revolut_fraudsters_detection.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import datetime as dt\n",
    "from sklearn.model_selection import StratifiedKFold, cross_validate, train_test_split\n",
    "from sklearn.tree import DecisionTreeClassifier, plot_tree\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "sns.set_style('darkgrid', {'axes.facecolor': '0.9'})\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_users = pd.read_csv('users.csv')\n",
    "df_transactions = pd.read_csv('transactions.csv')\n",
    "df_countries = pd.read_csv('countries.csv')\n",
    "df_currency_details = pd.read_csv('currency_details.csv')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CURRENCY</th>\n",
       "      <th>AMOUNT</th>\n",
       "      <th>STATE</th>\n",
       "      <th>CREATED_DATE</th>\n",
       "      <th>MERCHANT_CATEGORY</th>\n",
       "      <th>MERCHANT_COUNTRY</th>\n",
       "      <th>ENTRY_METHOD</th>\n",
       "      <th>USER_ID</th>\n",
       "      <th>TYPE</th>\n",
       "      <th>SOURCE</th>\n",
       "      <th>ID</th>\n",
       "      <th>AMOUNT_USD</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GBP</td>\n",
       "      <td>4420</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:38:55.577</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NLD</td>\n",
       "      <td>chip</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>ATM</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>367bf5f9-7cce-4683-90b9-d3c011bf4c87</td>\n",
       "      <td>3268.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PLN</td>\n",
       "      <td>1500</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:37:24.792</td>\n",
       "      <td>point_of_interest</td>\n",
       "      <td>POL</td>\n",
       "      <td>manu</td>\n",
       "      <td>76cbaad3-4721-4a3b-92b9-3eb9e9319565</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>ff6802b9-360d-4efe-b09b-f99c6cac3383</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>GBP</td>\n",
       "      <td>191</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:37:16.234</td>\n",
       "      <td>airport</td>\n",
       "      <td>PRT</td>\n",
       "      <td>chip</td>\n",
       "      <td>7bcaa34e-b889-4582-9c29-0b3bab34fb8c</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>ddb4a930-7d8a-4f38-9079-ddc4b0db980e</td>\n",
       "      <td>141.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>GBP</td>\n",
       "      <td>10000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:34:42.592</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>b3a6762b-a940-4459-bb1d-8e28e151f901</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>080ef0be-850a-4f14-ab03-28ff68e6b797</td>\n",
       "      <td>7394.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>EUR</td>\n",
       "      <td>1159</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:34:17.88</td>\n",
       "      <td>NaN</td>\n",
       "      <td>FRA</td>\n",
       "      <td>manu</td>\n",
       "      <td>f5723f5e-5a00-4851-ae1d-5c11146aa491</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>bd13abac-4004-4c67-9bb7-90a42b027e82</td>\n",
       "      <td>977.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  CURRENCY  AMOUNT      STATE             CREATED_DATE  MERCHANT_CATEGORY  \\\n",
       "0      GBP    4420  COMPLETED  2017-12-10 16:38:55.577                NaN   \n",
       "1      PLN    1500  COMPLETED  2017-12-10 16:37:24.792  point_of_interest   \n",
       "2      GBP     191  COMPLETED  2017-12-10 16:37:16.234            airport   \n",
       "3      GBP   10000  COMPLETED  2017-12-10 16:34:42.592                NaN   \n",
       "4      EUR    1159  COMPLETED   2017-12-10 16:34:17.88                NaN   \n",
       "\n",
       "  MERCHANT_COUNTRY ENTRY_METHOD                               USER_ID  \\\n",
       "0              NLD         chip  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "1              POL         manu  76cbaad3-4721-4a3b-92b9-3eb9e9319565   \n",
       "2              PRT         chip  7bcaa34e-b889-4582-9c29-0b3bab34fb8c   \n",
       "3              NaN         misc  b3a6762b-a940-4459-bb1d-8e28e151f901   \n",
       "4              FRA         manu  f5723f5e-5a00-4851-ae1d-5c11146aa491   \n",
       "\n",
       "           TYPE SOURCE                                    ID  AMOUNT_USD  \n",
       "0           ATM   GAIA  367bf5f9-7cce-4683-90b9-d3c011bf4c87      3268.0  \n",
       "1  CARD_PAYMENT   GAIA  ff6802b9-360d-4efe-b09b-f99c6cac3383         NaN  \n",
       "2  CARD_PAYMENT   GAIA  ddb4a930-7d8a-4f38-9079-ddc4b0db980e       141.0  \n",
       "3         TOPUP   HERA  080ef0be-850a-4f14-ab03-28ff68e6b797      7394.0  \n",
       "4  CARD_PAYMENT   GAIA  bd13abac-4004-4c67-9bb7-90a42b027e82       977.0  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_transactions.head()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- All datasets have column names in all caps. Let's convert them to lower case."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "for df in [df_users, df_transactions, df_currency_details, df_countries]:\n",
    "    df.columns = df.columns.str.lower()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Users\n",
    "- Useful features\n",
    "    - has_email, kyc, failed_sign_in_attempts\n",
    "- Create features:\n",
    "    - age of account (from created_date)\n",
    "- Numerical features:\n",
    "    - `has_email`, `birth_year`, `failed_sign_in_attempts`\n",
    "### Transactions\n",
    "- transaction_state = completed or failed can determine fraudster or not.\n",
    "- Create features:\n",
    "    - number of failed transactions\n",
    "    - number of transactions above median transaction value\n",
    "- Numerical features:\n",
    "    - `amount`, `amount_usd`\n",
    "### Currency Details\n",
    "- has a `IS_CRYPTO` column which might be useful?\n",
    "### Country\n",
    "- Let's have complete name of the country, so merge this table too.\n",
    "\n",
    "## Initial EDA on df_transactions\n",
    "- Let's create some aggregate metrics on the transactions dataset grouped by `USER_ID`. Let's call this `df_transaction_summary`\n",
    "- Specifically, create metrics specified [here](https://fraud-detection-handbook.github.io/fraud-detection-handbook/)\n",
    "- (amount_usd).\n",
    "    - `amount_usd` has missing values. Impute them using the average or median per user.\n",
    "    - Or use np.nansum()\n",
    "## Feature Engineering\n",
    "- A feature for whether a transaction occurs during a weekday or weekend.\n",
    "- A feature for whether a transaction occurs during the day or night.\n",
    "- *These features are useful because fraudulent patterns differ between weekdays and weekends, and between day and night.\n",
    "- Customer Spending behaviors using Recency, Frequency, Monetary value (RFM framework)\n",
    "    - Average spending amount and number of transactions for each customer for 3 window sizes. This will give us 6 new features.\n",
    "- Terminal 'risk' characteristics\n",
    "    - We don't have terminal IDs or any terminal identifiers, but if we did:\n",
    "    - We'll define risk as the average number of frauds that were observed on the terminal for 3 window sizes.\n",
    "\n",
    "## Initial EDA on df_users\n",
    "- Transform phone_country into an array of countries. Create a boolean to check whether country belongs in the phone_country array."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Date and time transformations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>currency</th>\n",
       "      <th>amount</th>\n",
       "      <th>state</th>\n",
       "      <th>created_date</th>\n",
       "      <th>merchant_category</th>\n",
       "      <th>merchant_country</th>\n",
       "      <th>entry_method</th>\n",
       "      <th>user_id</th>\n",
       "      <th>type</th>\n",
       "      <th>source</th>\n",
       "      <th>id</th>\n",
       "      <th>amount_usd</th>\n",
       "      <th>tx_time</th>\n",
       "      <th>tx_date</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GBP</td>\n",
       "      <td>4420</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:38:55.577</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NLD</td>\n",
       "      <td>chip</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>ATM</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>367bf5f9-7cce-4683-90b9-d3c011bf4c87</td>\n",
       "      <td>3268.0</td>\n",
       "      <td>16:38:55.577000</td>\n",
       "      <td>2017-12-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PLN</td>\n",
       "      <td>1500</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:37:24.792</td>\n",
       "      <td>point_of_interest</td>\n",
       "      <td>POL</td>\n",
       "      <td>manu</td>\n",
       "      <td>76cbaad3-4721-4a3b-92b9-3eb9e9319565</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>ff6802b9-360d-4efe-b09b-f99c6cac3383</td>\n",
       "      <td>NaN</td>\n",
       "      <td>16:37:24.792000</td>\n",
       "      <td>2017-12-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>GBP</td>\n",
       "      <td>191</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:37:16.234</td>\n",
       "      <td>airport</td>\n",
       "      <td>PRT</td>\n",
       "      <td>chip</td>\n",
       "      <td>7bcaa34e-b889-4582-9c29-0b3bab34fb8c</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>ddb4a930-7d8a-4f38-9079-ddc4b0db980e</td>\n",
       "      <td>141.0</td>\n",
       "      <td>16:37:16.234000</td>\n",
       "      <td>2017-12-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>GBP</td>\n",
       "      <td>10000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:34:42.592</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>b3a6762b-a940-4459-bb1d-8e28e151f901</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>080ef0be-850a-4f14-ab03-28ff68e6b797</td>\n",
       "      <td>7394.0</td>\n",
       "      <td>16:34:42.592000</td>\n",
       "      <td>2017-12-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>EUR</td>\n",
       "      <td>1159</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:34:17.880</td>\n",
       "      <td>NaN</td>\n",
       "      <td>FRA</td>\n",
       "      <td>manu</td>\n",
       "      <td>f5723f5e-5a00-4851-ae1d-5c11146aa491</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>bd13abac-4004-4c67-9bb7-90a42b027e82</td>\n",
       "      <td>977.0</td>\n",
       "      <td>16:34:17.880000</td>\n",
       "      <td>2017-12-10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  currency  amount      state            created_date  merchant_category  \\\n",
       "0      GBP    4420  COMPLETED 2017-12-10 16:38:55.577                NaN   \n",
       "1      PLN    1500  COMPLETED 2017-12-10 16:37:24.792  point_of_interest   \n",
       "2      GBP     191  COMPLETED 2017-12-10 16:37:16.234            airport   \n",
       "3      GBP   10000  COMPLETED 2017-12-10 16:34:42.592                NaN   \n",
       "4      EUR    1159  COMPLETED 2017-12-10 16:34:17.880                NaN   \n",
       "\n",
       "  merchant_country entry_method                               user_id  \\\n",
       "0              NLD         chip  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "1              POL         manu  76cbaad3-4721-4a3b-92b9-3eb9e9319565   \n",
       "2              PRT         chip  7bcaa34e-b889-4582-9c29-0b3bab34fb8c   \n",
       "3              NaN         misc  b3a6762b-a940-4459-bb1d-8e28e151f901   \n",
       "4              FRA         manu  f5723f5e-5a00-4851-ae1d-5c11146aa491   \n",
       "\n",
       "           type source                                    id  amount_usd  \\\n",
       "0           ATM   GAIA  367bf5f9-7cce-4683-90b9-d3c011bf4c87      3268.0   \n",
       "1  CARD_PAYMENT   GAIA  ff6802b9-360d-4efe-b09b-f99c6cac3383         NaN   \n",
       "2  CARD_PAYMENT   GAIA  ddb4a930-7d8a-4f38-9079-ddc4b0db980e       141.0   \n",
       "3         TOPUP   HERA  080ef0be-850a-4f14-ab03-28ff68e6b797      7394.0   \n",
       "4  CARD_PAYMENT   GAIA  bd13abac-4004-4c67-9bb7-90a42b027e82       977.0   \n",
       "\n",
       "           tx_time     tx_date  \n",
       "0  16:38:55.577000  2017-12-10  \n",
       "1  16:37:24.792000  2017-12-10  \n",
       "2  16:37:16.234000  2017-12-10  \n",
       "3  16:34:42.592000  2017-12-10  \n",
       "4  16:34:17.880000  2017-12-10  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_transactions['created_date'] = pd.to_datetime(\n",
    "    df_transactions['created_date'])\n",
    "df_transactions['tx_time'] = df_transactions['created_date'].dt.time\n",
    "df_transactions['tx_date'] = df_transactions['created_date'].dt.date\n",
    "df_transactions.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# function to get is_weekend\n",
    "def is_weekend(tx_date):\n",
    "    weekday = tx_date.weekday()\n",
    "    is_weekend = weekday >= 5\n",
    "\n",
    "    return int(is_weekend)\n",
    "\n",
    "# function to get is_night\n",
    "\n",
    "\n",
    "def is_night(tx_time):\n",
    "    hour = tx_time.hour\n",
    "    is_night = hour <= 6\n",
    "\n",
    "    return int(is_night)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>currency</th>\n",
       "      <th>amount</th>\n",
       "      <th>state</th>\n",
       "      <th>created_date</th>\n",
       "      <th>merchant_category</th>\n",
       "      <th>merchant_country</th>\n",
       "      <th>entry_method</th>\n",
       "      <th>user_id</th>\n",
       "      <th>type</th>\n",
       "      <th>source</th>\n",
       "      <th>id</th>\n",
       "      <th>amount_usd</th>\n",
       "      <th>tx_time</th>\n",
       "      <th>tx_date</th>\n",
       "      <th>is_weekend</th>\n",
       "      <th>is_night</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GBP</td>\n",
       "      <td>4420</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:38:55.577</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NLD</td>\n",
       "      <td>chip</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>ATM</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>367bf5f9-7cce-4683-90b9-d3c011bf4c87</td>\n",
       "      <td>3268.0</td>\n",
       "      <td>16:38:55.577000</td>\n",
       "      <td>2017-12-10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PLN</td>\n",
       "      <td>1500</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:37:24.792</td>\n",
       "      <td>point_of_interest</td>\n",
       "      <td>POL</td>\n",
       "      <td>manu</td>\n",
