UNB-Data-Sprint.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "328c2560",
   "metadata": {},
   "source": [
    "# Project Niagara\n",
    "The ever-dynamic Asian telecommunication industry is facing stiff competition amongst\n",
    "service providers with rapid introduction of innovative features and services such as airtime\n",
    "transfer, data sharing, mobile money, location aware apps, advanced social networking, and\n",
    "several OTT services. Fast growing telecom market in developing Asian countries is primarily\n",
    "a prepaid connections market. Among the various ways to reload a prepaid mobile phone\n",
    "through a refill card, online bank transfer, electronic reload at convenient store etc. peer to\n",
    "peer airtime transfer is emerging as a popular option. Peer to peer transfer involves a\n",
    "subscriber A transfer a part of their balance to subscriber B which can then be used for data,\n",
    "SMS, or voice call."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a96a4592",
   "metadata": {},
   "source": [
    "### Challenge\n",
    "The airtime transfer can be done in two ways: direct transfer and accepting a request for a\n",
    "transfer. Direct transfer involves a subscriber A initiating a transfer directly to subscriber B.\n",
    "Request transfer involves subscriber B raising a request for a certain amount from subscriber\n",
    "A and then subscriber A either accepts or rejects the transfer request. Typically, the transfer is\n",
    "done through an SMS in a specific format to a toll-free number. Many a times the transfer\n",
    "fails for variety of reasons such as incorrect amount (Only integer amount of $1 - $25 is\n",
    "allowed), incorrect mobile number, sending airtime to a post postpaid number, technical\n",
    "glitches etc. <b>The objective of this data challenge is to characterize the users and identify\n",
    "the drivers of airtime transfer to enhance the count and value of successful transactions.</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "9993add3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T17:34:20.378028Z",
     "start_time": "2021-11-26T17:34:19.253283Z"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import sys\n",
    "\n",
    "from tqdm.notebook import tqdm\n",
    "from scipy.stats.stats import pearsonr\n",
    "from sklearn import cluster, preprocessing, metrics, model_selection, tree\n",
    "from bioinfokit.analys import stat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5f9b2130",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T17:34:46.301880Z",
     "start_time": "2021-11-26T17:34:20.379021Z"
    }
   },
   "outputs": [],
   "source": [
    "df = pd.read_csv('data/Sprint Data Set.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "5284179a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T17:34:46.308333Z",
     "start_time": "2021-11-26T17:34:46.304239Z"
    }
   },
   "outputs": [],
   "source": [
    "# makes indexing easier\n",
    "df.columns = [i.lower() for i in df.columns]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aeff56bc",
   "metadata": {},
   "source": [
    "Let's start by dividing the columns into meta data (transactions) and customer data to avoid confusion."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "c62bc9a3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T17:34:46.644131Z",
     "start_time": "2021-11-26T17:34:46.310330Z"
    }
   },
   "outputs": [],
   "source": [
    "cust_df = pd.DataFrame()\n",
    "cust_df['uid'] = df.uid\n",
    "cust_df['cnt_uid'] = df.cnt_uid\n",
    "cust_df['activedays'] = df.activedays\n",
    "cust_df['total_amt'] = df.total_amount"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "1b7066a0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T17:34:48.071176Z",
     "start_time": "2021-11-26T17:34:46.644911Z"
    }
   },
   "outputs": [],
   "source": [
    "transaction_cols = [i for i in df.columns if 'mn' in i]\n",
    "transactions_df = df.loc[:, ['uid'] + transaction_cols[8:]]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d52c8861",
   "metadata": {},
   "source": [
    "Also, since the data was collected in a real-time environment, there are a few cases where there are inconsistencies in the columns like `tx` and `success` which indicate the total number of transactions and total number of successful transactions respectively. Therefore, these columns are re-calculated from the rest of the data provided\n",
    "\n",
    "*Note: This information was verified by the moderator of the competition.*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "86982a86",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T17:34:49.097201Z",
     "start_time": "2021-11-26T17:34:48.073310Z"
    }
   },
   "outputs": [],
   "source": [
    "cust_df['tx'] = transactions_df.iloc[:,1:9].sum(axis=1)\n",
    "cust_df['suc_tx'] = transactions_df.iloc[:,17:].sum(axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "178f0ec1",
   "metadata": {},
   "source": [
    "We start by investigating the transaction data, firstly let's check whether there are any users who don't have a single successful transaction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "f5990271",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T17:34:49.514155Z",
