Data Analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.lines as mlines\n",
    "import datetime\n",
    "from operator import add\n",
    "from numpy import where\n",
    "from sklearn.datasets import make_classification\n",
    "from matplotlib import pyplot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv('ecommerce-data/data.csv', encoding=\"ISO-8859-1\", dtype={'CustomerID': str,'InvoiceID': str})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Drop the rows that don't have and entry for the CustomerId field\n",
    "rows_nan = data[data.isnull().T.any().T]\n",
    "data = data.dropna(subset=[\"CustomerID\"])\n",
    "\n",
    "# Dropping duplicate rows\n",
    "duplicated_data = data[data.duplicated(keep=False)]\n",
    "data.drop_duplicates(inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Clean cancelled orders and descounts\n",
    "df_cleaned = data.copy(deep = True)\n",
    "df_cleaned['QuantityCanceled'] = 0\n",
    "\n",
    "entry_to_remove = [] ; doubtfull_entry = []\n",
    "\n",
    "for index, col in  data.iterrows():\n",
    "    if (col['Quantity'] > 0) or col['Description'] == 'Discount': continue        \n",
    "    df_test = data[(data['CustomerID'] == col['CustomerID']) &\n",
    "                         (data['StockCode']  == col['StockCode']) & \n",
    "                         (data['InvoiceDate'] < col['InvoiceDate']) & \n",
    "                         (data['Quantity']   > 0)].copy()\n",
    "    #_________________________________\n",
    "    # Cancelation WITHOUT counterpart\n",
    "    if (df_test.shape[0] == 0): \n",
    "        doubtfull_entry.append(index)\n",
    "    #________________________________\n",
    "    # Cancelation WITH a counterpart\n",
    "    elif (df_test.shape[0] == 1): \n",
    "        index_order = df_test.index[0]\n",
    "        df_cleaned.loc[index_order, 'QuantityCanceled'] = -col['Quantity']\n",
    "        entry_to_remove.append(index)        \n",
    "    #______________________________________________________________\n",
    "    # Various counterparts exist in orders: we delete the last one\n",
    "    elif (df_test.shape[0] > 1): \n",
    "        df_test.sort_index(axis=0 ,ascending=False, inplace = True)        \n",
    "        for ind, val in df_test.iterrows():\n",
    "            if val['Quantity'] < -col['Quantity']: continue\n",
    "            df_cleaned.loc[ind, 'QuantityCanceled'] = -col['Quantity']\n",
    "            entry_to_remove.append(index) \n",
    "            break \n",
    "df_cleaned.drop(entry_to_remove, axis = 0, inplace = True)\n",
    "df_cleaned.drop(doubtfull_entry, axis = 0, inplace = True)\n",
    "remaining_entries = df_cleaned[(df_cleaned['Quantity'] < 0) & (df_cleaned['StockCode'] != 'D')]\n",
    "df_cleaned = df_cleaned[df_cleaned['StockCode']!= 'POST']\n",
    "df_cleaned = df_cleaned[df_cleaned['StockCode']!= 'D']\n",
    "df_cleaned = df_cleaned[df_cleaned['StockCode']!= 'C2']\n",
    "df_cleaned = df_cleaned[df_cleaned['StockCode']!= 'M']\n",
    "df_cleaned = df_cleaned[df_cleaned['StockCode']!= 'BANK CHARGES']\n",
    "df_cleaned = df_cleaned[df_cleaned['StockCode']!= 'PADS']\n",
    "df_cleaned = df_cleaned[df_cleaned['StockCode']!= 'DOT']\n",
    "\n",
    "cleanedData = df_cleaned.copy(deep=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Base Assumption: In a general population, majority of the people would be medium-risk takers, while the outliers would \n",
    "# represent either low or high risk takers. The distrbution of people should follow an approximate bell-shaped curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Assumption: Consumers buying bulk items togeather are implulsive, and hence have a more risk-taking nature\n",
    "grp = cleanedData.groupby('CustomerID')\n",
    "avgSpending = dict()\n",
    "for index,item in grp:\n",
    "    mean = grp.get_group(index).mean(skipna=True, numeric_only=True) \n",
    "    avgSpending[int(index)] = round(mean['Quantity'] * mean['UnitPrice'])\n",
    "plt.bar(list(avgSpending.keys()), avgSpending.values(), color='g')\n",
    "plt.ylim(0,500)\n",
    "plt.xlim(12000, 18500)\n",
    "plt.xlabel('CustomerID')\n",
    "plt.ylabel('Average Spending per invoice')\n",
    "plt.show()\n",
    "# Conclusion: Most of the bars are small or medium in size, representing low and medium risk-takers respectively\n",
    "# There are some exceptionally tall bars, denoting the high-risk takers. The assumption is further supported by the distrbution\n",
    "# of the bars of various sizes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Assumption: Customers with longer duration between purchases have a more risky behaviour as they tend to wait\n",
    "# for the last moment to accomplish stuff\n",
    "dataDict = dict()\n",
    "avgTimeBetweenBuys = {}\n",
    "for index,item in grp:\n",
    "    dates = grp.get_group(index).drop_duplicates(subset=['InvoiceNo'], keep='first')\n",
    "    if len(dates.index > 1):\n",
    "        diff = [datetime.datetime.strptime(dates.iat[i, 4], '%m/%d/%Y %H:%M') - datetime.datetime.strptime(dates.iat[i - 1, 4], '%m/%d/%Y %H:%M') for i in range(len(dates.index)) if i > 0]\n",
    "        diffs = [x.total_seconds() for x in diff]\n",
    "        if len(diffs) > 0:\n",