       "      <td>76cbaad3-4721-4a3b-92b9-3eb9e9319565</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>ff6802b9-360d-4efe-b09b-f99c6cac3383</td>\n",
       "      <td>NaN</td>\n",
       "      <td>16:37:24.792000</td>\n",
       "      <td>2017-12-10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>GBP</td>\n",
       "      <td>191</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:37:16.234</td>\n",
       "      <td>airport</td>\n",
       "      <td>PRT</td>\n",
       "      <td>chip</td>\n",
       "      <td>7bcaa34e-b889-4582-9c29-0b3bab34fb8c</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>ddb4a930-7d8a-4f38-9079-ddc4b0db980e</td>\n",
       "      <td>141.0</td>\n",
       "      <td>16:37:16.234000</td>\n",
       "      <td>2017-12-10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>GBP</td>\n",
       "      <td>10000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:34:42.592</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>b3a6762b-a940-4459-bb1d-8e28e151f901</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>080ef0be-850a-4f14-ab03-28ff68e6b797</td>\n",
       "      <td>7394.0</td>\n",
       "      <td>16:34:42.592000</td>\n",
       "      <td>2017-12-10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>EUR</td>\n",
       "      <td>1159</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-10 16:34:17.880</td>\n",
       "      <td>NaN</td>\n",
       "      <td>FRA</td>\n",
       "      <td>manu</td>\n",
       "      <td>f5723f5e-5a00-4851-ae1d-5c11146aa491</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>bd13abac-4004-4c67-9bb7-90a42b027e82</td>\n",
       "      <td>977.0</td>\n",
       "      <td>16:34:17.880000</td>\n",
       "      <td>2017-12-10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  currency  amount      state            created_date  merchant_category  \\\n",
       "0      GBP    4420  COMPLETED 2017-12-10 16:38:55.577                NaN   \n",
       "1      PLN    1500  COMPLETED 2017-12-10 16:37:24.792  point_of_interest   \n",
       "2      GBP     191  COMPLETED 2017-12-10 16:37:16.234            airport   \n",
       "3      GBP   10000  COMPLETED 2017-12-10 16:34:42.592                NaN   \n",
       "4      EUR    1159  COMPLETED 2017-12-10 16:34:17.880                NaN   \n",
       "\n",
       "  merchant_country entry_method                               user_id  \\\n",
       "0              NLD         chip  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "1              POL         manu  76cbaad3-4721-4a3b-92b9-3eb9e9319565   \n",
       "2              PRT         chip  7bcaa34e-b889-4582-9c29-0b3bab34fb8c   \n",
       "3              NaN         misc  b3a6762b-a940-4459-bb1d-8e28e151f901   \n",
       "4              FRA         manu  f5723f5e-5a00-4851-ae1d-5c11146aa491   \n",
       "\n",
       "           type source                                    id  amount_usd  \\\n",
       "0           ATM   GAIA  367bf5f9-7cce-4683-90b9-d3c011bf4c87      3268.0   \n",
       "1  CARD_PAYMENT   GAIA  ff6802b9-360d-4efe-b09b-f99c6cac3383         NaN   \n",
       "2  CARD_PAYMENT   GAIA  ddb4a930-7d8a-4f38-9079-ddc4b0db980e       141.0   \n",
       "3         TOPUP   HERA  080ef0be-850a-4f14-ab03-28ff68e6b797      7394.0   \n",
       "4  CARD_PAYMENT   GAIA  bd13abac-4004-4c67-9bb7-90a42b027e82       977.0   \n",
       "\n",
       "           tx_time     tx_date  is_weekend  is_night  \n",
       "0  16:38:55.577000  2017-12-10           1         0  \n",
       "1  16:37:24.792000  2017-12-10           1         0  \n",
       "2  16:37:16.234000  2017-12-10           1         0  \n",
       "3  16:34:42.592000  2017-12-10           1         0  \n",
       "4  16:34:17.880000  2017-12-10           1         0  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Apply to all transactions\n",
    "df_transactions['is_weekend'] = df_transactions['tx_date'].apply(is_weekend)\n",
    "df_transactions['is_night'] = df_transactions['tx_time'].apply(is_night)\n",
    "df_transactions.head()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Customer ID Transformations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define a function that takes a set of transactions for a customer and a set of window sizes and returns a DataFrame with 6 new features.\n",
    "# We will use the rolling() function in pandas\n",
    "def get_customer_spending_behavior_features(df_transactions, window_size_in_days=[1, 7, 30]):\n",
    "\n",
    "    # Order transactions chronologically\n",
    "    df_transactions = df_transactions.sort_values(by='created_date')\n",
    "\n",
    "    # Transaction date and time is set as the index, which will allow for the use of the rolling function\n",
    "    df_transactions.index = df_transactions.created_date\n",
    "\n",
    "    # For each window size\n",
    "    for window_size in window_size_in_days:\n",
    "        # compute the sum of the transaction amounts and the number of transactions for the given window size\n",
    "        # Use np.nansum() because amount_usd has NaN values\n",
    "        sum_amount_tx_window = df_transactions['amount_usd'].rolling(\n",
    "            str(window_size) + 'd').sum()\n",
    "        num_tx_window = df_transactions['id'].rolling(\n",
    "            str(window_size) + 'd').count()\n",
    "\n",
    "        # Compute the average transaction amount for the given window size\n",
    "        avg_amount_tx_window = sum_amount_tx_window / num_tx_window\n",
    "\n",
    "        # Save feature values\n",
    "        df_transactions['customer_id_num_tx_' +\n",
    "                        str(window_size) + 'day_window'] = list(num_tx_window)\n",
    "        df_transactions['customer_id_avg_amount_' +\n",
    "                        str(window_size) + 'day_window'] = list(avg_amount_tx_window)\n",
    "\n",
    "    # Reindex according to transaction IDs\n",
    "    df_transactions.index = df_transactions.user_id\n",
    "\n",
    "    # Return the dataframe with new features\n",
    "    return df_transactions\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>currency</th>\n",
       "      <th>amount</th>\n",
       "      <th>state</th>\n",
       "      <th>created_date</th>\n",
       "      <th>merchant_category</th>\n",
       "      <th>merchant_country</th>\n",
       "      <th>entry_method</th>\n",
       "      <th>user_id</th>\n",
       "      <th>type</th>\n",
       "      <th>source</th>\n",
       "      <th>...</th>\n",
       "      <th>tx_time</th>\n",
       "      <th>tx_date</th>\n",
       "      <th>is_weekend</th>\n",
       "      <th>is_night</th>\n",
       "      <th>customer_id_num_tx_1day_window</th>\n",
       "      <th>customer_id_avg_amount_1day_window</th>\n",
       "      <th>customer_id_num_tx_7day_window</th>\n",
       "      <th>customer_id_avg_amount_7day_window</th>\n",
       "      <th>customer_id_num_tx_30day_window</th>\n",
       "      <th>customer_id_avg_amount_30day_window</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>user_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>100</td>\n",
       "      <td>REVERTED</td>\n",
       "      <td>2017-10-08 12:55:01.503</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>12:55:01.503000</td>\n",
       "      <td>2017-10-08</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>74.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>74.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>74.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>1000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-10-08 12:55:20.449</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>12:55:20.449000</td>\n",
       "      <td>2017-10-08</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>409.500000</td>\n",
       "      <td>2.0</td>\n",
       "      <td>409.500000</td>\n",
       "      <td>2.0</td>\n",
       "      <td>409.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>19000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-04 18:12:18.748</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>18:12:18.748000</td>\n",
       "      <td>2017-12-04</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>14130.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>14130.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>14130.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>1820</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-07 07:21:20.940</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NLD</td>\n",
       "      <td>chip</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>...</td>\n",
       "      <td>07:21:20.940000</td>\n",
       "      <td>2017-12-07</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1350.000000</td>\n",
       "      <td>2.0</td>\n",
       "      <td>7740.000000</td>\n",
       "      <td>2.0</td>\n",
       "      <td>7740.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>2147</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2017-12-07 12:12:24.170</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NLD</td>\n",
       "      <td>cont</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>...</td>\n",
       "      <td>12:12:24.170000</td>\n",
       "      <td>2017-12-07</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1464.500000</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5686.333333</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5686.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>2000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-04-02 13:09:07.556</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>BANK_TRANSFER</td>\n",
       "      <td>CRONUS</td>\n",
       "      <td>...</td>\n",
       "      <td>13:09:07.556000</td>\n",
       "      <td>2018-04-02</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1494.000000</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1865.500000</td>\n",
       "      <td>8.0</td>\n",
       "      <td>2894.125000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>1000</td>\n",
       "      <td>FAILED</td>\n",
       "      <td>2018-06-10 15:39:01.934</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>15:39:01.934000</td>\n",
       "      <td>2018-06-10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>744.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>744.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>744.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>100</td>\n",
       "      <td>REVERTED</td>\n",
       "      <td>2018-06-10 15:40:49.467</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>15:40:49.467000</td>\n",
       "      <td>2018-06-10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>409.000000</td>\n",
       "      <td>2.0</td>\n",
       "      <td>409.000000</td>\n",
       "      <td>2.0</td>\n",
       "      <td>409.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>1000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-06-10 15:41:06.624</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>15:41:06.624000</td>\n",
       "      <td>2018-06-10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>520.666667</td>\n",
       "      <td>3.0</td>\n",
       "      <td>520.666667</td>\n",
       "      <td>3.0</td>\n",
       "      <td>520.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3ff52b92-d416-4e22-8cad-018f500d4bbc</th>\n",
       "      <td>GBP</td>\n",
       "      <td>1629</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-07-10 10:07:28.219</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>3ff52b92-d416-4e22-8cad-018f500d4bbc</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>10:07:28.219000</td>\n",
       "      <td>2018-07-10</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1228.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1228.000000</td>\n",
       "      <td>4.0</td>\n",
       "      <td>697.500000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>62 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     currency  amount      state  \\\n",
       "user_id                                                            \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP     100   REVERTED   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP    1000  COMPLETED   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP   19000  COMPLETED   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP    1820  COMPLETED   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP    2147  COMPLETED   \n",
       "...                                       ...     ...        ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP    2000  COMPLETED   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP    1000     FAILED   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP     100   REVERTED   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP    1000  COMPLETED   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc      GBP    1629  COMPLETED   \n",
       "\n",
       "                                                created_date  \\\n",
       "user_id                                                        \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2017-10-08 12:55:01.503   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2017-10-08 12:55:20.449   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2017-12-04 18:12:18.748   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2017-12-07 07:21:20.940   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2017-12-07 12:12:24.170   \n",
       "...                                                      ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2018-04-02 13:09:07.556   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2018-06-10 15:39:01.934   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2018-06-10 15:40:49.467   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2018-06-10 15:41:06.624   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc 2018-07-10 10:07:28.219   \n",
       "\n",
       "                                     merchant_category merchant_country  \\\n",
       "user_id                                                                   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NaN   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NaN   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NaN   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NLD   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NLD   \n",
       "...                                                ...              ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NaN   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NaN   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NaN   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NaN   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc               NaN              NaN   \n",