     "start_time": "2021-11-26T17:34:49.098716Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.37119894148277777"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cust_df[cust_df.suc_tx == 0].shape[0] / cust_df.shape[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8aaa84e6",
   "metadata": {},
   "source": [
    "Interestingly, 37% users don't even have a single successful transaction"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7047e1ab",
   "metadata": {},
   "source": [
    "Since, the customer-level transaction data is given for a period of 8 months, it would be interesting to see the month-wise trend in the data. This calculation is a bit complex and involves grouping and slicing operations."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a51e5e25",
   "metadata": {},
   "source": [
    "Here's a look at the table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "99c7e93c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:48:00.335807Z",
     "start_time": "2021-11-26T18:47:57.878334Z"
    }
   },
   "outputs": [],
   "source": [
    "suc_cols = transactions_df.iloc[:,17:].columns\n",
    "\n",
    "tx_mn = transactions_df[suc_cols].sum().reset_index(name='cnt')\n",
    "tx_mn['tx_type'] = tx_mn['index'].apply(lambda x: x.split('_')[1])\n",
    "tx_mn['month'] = tx_mn['index'].apply(lambda x: x.split('n')[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "4bf0a6ca",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:48:00.344526Z",
     "start_time": "2021-11-26T18:48:00.337459Z"
    }
   },
   "outputs": [],
   "source": [
    "tx_mn = tx_mn.groupby(['tx_type', 'month'])['cnt'].sum().unstack('tx_type')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "id": "7353f074",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:48:00.350190Z",
     "start_time": "2021-11-26T18:48:00.345224Z"
    }
   },
   "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>tx_type</th>\n",
       "      <th>ra</th>\n",
       "      <th>rb</th>\n",
       "      <th>ta</th>\n",
       "      <th>tb</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>month</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>634246</td>\n",
       "      <td>634246</td>\n",
       "      <td>7372594</td>\n",
       "      <td>7373422</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>507185</td>\n",
       "      <td>507184</td>\n",
       "      <td>6342612</td>\n",
       "      <td>6343615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>538862</td>\n",
       "      <td>538862</td>\n",
       "      <td>6759832</td>\n",
       "      <td>6760975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>490808</td>\n",
       "      <td>490807</td>\n",
       "      <td>6034889</td>\n",
       "      <td>6035866</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>478420</td>\n",
       "      <td>478418</td>\n",
       "      <td>6081518</td>\n",
       "      <td>6082517</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>405210</td>\n",
       "      <td>405210</td>\n",
       "      <td>5340496</td>\n",
       "      <td>5341387</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>406012</td>\n",
       "      <td>407809</td>\n",
       "      <td>5246627</td>\n",
       "      <td>5248129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>52292</td>\n",
       "      <td>52528</td>\n",
       "      <td>673933</td>\n",
       "      <td>674155</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "tx_type      ra      rb       ta       tb\n",
       "month                                    \n",
       "1        634246  634246  7372594  7373422\n",
       "2        507185  507184  6342612  6343615\n",
       "3        538862  538862  6759832  6760975\n",
       "4        490808  490807  6034889  6035866\n",
       "5        478420  478418  6081518  6082517\n",
       "6        405210  405210  5340496  5341387\n",
       "7        406012  407809  5246627  5248129\n",
       "8         52292   52528   673933   674155"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tx_mn"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7965ed41",
   "metadata": {},
   "source": [
    "Now for the trend"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "id": "10490d86",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:48:00.504647Z",
     "start_time": "2021-11-26T18:48:00.351336Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(12, 6))\n",
    "ax1 = ax.twinx()\n",
    "\n",
    "p1 = ax.plot(tx_mn.ta, data=tx_mn, marker='', markersize=15, color='olive', linewidth=2, \n",
    "             linestyle='dashed', label='Direct Transfer Sent')\n",
    "p2 = ax.plot(tx_mn.tb, data=tx_mn, marker='o', color='blue', linewidth=2, \n",
    "             alpha=.25, label='Direct Transfer Received')\n",
    "\n",
    "p3 = ax1.plot(tx_mn.ra, data=tx_mn, marker='', color='green', linewidth=2, \n",
    "              linestyle=':', label='Request Transfer Sent')\n",
    "p4 = ax1.plot(tx_mn.rb, data=tx_mn, marker='*', markersize=15, color='red', \n",