    "            dataDict[max(diff).total_seconds()] = sum(diffs) / len(diffs)\n",
    "            avgTimeBetweenBuys[int(index)] = sum(diffs) / len(diffs)\n",
    "fig, ax = plt.subplots()\n",
    "ax.scatter(dataDict.keys(), dataDict.values())\n",
    "ax.set_xlabel('Maximum time between consicutive buys')\n",
    "ax.set_ylabel('Average time between consicutive buys')\n",
    "plt.show()\n",
    "# Conclusion: The dots lying at the top-right corner of the plot, as well as to some extent the ones in the bottom-right corner\n",
    "# represent high-risk takers, as they either have a high avg as well as max delay between two invoices(top-right corner), or \n",
    "# a high max delay between two invoices(bottom-right corner)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Assumption: Consumers buying the same item over and over again are less risk taking in nature\n",
    "freqDict = dict()\n",
    "for index,item in grp:\n",
    "    invoice_count = grp.get_group(index).duplicated(subset=['InvoiceNo']).sum()\n",
    "    # print(invoice_count)\n",
    "    # data = grp.get_group(index).groupby(by=[\"StockCode\"]).count().sort_values('InvoiceNo', ascending=False)\n",
    "    group = grp.get_group(index)\n",
    "    duplicates = group[group.duplicated(subset=[\"StockCode\"], keep=False)].groupby(by=[\"StockCode\"]).count()['InvoiceNo']\n",
    "    if len(duplicates) > 0:\n",
    "        freqDict[int(index)] = float(invoice_count) / float(sum(duplicates))\n",
    "plt.bar(list(freqDict.keys()), freqDict.values(), color='g')\n",
    "plt.ylim(0,40)\n",
    "plt.xlim(12000, 18500)\n",
    "plt.xlabel('CustomerID')\n",
    "plt.ylabel('1 / Chances of buying same items')\n",
    "plt.show()\n",
    "# Conclusion: Most of the bars are small or medium in size, representing low and medium risk-takers respectively\n",
    "# There are some exceptionally tall bars, denoting the high-risk takers. The assumption is further supported by the distrbution\n",
    "# of the bars of various sizes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Assumption: People which buy items with higher unit Prices are probably more risk taking in nature\n",
    "meanUnitPrice = cleanedData['UnitPrice'].mean()\n",
    "maxUnitPrice = cleanedData['UnitPrice'].max()\n",
    "# cleanedData.boxplot(column='UnitPrice')\n",
    "# Let us assume that the products with top 1% unit Prices in the distrbution can be assumed as expensive\n",
    "priceLimit = np.percentile(cleanedData['UnitPrice'], 99)\n",
    "percentageOverspent = dict()\n",
    "for index, item in grp:\n",
    "    uniqueProducts = grp.get_group(index).drop_duplicates(subset=['StockCode'], keep='first')\n",
    "    overspentCount = len(uniqueProducts[uniqueProducts['UnitPrice'] > priceLimit])\n",
    "    if(overspentCount != 0):\n",
    "        percentageOverspent[int(index)] = overspentCount / float(len(uniqueProducts))\n",
    "plt.bar(list(percentageOverspent.keys()), percentageOverspent.values(), color='g')\n",
    "plt.ylim(0,0.35)\n",
    "plt.xlim(12000, 18500)\n",
    "plt.xlabel('CustomerID')\n",
    "plt.ylabel('% of expensive items bought')\n",
    "plt.show()\n",
    "# Conclusion: Most of the bars are small or medium in size, representing low and medium risk-takers respectively\n",
    "# There are some exceptionally tall bars, denoting the high-risk takers. The assumption is further supported by the distrbution\n",
    "# of the bars of various sizes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Converting Observations into Probablities\n",
    "# Basic Assumption: Among all the users, around 15% are low-risk takers, 15% are high-risk takers, and the rest 60% of the\n",
    "# people are medium risk takers\n",
    "# Assumption: ALl the 4 parameters calculated above have equal weightage when it comes to determining the final probablity\n",
    "\n",
    "weight = [0.25] * 4\n",
    "user_probablity = {}\n",
    "total_users = len(grp)\n",
    "\n",
    "# Probablities based on average spending\n",
    "limits = np.percentile(list(avgSpending.values()), [15, 85])\n",
    "for key, item in avgSpending.items():\n",
    "    user_probablity[key] = [weight[0] * int(int(item) <= limits[0]), weight[0] * int(int(item) <= limits[1] and int(item) > limits[0]), weight[0] * int(int(item) > limits[1])]\n",
    "\n",
    "# Adding probablities based on longer duration between purchases\n",
    "limits = np.percentile(list(avgTimeBetweenBuys.values()), [15, 85])\n",
    "for key, item in avgTimeBetweenBuys.items():\n",
    "    new_list = [weight[1] * int(int(item) <= limits[0]), weight[1] * int(int(item) <= limits[1] and int(item) > limits[0]), weight[1] * int(int(item) > limits[1])]\n",
    "    if key in user_probablity.keys():\n",
    "        prev_list = user_probablity[key]\n",
    "        user_probablity[key] = [prev_list[i] + new_list[i] for i in range(len(prev_list))]\n",
    "    else:\n",
    "        user_probablity[key] = new_list\n",
    "        \n",
    "# Adding probablities based on buying the same item over and over again\n",
    "limits = np.percentile(list(freqDict.values()), [15, 85])\n",
    "for key, item in freqDict.items():\n",
    "    new_list = [weight[2] * int(int(item) <= limits[0]), weight[2] * int(int(item) <= limits[1] and int(item) > limits[0]), weight[2] * int(int(item) > limits[1])]\n",
