1FRoYqKCvfroqKi+ujWRa/1ZWWa894eBdldKiv10xNjW+vgtzWlic2fhuuL9xvryKFAxbas1JgpuXrq33t17/XtVVX505f/0DFHNe7POQoJc+ngt3Y9fEtbVVedPG5o8pxsffBqE337VahiWlSephdA3fjs7cbasz1ELyz/ptax+HblimlRqX+kNdf9Mw8pONSlt+Y307G8QB07Ur9//nZtDtXn7zXWE//cW6/3gfcoT5h3Qf0ETEtLk8PhcG/x8fFnPgm1HPrOrgkDOuj+oe31/j+bavJzB9WyfU3KdfW7kdr4SYQO7A7RhpUO/Xl0W13StkK/6ucZoK16K1ITBnbQpBsu1ff77Jo6/4ACf/w1N2LsUYWGO/XGC9Hn/L3BWvK+D9S8Ry/RlBcOKCi49l/0gEDpkZf36fvvgvXbzl00/NKu2rauka7+TZH8/OuvX/t3B2vabW00+n+PKLF3yZlPwLlh1MFmcRfUQMiHH35YDzzwgPt1UVERgcNZqK7ycw+E/ParUHW88oRGjsvX8w/V/iyP5QUq71Cgx0hzSTpR7K8Txf46vM+ur7eEamnWTvUaXKjP3o7Ulb1KdNlVJ/T+/q88zpn74Tda9VaknkltWX9vDpay56tQHT8aqHuv7+je53LatH19mN59pane379N7buWad7Hu1Va5KeqKpsaN3Fq4pD26tD158cvmHHgG7seuulSDR79g5JTj9TLPYCGckEFDXa7XXa7vaG7cVEKDDp9CB0eWa1mcVVnTuXaDPc1XnzkEi2cGes+1CS2Wmn/2qsZd7XS11tD66zPwJXXFmv+qq899j37vy0V365cN9+TJ///yiaERdRkwr7fG6Rvt4VqzIO5dd6f/buD9dBNl2rATcd02x/r/vowh/KEeRdU0ADzbvtjjjatClf+4SCFNHKqz4jj6npNif48uq2CQ51KmXxEaz5w6NiRQMXEV+q2h3NUeCzAPQI8tmWFeg8/rozV4So8FqCmsVW6+Z48VZb5aeMn4ZKk/O+DPO5ZXlqTpTh8wK6jOZ7HADNCG7nU+pTZDMGhLoVHOt37P3/PIUcTp6IvqdS+rGC99GgLJV1fqMQ+xe5zjuUFqCAvUIf31fz73Pd1sELDXGp2SaUiImsGMuYdqplJlPd9oFxO6bsdNWuOxLWpUEiYS/t3B2vKby9VYu9i3Xhnvo7l1fx59fM31LgJgyHPCzzl0jSCBotp3KxaD75wUFHR1TpR7K99WcH68+i22vJ5uIKCXWp9WZn6/7ZAYRFOHcsL0LYvG2nGXa1UVlrzk62ywk8JPUp1w/ijauRw6vjRAG1fH6b/HdFOhT8EnuHuwLl37Eig5k+7RMePBigqulr9bzpWq2zwwT+b6rVZP2XHJt9Qs5hZzWJnxyRJ/3ymuVa+GeVuM2FgTUnkqf/s0RXXlOiL9xqr8IdArXorSqve+qldTItK/XPjrnp7f8C5ZDOMhgudSkpKtGfPHklSt27dNGvWLPXt21dRUVFq2fLMde+ioiI5HA710QgF2PjCwsXpo8OZDd0FoN4UFbsU2WGvCgsLFRERUT/3+PG7ImnwEwoIrL1yrbeqq8q17sNH67Wv57sGzTRs3rxZffv2db8+OchxzJgxWrhwYQP1CgBwUWIZadMaNGjo06ePGjDRAQAAfMCYBgCAJTB7wjyCBgCANbiMms3M+RZH0AAAsAbGNJh2QS0jDQAAGg6ZBgCAJdhkckxDnfXkwkXQAACwBlaENI3yBAAA8AqZBgCAJTDl0jyCBgCANTB7wjTKEwAAwCtkGgAAlmAzDNlMDGY0c+7FgqABAGANrh83M+dbHOUJAADgFTINAABLoDxhHkEDAMAamD1hGkEDAMAaWBHSNMY0AAAAr5BpAABYAitCmkfQAACwBsoTplGeAAAAXiHTAACwBJurZjNzvtURNAAArIHyhGmUJwAAgFfINAAArIHFnUwjaAAAWALLSJtHeQIAAHiFTAMAwBoYCGkaQQMAwBoMSWamTRIzEDQAAKyBMQ3mMaYBAIB68Pnnn2vYsGGKi4uTzWbT22+/7XHcMAxNmzZNcXFxCgkJUZ8+fbRz506PNhUVFbrvvvvUtGlThYWFafjw4Tp06JBHm4KCAqWkpMjhcMjhcCglJUXHjx/3aHPw4EENGzZMYWFhatq0qSZOnKjKykqf3xNBAwDAGgz9NK7hrDbfbldaWqorrrhCc+fOPe3xp556SrNmzdLcuXO1adMmxcbGasCAASouLna3SU1N1bJly7RkyRKtWbNGJSUlGjp0qJxOp7tNcnKyMjMzlZ6ervT0dGVmZiolJcV93Ol0asiQISotLdWaNWu0ZMkSLV26VJMmTfLtDYnyBADAKupoIGRRUZHHbrvdLrvdXqv54MGDNXjw4J+5lKE5c+Zo6tSpuvHGGyVJixYtUkxMjF5//XXdeeedKiws1N///ne9+uqr6t+/vyTptddeU3x8vD7++GMNGjRIWVlZSk9P1/r169WjRw9J0oIFC5SUlKTdu3erY8eOWrFihXbt2qXs7GzFxcVJkp599lndeuutmj59uiIiIrz+CMg0AADgg/j4eHcpwOFwKC0tzedr7Nu3T7m5uRo4cKB7n91uV+/evbV27VpJUkZGhqqqqjzaxMXFKSEhwd1m3bp1cjgc7oBBknr27CmHw+HRJiEhwR0wSNKgQYNUUVGhjIwMn/pNpgEAYA0uSTaT50vKzs72+HV+uizDmeTm5kqSYmJiPPbHxMTowIED7jZBQUGKjIys1ebk+bm5uYqOjq51/ejoaI82p94nMjJSQUFB7jbeImgAAFhCXc2eiIiI8Cml/4vXtHlGMYZh1Np3qlPbnK792bTxBuUJAADOsdjYWEmq9Us/Ly/PnRWIjY1VZWWlCgoKfrHNkSNHal0/Pz/fo82p9ykoKFBVVVWtDMSZEDQAAKzB1MwJk4MoT9GmTRvFxsZq5cqV7n2VlZVavXq1rrnmGklSYmKiAgMDPdrk5ORox44d7jZJSUkqLCzUxo0b3W02bNigwsJCjzY7duxQTk6Ou82KFStkt9uVmJjoU78pTwAArOEcLyNdUlKiPXv2uF/v27dPmZmZioqKUsuWLZWamqoZM2aoffv2at++vWbMmKHQ0FAlJydLkhwOh8aOHatJkyapSZMmioqK0uTJk9WlSxf3bIpOnTrp+uuv1/jx4zV//nxJ0h133KGhQ4eqY8eOkqSBAweqc+fOSklJ0dNPP61jx45p8uTJGj9+vM9lFoIGAADqwebNm9W3b1/36wceeECSNGbMGC1cuFBTpkxRWVmZJkyYoIKCAvXo0UMrVqxQeHi4+5zZs2crICBAN998s8rKytSvXz8tXLhQ/v7+7jaLFy/WxIkT3bMshg8f7rE2hL+/vz744ANNmDBBvXr1UkhIiJKTk/XMM8/4/J5shnHhrotZVFQkh8OhPhqhAFtgQ3cHqBcfHc5s6C4A9aao2KXIDntVWFhYZ4MLa93jx++Kfp0mKcDf95kOJ1U7K/RJ1rP12tfzHZkGAIA11NGUSysjaAAAWAIPrDKP2RMAAMArZBoAANZwjmdPXIwIGgAA1uAyJJuJL34XQQPlCQAA4BUyDQAAa6A8YRpBAwDAIswuBU3QQHkCAAB4hUwDAMAaKE+YRtAAALAGlyFTJQZmT1CeAAAA3iHTAACwBsNVs5k53+IIGgAA1sCYBtMIGgAA1sCYBtMY0wAAALxCpgEAYA2UJ0wjaAAAWIMhk0FDnfXkgkV5AgAAeIVMAwDAGihPmEbQAACwBpdLkom1Flys00B5AgAAeIVMAwDAGihPmEbQAACwBoIG0yhPAAAAr5BpAABYA8tIm0bQAACwBMNwyTDxpEoz514sCBoAANZgGOayBYxpYEwDAADwDpkGAIA1GCbHNJBpIGgAAFiEyyXZTIxLYEwD5QkAAOAdMg0AAGugPGEaQQMAwBIMl0uGifIEUy4pTwAAAC+RaQAAWAPlCdMIGgAA1uAyJBtBgxmUJwAAgFfINAAArMEwJJlZp4FMA0EDAMASDJchw0R5wiBoIGgAAFiE4ZK5TANTLhnTAAAAvEKmAQBgCZQnzCNoAABYA+UJ0y7ooOFk1FetKlPrdQDns6Ji/lDh4lVUUvPv+1z8ijf7XVGtqrrrzAXqgg4aiouLJUlrtLyBewLUn8gODd0DoP4VFxfL4XDUy7WDgoIUGxurNbnmvytiY2MVFBRUB726MNmMC7hI43K5dPjwYYWHh8tmszV0dyyhqKhI8fHxys7OVkREREN3B6hT/Ps+9wzDUHFxseLi4uTnV39j88vLy1VZWWn6OkFBQQoODq6DHl2YLuhMg5+fn1q0aNHQ3bCkiIgI/qjiosW/73OrvjIM/y04ONjSX/Z1hSmXAADAKwQNAADAKwQN8Indbtdjjz0mu93e0F0B6hz/voFfdkEPhAQAAOcOmQYAAOAVggYAAOAVggYAAOAVggYAAOAVggZ47cUXX1SbNm0UHBysxMREffHFFw3dJaBOfP755xo2bJji4uJks9n09ttvN3SXgPMSQQO88sYbbyg1NVVTp07V1q1bde2112rw4ME6ePBgQ3cNMK20tFRXXHGF5s6d29BdAc5rTLmEV3r06KGrrrpK8+bNc+/r1KmTRo4cqbS0tAbsGVC3bDabli1bppEjRzZ0V4DzDpkGnFFlZaUyMjI0cOBAj/0DBw7U2rVrG6hXAIBzjaABZ3T06FE5nU7FxMR47I+JiVFubm4D9QoAcK4RNMBrpz5+3DAMHkkOABZC0IAzatq0qfz9/WtlFfLy8mplHwAAFy+CBpxRUFCQEhMTtXLlSo/9K1eu1DXXXNNAvQIAnGsBDd0BXBgeeOABpaSkqHv37kpKStLf/vY3HTx4UHfddVdDdw0wraSkRHv27HG/3rdvnzIzMxUVFaWWLVs2YM+A8wtTLuG1F198UU899ZRycnKUkJCg2bNn67rrrmvobgGmffbZZ+rbt2+t/WPGjNHChQvPfYeA8xRBAwAA8ApjGgAAgFcIGgAAgFcIGgAAgFcIGgAAgFcIGgAAgFcIGgAAgFcIGgAAgFcIGgAAgFcIGgCTpk2bpiuvvNL9+tZbb9XIkSPPeT/2798vm82mzMzMn23TunVrzZkzx+trLly4UI0bNzbdN5vNprffftv0dQA0LIIGXJRuvfVW2Ww22Ww2BQYGqm3btpo8ebJKS0vr/d7PPfec10sPe/NFDwDnCx5YhYvW9ddfr1deeUVVVVX64osvNG7cOJWWlmrevHm12lZVVSkwMLBO7utwOOrkOgBwviHTgIuW3W5XbGys4uPjlZycrNGjR7tT5CdLCv/4xz/Utm1b2e12GYahwsJC3XHHHYqOjlZERIR+85vfaNu2bR7XffLJJxUTE6Pw8HCNHTtW5eXlHsdPLU+4XC7NnDlT7dq1k91uV8uWLTV9+nRJUps2bSRJ3bp1k81mU58+fdznvfLKK+rUqZOCg4N12WWX6cUXX/S4z8aNG9WtWzcFBwere/fu2rp1q8+f0axZs9SlSxeFhYUpPj5eEyZMUElJSa12b7/9tjp06KDg4GANGDBA2dnZHsffe+89JSYmKjg4WG3bttXjjz+u6upqn/sD4PxG0ADLCAkJUVVVlfv1nj179Oabb2rp0qXu8sCQIUOUm5ur5cuXKyMjQ1dddZX69eunY8eOSZLefPNNPfbYY5o+fbo2b96s5s2b1/oyP9XDDz+smTNn6pFHHtGuXbv0+uuvKyYmRlLNF78kffzxx8rJydFbb70lSVqwYIGmTp2q6dOnKysrSzNmzNAjjzyiRYsWSZJKS0s1dOhQdezYURkZGZo2bZomT57s82fi5+en559/Xjt27NCiRYu0atUqTZkyxaPNiRMnNH36dC1atEhffvmlioqKdMstt7iPf/TRR/r973+viRMnateuXZo/f74WLlzoDowAXEQM4CI0ZswYY8SIEe7XGzZsMJo0aWLcfPPNhmEYxmOPPWYEBgYaeXl57jaffPKJERERYZSXl3tc69JLLzXmz59vGIZhJCUlGXfddZfH8R49ehhXXHHFae9dVFRk2O12Y8GCBaft5759+wxJxtatWz32x8fHG6+//rrHvr/85S9GUlKSYRiGMX/+fCMqKsooLS11H583b95pr/XfWrVqZcyePftnj7/55ptGkyZN3K9feeUVQ5Kxfv16976srCxDkrFhwwbDMAzj2muvNWbMmOFxnVdffdVo3ry5+7UkY9myZT97XwAXBsY04KL1/vvvq1GjRqqurlZVVZVGjBihF154wX28VatWatasmft1RkaGSkpK1KRJE4/rlJWV6bvvvpMkZWVl6a677vI4npSUpE8//fS0fcjKylJFRYX69evndb/z8/OVnZ2tsWPHavz48e791dXV7vESWVlZuuKKKxQaGurRD199+umnmjFjhnbt2qWioiJVV1ervLxcpaWlCgsLkyQFBASoe/fu7nMuu+wyNW7cWFlZWfrVr36ljIwMbdq0ySOz4HQ6VV5erhMnTnj0EcCFjaABF62+fftq3rx5CgwMVFxcXK2Bjie/FE9yuVxq3ry5Pvvss1rXOttphyEhIT6f43K5JNWUKHr06OFxzN/fX5JkGMZZ9ee/HThwQP/zP/+ju+66S3/5y18UFRWlNWvWaOzYsR5lHKlmyuSpTu5zuVx6/PHHdeONN9ZqExwcbLqfAM4fBA24aIWFhaldu3Zet7/qqquUm5urgIAAtW7d+rRtOnXqpPXr1+sPf/iDe9/69et/9prt27dXSEiIPvnkE40bN67W8aCgIEk1v8xPiomJ0SWXXKK9e/dq9OjRp71u586d9eqrr6qsrMwdmPxSP05n8+bNqq6u1rPPPis/v5rhTW+++WatdtXV1dq8ebN+9atfSZJ2796t48eP67LLLpNU87nt3r3bp88awIWJoAH4Uf/+/ZWUlKSRI0dq5syZ6tixow4fPqzly5dr5MiR6t69u+6//36NGTNG3bt3169//WstXrxYO3fuVNu2bU97zeDgYD300EOaMmWKgoKC1KtXL+Xn52vnzp0aO3asoqOjFRISovT0dLVo0ULBwcFyOByaNm2aJk6cqIiICA0ePFgVFRXavHmzCgoK9MADDyg5OVlTp07V2LFj9ec//1n79+/XM88849P7vfTSS1VdXa0XXnhBw4YN05dffqmXXnqpVrvAwEDdd999ev755xUYGKh7771XPXv2dAcRjz76qIYOHar4+HjddNNN8vPz01dffaXt27frr3/9q+//IQCct5g9AfzIZrNp+fLluu6663T77berQ4cOuuWWW7R//373bIdRo0bp