       "\n",
       "                                     entry_method  \\\n",
       "user_id                                             \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         misc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         misc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         misc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         chip   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         cont   \n",
       "...                                           ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         misc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         misc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         misc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         misc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc         misc   \n",
       "\n",
       "                                                                   user_id  \\\n",
       "user_id                                                                      \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "...                                                                    ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  3ff52b92-d416-4e22-8cad-018f500d4bbc   \n",
       "\n",
       "                                               type  source  ...  \\\n",
       "user_id                                                      ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc          TOPUP    HERA  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc          TOPUP    HERA  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc          TOPUP    HERA  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc   CARD_PAYMENT    GAIA  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc   CARD_PAYMENT    GAIA  ...   \n",
       "...                                             ...     ...  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  BANK_TRANSFER  CRONUS  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc          TOPUP    HERA  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc          TOPUP    HERA  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc          TOPUP    HERA  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc          TOPUP    HERA  ...   \n",
       "\n",
       "                                              tx_time     tx_date is_weekend  \\\n",
       "user_id                                                                        \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  12:55:01.503000  2017-10-08          1   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  12:55:20.449000  2017-10-08          1   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  18:12:18.748000  2017-12-04          0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  07:21:20.940000  2017-12-07          0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  12:12:24.170000  2017-12-07          0   \n",
       "...                                               ...         ...        ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  13:09:07.556000  2018-04-02          0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  15:39:01.934000  2018-06-10          1   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  15:40:49.467000  2018-06-10          1   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  15:41:06.624000  2018-06-10          1   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc  10:07:28.219000  2018-07-10          0   \n",
       "\n",
       "                                     is_night  customer_id_num_tx_1day_window  \\\n",
       "user_id                                                                         \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             2.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             2.0   \n",
       "...                                       ...                             ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             2.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             3.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc        0                             1.0   \n",
       "\n",
       "                                      customer_id_avg_amount_1day_window  \\\n",
       "user_id                                                                    \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                           74.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          409.500000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                        14130.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                         1350.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                         1464.500000   \n",
       "...                                                                  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                         1494.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          744.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          409.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          520.666667   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                         1228.000000   \n",
       "\n",
       "                                      customer_id_num_tx_7day_window  \\\n",
       "user_id                                                                \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             2.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             2.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             3.0   \n",
       "...                                                              ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             2.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             2.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             3.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                             1.0   \n",
       "\n",
       "                                      customer_id_avg_amount_7day_window  \\\n",
       "user_id                                                                    \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                           74.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          409.500000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                        14130.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                         7740.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                         5686.333333   \n",
       "...                                                                  ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                         1865.500000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          744.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          409.000000   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          520.666667   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                         1228.000000   \n",
       "\n",
       "                                      customer_id_num_tx_30day_window  \\\n",
       "user_id                                                                 \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              2.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              2.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              3.0   \n",
       "...                                                               ...   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              8.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              1.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              2.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              3.0   \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                              4.0   \n",
       "\n",
       "                                      customer_id_avg_amount_30day_window  \n",
       "user_id                                                                    \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                            74.000000  \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                           409.500000  \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                         14130.000000  \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          7740.000000  \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          5686.333333  \n",
       "...                                                                   ...  \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                          2894.125000  \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                           744.000000  \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                           409.000000  \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                           520.666667  \n",
       "3ff52b92-d416-4e22-8cad-018f500d4bbc                           697.500000  \n",
       "\n",
       "[62 rows x 22 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Test the function on one user_id\n",
    "spending_behavior_customer_0 = get_customer_spending_behavior_features(\n",
    "    df_transactions[df_transactions['user_id'] == '3ff52b92-d416-4e22-8cad-018f500d4bbc'])\n",
    "spending_behavior_customer_0\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1 µs, sys: 0 ns, total: 1 µs\n",
      "Wall time: 2.86 µs\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>currency</th>\n",
       "      <th>amount</th>\n",
       "      <th>state</th>\n",
       "      <th>created_date</th>\n",
       "      <th>merchant_category</th>\n",
       "      <th>merchant_country</th>\n",
       "      <th>entry_method</th>\n",
       "      <th>user_id</th>\n",
       "      <th>type</th>\n",
       "      <th>source</th>\n",
       "      <th>...</th>\n",
       "      <th>tx_time</th>\n",
       "      <th>tx_date</th>\n",
       "      <th>is_weekend</th>\n",
       "      <th>is_night</th>\n",
       "      <th>customer_id_num_tx_1day_window</th>\n",
       "      <th>customer_id_avg_amount_1day_window</th>\n",
       "      <th>customer_id_num_tx_7day_window</th>\n",
       "      <th>customer_id_avg_amount_7day_window</th>\n",
       "      <th>customer_id_num_tx_30day_window</th>\n",
       "      <th>customer_id_avg_amount_30day_window</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>user_id</th>\n",
       "      <th>user_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"5\" valign=\"top\">000e88bb-d302-4fdc-b757-2b1a2c33e7d6</th>\n",
       "      <th>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</th>\n",
       "      <td>DKK</td>\n",
       "      <td>100</td>\n",
       "      <td>REVERTED</td>\n",
       "      <td>2018-03-07 19:34:44.503</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>19:34:44.503000</td>\n",
       "      <td>2018-03-07</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</th>\n",
       "      <td>DKK</td>\n",
       "      <td>10000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-03-07 19:35:01.988</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>19:35:01.988000</td>\n",
       "      <td>2018-03-07</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</th>\n",
       "      <td>DKK</td>\n",
       "      <td>50000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-04-20 11:41:47.677</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>11:41:47.677000</td>\n",
       "      <td>2018-04-20</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</th>\n",
       "      <td>DKK</td>\n",
       "      <td>59589</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-04-20 11:47:23.258</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</td>\n",
       "      <td>BANK_TRANSFER</td>\n",
       "      <td>CRONUS</td>\n",
       "      <td>...</td>\n",
       "      <td>11:47:23.258000</td>\n",
       "      <td>2018-04-20</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</th>\n",
       "      <td>DKK</td>\n",
       "      <td>50000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-05-02 17:41:01.170</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>17:41:01.170000</td>\n",
       "      <td>2018-05-02</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"3\" valign=\"top\">ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33</th>\n",
       "      <th>ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33</th>\n",
       "      <td>EUR</td>\n",
       "      <td>7786</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-07-08 08:42:39.861</td>\n",
       "      <td>NaN</td>\n",
       "      <td>GBR</td>\n",
       "      <td>manu</td>\n",
       "      <td>ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>...</td>\n",
       "      <td>08:42:39.861000</td>\n",
       "      <td>2018-07-08</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>6639.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>6639.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>6639.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33</th>\n",
       "      <td>EUR</td>\n",
       "      <td>7786</td>\n",
       "      <td>DECLINED</td>\n",
       "      <td>2018-07-08 08:43:01.388</td>\n",
       "      <td>NaN</td>\n",
       "      <td>GBR</td>\n",
       "      <td>manu</td>\n",
       "      <td>ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>...</td>\n",
       "      <td>08:43:01.388000</td>\n",
       "      <td>2018-07-08</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6639.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6639.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6639.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33</th>\n",
       "      <td>EUR</td>\n",
       "      <td>1800</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-07-26 07:17:53.585</td>\n",
       "      <td>NaN</td>\n",
       "      <td>FRA</td>\n",
       "      <td>manu</td>\n",
       "      <td>ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33</td>\n",
       "      <td>CARD_PAYMENT</td>\n",
       "      <td>GAIA</td>\n",
       "      <td>...</td>\n",
       "      <td>07:17:53.585000</td>\n",
       "      <td>2018-07-26</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1547.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1547.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>4941.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">fff200b0-5204-4f7d-86eb-3bb5654b4636</th>\n",