    "    if key in user_probablity.keys():\n",
    "        prev_list = user_probablity[key]\n",
    "        user_probablity[key] = [prev_list[i] + new_list[i] for i in range(len(prev_list))]\n",
    "    else:\n",
    "        user_probablity[key] = new_list\n",
    "        \n",
    "# Adding probablities based on big bucket prices\n",
    "limits = np.percentile(list(percentageOverspent.values()), [15, 85])\n",
    "for key, item in percentageOverspent.items():\n",
    "    new_list = [weight[3] * int(int(item) <= limits[0]), weight[3] * int(int(item) <= limits[1] and int(item) > limits[0]), weight[3] * int(int(item) > limits[1])]\n",
    "    if key in user_probablity.keys():\n",
    "        prev_list = user_probablity[key]\n",
    "        user_probablity[key] = [prev_list[i] + new_list[i] for i in range(len(prev_list))]\n",
    "    else:\n",
    "        user_probablity[key] = new_list\n",
    "        \n",
    "for key, item in user_probablity.items():\n",
    "    user_probablity[key] = np.argmax(np.array(user_probablity[key]))\n",
    "\n",
    "high_risk_customer_id = [key for key, item in user_probablity.items() if item == 2]\n",
    "low_risk_customer_id = [key for key, item in user_probablity.items() if item == 0]\n",
    "medium_risk_customer_id = [key for key, item in user_probablity.items() if item == 1]\n",
    "\n",
    "# Display a pie-chart corrosponding to the % of users\n",
    "labels = 'Low-Risk', 'Medium-Risk', 'High-Risk'\n",
    "sizes = [(len(low_risk_customer_id) * 100) / total_users, (len(medium_risk_customer_id) * 100) / total_users, (len(high_risk_customer_id) * 100) / total_users]\n",
    "explode = (0, 0, 0.1)  # only \"explode\" the 2nd slice (i.e. 'Hogs')\n",
    "fig1, ax1 = plt.subplots()\n",
    "ax1.pie(sizes, explode=explode, labels=labels, autopct='%1.1f%%',shadow=True, startangle=90)\n",
    "ax1.axis('equal')  # Equal aspect ratio ensures that pie is drawn as a circle.\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/prarabdh/.local/lib/python2.7/site-packages/ipykernel_launcher.py:5: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  \"\"\"\n",
      "/home/prarabdh/.local/lib/python2.7/site-packages/ipykernel_launcher.py:7: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  import sys\n",
      "/home/prarabdh/.local/lib/python2.7/site-packages/ipykernel_launcher.py:9: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  if __name__ == '__main__':\n",
      "/home/prarabdh/.local/lib/python2.7/site-packages/ipykernel_launcher.py:11: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  # This is added back by InteractiveShellApp.init_path()\n"
     ]
    }
   ],
   "source": [
    "# Generating data for applying clustering algorithms\n",
    "\n",
    "data_frame = pd.DataFrame(columns=['averageSpending', 'Duration Between Purchases', 'Same Item Over and Over', 'Big Bucket Price', 'Label'], index=list(map(int, grp.groups.keys())))\n",
    "for key, item in avgSpending.items():\n",
    "    data_frame.ix[key]['averageSpending'] = item\n",
    "for key, item in avgTimeBetweenBuys.items():\n",
    "    data_frame.ix[key]['Duration Between Purchases'] = item\n",
    "for key, item in freqDict.items():\n",
    "    data_frame.ix[key]['Same Item Over and Over'] = item\n",
    "for key, item in percentageOverspent.items():\n",
    "    data_frame.ix[key]['Big Bucket Price'] = item\n",
    "\n",
    "# print(data_frame.info())\n",
    "data_frame.fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.cluster import Birch\n",
    "from sklearn.cluster import DBSCAN\n",
    "from sklearn.cluster import AgglomerativeClustering\n",
    "from sklearn.cluster import AffinityPropagation\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.cluster import MeanShift\n",
    "from sklearn.mixture import GaussianMixture\n",
    "from sklearn.cluster import SpectralClustering\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def print_percentage_of_users_in_clusters(model):\n",
    "    model.fit(data_frame)\n",
    "    try:\n",
    "        yhat = model.predict(data_frame)\n",
    "    except:\n",
    "        yhat = model.fit_predict(data_frame)\n",
    "    clusters = np.unique(yhat)\n",
    "    list1 = []\n",
    "    for cluster in clusters:\n",
    "        row_ix = where(yhat == cluster)\n",
    "        list1.append((len(row_ix[0]) * 100 / total_users))\n",
    "    print(list1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[80, 15, 3]\n",
      "[80, 15, 3]\n",
      "[29, 7, 63]\n"
     ]
    }
   ],
   "source": [
    "# Testing various models with 3 clusters: Low, Medium, and High\n",
    "\n",
    "model = Birch(threshold=0.01, n_clusters=3)\n",
    "print_percentage_of_users_in_clusters(model)\n",
    "\n",
    "model = AgglomerativeClustering(n_clusters=3)\n",
    "print_percentage_of_users_in_clusters(model)\n",
    "\n",
    "model = KMeans(n_clusters=3, init='k-means++', max_iter=300, n_init=10, random_state=0)\n",
    "print_percentage_of_users_in_clusters(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2601, 4)\n",
      "(1734, 4)\n"
     ]
    }
   ],
   "source": [
    "# Labeling Data based on the k-means clustering algorithms, as the percentage of customers in were closest to the manual analysis\n",