    "              linewidth=2, alpha=.3, label='Request Transfer Received')\n",
    "\n",
    "lns = p1 + p2 + p3 + p4\n",
    "labs = [l.get_label() for l in lns]\n",
    "ax.legend(lns, labs, loc=[.2,.3])\n",
    "\n",
    "plt.title('Trend in transfers by month');\n",
    "ax.set_xlabel('Months');\n",
    "\n",
    "plt.savefig('assets/month-trend-fig1.jpeg', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a9acd10",
   "metadata": {},
   "source": [
    "Overall there is a decreasing trend in the number of transactions across months"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a45f1ae",
   "metadata": {},
   "source": [
    "Previously created table would also help us determine what proportion of transactions are successful"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "d8ee5db8",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T17:21:21.664165Z",
     "start_time": "2021-11-26T17:21:21.661830Z"
    }
   },
   "outputs": [],
   "source": [
    "tx_mn['suc_tx'] = tx_mn.sum(axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "caaab045",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T17:21:21.760959Z",
     "start_time": "2021-11-26T17:21:21.665191Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.2586547689367111"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total_tx = transactions_df.iloc[:,1:9].sum()\n",
    "tx_mn.suc_tx.sum() / total_tx.sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d222bb5a",
   "metadata": {},
   "source": [
    "On average, only 1 in 4 transactions is successful "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eb6add72",
   "metadata": {},
   "source": [
    "Let's also have a look at the proportion of direct and request transfers in our dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "4a54c0aa",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T15:28:25.107033Z",
     "start_time": "2021-11-26T15:28:25.096186Z"
    }
   },
   "outputs": [],
   "source": [
    "tx_mn.columns = ['sent_rqtx', 'rec_rqtx', 'sent_dttx', 'rec_dttx', 'suc_tx']\n",
    "tx_mn['dtx_total'] = tx_mn['sent_dttx'] + tx_mn['rec_dttx']\n",
    "tx_mn['rqtx_total'] = tx_mn['sent_rqtx'] + tx_mn['rec_rqtx']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "7953e70c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T15:28:25.291400Z",
     "start_time": "2021-11-26T15:28:25.281430Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.9258175, 0.0741825])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(tx_mn.dtx_total.sum(), tx_mn.rqtx_total.sum()) / tx_mn.suc_tx.sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cfe6e64c",
   "metadata": {},
   "source": [
    "Here the values denote number of direct and request transfers respectively. Large proportion of customers prefer direct transfers over request transfers"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c2ac9474",
   "metadata": {},
   "source": [
    "Now that we understand the customer transactions better, let's have a look at the customer table and analyze if there's any relationship between the variables. More specifically we're aiming to see if any linear relationship exists among the columns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "807729d7",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:42:32.346259Z",
     "start_time": "2021-11-26T18:42:30.961660Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8));\n",
    "sns.heatmap(cust_df.iloc[:, 1:].corr(), annot=True);\n",
    "\n",
    "plt.savefig('assets/cust-corr-fig2.jpeg', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "a42b214c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T15:45:52.830381Z",
     "start_time": "2021-11-26T15:45:52.563375Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.7827031132599644, 0.0)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pearsonr(cust_df.suc_tx, cust_df.total_amt)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ff7c7261",
   "metadata": {},
   "source": [
    "- The above values indicate the correlation between successful transactions and the total amount transferred for a particular customer, and its corresponding p-value respectively. The values indicate that there is a significantly strong positive correlation between these columns.\n",
    "\n",
    "- Also, from the heatmap it can be observed that majority of the transactions involve new people as `cnt_uid` column has a positive correlation with the total number of transactions."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4ed2641",
   "metadata": {},
   "source": [
    "Let's characterize the users using clustering. More specifically using KMeans clustering methodology. It is crucial to normalize the columns during preprocessing due to different scales, as KMeans is a distance-based algorithm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "ccb7d96e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:42:37.095058Z",
     "start_time": "2021-11-26T18:42:36.622013Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9371600, 6)"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scaler = preprocessing.MinMaxScaler()\n",
    "scaled_arr = scaler.fit_transform(cust_df)\n",
    "\n",