       "      <th>fff200b0-5204-4f7d-86eb-3bb5654b4636</th>\n",
       "      <td>RON</td>\n",
       "      <td>100</td>\n",
       "      <td>REVERTED</td>\n",
       "      <td>2018-07-22 09:19:44.117</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>fff200b0-5204-4f7d-86eb-3bb5654b4636</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>09:19:44.117000</td>\n",
       "      <td>2018-07-22</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fff200b0-5204-4f7d-86eb-3bb5654b4636</th>\n",
       "      <td>RON</td>\n",
       "      <td>10000</td>\n",
       "      <td>COMPLETED</td>\n",
       "      <td>2018-07-22 09:20:10.612</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>misc</td>\n",
       "      <td>fff200b0-5204-4f7d-86eb-3bb5654b4636</td>\n",
       "      <td>TOPUP</td>\n",
       "      <td>HERA</td>\n",
       "      <td>...</td>\n",
       "      <td>09:20:10.612000</td>\n",
       "      <td>2018-07-22</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>688651 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                          currency  \\\n",
       "user_id                              user_id                                         \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6      DKK   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6      DKK   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6      DKK   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6      DKK   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6      DKK   \n",
       "...                                                                            ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33      EUR   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33      EUR   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33      EUR   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636      RON   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636      RON   \n",
       "\n",
       "                                                                           amount  \\\n",
       "user_id                              user_id                                        \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6     100   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6   10000   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6   50000   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6   59589   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6   50000   \n",
       "...                                                                           ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33    7786   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33    7786   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33    1800   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636     100   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636   10000   \n",
       "\n",
       "                                                                               state  \\\n",
       "user_id                              user_id                                           \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6   REVERTED   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  COMPLETED   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  COMPLETED   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  COMPLETED   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  COMPLETED   \n",
       "...                                                                              ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  COMPLETED   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33   DECLINED   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  COMPLETED   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636   REVERTED   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636  COMPLETED   \n",
       "\n",
       "                                                                                     created_date  \\\n",
       "user_id                              user_id                                                        \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6 2018-03-07 19:34:44.503   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6 2018-03-07 19:35:01.988   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6 2018-04-20 11:41:47.677   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6 2018-04-20 11:47:23.258   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6 2018-05-02 17:41:01.170   \n",
       "...                                                                                           ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 2018-07-08 08:42:39.861   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 2018-07-08 08:43:01.388   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 2018-07-26 07:17:53.585   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636 2018-07-22 09:19:44.117   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636 2018-07-22 09:20:10.612   \n",
       "\n",
       "                                                                          merchant_category  \\\n",
       "user_id                              user_id                                                  \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6               NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6               NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6               NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6               NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6               NaN   \n",
       "...                                                                                     ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33               NaN   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33               NaN   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33               NaN   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636               NaN   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636               NaN   \n",
       "\n",
       "                                                                          merchant_country  \\\n",
       "user_id                              user_id                                                 \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6              NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6              NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6              NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6              NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6              NaN   \n",
       "...                                                                                    ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33              GBR   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33              GBR   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33              FRA   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636              NaN   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636              NaN   \n",
       "\n",
       "                                                                          entry_method  \\\n",
       "user_id                              user_id                                             \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6         misc   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6         misc   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6         misc   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6         misc   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6         misc   \n",
       "...                                                                                ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33         manu   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33         manu   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33         manu   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636         misc   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636         misc   \n",
       "\n",
       "                                                                                                        user_id  \\\n",
       "user_id                              user_id                                                                      \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6  000e88bb-d302-4fdc-b757-2b1a2c33e7d6   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  000e88bb-d302-4fdc-b757-2b1a2c33e7d6   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  000e88bb-d302-4fdc-b757-2b1a2c33e7d6   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  000e88bb-d302-4fdc-b757-2b1a2c33e7d6   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  000e88bb-d302-4fdc-b757-2b1a2c33e7d6   \n",
       "...                                                                                                         ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636  fff200b0-5204-4f7d-86eb-3bb5654b4636   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636  fff200b0-5204-4f7d-86eb-3bb5654b4636   \n",
       "\n",
       "                                                                                    type  \\\n",
       "user_id                              user_id                                               \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6          TOPUP   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6          TOPUP   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6          TOPUP   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  BANK_TRANSFER   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6          TOPUP   \n",
       "...                                                                                  ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33   CARD_PAYMENT   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33   CARD_PAYMENT   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33   CARD_PAYMENT   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636          TOPUP   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636          TOPUP   \n",
       "\n",
       "                                                                           source  \\\n",
       "user_id                              user_id                                        \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6    HERA   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6    HERA   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6    HERA   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  CRONUS   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6    HERA   \n",
       "...                                                                           ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33    GAIA   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33    GAIA   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33    GAIA   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636    HERA   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636    HERA   \n",
       "\n",
       "                                                                           ...  \\\n",
       "user_id                              user_id                               ...   \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6  ...   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  ...   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  ...   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  ...   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  ...   \n",
       "...                                                                        ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  ...   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  ...   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  ...   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636  ...   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636  ...   \n",
       "\n",
       "                                                                                   tx_time  \\\n",
       "user_id                              user_id                                                 \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6  19:34:44.503000   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  19:35:01.988000   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  11:41:47.677000   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  11:47:23.258000   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  17:41:01.170000   \n",
       "...                                                                                    ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  08:42:39.861000   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  08:43:01.388000   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  07:17:53.585000   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636  09:19:44.117000   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636  09:20:10.612000   \n",
       "\n",
       "                                                                              tx_date  \\\n",
       "user_id                              user_id                                            \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6  2018-03-07   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  2018-03-07   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  2018-04-20   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  2018-04-20   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6  2018-05-02   \n",
       "...                                                                               ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  2018-07-08   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  2018-07-08   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  2018-07-26   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636  2018-07-22   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636  2018-07-22   \n",
       "\n",
       "                                                                          is_weekend  \\\n",
       "user_id                              user_id                                           \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6          0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6          0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6          0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6          0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6          0   \n",