    "# 0 means low risk-taker, 2 means high-risk taker, 1 means average risk taker\n",
    "labels = [2, 1, 0]\n",
    "model = KMeans(n_clusters=3)\n",
    "model.fit(data_frame)\n",
    "yhat = model.predict(data_frame)\n",
    "# model = AgglomerativeClustering(n_clusters=3)\n",
    "# model.fit(data_frame)\n",
    "# yhat = model.fit_predict(data_frame)\n",
    "clusters = np.unique(yhat)\n",
    "i = 0\n",
    "remove = []\n",
    "pd.set_option('display.max_rows', 5000)\n",
    "pd.set_option('display.max_columns', 6)\n",
    "for cluster in clusters:\n",
    "    row_ix = where(yhat == cluster)\n",
    "    for index in row_ix[0]:\n",
    "        cust_id = grp.groups.keys()[index]\n",
    "        data_frame.at[int(cust_id), 'Label'] = labels[i]\n",
    "    i += 1\n",
    "\n",
    "y = data_frame['Label']\n",
    "x = data_frame.drop('Label', axis=1)\n",
    "x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.4)\n",
    "print(x_train.shape)\n",
    "print(x_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 487    0    0]\n",
      " [   0  136    0]\n",
      " [   0    0 1111]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      1.00      1.00       487\n",
      "           1       1.00      1.00      1.00       136\n",
      "           2       1.00      1.00      1.00      1111\n",
      "\n",
      "   micro avg       1.00      1.00      1.00      1734\n",
      "   macro avg       1.00      1.00      1.00      1734\n",
      "weighted avg       1.00      1.00      1.00      1734\n",
      "\n",
      "1.0\n"
     ]
    }
   ],
   "source": [
    "# from sklearn import linear_model\n",
    "# regr = linear_model.LinearRegression()\n",
    "# regr.fit(x_train, y_train)\n",
    "\n",
    "# plt.scatter(x_train['Big Bucket Price'], y_train, color='red')\n",
    "# plt.scatter(x_train['averageSpending'], y_train, color='red')\n",
    "# plt.scatter(x_train['Duration Between Purchases'], y_train, color='red')\n",
    "# plt.scatter(x_train['Same Item Over and Over'], y_train, color='red')\n",
    "# plt.show()\n",
    "# print('Intercept: \\n', regr.intercept_)\n",
    "# print('Coefficients: \\n', regr.coef_)\n",
    "# print(list(map(round, regr.predict(x_test))))\n",
    "\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import classification_report, confusion_matrix, accuracy_score\n",
    "\n",
    "classifier = RandomForestClassifier(n_estimators=4, random_state=0)\n",
    "classifier.fit(x_train, y_train)\n",
    "y_pred = classifier.predict(x_test)\n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))\n",
    "print(accuracy_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Testing the model with a new dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10       20-02-2014\n",
      "11       20-02-2014\n",
      "12       20-02-2014\n",
      "14       20-02-2014\n",
      "15       20-02-2014\n",
      "16       20-02-2014\n",
      "17       20-02-2014\n",
      "18       20-02-2014\n",
      "19       20-02-2014\n",
      "20       20-02-2014\n",
      "21       20-02-2014\n",
      "22       20-02-2014\n",
      "23       20-02-2014\n",
      "24       20-02-2014\n",
      "25       20-02-2014\n",
      "26       20-02-2014\n",
      "27       20-02-2014\n",
      "29       20-02-2014\n",
      "30       19-02-2014\n",
      "31       19-02-2014\n",
      "32       19-02-2014\n",
      "33       19-02-2014\n",
      "34       19-02-2014\n",
      "35       19-02-2014\n",
      "36       19-02-2014\n",
      "37       19-02-2014\n",
      "38       19-02-2014\n",
      "39       19-02-2014\n",
      "40       19-02-2014\n",
      "41       19-02-2014\n",
      "42       19-02-2014\n",
      "43       19-02-2014\n",
      "44       19-02-2014\n",
      "45       19-02-2014\n",
      "46       19-02-2014\n",
      "47       18-02-2014\n",
      "48       18-02-2014\n",
      "49       18-02-2014\n",
      "50       18-02-2014\n",
      "51       18-02-2014\n",
      "52       18-02-2014\n",
      "53       18-02-2014\n",
      "54       18-02-2014\n",
      "55       18-02-2014\n",
      "57       18-02-2014\n",
      "58       18-02-2014\n",
      "59       18-02-2014\n",
      "60       18-02-2014\n",
      "61       18-02-2014\n",
      "62       18-02-2014\n",
      "63       17-02-2014\n",
      "64       17-02-2014\n",
      "65       17-02-2014\n",
      "66       17-02-2014\n",
      "68       17-02-2014\n",
      "69       17-02-2014\n",
      "71       17-02-2014\n",
      "72       17-02-2014\n",
      "73       17-02-2014\n",
      "74       17-02-2014\n",
      "75       17-02-2014\n",
      "76       17-02-2014\n",
      "77       16-02-2014\n",
      "78       16-02-2014\n",
      "79       16-02-2014\n",
      "80       16-02-2014\n",
      "81       16-02-2014\n",
      "82       16-02-2014\n",
      "83       16-02-2014\n",
      "84       16-02-2014\n",
      "85       16-02-2014\n",
      "86       16-02-2014\n",
      "87       16-02-2014\n",
      "88       16-02-2014\n",
      "89       16-02-2014\n",
      "90       16-02-2014\n",
      "91       16-02-2014\n",
      "92       16-02-2014\n",
      "93       15-02-2014\n",
      "94       15-02-2014\n",
      "95       15-02-2014\n",
      "96       15-02-2014\n",
      "97       15-02-2014\n",