    "scaled_arr.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "67b29acc",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:42:37.107248Z",
     "start_time": "2021-11-26T18:42:37.095980Z"
    }
   },
   "outputs": [],
   "source": [
    "scaled_df = pd.DataFrame(scaled_arr, columns=cust_df.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "41c0664b",
   "metadata": {},
   "source": [
    "We use elbow-plot to determine the optimal number of clusters for KMeans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "18218377",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:46:57.353842Z",
     "start_time": "2021-11-26T18:43:39.053944Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c087ff11d3244113a8d682fca6a12451",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/10 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cs = []\n",
    "\n",
    "values = range(1, 11)\n",
    "with tqdm(total=len(values), file=sys.stdout) as pbar:\n",
    "    for i in values:\n",
    "        pbar.set_description(f\"processing k={i}\")\n",
    "        pbar.update()\n",
    "        \n",
    "        kmeans = cluster.KMeans(n_clusters=i, max_iter=300, n_init=10, random_state=42)\n",
    "        kmeans.fit(scaled_df.iloc[:, 1:])\n",
    "        cs.append(kmeans.inertia_)\n",
    "\n",
    "plt.figure(figsize=(8, 6))\n",
    "plt.plot(range(1, 11), cs)\n",
    "plt.title('The Elbow Method', fontsize=16)\n",
    "plt.xlabel('Number of clusters', fontsize=12)\n",
    "plt.ylabel('CS', fontsize=12)\n",
    "\n",
    "plt.savefig('assets/optimal-cluster-fig3.jpeg', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cd596777",
   "metadata": {},
   "source": [
    "We observe from the above plot that 3 would be the optimal number of clusters for our case."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6c659f80",
   "metadata": {},
   "source": [
    "Fitting the algorithm..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "id": "de49973f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T19:25:58.477611Z",
     "start_time": "2021-11-26T19:25:47.506133Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KMeans(n_clusters=3, random_state=42)"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kmeans = cluster.KMeans(n_clusters=3, random_state=42) \n",
    "kmeans.fit(scaled_df.iloc[:,1:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "874ba7dc",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T19:25:58.489973Z",
     "start_time": "2021-11-26T19:25:58.479002Z"
    }
   },
   "outputs": [],
   "source": [
    "cust_df['cluster'] = kmeans.labels_\n",
    "scaled_df['cluster'] = kmeans.labels_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "e3ee24e3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T19:25:58.523455Z",
     "start_time": "2021-11-26T19:25:58.490766Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    0.473282\n",
       "1    0.357523\n",
       "2    0.169195\n",
       "Name: cluster, dtype: float64"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cust_df.cluster.value_counts(normalize=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fff2f004",
   "metadata": {},
   "source": [
    "Here's a look at the clusters and its characterics for different columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "a7aeda14",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T15:46:37.836925Z",
     "start_time": "2021-11-26T15:46:37.029689Z"
    }
   },
   "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>cnt_uid</th>\n",
       "      <th>activedays</th>\n",
       "      <th>total_amt</th>\n",
       "      <th>tx</th>\n",
       "      <th>suc_tx</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cluster</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>0</th>\n",
       "      <td>4.327881</td>\n",
       "      <td>7.333546</td>\n",
       "      <td>6.427707</td>\n",
       "      <td>4.325501</td>\n",
       "      <td>1.112377</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>92.512831</td>\n",
       "      <td>193.516409</td>\n",
       "      <td>55.570050</td>\n",
       "      <td>91.677877</td>\n",
       "      <td>23.953850</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>25.305886</td>\n",
       "      <td>96.190044</td>\n",
       "      <td>20.740651</td>\n",
       "      <td>25.179140</td>\n",
       "      <td>6.021600</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           cnt_uid  activedays  total_amt         tx     suc_tx\n",
       "cluster                                                        \n",
       "0         4.327881    7.333546   6.427707   4.325501   1.112377\n",
       "1        92.512831  193.516409  55.570050  91.677877  23.953850\n",
       "2        25.305886   96.190044  20.740651  25.179140   6.021600"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_mean = cust_df.groupby('cluster')[scaled_df.columns[1:-1]].mean()\n",
    "df_mean"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5d7db211",
   "metadata": {},
   "source": [
    "The clusters are quite different from each other with respect to the mean value of the features. However let's check if these differences are significant using ANOVA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "id": "5c7797f2",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-19T14:32:26.766824Z",