       "...                                                                              ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33          1   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33          1   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33          0   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636          1   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636          1   \n",
       "\n",
       "                                                                          is_night  \\\n",
       "user_id                              user_id                                         \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6        0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6        0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6        0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6        0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6        0   \n",
       "...                                                                            ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33        0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33        0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33        0   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636        0   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636        0   \n",
       "\n",
       "                                                                           customer_id_num_tx_1day_window  \\\n",
       "user_id                              user_id                                                                \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             1.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             2.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             1.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             2.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             1.0   \n",
       "...                                                                                                   ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                             1.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                             2.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                             1.0   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636                             1.0   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636                             2.0   \n",
       "\n",
       "                                                                           customer_id_avg_amount_1day_window  \\\n",
       "user_id                              user_id                                                                    \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "...                                                                                                       ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                              6639.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                              6639.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                              1547.0   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636                                 NaN   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636                                 NaN   \n",
       "\n",
       "                                                                           customer_id_num_tx_7day_window  \\\n",
       "user_id                              user_id                                                                \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             1.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             2.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             1.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             2.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                             1.0   \n",
       "...                                                                                                   ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                             1.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                             2.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                             1.0   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636                             1.0   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636                             2.0   \n",
       "\n",
       "                                                                           customer_id_avg_amount_7day_window  \\\n",
       "user_id                              user_id                                                                    \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                 NaN   \n",
       "...                                                                                                       ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                              6639.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                              6639.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                              1547.0   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636                                 NaN   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636                                 NaN   \n",
       "\n",
       "                                                                           customer_id_num_tx_30day_window  \\\n",
       "user_id                              user_id                                                                 \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6                              1.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                              2.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                              1.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                              2.0   \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                              3.0   \n",
       "...                                                                                                    ...   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                              1.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                              2.0   \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                              3.0   \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636                              1.0   \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636                              2.0   \n",
       "\n",
       "                                                                           customer_id_avg_amount_30day_window  \n",
       "user_id                              user_id                                                                    \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6 000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                  NaN  \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                  NaN  \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                  NaN  \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                  NaN  \n",
       "                                     000e88bb-d302-4fdc-b757-2b1a2c33e7d6                                  NaN  \n",
       "...                                                                                                        ...  \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33 ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                          6639.000000  \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                          6639.000000  \n",
       "                                     ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33                          4941.666667  \n",
       "fff200b0-5204-4f7d-86eb-3bb5654b4636 fff200b0-5204-4f7d-86eb-3bb5654b4636                                  NaN  \n",
       "                                     fff200b0-5204-4f7d-86eb-3bb5654b4636                                  NaN  \n",
       "\n",
       "[688651 rows x 22 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Applying to all users\n",
    "%time\n",
    "df_transactions_mid = df_transactions.groupby('user_id').apply(lambda x: get_customer_spending_behavior_features(x, window_size_in_days=[1, 7, 30]))\n",
    "# df_transactions = df_transactions.reset_index(drop=True)\n",
    "df_transactions_mid\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create df_transactions_summary\n",
    "- Create df_transactions_summary by grouping by user_id and calculating some metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>most_recent_tx_date</th>\n",
       "      <th>num_tx_weekend</th>\n",
       "      <th>num_tx_night</th>\n",
       "      <th>max_tx_1d</th>\n",
       "      <th>max_amt_1d</th>\n",
       "      <th>max_tx_7d</th>\n",
       "      <th>max_amt_7d</th>\n",
       "      <th>max_tx_30d</th>\n",
       "      <th>max_amt_30d</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>user_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>000e88bb-d302-4fdc-b757-2b1a2c33e7d6</th>\n",
       "      <td>2018-06-30</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>165.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>127.666667</td>\n",
       "      <td>10.0</td>\n",
       "      <td>127.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>001032e0-8071-4baf-95b9-e50214665c2e</th>\n",
       "      <td>2018-01-09</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>7.0</td>\n",
       "      <td>20524.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>20524.000000</td>\n",
       "      <td>25.0</td>\n",
       "      <td>20524.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>00131af8-66f0-4526-8b5f-dc2fdb26c7d7</th>\n",
       "      <td>2018-07-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2828.5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2828.500000</td>\n",
       "      <td>7.0</td>\n",
       "      <td>1645.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>001926be-3245-43fa-86dd-b40ee160b6f9</th>\n",
       "      <td>2018-07-30</td>\n",
       "      <td>124</td>\n",
       "      <td>5</td>\n",
       "      <td>11.0</td>\n",
       "      <td>58813.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>22585.000000</td>\n",
       "      <td>35.0</td>\n",
       "      <td>11665.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>001cc034-5730-47c6-a70c-25f42249c9ee</th>\n",
       "      <td>2018-07-20</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1064.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1064.000000</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1064.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ffc8a89d-8e53-4066-b622-36a60e2e9983</th>\n",
       "      <td>2016-08-15</td>\n",
       "      <td>20</td>\n",
       "      <td>1</td>\n",
       "      <td>5.0</td>\n",
       "      <td>36905.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>18813.250000</td>\n",
       "      <td>20.0</td>\n",
       "      <td>18813.250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ffcc7fda-bbb0-4dba-b5e0-832ca5823f39</th>\n",
       "      <td>2018-02-28</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>85.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>85.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>85.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ffd20bab-7b39-4d22-bd24-08bd36f3dbb2</th>\n",
       "      <td>2018-04-22</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>84510.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>84510.000000</td>\n",
       "      <td>4.0</td>\n",
       "      <td>84510.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ffd3ee80-dec1-4603-bb3b-ef19d8bcc147</th>\n",
       "      <td>2018-07-09</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>5624.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5624.000000</td>\n",
       "      <td>8.0</td>\n",
       "      <td>2916.857143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33</th>\n",
       "      <td>2018-07-26</td>\n",
       "      <td>6</td>\n",
       "      <td>8</td>\n",
       "      <td>5.0</td>\n",
       "      <td>33755.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>33755.000000</td>\n",
       "      <td>19.0</td>\n",
       "      <td>33742.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7459 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     most_recent_tx_date  num_tx_weekend  \\\n",
       "user_id                                                                    \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6          2018-06-30               6   \n",
       "001032e0-8071-4baf-95b9-e50214665c2e          2018-01-09               1   \n",
       "00131af8-66f0-4526-8b5f-dc2fdb26c7d7          2018-07-06               0   \n",
       "001926be-3245-43fa-86dd-b40ee160b6f9          2018-07-30             124   \n",
       "001cc034-5730-47c6-a70c-25f42249c9ee          2018-07-20               0   \n",
       "...                                                  ...             ...   \n",
       "ffc8a89d-8e53-4066-b622-36a60e2e9983          2016-08-15              20   \n",
       "ffcc7fda-bbb0-4dba-b5e0-832ca5823f39          2018-02-28               0   \n",
       "ffd20bab-7b39-4d22-bd24-08bd36f3dbb2          2018-04-22               1   \n",
       "ffd3ee80-dec1-4603-bb3b-ef19d8bcc147          2018-07-09               2   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33          2018-07-26               6   \n",
       "\n",
       "                                      num_tx_night  max_tx_1d  max_amt_1d  \\\n",
       "user_id                                                                     \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6             0        6.0       165.0   \n",
       "001032e0-8071-4baf-95b9-e50214665c2e             3        7.0     20524.0   \n",
       "00131af8-66f0-4526-8b5f-dc2fdb26c7d7             0        5.0      2828.5   \n",
       "001926be-3245-43fa-86dd-b40ee160b6f9             5       11.0     58813.0   \n",
       "001cc034-5730-47c6-a70c-25f42249c9ee             1        4.0      1064.0   \n",
       "...                                            ...        ...         ...   \n",
       "ffc8a89d-8e53-4066-b622-36a60e2e9983             1        5.0     36905.0   \n",
       "ffcc7fda-bbb0-4dba-b5e0-832ca5823f39             0        1.0        85.0   \n",