      "98       15-02-2014\n",
      "99       15-02-2014\n",
      "100      15-02-2014\n",
      "101      15-02-2014\n",
      "102      15-02-2014\n",
      "103      15-02-2014\n",
      "104      15-02-2014\n",
      "105      15-02-2014\n",
      "106      15-02-2014\n",
      "107      15-02-2014\n",
      "108      15-02-2014\n",
      "109      15-02-2014\n",
      "110      14-02-2014\n",
      "111      14-02-2014\n",
      "112      14-02-2014\n",
      "113      14-02-2014\n",
      "115      14-02-2014\n",
      "116      14-02-2014\n",
      "117      14-02-2014\n",
      "118      14-02-2014\n",
      "119      14-02-2014\n",
      "120      14-02-2014\n",
      "121      14-02-2014\n",
      "122      14-02-2014\n",
      "123      14-02-2014\n",
      "124      14-02-2014\n",
      "125      14-02-2014\n",
      "126      14-02-2014\n",
      "127      14-02-2014\n",
      "128      14-02-2014\n",
      "129      13-02-2014\n",
      "130      13-02-2014\n",
      "131      13-02-2014\n",
      "132      13-02-2014\n",
      "133      13-02-2014\n",
      "134      13-02-2014\n",
      "135      13-02-2014\n",
      "136      13-02-2014\n",
      "137      13-02-2014\n",
      "138      13-02-2014\n",
      "140      13-02-2014\n",
      "141      13-02-2014\n",
      "142      13-02-2014\n",
      "143      13-02-2014\n",
      "144       12/2/2014\n",
      "145       12/2/2014\n",
      "146       12/2/2014\n",
      "147       12/2/2014\n",
      "148       12/2/2014\n",
      "149       12/2/2014\n",
      "150       12/2/2014\n",
      "151       12/2/2014\n",
      "152       12/2/2014\n",
      "153       12/2/2014\n",
      "154       12/2/2014\n",
      "156       12/2/2014\n",
      "157       12/2/2014\n",
      "158       12/2/2014\n",
      "159       12/2/2014\n",
      "160       12/2/2014\n",
      "161       12/2/2014\n",
      "162       12/2/2014\n",
      "163       11/2/2014\n",
      "164       11/2/2014\n",
      "166       11/2/2014\n",
      "167       11/2/2014\n",
      "168       11/2/2014\n",
      "169       11/2/2014\n",
      "170       11/2/2014\n",
      "171       11/2/2014\n",
      "172       11/2/2014\n",
      "173       11/2/2014\n",
      "174       11/2/2014\n",
      "175       11/2/2014\n",
      "177       11/2/2014\n",
      "178       11/2/2014\n",
      "179       11/2/2014\n",
      "180       11/2/2014\n",
      "181       11/2/2014\n",
      "182       11/2/2014\n",
      "183       10/2/2014\n",
      "184       10/2/2014\n",
      "185       10/2/2014\n",
      "186       10/2/2014\n",
      "187       10/2/2014\n",
      "188       10/2/2014\n",
      "190       10/2/2014\n",
      "191       10/2/2014\n",
      "192       10/2/2014\n",
      "193       10/2/2014\n",
      "194       10/2/2014\n",
      "195       10/2/2014\n",
      "196       10/2/2014\n",
      "197       10/2/2014\n",
      "198       10/2/2014\n",
      "199       10/2/2014\n",
      "200       10/2/2014\n",
      "201        9/2/2014\n",
      "203        9/2/2014\n",
      "204        9/2/2014\n",
      "205        9/2/2014\n",
      "207        9/2/2014\n",
      "208        9/2/2014\n",
      "209        9/2/2014\n",
      "210        9/2/2014\n",
      "211        9/2/2014\n",
      "212        9/2/2014\n",
      "214        9/2/2014\n",
      "215        9/2/2014\n",
      "216        9/2/2014\n",
      "217        9/2/2014\n",
      "218        9/2/2014\n",
      "219        9/2/2014\n",
      "220        9/2/2014\n",
      "222        8/2/2014\n",
      "223        8/2/2014\n",
      "224        8/2/2014\n",
      "225        8/2/2014\n",
      "226        8/2/2014\n",
      "227        8/2/2014\n",
      "228        8/2/2014\n",
      "229        8/2/2014\n",
      "230        8/2/2014\n",
      "231        8/2/2014\n",
      "232        8/2/2014\n",
      "233        8/2/2014\n",
      "234        8/2/2014\n",
      "235        8/2/2014\n",
      "236        8/2/2014\n",
      "238        8/2/2014\n",
      "239        8/2/2014\n",
      "241        7/2/2014\n",
      "242        7/2/2014\n",
      "243        7/2/2014\n",
      "244        7/2/2014\n",
      "245        7/2/2014\n",
      "246        7/2/2014\n",
      "247        7/2/2014\n",
      "248        7/2/2014\n",
      "249        7/2/2014\n",
      "250        7/2/2014\n",
      "251        7/2/2014\n",
      "252        7/2/2014\n",
      "253        7/2/2014\n",
      "254        7/2/2014\n",
      "255        7/2/2014\n",
      "256        7/2/2014\n",
      "257        7/2/2014\n",
      "258        7/2/2014\n",
      "259        7/2/2014\n",
      "261        7/2/2014\n",
      "263        7/2/2014\n",
      "265        7/2/2014\n",
      "266        6/2/2014\n",
      "267        6/2/2014\n",
      "268        6/2/2014\n",
      "269        6/2/2014\n",
      "270        6/2/2014\n",
      "271        6/2/2014\n",
      "272        6/2/2014\n",
      "273        6/2/2014\n",
      "274        6/2/2014\n",
      "275        6/2/2014\n",
      "276        6/2/2014\n",
      "277        6/2/2014\n",
      "278        6/2/2014\n",
      "279        6/2/2014\n",
      "            ...    \n",
      "22783      7/2/2011\n",
      "22784      7/2/2011\n",
      "22785      6/2/2011\n",