     "start_time": "2021-11-19T14:29:40.732368Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "64f914f29e1146168070413bcf850bed",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/5 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m Col = cnt_uid \u001b[0m\n",
      "                   df        sum_sq       mean_sq             F  PR(>F)\n",
      "C(cluster)        2.0  1.522260e+10  7.611302e+09  47570.489501     0.0\n",
      "Residual    9371597.0  1.499460e+12  1.600005e+05           NaN     NaN\n",
      "\n",
      "   group1  group2       Diff      Lower      Upper     q-value  p-value\n",
      "0       2       0  20.978005  20.110581  21.845429   80.158685    0.001\n",
      "1       2       1  67.206945  66.303296  68.110595  246.508642    0.001\n",
      "2       0       1  88.184951  87.506381  88.863520  430.742777    0.001\n",
      "\n",
      "\n",
      "\u001b[1m Col = activedays \u001b[0m\n",
      "                   df        sum_sq       mean_sq             F  PR(>F)\n",
      "C(cluster)        2.0  6.626394e+10  3.313197e+10  9.790829e+07     0.0\n",
      "Residual    9371597.0  3.171330e+09  3.383980e+02           NaN     NaN\n",
      "\n",
      "   group1  group2        Diff       Lower       Upper       q-value  p-value\n",
      "0       2       0   88.856498   88.816606   88.896390   7382.821651    0.001\n",
      "1       2       1   97.326366   97.284808   97.367923   7762.387032    0.001\n",
      "2       0       1  186.182864  186.151657  186.214070  19774.705623    0.001\n",
      "\n",
      "\n",
      "\u001b[1m Col = total_amt \u001b[0m\n",
      "                   df        sum_sq       mean_sq              F  PR(>F)\n",
      "C(cluster)        2.0  4.670997e+09  2.335498e+09  472249.958543     0.0\n",
      "Residual    9371597.0  4.634696e+10  4.945471e+03            NaN     NaN\n",
      "\n",
      "   group1  group2       Diff      Lower      Upper      q-value  p-value\n",
      "0       2       0  14.312944  14.160442  14.465445   311.080081    0.001\n",
      "1       2       1  34.829399  34.670529  34.988270   726.642547    0.001\n",
      "2       0       1  49.142343  49.023044  49.261642  1365.325600    0.001\n",
      "\n",
      "\n",
      "\u001b[1m Col = tx \u001b[0m\n",
      "                   df        sum_sq       mean_sq             F  PR(>F)\n",
      "C(cluster)        2.0  1.493326e+10  7.466632e+09  46680.592349     0.0\n",
      "Residual    9371597.0  1.499001e+12  1.599515e+05           NaN     NaN\n",
      "\n",
      "   group1  group2       Diff      Lower      Upper     q-value  p-value\n",
      "0       2       0  20.853640  19.986348  21.720931   79.695670    0.001\n",
      "1       2       1  66.498737  65.595226  67.402248  243.948337    0.001\n",
      "2       0       1  87.352376  86.673911  88.030842  426.741347    0.001\n",
      "\n",
      "\n",
      "\u001b[1m Col = suc_tx \u001b[0m\n",
      "                   df        sum_sq       mean_sq             F  PR(>F)\n",
      "C(cluster)        2.0  1.027724e+09  5.138621e+08  1.059767e+06     0.0\n",
      "Residual    9371597.0  4.544122e+09  4.848823e+02           NaN     NaN\n",
      "\n",
      "   group1  group2       Diff      Lower      Upper      q-value  p-value\n",
      "0       2       0   4.909222   4.861471   4.956974   340.754495    0.001\n",
      "1       2       1  17.932250  17.882504  17.981996  1194.800167    0.001\n",
      "2       0       1  22.841473  22.804117  22.878828  2026.703045    0.001\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "values = cust_df.columns[1:-1]\n",
    "with tqdm(total=len(values), file=sys.stdout) as pbar:\n",
    "    for i in values:\n",
    "        pbar.set_description(f\"processing col={i}\")\n",
    "        pbar.update()\n",
    "        \n",
    "        res = stat()\n",
    "        res.anova_stat(df=cust_df, res_var=i, anova_model=f'{i} ~ C(cluster)')\n",
    "        print(f'\\033[1m Col = {i} \\033[0m')\n",
    "        print(f'{res.anova_summary}\\n')\n",
    "        \n",
    "        res.tukey_hsd(df=cust_df, res_var=i, xfac_var='cluster', anova_model=f'{i} ~ C(cluster)')\n",
    "        print(f'{res.tukey_summary}\\n\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9337304c",
   "metadata": {},
   "source": [
    "The above table shows the signifance of the difference between clusters for various columns and also shows which clusters are different from each other using Tukey's HSD Test."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5c9083bd",
   "metadata": {},
   "source": [
    "Let's move on to the second part of the problem where we determine the factors that affect the count and value of a successful transaction. \n",
    "\n",
    "We have already identified that count and value of a successful transaction are correlated therefore we exclude `total_amt` column from the analysis and build a model on the following features to study its effect on the target (num of successful transactions):\n",
    "- `cnt_uid`\n",
    "- `activedays`\n",