       "ffd20bab-7b39-4d22-bd24-08bd36f3dbb2             2        1.0     84510.0   \n",
       "ffd3ee80-dec1-4603-bb3b-ef19d8bcc147             2        4.0      5624.0   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33             8        5.0     33755.0   \n",
       "\n",
       "                                      max_tx_7d    max_amt_7d  max_tx_30d  \\\n",
       "user_id                                                                     \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6        8.0    127.666667        10.0   \n",
       "001032e0-8071-4baf-95b9-e50214665c2e       15.0  20524.000000        25.0   \n",
       "00131af8-66f0-4526-8b5f-dc2fdb26c7d7        5.0   2828.500000         7.0   \n",
       "001926be-3245-43fa-86dd-b40ee160b6f9       17.0  22585.000000        35.0   \n",
       "001cc034-5730-47c6-a70c-25f42249c9ee        4.0   1064.000000         4.0   \n",
       "...                                         ...           ...         ...   \n",
       "ffc8a89d-8e53-4066-b622-36a60e2e9983       10.0  18813.250000        20.0   \n",
       "ffcc7fda-bbb0-4dba-b5e0-832ca5823f39        1.0     85.000000         1.0   \n",
       "ffd20bab-7b39-4d22-bd24-08bd36f3dbb2        2.0  84510.000000         4.0   \n",
       "ffd3ee80-dec1-4603-bb3b-ef19d8bcc147        5.0   5624.000000         8.0   \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33       10.0  33755.000000        19.0   \n",
       "\n",
       "                                       max_amt_30d  \n",
       "user_id                                             \n",
       "000e88bb-d302-4fdc-b757-2b1a2c33e7d6    127.666667  \n",
       "001032e0-8071-4baf-95b9-e50214665c2e  20524.000000  \n",
       "00131af8-66f0-4526-8b5f-dc2fdb26c7d7   1645.500000  \n",
       "001926be-3245-43fa-86dd-b40ee160b6f9  11665.500000  \n",
       "001cc034-5730-47c6-a70c-25f42249c9ee   1064.000000  \n",
       "...                                            ...  \n",
       "ffc8a89d-8e53-4066-b622-36a60e2e9983  18813.250000  \n",
       "ffcc7fda-bbb0-4dba-b5e0-832ca5823f39     85.000000  \n",
       "ffd20bab-7b39-4d22-bd24-08bd36f3dbb2  84510.000000  \n",
       "ffd3ee80-dec1-4603-bb3b-ef19d8bcc147   2916.857143  \n",
       "ffd5d3f9-2f87-4f38-8e6d-f008ac79bb33  33742.000000  \n",
       "\n",
       "[7459 rows x 9 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_transactions_summary = df_transactions_mid.reset_index(drop=True).groupby('user_id').agg({\n",
    "    'tx_date': 'max',\n",
    "    'is_weekend': 'sum',\n",
    "    'is_night': 'sum',\n",
    "    'customer_id_num_tx_1day_window': 'max',\n",
    "    'customer_id_avg_amount_1day_window': 'max',\n",
    "    'customer_id_num_tx_7day_window': 'max',\n",
    "    'customer_id_avg_amount_7day_window': 'max',\n",
    "    'customer_id_num_tx_30day_window': 'max',\n",
    "    'customer_id_avg_amount_30day_window': 'max',\n",
    "})\n",
    "\n",
    "df_transactions_summary = df_transactions_summary.rename(columns={'tx_date': 'most_recent_tx_date', 'is_weekend': 'num_tx_weekend', 'is_night': 'num_tx_night',\n",
    "                                                                  'customer_id_num_tx_1day_window': 'max_tx_1d', 'customer_id_avg_amount_1day_window': 'max_amt_1d', 'customer_id_num_tx_7day_window':\n",
    "                                                                  'max_tx_7d', 'customer_id_avg_amount_7day_window': 'max_amt_7d', 'customer_id_num_tx_30day_window': 'max_tx_30d',\n",
    "                                                                  'customer_id_avg_amount_30day_window': 'max_amt_30d'})\n",
    "df_transactions_summary = df_transactions_summary.dropna()\n",
    "df_transactions_summary\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- There are only 8021 unique users in the df_transactions_summary table compared to 9944 users in the df_users.\n",
    "- Need to check we're not losing valuable is_fraudster data when we merge the two dataframes.\n",
    "- Performed the check and noticed that only 1 example of is_fraudster = True is lost during the merge which is acceptable.\n",
    "- Will proceed with merging the two DataFrames."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- the df_transactions_summary dataset is ready to be used.\n",
    "- There are some NULLs in some columns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>has_email</th>\n",
       "      <th>phone_country</th>\n",
       "      <th>is_fraudster</th>\n",
       "      <th>terms_version</th>\n",
       "      <th>created_date</th>\n",
       "      <th>state</th>\n",
       "      <th>country</th>\n",
       "      <th>birth_year</th>\n",
       "      <th>kyc</th>\n",
       "      <th>failed_sign_in_attempts</th>\n",
       "      <th>most_recent_tx_date</th>\n",
       "      <th>num_tx_weekend</th>\n",
       "      <th>num_tx_night</th>\n",
       "      <th>max_tx_1d</th>\n",
       "      <th>max_amt_1d</th>\n",
       "      <th>max_tx_7d</th>\n",
       "      <th>max_amt_7d</th>\n",
       "      <th>max_tx_30d</th>\n",
       "      <th>max_amt_30d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1872820f-e3ac-4c02-bdc7-727897b60043</td>\n",
       "      <td>1</td>\n",
       "      <td>GB||JE||IM||GG</td>\n",
       "      <td>False</td>\n",
       "      <td>2018-05-25</td>\n",
       "      <td>2017-08-06 07:33:33.341000</td>\n",
       "      <td>ACTIVE</td>\n",
       "      <td>GB</td>\n",
       "      <td>1971</td>\n",
       "      <td>PASSED</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-06-03</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>37130.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.713000e+04</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.713000e+04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>545ff94d-66f8-4bea-b398-84425fb2301e</td>\n",
       "      <td>1</td>\n",
       "      <td>GB||JE||IM||GG</td>\n",
       "      <td>False</td>\n",
       "      <td>2018-01-01</td>\n",
       "      <td>2017-03-07 10:18:59.427000</td>\n",
       "      <td>ACTIVE</td>\n",
       "      <td>GB</td>\n",
       "      <td>1982</td>\n",
       "      <td>PASSED</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-07-03</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>29607.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>9.664750e+03</td>\n",
       "      <td>7.0</td>\n",
       "      <td>9.664750e+03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10376f1a-a28a-4885-8daa-c8ca496026bb</td>\n",
       "      <td>1</td>\n",
       "      <td>ES</td>\n",
       "      <td>False</td>\n",
       "      <td>2018-09-20</td>\n",
       "      <td>2018-05-31 04:41:24.672000</td>\n",
       "      <td>ACTIVE</td>\n",
       "      <td>ES</td>\n",
       "      <td>1973</td>\n",
       "      <td>PASSED</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-07-30</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>13.0</td>\n",
       "      <td>34355.0</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3.641263e+03</td>\n",
       "      <td>52.0</td>\n",
       "      <td>2.340333e+03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>fd308db7-0753-4377-879f-6ecf2af14e4f</td>\n",
       "      <td>1</td>\n",
       "      <td>FR</td>\n",
       "      <td>False</td>\n",
       "      <td>2018-05-25</td>\n",
       "      <td>2018-06-01 17:24:23.852000</td>\n",
       "      <td>ACTIVE</td>\n",
       "      <td>FR</td>\n",
       "      <td>1986</td>\n",
       "      <td>PASSED</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-06-04</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>680.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6.800000e+02</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6.800000e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>755fe256-a34d-4853-b7ca-d9bb991a86d3</td>\n",
       "      <td>1</td>\n",
       "      <td>GB||JE||IM||GG</td>\n",
       "      <td>False</td>\n",
       "      <td>2018-09-20</td>\n",
       "      <td>2017-08-09 15:03:33.945000</td>\n",
       "      <td>ACTIVE</td>\n",
       "      <td>GB</td>\n",
       "      <td>1989</td>\n",
       "      <td>PASSED</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-07-27</td>\n",
       "      <td>30</td>\n",
       "      <td>33</td>\n",
       "      <td>19.0</td>\n",
       "      <td>4497500.0</td>\n",
       "      <td>65.0</td>\n",
       "      <td>3.483710e+06</td>\n",
       "      <td>78.0</td>\n",
       "      <td>1.671384e+06</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     id  has_email   phone_country  \\\n",
       "0  1872820f-e3ac-4c02-bdc7-727897b60043          1  GB||JE||IM||GG   \n",
       "1  545ff94d-66f8-4bea-b398-84425fb2301e          1  GB||JE||IM||GG   \n",
       "2  10376f1a-a28a-4885-8daa-c8ca496026bb          1              ES   \n",
       "3  fd308db7-0753-4377-879f-6ecf2af14e4f          1              FR   \n",
       "4  755fe256-a34d-4853-b7ca-d9bb991a86d3          1  GB||JE||IM||GG   \n",
       "\n",
       "   is_fraudster terms_version                created_date   state country  \\\n",
       "0         False    2018-05-25  2017-08-06 07:33:33.341000  ACTIVE      GB   \n",
       "1         False    2018-01-01  2017-03-07 10:18:59.427000  ACTIVE      GB   \n",
       "2         False    2018-09-20  2018-05-31 04:41:24.672000  ACTIVE      ES   \n",
       "3         False    2018-05-25  2018-06-01 17:24:23.852000  ACTIVE      FR   \n",
       "4         False    2018-09-20  2017-08-09 15:03:33.945000  ACTIVE      GB   \n",
       "\n",
       "   birth_year     kyc  failed_sign_in_attempts most_recent_tx_date  \\\n",
       "0        1971  PASSED                        0          2018-06-03   \n",
       "1        1982  PASSED                        0          2018-07-03   \n",
       "2        1973  PASSED                        0          2018-07-30   \n",
       "3        1986  PASSED                        0          2018-06-04   \n",
       "4        1989  PASSED                        0          2018-07-27   \n",
       "\n",
       "   num_tx_weekend  num_tx_night  max_tx_1d  max_amt_1d  max_tx_7d  \\\n",
       "0               4             0        2.0     37130.0        4.0   \n",
       "1               2             0        2.0     29607.0        6.0   \n",
       "2              10             1       13.0     34355.0       19.0   \n",
       "3               0             0        2.0       680.0        2.0   \n",
       "4              30            33       19.0   4497500.0       65.0   \n",
       "\n",
       "     max_amt_7d  max_tx_30d   max_amt_30d  \n",
       "0  3.713000e+04         4.0  3.713000e+04  \n",
       "1  9.664750e+03         7.0  9.664750e+03  \n",
       "2  3.641263e+03        52.0  2.340333e+03  \n",
       "3  6.800000e+02         2.0  6.800000e+02  \n",
       "4  3.483710e+06        78.0  1.671384e+06  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_final = df_users.merge(df_transactions_summary.dropna(\n",
    "), left_on='id', right_on='user_id', how='inner')\n",
    "df_final.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_final['created_date'] = pd.to_datetime(df_final['created_date']).dt.date\n",
    "df_final['account_age'] = (df_final['most_recent_tx_date'] - df_final['created_date']).dt.days\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>has_email</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>0.964097</td>\n",
       "      <td>0.186062</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.000</td>\n",
       "      <td>1.00</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>birth_year</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>1983.388951</td>\n",
       "      <td>11.387634</td>\n",
       "      <td>1927.0</td>\n",
       "      <td>1978.000</td>\n",
       "      <td>1986.00</td>\n",
       "      <td>1992.000000</td>\n",
       "      <td>2000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>failed_sign_in_attempts</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>0.009880</td>\n",
       "      <td>0.144632</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>6.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_tx_weekend</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>23.919148</td>\n",
       "      <td>46.427258</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.000</td>\n",
       "      <td>7.00</td>\n",
       "      <td>25.000000</td>\n",
       "      <td>684.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_tx_night</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>10.478848</td>\n",
       "      <td>25.302120</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000</td>\n",
       "      <td>2.00</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>636.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_tx_1d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>7.351934</td>\n",
       "      <td>5.982309</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.000</td>\n",
       "      <td>6.00</td>\n",
       "      <td>10.000000</td>\n",
       "      <td>95.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_amt_1d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>62238.908708</td>\n",
       "      <td>296612.273809</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2535.500</td>\n",
       "      <td>15246.00</td>\n",
       "      <td>42712.000000</td>\n",
       "      <td>10000000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_tx_7d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>16.259672</td>\n",
       "      <td>16.286849</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.000</td>\n",
       "      <td>11.00</td>\n",
       "      <td>24.000000</td>\n",
       "      <td>238.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_amt_7d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>44969.851189</td>\n",
       "      <td>228655.254665</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2132.875</td>\n",
       "      <td>11477.50</td>\n",
       "      <td>33982.750000</td>\n",
       "      <td>9762461.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_tx_30d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>30.877818</td>\n",
       "      <td>38.492401</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.000</td>\n",
       "      <td>17.00</td>\n",
       "      <td>43.000000</td>\n",
       "      <td>585.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_amt_30d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>27331.487589</td>\n",
       "      <td>103818.109528</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1679.250</td>\n",