      "22786      6/2/2011\n",
      "22787      6/2/2011\n",
      "22788      6/2/2011\n",
      "22789      6/2/2011\n",
      "22790      6/2/2011\n",
      "22791      6/2/2011\n",
      "22792      6/2/2011\n",
      "22793      6/2/2011\n",
      "22795      6/2/2011\n",
      "22796      6/2/2011\n",
      "22797      6/2/2011\n",
      "22798      6/2/2011\n",
      "22799      6/2/2011\n",
      "22800      6/2/2011\n",
      "22802      5/2/2011\n",
      "22803      5/2/2011\n",
      "22804      5/2/2011\n",
      "22805      5/2/2011\n",
      "22806      5/2/2011\n",
      "22807      5/2/2011\n",
      "22808      5/2/2011\n",
      "22809      5/2/2011\n",
      "22810      5/2/2011\n",
      "22811      5/2/2011\n",
      "22812      5/2/2011\n",
      "22813      5/2/2011\n",
      "22814      5/2/2011\n",
      "22815      5/2/2011\n",
      "22817      5/2/2011\n",
      "22819      5/2/2011\n",
      "22820      5/2/2011\n",
      "22821      5/2/2011\n",
      "22822      4/2/2011\n",
      "22823      4/2/2011\n",
      "22824      4/2/2011\n",
      "22826      4/2/2011\n",
      "22827      4/2/2011\n",
      "22828      4/2/2011\n",
      "22829      4/2/2011\n",
      "22830      4/2/2011\n",
      "22831      4/2/2011\n",
      "22832      4/2/2011\n",
      "22833      4/2/2011\n",
      "22834      4/2/2011\n",
      "22835      4/2/2011\n",
      "22837      4/2/2011\n",
      "22838      4/2/2011\n",
      "22839      4/2/2011\n",
      "22841      4/2/2011\n",
      "22842      4/2/2011\n",
      "22843      4/2/2011\n",
      "22844      4/2/2011\n",
      "22845      4/2/2011\n",
      "22846      4/2/2011\n",
      "22847      4/2/2011\n",
      "22848      3/2/2011\n",
      "22849      3/2/2011\n",
      "22850      3/2/2011\n",
      "22851      3/2/2011\n",
      "22852      3/2/2011\n",
      "22853      3/2/2011\n",
      "22854      3/2/2011\n",
      "22855      3/2/2011\n",
      "22856      3/2/2011\n",
      "22857      3/2/2011\n",
      "22858      3/2/2011\n",
      "22859      3/2/2011\n",
      "22860      3/2/2011\n",
      "22861      3/2/2011\n",
      "22862      3/2/2011\n",
      "22864      2/2/2011\n",
      "22865      2/2/2011\n",
      "22866      2/2/2011\n",
      "22868      2/2/2011\n",
      "22869      2/2/2011\n",
      "22870      2/2/2011\n",
      "22871      2/2/2011\n",
      "22872      2/2/2011\n",
      "22873      2/2/2011\n",
      "22875      2/2/2011\n",
      "22876      2/2/2011\n",
      "22877      2/2/2011\n",
      "22878      2/2/2011\n",
      "22879      2/2/2011\n",
      "22880      2/2/2011\n",
      "22881      2/2/2011\n",
      "22882      2/2/2011\n",
      "22883      2/2/2011\n",
      "22885      1/2/2011\n",
      "22886      1/2/2011\n",
      "22887      1/2/2011\n",
      "22888      1/2/2011\n",
      "22889      1/2/2011\n",
      "22890      1/2/2011\n",
      "22891      1/2/2011\n",
      "22892      1/2/2011\n",
      "22893      1/2/2011\n",
      "22894      1/2/2011\n",
      "22896      1/2/2011\n",
      "22897      1/2/2011\n",
      "22900      1/2/2011\n",
      "22901      1/2/2011\n",
      "22902      1/2/2011\n",
      "22903      1/2/2011\n",
      "22904      1/2/2011\n",
      "22906      1/2/2011\n",
      "22907      1/2/2011\n",
      "22908      1/2/2011\n",
      "22909      1/2/2011\n",
      "22910      1/2/2011\n",
      "22911      1/2/2011\n",
      "22912    31-01-2011\n",
      "22913    31-01-2011\n",
      "22914    31-01-2011\n",
      "22915    31-01-2011\n",
      "22916    31-01-2011\n",
      "22917    31-01-2011\n",
      "22918    31-01-2011\n",
      "22920    31-01-2011\n",
      "22921    31-01-2011\n",
      "22922    31-01-2011\n",
      "22923    31-01-2011\n",
      "22924    31-01-2011\n",
      "22925    31-01-2011\n",
      "22927    31-01-2011\n",
      "22928    31-01-2011\n",
      "22930    31-01-2011\n",
      "22931    31-01-2011\n",
      "22932    31-01-2011\n",
      "22933    31-01-2011\n",
      "22934    30-01-2011\n",
      "22935    30-01-2011\n",
      "22936    30-01-2011\n",
      "22937    30-01-2011\n",
      "22938    30-01-2011\n",
      "22939    30-01-2011\n",
      "22940    30-01-2011\n",
      "22941    30-01-2011\n",
      "22942    30-01-2011\n",
      "22943    30-01-2011\n",
      "22944    30-01-2011\n",
      "22945    30-01-2011\n",
      "22946    30-01-2011\n",
      "22947    30-01-2011\n",
      "22948    30-01-2011\n",
      "22949    30-01-2011\n",
      "22950    30-01-2011\n",
      "22951    30-01-2011\n",
      "22952    30-01-2011\n",
      "22953    30-01-2011\n",
      "22954    30-01-2011\n",
      "22955    30-01-2011\n",
      "22956    30-01-2011\n",
      "22957    30-01-2011\n",
      "22958    30-01-2011\n",
      "22959    30-01-2011\n",
      "22960    30-01-2011\n",
      "22961    29-01-2011\n",
      "22962    29-01-2011\n",
      "22963    29-01-2011\n",
      "22964    29-01-2011\n",
      "22965    29-01-2011\n",
      "22966    29-01-2011\n",
      "22967    29-01-2011\n",
      "22968    29-01-2011\n",
      "22969    29-01-2011\n",
      "22970    29-01-2011\n",
      "22971    29-01-2011\n",
      "22972    29-01-2011\n",
      "22973    29-01-2011\n",
      "22974    29-01-2011\n",