    "- `tx`"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b16fcf78",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:23:20.596929Z",
     "start_time": "2021-11-26T18:23:20.585339Z"
    }
   },
   "source": [
    "In order to study the effect of different features on the number of successful transactions, a transaction first has to be successful. Therefore we'll subset the dataframe to include only those transactions which were successful."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "baebfedc",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:25:12.222438Z",
     "start_time": "2021-11-26T18:25:11.763324Z"
    }
   },
   "outputs": [],
   "source": [
    "suc_tx = cust_df[cust_df.suc_tx > 0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "230a99e2",
   "metadata": {},
   "source": [
    "Let's plot the distribution of our target."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "8976e688",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:42:24.088054Z",
     "start_time": "2021-11-26T18:42:23.339785Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Pass the following variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdAAAAFSCAYAAABCJY7TAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAAAc+UlEQVR4nO3de5ycVZ3n8c+vu3ODwMuQEIIQaLIhTFiRGUVHFJVhcSVBwcuw8nIlyWtF3HUHEJURSAOJZF6OF8DIrOOyOEsEHS8zooAEJTrMiOtlg4LDDMPSaiNgICEY5RrTnbN/PE81T1dX9eWkOpXL5/161avrOXWec85zqtLfek49nYqUEpIkaXw62j0ASZJ2RwaoJEkZDFBJkjIYoJIkZTBAJUnKYIBKkpTBAFVDEbEsIlLl9kxE9EXETRHxnyKio65+d1lv2Tj6ODEiVtS3NcZxdVfK+iLixrG2kTuunGPcmSKiIyI+FREbImJ7RHy9Sb0Xlcf3sgaP3RkRd03gGNMYbn0T1X+OiHh/RLyt3ePQrqer3QPQLu8M4BFgCnAYcCrwt8A5EfHmlNJzZb0NwPHAz8fR9onA5cAqYPsY9/lm2c+GcfQzXifSeFw5x7gz/SlwPvBB4AfA5ib1XkRxfI8AP9kpI3vB8XXbNwH3AisqZVt32mjG5v3AXcDX2jwO7WIMUI3mnpRSb2X7hoj4KvBV4OPAuQAppa3ADydqEBExCehPKW0CNk1UPyOZ6GNsgYXlz0+llMb6hmSnSikNmb+I2Ao8UV+eIyKmlM+RtFO4hKtxSyn9PfAN4D0RsQ80Xt6MiFdExB0RsTkino2IX0TEZ8rHVlCcBQFsqy3f1bX1voj4eET8muKs5EWNlnAr/b0nInoj4vmI+ElE/End43dGxJ0N9uuLiOvHMa5ldfu/KyLuLft9IiJuiIiDG/RxY0ScGRH3l0vi6yPihFEnvNj/lIj4QUQ8FxG/jYivR8RR1fZ54SxuoNlSczlvvyw3/1dl2bT+mE4u5/DZiLgvIt7SoK1jI+LmiPhNOa7vR8Rrx3I8Ixzn1Ii4uuzz6Yh4LCJuiYg/qKtXex28LiK+GhFbgB+Vj+0TEX9dvu6eiuJjh1c3Oc7XR8R3ynrPRMS3IuIllcf7gMOB/1yZq+t35Bi15zBAles2imXd4xo9GBHTgW8BA8AyYDHwEV5Y9bgO+Fx5/wSKpb365b3lwALgHOCtwPMjjOf1wAfKfc6kCNy11ZAZo7GMa1BEnAPcANwPvA24CHgj8I/lHFS9lmJ59VLgHUAncGtEvGikAUXEKRRL10+X+/034CXAXRFxSFntrcD15f3amL/ZoLkN5TgBPtqk7r8DVgNXlXU3AH8XEfMrY3oZ8H+AA4D3AG+nWDJeFxEvH+l4RjEF2I9i+fxUimOdCvwwIuY0qP8FijcEf0ox9wDXAv8F+GQ5/gfKekNExKnAdyjm9V3AO8u+vxcRc8tqbwUeo3gt1+bqih04Pu1JUkrevA27UYReAuY3efyN5ePvKLe7y+1l5fZx5fZLR+hjRVmnq6681tZPgGgyru5KWR/we+CwStl+wJPADZWyO4E7G4yjD7h+HOOqHWMn8DjwD3X1TijrnVfXx2+AGZWy2hy9c5TnYj3wYHU8wBHANuCqStmq4p/0qM9t7TjObvDYnWW7R1bKZlO8EbqkUvYdijcNkytlnWXZ18fxOusDbhzh8U5gH+Ap4IIGr4Or6+ofRfG59Z/XlX+6+tyVZb3Ad+rq7Q88QbEMPqYxett7b56BKleUP5t9G8GDwBbgf5ZLnHOb1BvJ11NKY/22gx+mlH5V20gpPcULFxxNlKMowmXI2U1K6S7gIYqz4qofpJR+U9n+5/LnYc06iIh9gZcBX04p9Vf6+CXw/QZ9tMKDKaUHK31tBDbWxhkR08p+vwpsj4iuiOiieE2sA163I51HcZX3j8pl2X7gGWA6xXzXu6lu+4/LcXy1rvzv6vo4kuJM+wu18ZfH8CzFBVg7dAzaOxigylULxIZXw6aUfgv8CfBr4DPAr8rPtd4+jj7Gc6Xt403KDmlQ3ioHlD8bjfOxyuM1T1Y30gsXvEwdoY8ZFIEw1j5a4ckGZVt5YZwHUJwZXkpxtlq9/RkwI8bxp0lVEfFm4MsUZ7LvpAjEV1BcONZonurnpfbZ88a68vrXx+zy5+cYfgxvAmZmDF97Ga/CVa5TKT6TvLtZhZTSPcDby3f2xwEXA1+JiGNTSveNoY/xfNfeQU3KHq1sP0+xRFcvN4RqQdPos7k5FEuvO+o3FPPQrI9mf6oykbZQLJP+D+DzjSqk/KuAzwR6U0rLagVRXIHd7Dmqf43UAnU2L1wsBcNfH7V5u5jirLne78cyWO3dPAPVuEXxR+WnAZ9NKT07Wv2UUn8q/kzhUorXXO3PLWpnYNNaMKxXVZeJI2I/ipD/QaXOQ8CCiJhcqfc6is9Lq8Y6rgcozmzOrBZGxKsprtz8x/EcQCMppWco3qScERGdlT4OB16d2ccOzXs5pu8BxwI/SSmtr7/ltFvah2LZtuosijPesfgRRaieUVdev/0AxWeb/77R+FNKP6vU3UprXqPaw3gGqtH8YUTMAiZTfAb2JopfRndQvHtvKCLeRHH17NcpzgT2Bc6juBikFmr/Wv78YESsBQZ24Jfv48C3yz9D2Qp8uOyzesXkl8ox/U35pwhHUFy5+9u6tsY0rpTSQERcRvE5743AjRRLxn9B8Rnw/848lnqXUnyee2sUfwY0HVhZjvvKjPYepzgDOzMifkbxGeMvU0rjOZv9APBPwLci4nMUZ36zKD6v7UwpXTTSziO4HXhLRFwN3Aq8nOJ1s2UsO6eUHoiILwJXlMvIdwMnAW8uq2wv66WI+O/AN8o3VF+huHjoIIo3Jr9KKV1V7vOvwGvL1/RjFH+32pd5fNqTtPsqJm+75o0XrnKs3Z6jOIO7iSJA66+O7WboFapHUXyW9UuKpdNNFH/68seVfToplgE3UvxiS3VtNbpKtDau7kpZH0V4nU3xvwRtBX4KnNRg//dShNtzFH+G8XKGX4U72riW1bX5Lor/TWcrRTDdABxcV6ePBldylu2tGMPzcQrFG4/nKILzG8BRdXXGdBVuWfctFMGwre55uxO4q0H9IXNUli2keFOysTz2R4CbgcXjeJ0NmReKFYpVFJ+dP0txhv1HDZ6j2utg2FXiFGexf02xxP50OaZTy/qn19U9niKof1O+TvvKYzq+UucPKM64ny3buH6sx+dtz75FSuP5mEmSdj8RcSHwMYo3Xr8arb40Fi7hStqjlEutLwHuoVhBeC3wIeArhqdayQCVtKd5imKJ+iKKz8EfpfiPFC4fYR9p3FzClSQpg3/GIklSBgNUkqQM4/oMdNasWam7u3uChiJJ0q7l7rvvfiKldGCjx8YVoN3d3axf34r/nUySpF1fRDzU7DGXcCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjIYoJIkZTBAJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGXoalfHZ599Nlu2bOH1r3895557bruGIUlSlrYF6IYNG3jmmWfo7e1t1xAkScrmEq4kSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjIYoJIkZTBAJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjIYoJIkZTBAJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlKFtAbp169Yh29dccw3XXHNNm0YjSdL4dLWr4+3btw/Z7u3tbdNIJEkaP5dwJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjIYoJIkZTBAJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjIYoJIkZTBAJUnK0NXuAdTce++9AJx44ontHcgubP/99+d3v/sdALNnz2bjxo0AzJgxgy1btjBp0iQ6Ojp43/vex9VXX01XVxednZ3st99+bNq0aVh7kydPZs6cOWzatIk5c+YwMDDAww8/zEEHHcTjjz9OV1cXHR0ddHR0cNZZZ3Httdcya9YsNm/ezItf/GKefPJJLrzwQq688kpWr17NjBkz6OnpISK44oorALjwwgv5xS9+weTJk+nu7ua9730vy5cv5/nnn2fevHl84hOf4N577+UjH/kIl19+OXPnzuX8889n9erVzJ8/f3Csmzdv5uKLL+bhhx9mzpw5TJs2jTPOOIMrrriCSy+9lJtuuoklS5Zw2WWXccghh/DhD3+YK6+8koGBAQA6Ozt597vfzeWXXz6sbYDe3l7OP/98PvjBD3LllVfyoQ99iE9+8pODx7Vy5UrOO+88Pv3pT3P55Zczc+bMYfPZ29vLueeey9y5c/noRz/KzJkz2bx5MytXrhy2T6PyahlAT08P/f39DAwM8Oijj9LR0cGqVau47rrrGBgYoLOzk1WrVg0bS307tbHX5qPRfrV9lixZMjhHteNuNsaRykYy0pxUXz/N2hprf/X1xjvO3VmrnpOJ7rfVdnb/kVIac+XjjjsurV+/viUdn3TSSWzfvp1jjz2W1atXG5wtFBGM53ndEV1dXfT399Pd3c1LX/pSbr75ZgBOP/10UkqD2zXTp0/n6aefHtw+/fTT+eY3v0l/fz9dXV0ceuih9PX10d3dzfXXXz9Y76qrrhrWVq3vrq4uBgYG2HfffQfb7u7upq+vr2Hf9W0DLFu2jL6+viFtVo/rlltu4fDDD+ehhx7itNNO44ILLhg2F7U2asd1wQUXcNVVV3HLLbcM26dRebWs0dw1m7/6sdS3Uxt7dT7q96vtU5vD6nE3G+NIZSMZaU6qr59mbY21v/p64x3n7qxVz8lE99tqE9F/RNydUjqu0WO7xBKu4dlaOys8Afr7+wHo6+vjtttuGyy/7bbbhmzXVH/5A9x8882DbfT39w/+ku/r66O3txco3lWuXbu2ad/9/f2klIa0XR+e1b6rbUNx5lirX22zVnft2rWklOjr6yOlxO23387mzZuHtF1to3b8vb293H777cP22bx587DyatnatWsbHm+j+Vu7du2QsTRqpzb2ZvtV96nO0W233dZ0jCOVjaRZ/frnuP64Rtt/tHrNnos9Uauek4nut9Xa0X/bA7T6i0y7t1roAGzbtm3IdjMjhf2qVasAWLNmDdu2bdvxATZou/5+I/V9DwwM8PnPf75pe7V9Vq1axfbt24fts2bNmmHl1bJt27aN+Xi3bds2ZCz17TR7Dqr7Vfepqu3baIwjlY2kWf01a9YMe/00amus/dXXa/Zc7Ila9ZxMdL+t1o7+Rw3QiDgnItZHxPpGn6NJE6V21rRu3boJa7v+/lj09/dzxx13NG2vWlY9o63ts27dumHl1bLxrCCklIaMpb6dZm1V96vu00ijMY5UNpJm9detWzdkrPXHNdr+o9Vr9lzsiVr1nEx0v63Wjv5HDdCU0rUppeNSSscdeOCBLR9A/cUcUk13dzcAJ5988oS1XX9/LLq6unjDG97QtL1qWVdX17B9Tj755GHl1bKIGPNYImLIWOrbadZWdb/qPo00GuNIZSNpVv/kk08eMtb64xpt/9HqNXsu9kStek4mut9Wa0f/bV/C1Z6j+kt40qRJI/5SrhkpLHp6egBYunQpkyZN2vEBNmi7/n4j9X13dnayZMmSpu3V9unp6aGjo2PYPkuXLh1WXi2bNGnSmI930qRJQ8ZS306z56C6X3Wfqtq+jcY4UtlImtVfunTpsNdPo7bG2l99vWbPxZ6oVc/JRPfbau3of5cI0DvvvLPdQ9ijjOcMZkfVful1d3ezePHiwfLFixcP2a6ZPn36kO3TTjtt2JlCrb3a6sTMmTNZtGhR0767urqIiCFtNzojrD1ebRuKVZBa/WqbtbqLFi0iIuju7iYiOOWUU4ZdIl9to3b88+fP55RTThm2z8yZM4eVV8sWLVrU8Hgbzd+iRYuGjKVRO7WxN9uvuk91jhYvXtx0jCOVjaRZ/frnuP64Rtt/tHrNnos9Uauek4nut9Xa0f8uEaAam/3333/w/uzZswfvz5gxg4hg8uTJTJ06lQsuuICIYNKkSUydOpVmS++TJ0/msMMOY9q0aRxxxBEcdthhRARz5swZ3H/KlClMmzaNc845B4BZs2YRERxyyCFMmzaNSy65hH333Zeenh6WLl3KwoULOfroowfPTubNmzfY14IFC1i5ciVTp04FYN68eSxZsoRLLrkEgOXLl9PT0zPYXtXSpUtZsGDB4FiPPvpoli9fTkdHB8uXL+eYY45hxYoV7LPPPhx55JH09PSwcOFCFixYwIIFC1i4cCErVqxo2DYw2G/teJYvXz7kuI455hh6eno45phjmr6z7enpYdq0aSxYsGDImVWjfRqVV8tqc3nkkUcyb968wedh5cqVg8e1cOHCpmdp1XZqYx9pv1q96hyNNsaRykYy0pxUXz/j3X+0euMd5+6sVc/JRPfbaju7/13m70DPP/98AFavXt2S9iVJ2lG7/N+BSpK0uzFAJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjIYoJIkZTBAJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjJ0tavjjo4Otm/fPrg9f/78dg1FkqRxa1uATpkyhf7+/sHtc889t11DkSRp3FzClSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjIYoJIkZTBAJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjIYoJIkZTBAJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIydLWr44MPPpgtW7Ywf/78dg1BkqRsbQvQ6667rl1dS5K0w1zClSQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASpKUwQCVJCmDASpJUgYDVJKkDAaoJEkZDFBJkjIYoJIkZTBAJUnKYIBKkpTBAJUkKYMBKklSBgNUkqQMBqgkSRkMUEmSMhigkiRlMEAlScpggEqSlMEAlSQpQ6SUxl45YhPwUAv7nwU80cL29lbOY+s4l63hPLaOc9kaufN4eErpwEYPjCtAWy0i1qeUjmvbAPYQzmPrOJet4Ty2jnPZGhMxjy7hSpKUwQCVJClDuwP02jb3v6dwHlvHuWwN57F1nMvWaPk8tvUzUEmSdlftPgOVJGm31JYAjYhTIuKBiOiNiIvaMYZdXUT8TURsjIj7KmUHRMQdEfFg+XNG5bGLy/l8ICLeWCl/eUT8c/nYpyMidvaxtFNEzI2If4iI+yPiXyLi/LLcuRyHiJgaET+OiHvLeVxZljuPmSKiMyJ+GhG3ltvOZYaI6Cvn4J6IWF+W7Zy5TCnt1BvQCfwcmAdMBu4Fjt7Z49jVb8DrgJcB91XKPg5cVN6/CPhYef/och6nAEeU89tZPvZj4HgggLXAonYf206ex4OBl5X39wP+XzlfzuX45jGA6eX9ScCPgFc5jzs0px8AvgjcWm47l3nz2AfMqivbKXPZjjPQVwK9KaVfpJR+D3wJOL0N49ilpZT+CXiyrvh0YE15fw3wlkr5l1JKW1NKvwR6gVdGxMHA/imlH6TiFfL5yj57hZTShpTST8r7TwH3A4fgXI5LKjxdbk4qbwnnMUtEHAqcClxXKXYuW2enzGU7AvQQ4OHK9iNlmUZ3UEppAxTBAMwuy5vN6SHl/fryvVJEdAN/RHH25FyOU7nkeA+wEbgjpeQ85vsU8OfA9kqZc5knAd+OiLsj4pyybKfMZdcODjxHo3VlLwXeMc3m1LkuRcR04O+B96eUfjfCxxvOZRMppQHgDyPiRcBNEfGSEao7j01ExJuAjSmluyPixLHs0qDMuXzBa1JKv46I2cAdEfFvI9Rt6Vy24wz0EWBuZftQ4NdtGMfu6PFyqYHy58ayvNmcPlLery/fq0TEJIrw/EJK6WtlsXOZKaW0BbgTOAXnMcdrgNMioo/iI6yTIuJGnMssKaVflz83AjdRfEy4U+ayHQH6f4EjI+KIiJgMnAnc3IZx7I5uBpaW95cC36iUnxkRUyLiCOBI4Mfl0sVTEfGq8oqyJZV99grlcX8OuD+ldFXlIedyHCLiwPLMk4iYBpwM/BvO47illC5OKR2aUuqm+P333ZTSu3Auxy0i9o2I/Wr3gf8I3MfOmss2XTW1mOJqyJ8Dy9sxhl39BvwtsAHYRvHu6N3ATOA7wIPlzwMq9ZeX8/kAlavHgOPKF9TPgb+i/M8z9pYbcALFUszPgHvK22Lnctzz+FLgp+U83gdcVpY7jzs2ryfywlW4zuX4528exVW19wL/UsuTnTWX/k9EkiRl8H8ikiQpgwEqSVIGA1SSpAwGqCRJGQxQSZIyGKCSJGUwQCVJymCASru5iFgWEXe1exzS3sYAlSQpgwEqTaCI+HBEPBoRT0XEAxHxHyLi+ohYValzYkQ8UtmeGxFfi4hNEbE5Iv5qhPYXAp8Fjo+IpyNiS0RMjoh7IuLcsk5nRHw/Ii6byGOV9jbt+Dozaa8QEUcBfwa8IhVft9QNdI6yTydwK/Bd4CxggOL/6GwopXR/RPxX4OyU0gmVdt4FfC8i1gFvK/v9ix07IklVBqg0cQaAKcDREbEppdQHMMJ3kULxVUwvBi5MKfWXZeP+fDOldF95lnsTcBDwylR8n6ekFnEJV5ogKaVe4P3ACmBjRHwpIl48ym5zgYcq4bkj1gDdwG0ppQdb0J6kCgNUmkAppS+WS6uHU3yt2seAZ4B9KtXmVO4/DBwWEeNZHWr2lUqfoVgOfmNEnNCkjqRMBqg0QSLiqIg4KSKmAM8Dz1Es694DLI6IAyJiDsVZas2PKb4H9i/LLwueGhGvGaWrx4FDyy+or/V9FvByYBlwHrAmIqa35sgkgQEqTaQpwF8CTwCPAbOBS4AbKL4AuA/4NvDl2g7l55RvBuYDv6L4MvV3jNLPdym+TPixiHgiIg4DPgUsSSk9nVL6IrAeuLpVByYJv1BbkqQcnoFKkpTBAJV2AxHx2fI/Sqi/fbbdY5P2Vi7hSpKUwTNQSZIyGKCSJGUwQCVJymCASpKUwQCVJCnD/wdzukJmjacC+gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,5))\n",