       "      <td>7565.75</td>\n",
       "      <td>21240.340909</td>\n",
       "      <td>3255419.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>account_age</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>206.099917</td>\n",
       "      <td>223.966473</td>\n",
       "      <td>0.0</td>\n",
       "      <td>31.000</td>\n",
       "      <td>126.00</td>\n",
       "      <td>302.000000</td>\n",
       "      <td>939.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          count          mean            std     min  \\\n",
       "has_email                7186.0      0.964097       0.186062     0.0   \n",
       "birth_year               7186.0   1983.388951      11.387634  1927.0   \n",
       "failed_sign_in_attempts  7186.0      0.009880       0.144632     0.0   \n",
       "num_tx_weekend           7186.0     23.919148      46.427258     0.0   \n",
       "num_tx_night             7186.0     10.478848      25.302120     0.0   \n",
       "max_tx_1d                7186.0      7.351934       5.982309     1.0   \n",
       "max_amt_1d               7186.0  62238.908708  296612.273809     0.0   \n",
       "max_tx_7d                7186.0     16.259672      16.286849     1.0   \n",
       "max_amt_7d               7186.0  44969.851189  228655.254665     0.0   \n",
       "max_tx_30d               7186.0     30.877818      38.492401     1.0   \n",
       "max_amt_30d              7186.0  27331.487589  103818.109528     0.0   \n",
       "account_age              7186.0    206.099917     223.966473     0.0   \n",
       "\n",
       "                              25%       50%           75%         max  \n",
       "has_email                   1.000      1.00      1.000000         1.0  \n",
       "birth_year               1978.000   1986.00   1992.000000      2000.0  \n",
       "failed_sign_in_attempts     0.000      0.00      0.000000         6.0  \n",
       "num_tx_weekend              1.000      7.00     25.000000       684.0  \n",
       "num_tx_night                0.000      2.00      9.000000       636.0  \n",
       "max_tx_1d                   3.000      6.00     10.000000        95.0  \n",
       "max_amt_1d               2535.500  15246.00  42712.000000  10000000.0  \n",
       "max_tx_7d                   4.000     11.00     24.000000       238.0  \n",
       "max_amt_7d               2132.875  11477.50  33982.750000   9762461.5  \n",
       "max_tx_30d                  5.000     17.00     43.000000       585.0  \n",
       "max_amt_30d              1679.250   7565.75  21240.340909   3255419.0  \n",
       "account_age                31.000    126.00    302.000000       939.0  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_final.describe().T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2a54e6440>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 500x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig_distribution, ax = plt.subplots(1, 1, figsize=(5,5))\n",
    "\n",
    "groups = df_final.groupby('is_fraudster')\n",
    "for name, group in groups:\n",
    "    ax.scatter(group.max_tx_30d, group.max_tx_30d, edgecolors='k', label=name, alpha=1, marker='o')\n",
    "ax.legend(loc='upper left', bbox_to_anchor=(1.05,1), title = 'Is Fraudster')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Is Fraudster: [0 1]\n",
      "int64\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[Text(0, 0, '6889'), Text(0, 0, '297')]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x900 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Change the is_fraudster column dtype from bool to integer i.e., 0 and 1\n",
    "df_final['is_fraudster'] = df_final['is_fraudster'].replace([False, True], [0,1])\n",
    "\n",
    "# Visualize the target variable\n",
    "print(f'Is Fraudster: {df_final.is_fraudster.unique()}')\n",
    "print(df_final['is_fraudster'].dtype)\n",
    "\n",
    "ax = df_final['is_fraudster'].value_counts().plot(kind='bar', figsize=(12, 9))\n",
    "ax.bar_label(ax.containers[0])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "KYC: [1 0]\n",
      "int64\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[Text(0, 0, '6296'), Text(0, 0, '890')]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x900 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Replace all values of KYC column. Replace PASSED with 1, everything else to be 0\n",
    "df_final['kyc'] = np.where(df_final['kyc'].isin(['NONE', 'FAILED', 'PENDING']), 0, 1)\n",
    "\n",
    "# Visualize the KYC variable\n",
    "print(f'KYC: {df_final.kyc.unique()}')\n",
    "print(df_final['kyc'].dtype)\n",
    "\n",
    "ax = df_final['kyc'].value_counts().plot(kind='bar', figsize=(12, 9))\n",
    "ax.bar_label(ax.containers[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>has_email</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>0.964097</td>\n",
       "      <td>0.186062</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.000</td>\n",
       "      <td>1.00</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>is_fraudster</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>0.041330</td>\n",
       "      <td>0.199067</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>birth_year</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>1983.388951</td>\n",
       "      <td>11.387634</td>\n",
       "      <td>1927.0</td>\n",
       "      <td>1978.000</td>\n",
       "      <td>1986.00</td>\n",
       "      <td>1992.000000</td>\n",
       "      <td>2000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>kyc</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>0.876148</td>\n",
       "      <td>0.329435</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.000</td>\n",
       "      <td>1.00</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>failed_sign_in_attempts</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>0.009880</td>\n",
       "      <td>0.144632</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>6.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_tx_weekend</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>23.919148</td>\n",
       "      <td>46.427258</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.000</td>\n",
       "      <td>7.00</td>\n",
       "      <td>25.000000</td>\n",
       "      <td>684.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_tx_night</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>10.478848</td>\n",
       "      <td>25.302120</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000</td>\n",
       "      <td>2.00</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>636.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_tx_1d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>7.351934</td>\n",
       "      <td>5.982309</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.000</td>\n",
       "      <td>6.00</td>\n",
       "      <td>10.000000</td>\n",
       "      <td>95.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_amt_1d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>62238.908708</td>\n",
       "      <td>296612.273809</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2535.500</td>\n",
       "      <td>15246.00</td>\n",
       "      <td>42712.000000</td>\n",
       "      <td>10000000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_tx_7d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>16.259672</td>\n",
       "      <td>16.286849</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.000</td>\n",
       "      <td>11.00</td>\n",
       "      <td>24.000000</td>\n",
       "      <td>238.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_amt_7d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>44969.851189</td>\n",
       "      <td>228655.254665</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2132.875</td>\n",
       "      <td>11477.50</td>\n",
       "      <td>33982.750000</td>\n",
       "      <td>9762461.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_tx_30d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>30.877818</td>\n",
       "      <td>38.492401</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.000</td>\n",
       "      <td>17.00</td>\n",
       "      <td>43.000000</td>\n",
       "      <td>585.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_amt_30d</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>27331.487589</td>\n",
       "      <td>103818.109528</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1679.250</td>\n",
       "      <td>7565.75</td>\n",
       "      <td>21240.340909</td>\n",
       "      <td>3255419.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>account_age</th>\n",
       "      <td>7186.0</td>\n",
       "      <td>206.099917</td>\n",
       "      <td>223.966473</td>\n",
       "      <td>0.0</td>\n",
       "      <td>31.000</td>\n",
       "      <td>126.00</td>\n",
       "      <td>302.000000</td>\n",
       "      <td>939.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          count          mean            std     min  \\\n",
       "has_email                7186.0      0.964097       0.186062     0.0   \n",
       "is_fraudster             7186.0      0.041330       0.199067     0.0   \n",
       "birth_year               7186.0   1983.388951      11.387634  1927.0   \n",
       "kyc                      7186.0      0.876148       0.329435     0.0   \n",
       "failed_sign_in_attempts  7186.0      0.009880       0.144632     0.0   \n",
       "num_tx_weekend           7186.0     23.919148      46.427258     0.0   \n",
       "num_tx_night             7186.0     10.478848      25.302120     0.0   \n",
       "max_tx_1d                7186.0      7.351934       5.982309     1.0   \n",
       "max_amt_1d               7186.0  62238.908708  296612.273809     0.0   \n",
       "max_tx_7d                7186.0     16.259672      16.286849     1.0   \n",
       "max_amt_7d               7186.0  44969.851189  228655.254665     0.0   \n",
       "max_tx_30d               7186.0     30.877818      38.492401     1.0   \n",
       "max_amt_30d              7186.0  27331.487589  103818.109528     0.0   \n",
       "account_age              7186.0    206.099917     223.966473     0.0   \n",
       "\n",
       "                              25%       50%           75%         max  \n",
       "has_email                   1.000      1.00      1.000000         1.0  \n",
       "is_fraudster                0.000      0.00      0.000000         1.0  \n",
       "birth_year               1978.000   1986.00   1992.000000      2000.0  \n",
       "kyc                         1.000      1.00      1.000000         1.0  \n",
       "failed_sign_in_attempts     0.000      0.00      0.000000         6.0  \n",
       "num_tx_weekend              1.000      7.00     25.000000       684.0  \n",
       "num_tx_night                0.000      2.00      9.000000       636.0  \n",
       "max_tx_1d                   3.000      6.00     10.000000        95.0  \n",
       "max_amt_1d               2535.500  15246.00  42712.000000  10000000.0  \n",
       "max_tx_7d                   4.000     11.00     24.000000       238.0  \n",
       "max_amt_7d               2132.875  11477.50  33982.750000   9762461.5  \n",
       "max_tx_30d                  5.000     17.00     43.000000       585.0  \n",
       "max_amt_30d              1679.250   7565.75  21240.340909   3255419.0  \n",
       "account_age                31.000    126.00    302.000000       939.0  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_final.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Let us check for collinearity in the features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/wg/8yz1tny10yl9ct8bq7zl3pp00000gn/T/ipykernel_20873/3211041764.py:2: FutureWarning: The default value of numeric_only in DataFrame.corr is deprecated. In a future version, it will default to False. Select only valid columns or specify the value of numeric_only to silence this warning.\n",
      "  corr = df_final.corr()\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Axes: >"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1100x900 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Generate a mask for the upper triangle\n",
    "corr = df_final.corr()\n",
    "mask = np.triu(np.ones_like(corr, dtype=bool))\n",
    "\n",
    "# Set up the matplotlib figure\n",
    "f, ax = plt.subplots(figsize=(11,9))\n",
    "\n",
    "# Generate a custom diverging colormap\n",
    "cmap = sns.diverging_palette(230,20, as_cmap=True)\n",
    "\n",
    "# Draw the heatmap with the mask and correct aspect ratio\n",
    "sns.heatmap(corr, mask=mask, cmap=cmap, vmax=3, center=0, square=True, linewidths=.5, cbar_kws={'shrink': .5})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_numerical = df_final[['has_email', 'kyc', 'num_tx_night', 'max_tx_1d', 'max_amt_1d', 'account_age']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Printing the top k correlations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_redundant_pairs(df):\n",
    "    \"\"\"Get diagonal and lower triangular pairs of correlation matrix\"\"\"\n",
    "    pairs_to_drop = set()\n",
    "    cols = df.columns\n",
    "    for i in range(0, df.shape[1]):\n",
    "        for j in range(0, i+1):\n",
    "            pairs_to_drop.add((cols[i], cols[j]))\n",
    "    return pairs_to_drop\n",
    "\n",
    "def get_top_abs_correlations(df, n=5):\n",
    "    \"\"\"Sort correlations in the descending order and return n highest results\"\"\"\n",
    "    au_corr = df.corr().abs().unstack()\n",
    "    labels_to_drop = get_redundant_pairs(df)\n",
    "    au_corr = au_corr.drop(labels=labels_to_drop).sort_values(ascending=False)\n",
    "    return au_corr[0:n]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top Absolute Correlations\n",
      "num_tx_night  max_tx_1d       0.496761\n",
      "has_email     kyc             0.377029\n",
      "num_tx_night  account_age     0.356550\n",
      "max_tx_1d     account_age     0.333060\n",
      "kyc           max_tx_1d       0.249308\n",
      "              account_age     0.213524\n",
      "              num_tx_night    0.140611\n",
      "max_tx_1d     max_amt_1d      0.107204\n",
      "has_email     max_tx_1d       0.106634\n",
      "max_amt_1d    account_age     0.090724\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "print('Top Absolute Correlations')\n",
    "print(get_top_abs_correlations(df_numerical, 10))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Decision Tree with Cross-Validation\n",
    "- We will divide the dataset into train and test splits.\n",
    "- We will perform cross validation on the training dataset and measure:\n",
    "    - AUC ROC, Average Precision and Balanced Accuracy\n",