0Ucf1UMPPaTExEQdOHBAd9999y9e95FHHtGkSZP06KOPqlOnTho1apTy8vIk1YwXeP755zV//nzFxcVpxIgRkqRx48bp5Zdf1sKFC9WlSxf17t1bCxcudE/RbNSokd577z3t2rVL3bp109SpUzVz5kyf3u+VV16pWbNmaebMmUpISNDixYuVlpZWq11oaKgeeughJScnKykpSSEhIVqyZIn7+KBBg/T+++9r5cqVuvrqq9WzZ0/NmjVLrVq18qk/AM5/NqMuiqMAAOCiR6YBAAB4haABAAB4haABAAB4haABAAB4haABAAB4haABAAB4haABAAB4haABAAB4haABAAB4haABAAB4haABAAB45f8D7pIQGCCbyHsAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "val_pred_df['final_preds'] = val_pred_df.pred_prob.apply(lambda x : 1 if x>best_cutoff else 0)\n",
    "val_pred_df.head()\n",
    "\n",
    "val_acc = metrics.accuracy_score(val_pred_df['actual'], val_pred_df['final_preds'])\n",
    "\n",
    "val_f1 = metrics.f1_score(val_pred_df['actual'], val_pred_df['final_preds'])\n",
    "\n",
    "val_cm = confusion_matrix(val_pred_df['actual'], val_pred_df['final_preds'], labels=classes)\n",
    "ConfusionMatrixDisplay(val_cm, display_labels=classes).plot()\n",
    "\n",
    "\n",
    "val_tp = val_cm[1,1]\n",
    "val_tn = val_cm[0,0]\n",
    "val_fp = val_cm[0,1]\n",
    "val_fn = val_cm[1,0]\n",
    "\n",
    "val_sensitivity = val_tp/float(val_tp + val_fn)\n",
    "val_specificity = val_tp/float(val_tp + val_fp)\n",
    "val_precision = precision_score(val_pred_df['actual'], val_pred_df['final_preds'])\n",
    "\n",
    "print('val acc: {:.2f}'.format(val_acc))\n",
    "print('val f1: {:.2f}'.format(val_f1))\n",
    "\n",
    "print('\\nValidation ==> Sensitivity/Recall: {:.2f} | Specificity: {:.2f}'.format(val_sensitivity, val_specificity))\n",
    "\n",
    "print('Validation ==> Precision: {:.2f} | Recall: {:.2f}\\n'.format(val_precision,val_sensitivity))\n",
    "\n",
    "print(val_cm)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "32ff0dda-6db6-4845-b0d4-61bb7c336229",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}