      "22975    29-01-2011\n",
      "22976    29-01-2011\n",
      "22978    29-01-2011\n",
      "22979    28-01-2011\n",
      "22980    28-01-2011\n",
      "22981    28-01-2011\n",
      "22982    28-01-2011\n",
      "22983    28-01-2011\n",
      "22984    28-01-2011\n",
      "22985    28-01-2011\n",
      "22986    28-01-2011\n",
      "22987    28-01-2011\n",
      "22988    28-01-2011\n",
      "22989    28-01-2011\n",
      "22990    28-01-2011\n",
      "22991    28-01-2011\n",
      "22992    27-01-2011\n",
      "22993    27-01-2011\n",
      "22994    27-01-2011\n",
      "22995    27-01-2011\n",
      "22996    27-01-2011\n",
      "22997    27-01-2011\n",
      "22998    27-01-2011\n",
      "22999    27-01-2011\n",
      "23000    27-01-2011\n",
      "23001    27-01-2011\n",
      "23002    27-01-2011\n",
      "23003    27-01-2011\n",
      "23004    27-01-2011\n",
      "23005    27-01-2011\n",
      "23006    27-01-2011\n",
      "23008    27-01-2011\n",
      "23009    27-01-2011\n",
      "23010    26-01-2011\n",
      "23011    26-01-2011\n",
      "23012    26-01-2011\n",
      "23013    26-01-2011\n",
      "23014    26-01-2011\n",
      "23015    26-01-2011\n",
      "23016    26-01-2011\n",
      "23017    26-01-2011\n",
      "23018    26-01-2011\n",
      "23019    26-01-2011\n",
      "23020    26-01-2011\n",
      "23021    26-01-2011\n",
      "23022    26-01-2011\n",
      "23023    26-01-2011\n",
      "23024    26-01-2011\n",
      "23025    26-01-2011\n",
      "23026    26-01-2011\n",
      "23027    26-01-2011\n",
      "23028    26-01-2011\n",
      "23029    26-01-2011\n",
      "23030    26-01-2011\n",
      "23031    26-01-2011\n",
      "23032    26-01-2011\n",
      "23033    25-01-2011\n",
      "23034    25-01-2011\n",
      "23035    25-01-2011\n",
      "23036    25-01-2011\n",
      "23037    25-01-2011\n",
      "23038    25-01-2011\n",
      "23039    25-01-2011\n",
      "23040    25-01-2011\n",
      "23041    25-01-2011\n",
      "23042    25-01-2011\n",
      "23043    25-01-2011\n",
      "23044    25-01-2011\n",
      "23045    25-01-2011\n",
      "23046    25-01-2011\n",
      "23047    25-01-2011\n",
      "23048    25-01-2011\n",
      "23049    25-01-2011\n",
      "23050    25-01-2011\n",
      "23051    25-01-2011\n",
      "23052    25-01-2011\n",
      "Name: InvoiceDate, Length: 20876, dtype: object\n",
      "          InvoiceNo     CustomerID      Quantity     UnitPrice     StockCode\n",
      "count  2.087600e+04   20876.000000  20876.000000  20876.000000  20876.000000\n",
      "mean   5.000701e+10  271027.141981      3.003497    785.927764     81.386568\n",
      "std    2.896115e+10    2433.338935      1.420479    415.586498     35.737376\n",
      "min    3.268991e+06  266783.000000      1.000000     70.000000     21.000000\n",
      "25%    2.487089e+10  268938.000000      2.000000    426.000000     44.000000\n",
      "50%    5.007968e+10  271004.000000      3.000000    788.000000     81.000000\n",
      "75%    7.512409e+10  273122.250000      4.000000   1148.000000    111.000000\n",
      "max    9.998755e+10  275265.000000      5.000000   1500.000000    132.000000\n"
     ]
    }
   ],
   "source": [
    "new_data = pd.read_csv('Retail Case Study/Transactions.csv', encoding=\"ISO-8859-1\")\n",
    "pd.set_option('display.max_rows', 500)\n",
    "pd.set_option('display.max_columns', 500)\n",
    "pd.set_option('display.width', 1000)\n",
    "new_data['StockCode'] = new_data['prod_cat_code'] * 20 + new_data['prod_subcat_code']\n",
    "new_data = new_data.drop(['Tax', 'Store_type', 'total_amt', 'prod_cat_code', 'prod_subcat_code'], axis=1)\n",
    "new_data = new_data.rename(columns={'transaction_id': 'InvoiceNo',\n",
    "                                   'cust_id': 'CustomerID',\n",
    "                                   'tran_date': 'InvoiceDate',\n",
    "                                   'Qty': 'Quantity',\n",
    "                                   'Rate': 'UnitPrice'})\n",
    "new_data = new_data.drop(index=new_data[new_data['Quantity'] < 0].index)\n",
    "print(new_data.InvoiceDate)\n",
    "print(new_data.describe())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "      InvoiceNo  CustomerID InvoiceDate  Quantity  UnitPrice  StockCode\n",
      "10  29258453508      270384  20-02-2014         5       1497         65\n",
      "11  25455265351      267750  20-02-2014         3       1360        132\n",
      "12   1571002198      275023  20-02-2014         4        587        106\n",
      "14  36554696014      269345  20-02-2014         3       1253        103\n",
      "15  56814940239      268799  20-02-2014         5        368        107\n"
     ]
    }
   ],
   "source": [
    "def convertDate(date):\n",
    "    try:\n",
    "        return datetime.datetime.strptime(date, '%d-%m-%Y')\n",
    "    except:\n",
    "        return datetime.datetime.strptime(date, '%d/%m/%Y')\n",
    "\n",
    "cleanedData = new_data.copy(deep=True)\n",
    "print(cleanedData.head())\n",
    "grp = cleanedData.groupby('CustomerID')\n",
    "dataDict = dict()\n",
    "avgTimeBetweenBuys = {}\n",
    "freqDict = dict()\n",
    "avgSpending = dict()\n",
    "meanUnitPrice = cleanedData['UnitPrice'].mean()\n",
    "maxUnitPrice = cleanedData['UnitPrice'].max()\n",