    "- We will do this for baseline (class weights of 1 for each class) and cost-sensitive (class weight = Imbalance ratio for class 0 because they're higher in number)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Size of training data is (5389, 6) and size of the test data is (1797, 6)\n"
     ]
    }
   ],
   "source": [
    "# Creating the train test split\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(df_final[df_numerical.columns.tolist()],\n",
    "                                                    df_final['is_fraudster'], test_size=0.25, random_state=42)\n",
    "\n",
    "print(f'Size of training data is {X_train.shape} and size of the test data is {X_test.shape}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a function to do cross validation and return results\n",
    "\n",
    "def kfold_cv_with_classifier(classifier, X, y, n_splits=5, strategy_name='Baseline Classifier'):\n",
    "    cv = StratifiedKFold(n_splits=n_splits, shuffle=True, random_state=42)\n",
    "    cv_results_ = cross_validate(classifier, X, y, cv=cv,\n",
    "                                 scoring=['roc_auc', 'average_precision', 'balanced_accuracy'], return_estimator=True)\n",
    "    \n",
    "    results = round(pd.DataFrame(cv_results_), 3)\n",
    "    results_mean = list(results.mean().values)\n",
    "    results_std = list(results.std().values)\n",
    "    results_df = pd.DataFrame([[str(round(results_mean[i], 3)) + '+/-' + str(round(results_std[i], 3)) for i in range(len(results))]],\n",
    "                              columns = ['Fit time (s)', 'Score time (s)', 'AUC ROC', 'Average Precision', 'Balanced Accuracy'])\n",
    "    results_df.rename(index={0:strategy_name}, inplace=True)\n",
    "\n",
    "    return results_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/wg/8yz1tny10yl9ct8bq7zl3pp00000gn/T/ipykernel_20873/550837358.py:9: FutureWarning: The default value of numeric_only in DataFrame.mean is deprecated. In a future version, it will default to False. In addition, specifying 'numeric_only=None' is deprecated. Select only valid columns or specify the value of numeric_only to silence this warning.\n",
      "  results_mean = list(results.mean().values)\n",
      "/var/folders/wg/8yz1tny10yl9ct8bq7zl3pp00000gn/T/ipykernel_20873/550837358.py:10: FutureWarning: The default value of numeric_only in DataFrame.std is deprecated. In a future version, it will default to False. In addition, specifying 'numeric_only=None' is deprecated. Select only valid columns or specify the value of numeric_only to silence this warning.\n",
      "  results_std = list(results.std().values)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Fit time (s)</th>\n",
       "      <th>Score time (s)</th>\n",
       "      <th>AUC ROC</th>\n",
       "      <th>Average Precision</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Decision Tree - Baseline</th>\n",
       "      <td>0.005+/-0.001</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.773+/-0.024</td>\n",
       "      <td>0.272+/-0.083</td>\n",
       "      <td>0.597+/-0.054</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                           Fit time (s) Score time (s)        AUC ROC  \\\n",
       "Decision Tree - Baseline  0.005+/-0.001    0.002+/-0.0  0.773+/-0.024   \n",
       "\n",
       "                         Average Precision Balanced Accuracy  \n",
       "Decision Tree - Baseline     0.272+/-0.083     0.597+/-0.054  "
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree = DecisionTreeClassifier(max_depth=5, class_weight={0:1,1:1}, random_state=42)\n",
    "\n",
    "results_df_dt_baseline = kfold_cv_with_classifier(tree, X_train, y_train, n_splits=5, strategy_name='Decision Tree - Baseline')\n",
    "\n",
    "results_df_dt_baseline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- The performances are rather good as the AUC ROC is > 0.5 and Average Precision is > 0.05. The balanced accuracy is however not so high.\n",
    "- Let us try to improve it by assigning class weights.\n",
    "- We know from the above result of the .describe() function on df_final that the mean of is_fraudster column = 0.041330\n",
    "- We will set that as the Imbalance Ratio (IR) = 0.041330"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "IR = 0.041330\n",
    "class_weight = {0:IR,1:1}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/wg/8yz1tny10yl9ct8bq7zl3pp00000gn/T/ipykernel_20873/550837358.py:9: FutureWarning: The default value of numeric_only in DataFrame.mean is deprecated. In a future version, it will default to False. In addition, specifying 'numeric_only=None' is deprecated. Select only valid columns or specify the value of numeric_only to silence this warning.\n",
      "  results_mean = list(results.mean().values)\n",
      "/var/folders/wg/8yz1tny10yl9ct8bq7zl3pp00000gn/T/ipykernel_20873/550837358.py:10: FutureWarning: The default value of numeric_only in DataFrame.std is deprecated. In a future version, it will default to False. In addition, specifying 'numeric_only=None' is deprecated. Select only valid columns or specify the value of numeric_only to silence this warning.\n",
      "  results_std = list(results.std().values)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Fit time (s)</th>\n",
       "      <th>Score time (s)</th>\n",
       "      <th>AUC ROC</th>\n",
       "      <th>Average Precision</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Decision Tree - Baseline</th>\n",
       "      <td>0.005+/-0.001</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.773+/-0.024</td>\n",
       "      <td>0.272+/-0.083</td>\n",
       "      <td>0.597+/-0.054</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Decision Tree - Cost-sensitive</th>\n",
       "      <td>0.005+/-0.0</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.783+/-0.015</td>\n",
       "      <td>0.253+/-0.048</td>\n",
       "      <td>0.72+/-0.02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 Fit time (s) Score time (s)        AUC ROC  \\\n",
       "Decision Tree - Baseline        0.005+/-0.001    0.002+/-0.0  0.773+/-0.024   \n",
       "Decision Tree - Cost-sensitive    0.005+/-0.0    0.002+/-0.0  0.783+/-0.015   \n",
       "\n",
       "                               Average Precision Balanced Accuracy  \n",
       "Decision Tree - Baseline           0.272+/-0.083     0.597+/-0.054  \n",
       "Decision Tree - Cost-sensitive     0.253+/-0.048       0.72+/-0.02  "
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree_cc = DecisionTreeClassifier(max_depth=5,class_weight=class_weight, random_state=42)\n",
    "\n",
    "results_df_dt_cost_sensitive = kfold_cv_with_classifier(tree_cc, X_train, y_train, n_splits=5, strategy_name='Decision Tree - Cost-sensitive')\n",
    "\n",
    "pd.concat([results_df_dt_baseline, results_df_dt_cost_sensitive])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Logistic Regression with Cross-Validation\n",
    "- Let us repeat the process for a Logistic Regression classifer (2 models: baseline, cost-sensitive)\n",
    "- Final comparison of the result for all 4 models is shown"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/wg/8yz1tny10yl9ct8bq7zl3pp00000gn/T/ipykernel_20873/550837358.py:9: FutureWarning: The default value of numeric_only in DataFrame.mean is deprecated. In a future version, it will default to False. In addition, specifying 'numeric_only=None' is deprecated. Select only valid columns or specify the value of numeric_only to silence this warning.\n",
      "  results_mean = list(results.mean().values)\n",
      "/var/folders/wg/8yz1tny10yl9ct8bq7zl3pp00000gn/T/ipykernel_20873/550837358.py:10: FutureWarning: The default value of numeric_only in DataFrame.std is deprecated. In a future version, it will default to False. In addition, specifying 'numeric_only=None' is deprecated. Select only valid columns or specify the value of numeric_only to silence this warning.\n",
      "  results_std = list(results.std().values)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Fit time (s)</th>\n",
       "      <th>Score time (s)</th>\n",
       "      <th>AUC ROC</th>\n",
       "      <th>Average Precision</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Decision Tree - Baseline</th>\n",
       "      <td>0.005+/-0.001</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.773+/-0.024</td>\n",
       "      <td>0.272+/-0.083</td>\n",
       "      <td>0.597+/-0.054</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Decision Tree - Cost-sensitive</th>\n",
       "      <td>0.005+/-0.0</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.783+/-0.015</td>\n",
       "      <td>0.253+/-0.048</td>\n",
       "      <td>0.72+/-0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Logistic Regression - Baseline</th>\n",
       "      <td>0.059+/-0.043</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.639+/-0.046</td>\n",
       "      <td>0.054+/-0.007</td>\n",
       "      <td>0.5+/-0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 Fit time (s) Score time (s)        AUC ROC  \\\n",
       "Decision Tree - Baseline        0.005+/-0.001    0.002+/-0.0  0.773+/-0.024   \n",
       "Decision Tree - Cost-sensitive    0.005+/-0.0    0.002+/-0.0  0.783+/-0.015   \n",
       "Logistic Regression - Baseline  0.059+/-0.043    0.002+/-0.0  0.639+/-0.046   \n",
       "\n",
       "                               Average Precision Balanced Accuracy  \n",
       "Decision Tree - Baseline           0.272+/-0.083     0.597+/-0.054  \n",
       "Decision Tree - Cost-sensitive     0.253+/-0.048       0.72+/-0.02  \n",
       "Logistic Regression - Baseline     0.054+/-0.007         0.5+/-0.0  "
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "logistic = LogisticRegression(C=1, class_weight={0:1,1:1}, random_state=42)\n",
    "results_df_lr_baseline = kfold_cv_with_classifier(logistic, X_train, y_train, n_splits=5, strategy_name='Logistic Regression - Baseline')\n",
    "\n",
    "pd.concat([results_df_dt_baseline, results_df_dt_cost_sensitive, results_df_lr_baseline])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- The logistic regression baseline performs worse than the decision tree\n",
    "- Let us adjust the class weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/wg/8yz1tny10yl9ct8bq7zl3pp00000gn/T/ipykernel_20873/550837358.py:9: FutureWarning: The default value of numeric_only in DataFrame.mean is deprecated. In a future version, it will default to False. In addition, specifying 'numeric_only=None' is deprecated. Select only valid columns or specify the value of numeric_only to silence this warning.\n",
      "  results_mean = list(results.mean().values)\n",
      "/var/folders/wg/8yz1tny10yl9ct8bq7zl3pp00000gn/T/ipykernel_20873/550837358.py:10: FutureWarning: The default value of numeric_only in DataFrame.std is deprecated. In a future version, it will default to False. In addition, specifying 'numeric_only=None' is deprecated. Select only valid columns or specify the value of numeric_only to silence this warning.\n",
      "  results_std = list(results.std().values)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Fit time (s)</th>\n",
       "      <th>Score time (s)</th>\n",
       "      <th>AUC ROC</th>\n",
       "      <th>Average Precision</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Decision Tree - Baseline</th>\n",
       "      <td>0.005+/-0.001</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.773+/-0.024</td>\n",
       "      <td>0.272+/-0.083</td>\n",
       "      <td>0.597+/-0.054</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Decision Tree - Cost-sensitive</th>\n",
       "      <td>0.005+/-0.0</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.783+/-0.015</td>\n",
       "      <td>0.253+/-0.048</td>\n",
       "      <td>0.72+/-0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Logistic Regression - Baseline</th>\n",
       "      <td>0.059+/-0.043</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.639+/-0.046</td>\n",
       "      <td>0.054+/-0.007</td>\n",
       "      <td>0.5+/-0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Logistic Regression - Cost-Sensitive</th>\n",
       "      <td>0.023+/-0.013</td>\n",
       "      <td>0.002+/-0.0</td>\n",
       "      <td>0.795+/-0.034</td>\n",
       "      <td>0.19+/-0.046</td>\n",
       "      <td>0.696+/-0.029</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                       Fit time (s) Score time (s)  \\\n",
       "Decision Tree - Baseline              0.005+/-0.001    0.002+/-0.0   \n",
       "Decision Tree - Cost-sensitive          0.005+/-0.0    0.002+/-0.0   \n",
       "Logistic Regression - Baseline        0.059+/-0.043    0.002+/-0.0   \n",
       "Logistic Regression - Cost-Sensitive  0.023+/-0.013    0.002+/-0.0   \n",
       "\n",
       "                                            AUC ROC Average Precision  \\\n",
       "Decision Tree - Baseline              0.773+/-0.024     0.272+/-0.083   \n",
       "Decision Tree - Cost-sensitive        0.783+/-0.015     0.253+/-0.048   \n",
       "Logistic Regression - Baseline        0.639+/-0.046     0.054+/-0.007   \n",
       "Logistic Regression - Cost-Sensitive  0.795+/-0.034      0.19+/-0.046   \n",
       "\n",
       "                                     Balanced Accuracy  \n",
       "Decision Tree - Baseline                 0.597+/-0.054  \n",
       "Decision Tree - Cost-sensitive             0.72+/-0.02  \n",
       "Logistic Regression - Baseline               0.5+/-0.0  \n",
       "Logistic Regression - Cost-Sensitive     0.696+/-0.029  "
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "logistic_cc = LogisticRegression(C=1, class_weight=class_weight, random_state=42)\n",
    "results_df_lr_cost_sensitive = kfold_cv_with_classifier(logistic_cc, X_train, y_train, n_splits=5, strategy_name='Logistic Regression - Cost-Sensitive')\n",
    "\n",
    "pd.concat([results_df_dt_baseline, results_df_dt_cost_sensitive, results_df_lr_baseline, results_df_lr_cost_sensitive])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- It looks like the cost-sensitive version of the Logistic Regression classifier is performing the best out of all.\n",
    "- Let us fit our training data to it and carry out predictions on test data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Before feature selection\n",
    "\n",
    "![Alt text](image.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.8 64-bit",
   "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": "3.10.8"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "aee8b7b246df8f9039afb4144a1f6fd8d2ca17a180786b69acc140d282b71a49"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}