    "sns.boxplot(suc_tx.suc_tx)\n",
    "\n",
    "plt.title('Distribution of the Target', fontsize=16)\n",
    "plt.xlabel('suc_tx', fontsize=12);\n",
    "\n",
    "plt.savefig('assets/target-dist-fig4.jpeg', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fa60571b",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:12:19.925078Z",
     "start_time": "2021-11-26T18:12:19.748650Z"
    }
   },
   "source": [
    "The target seems to be heavily skewed and there are a lot of outliers in the data. However, upon investigation it was found that these outliers were actual data points and not errors therefore it doesn't make sense to exlcude them from the analysis.\n",
    "\n",
    "Moreover, since the data is not normally distributed and due to lack of linear relationships between the target and the independant variables, we'll employ a tree based model."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c60bbf9f",
   "metadata": {},
   "source": [
    "Let's split the dataset into training and testing sets. Due to a huge amount of data and limited time, we'l only split the dataset into a train and test set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "06e24627",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:32:02.246062Z",
     "start_time": "2021-11-26T18:32:01.695809Z"
    }
   },
   "outputs": [],
   "source": [
    "x_train, x_test, y_train, y_test = model_selection.train_test_split(x, y, \n",
    "                                                    test_size=0.30, \n",
    "                                                    random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6cd5db79",
   "metadata": {},
   "source": [
    "Now, training our Decision Tree Regressor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "d0bea93f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:32:41.578410Z",
     "start_time": "2021-11-26T18:32:38.781026Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DecisionTreeRegressor(max_depth=7, random_state=49)"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dt = tree.DecisionTreeRegressor(max_depth=7, random_state=49)\n",
    "dt.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "33893f75",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:32:51.837486Z",
     "start_time": "2021-11-26T18:32:51.704873Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20.63432191870685"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.sqrt(metrics.mean_squared_error(y_test, dt.predict(x_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4906d10c",
   "metadata": {},
   "source": [
    "Here's look at the feature importances"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "5904a4f3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-26T18:47:45.800666Z",
     "start_time": "2021-11-26T18:47:45.681442Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature: cnt_uid || Score: 0.804\n",
      "Feature: activedays || Score: 0.049\n",
      "Feature: tx || Score: 0.147\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "importance = dt.feature_importances_\n",
    "for i,v in enumerate(importance):\n",
    "    print(f'Feature: {x_train.columns[i]} || Score: {v:.3f}')\n",
    "\n",
    "plt.figure(figsize=(12, 5))\n",
    "plt.bar([x_train.columns[i] for i in range(len(importance))], importance)\n",
    "\n",
    "plt.title('Feature Importances', fontsize=16)\n",
    "plt.xlabel('Features', fontsize=12)\n",
    "plt.ylabel('Importance', fontsize=12)\n",
    "plt.savefig('assets/feature_imp-fig5.jpeg', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "93e94f8a",
   "metadata": {},
   "source": [
    "Interestingly, the number of distinct persons with whom a customer has engaged is ranked highest, while the number of days a it has been active on the network is ranked lowest."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "014e7273",
   "metadata": {},
   "source": [
    "### Conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c34f6d9",
   "metadata": {},
   "source": [
    "- Cluster 1 was identified as a persona with the high value and more distinct transactions. This persona also interacts with new people more.\n",
    "- Cluster 2 consists of moderately engaged customers, an important goal for the business here would be to try to move these users into Cluster 1 by leveraging marketing efforts.\n",
    "- My first suggestion in order to drive the count and value of a successful transactions would be to improve the quality of the data by adding in features that determines the major reason behind an unsuccesful transaction.\n",
    "- Furthermore, as seen above, the value of a transaction has a strong positive correlation with the count, implying that enhancing one can boost the other.\n",
    "- Encouraging a customer to make more unique transactions will help boost the number of successful transactions, therefore enhancing the value of a transaction."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.9.7 64-bit ('base': conda)",
   "language": "python",
   "name": "python397jvsc74a57bd0c6e4e9f98eb68ad3b7c296f83d20e6de614cb42e90992a65aa266555a3137d0d"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}