    "priceLimit = np.percentile(cleanedData['UnitPrice'], 99)\n",
    "percentageOverspent = dict()\n",
    "for index,item in grp:\n",
    "    mean = grp.get_group(index).mean(skipna=True, numeric_only=True) \n",
    "    avgSpending[int(index)] = round(mean['Quantity'] * mean['UnitPrice'])\n",
    "    dates = grp.get_group(index).drop_duplicates(subset=['InvoiceNo'], keep='first')\n",
    "    if len(dates.index > 1):\n",
    "        diff = [convertDate(dates.iat[i, 2]) - convertDate(dates.iat[i - 1, 2]) for i in range(len(dates.index)) if i > 0]\n",
    "        diffs = [x.total_seconds() for x in diff]\n",
    "        if len(diffs) > 0:\n",
    "            dataDict[max(diff).total_seconds()] = sum(diffs) / len(diffs)\n",
    "            avgTimeBetweenBuys[int(index)] = sum(diffs) / len(diffs)\n",
    "    invoice_count = grp.get_group(index).duplicated(subset=['InvoiceNo']).sum()\n",
    "    group = grp.get_group(index)\n",
    "    duplicates = group[group.duplicated(subset=[\"StockCode\"], keep=False)].groupby(by=[\"StockCode\"]).count()['InvoiceNo']\n",
    "    if len(duplicates) > 0:\n",
    "        freqDict[int(index)] = float(invoice_count) / float(sum(duplicates))\n",
    "    uniqueProducts = grp.get_group(index).drop_duplicates(subset=['StockCode'], keep='first')\n",
    "    overspentCount = len(uniqueProducts[uniqueProducts['UnitPrice'] > priceLimit])\n",
    "    if(overspentCount != 0):\n",
    "        percentageOverspent[int(index)] = overspentCount / float(len(uniqueProducts))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/prarabdh/.local/lib/python2.7/site-packages/ipykernel_launcher.py:5: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  \"\"\"\n",
      "/home/prarabdh/.local/lib/python2.7/site-packages/ipykernel_launcher.py:7: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  import sys\n",
      "/home/prarabdh/.local/lib/python2.7/site-packages/ipykernel_launcher.py:9: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  if __name__ == '__main__':\n",
      "/home/prarabdh/.local/lib/python2.7/site-packages/ipykernel_launcher.py:11: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  # This is added back by InteractiveShellApp.init_path()\n"
     ]
    }
   ],
   "source": [
    "# Generating data for applying clustering algorithms\n",
    "\n",
    "data_frame = pd.DataFrame(columns=['averageSpending', 'Duration Between Purchases', 'Same Item Over and Over', 'Big Bucket Price', 'Label'], index=list(map(int, grp.groups.keys())))\n",
    "for key, item in avgSpending.items():\n",
    "    data_frame.ix[key]['averageSpending'] = item\n",
    "for key, item in avgTimeBetweenBuys.items():\n",
    "    data_frame.ix[key]['Duration Between Purchases'] = item\n",
    "for key, item in freqDict.items():\n",
    "    data_frame.ix[key]['Same Item Over and Over'] = item\n",
    "for key, item in percentageOverspent.items():\n",
    "    data_frame.ix[key]['Big Bucket Price'] = item\n",
    "\n",
    "# print(data_frame.info())\n",
    "data_frame.fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3303, 4)\n",
      "(2203, 4)\n"
     ]
    }
   ],
   "source": [
    "# Labeling Data based on the k-means clustering algorithms, as the percentage of customers in were closest to the manual analysis\n",
    "# 0 means low risk-taker, 2 means high-risk taker, 1 means average risk taker\n",
    "labels = [2, 1, 0]\n",
    "model = KMeans(n_clusters=3)\n",
    "model.fit(data_frame)\n",
    "yhat = model.predict(data_frame)\n",
    "# model = AgglomerativeClustering(n_clusters=3)\n",
    "# model.fit(data_frame)\n",
    "# yhat = model.fit_predict(data_frame)\n",
    "clusters = np.unique(yhat)\n",
    "i = 0\n",
    "remove = []\n",
    "pd.set_option('display.max_rows', 5000)\n",
    "pd.set_option('display.max_columns', 6)\n",
    "for cluster in clusters:\n",
    "    row_ix = where(yhat == cluster)\n",
    "    for index in row_ix[0]:\n",
    "        cust_id = grp.groups.keys()[index]\n",
    "        data_frame.at[int(cust_id), 'Label'] = labels[i]\n",
    "    i += 1\n",
    "\n",
    "y = data_frame['Label']\n",
    "x = data_frame.drop('Label', axis=1)\n",
    "x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.4)\n",
    "print(x_train.shape)\n",
    "print(x_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 165    0    0]\n",
      " [   2  873    0]\n",
      " [   0    0 1163]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.99      1.00      0.99       165\n",
      "           1       1.00      1.00      1.00       875\n",
      "           2       1.00      1.00      1.00      1163\n",
      "\n",
      "   micro avg       1.00      1.00      1.00      2203\n",
      "   macro avg       1.00      1.00      1.00      2203\n",
      "weighted avg       1.00      1.00      1.00      2203\n",
      "\n",
      "0.9990921470721743\n"
     ]
    }
   ],
   "source": [
    "classifier = RandomForestClassifier(n_estimators=4, random_state=0)\n",
    "classifier.fit(x_train, y_train)\n",
    "y_pred = classifier.predict(x_test)\n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))\n",
    "print(accuracy_score(y_test, y_pred))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.17"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}