CustomerSegmentation/3_DBSCAN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Customer Segmentation using DBSCAN\n",
    "Density-based spatial clustering (DBSCAN) is the other clustering algorithms we tested with the order history data. The algorithm groups together points that are close to each other based on a distance measurement, and marks the points in low-density regions as outliers.\n",
    "\n",
    "Model tweaking is done by tuning two key parameters:\n",
    "1. clustering radius (eps): maximum distance between two data points to be considered as the same group\n",
    "2. minimum sample size (min_samples): minimum number of points needed to form a group"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Import Depdendencies and Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import gridspec\n",
    "import seaborn as sns\n",
    "sns.set(font_scale=1.1)\n",
    "\n",
    "from sklearn import datasets\n",
    "from sklearn.cluster import DBSCAN\n",
    "from sklearn import metrics\n",
    "from scipy.spatial.distance import pdist, cdist, squareform"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>61</th>\n",
       "      <th>62</th>\n",
       "      <th>63</th>\n",
       "      <th>64</th>\n",
       "      <th>65</th>\n",
       "      <th>66</th>\n",
       "      <th>67</th>\n",
       "      <th>68</th>\n",
       "      <th>69</th>\n",
       "      <th>70</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>user_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>19510</th>\n",
       "      <td>-5.910101</td>\n",
       "      <td>-1.817756</td>\n",
       "      <td>-3.132463</td>\n",
       "      <td>-0.107840</td>\n",
       "      <td>2.262797</td>\n",
       "      <td>-1.085222</td>\n",
       "      <td>-0.343117</td>\n",
       "      <td>1.695994</td>\n",
       "      <td>0.221188</td>\n",
       "      <td>-0.033039</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.272325</td>\n",
       "      <td>0.211194</td>\n",
       "      <td>0.998564</td>\n",
       "      <td>0.661929</td>\n",
       "      <td>-0.246326</td>\n",
       "      <td>-0.339113</td>\n",
       "      <td>0.429339</td>\n",
       "      <td>0.863432</td>\n",
       "      <td>-0.747858</td>\n",
       "      <td>-0.352087</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>147319</th>\n",
       "      <td>-6.179444</td>\n",
       "      <td>-2.108571</td>\n",
       "      <td>-3.644230</td>\n",
       "      <td>0.078246</td>\n",
       "      <td>4.048101</td>\n",
       "      <td>-1.165402</td>\n",
       "      <td>1.540327</td>\n",
       "      <td>-0.342280</td>\n",
       "      <td>-0.159625</td>\n",
       "      <td>-0.925080</td>\n",
       "      <td>...</td>\n",
       "      <td>0.326506</td>\n",
       "      <td>0.172667</td>\n",
       "      <td>-1.231313</td>\n",
       "      <td>-0.543753</td>\n",
       "      <td>0.244803</td>\n",
       "      <td>0.049766</td>\n",
       "      <td>-0.890263</td>\n",
       "      <td>1.930898</td>\n",
       "      <td>0.833700</td>\n",
       "      <td>1.178497</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85028</th>\n",
       "      <td>175.175162</td>\n",
       "      <td>-2.025396</td>\n",
       "      <td>-0.748486</td>\n",
       "      <td>1.191982</td>\n",
       "      <td>5.387377</td>\n",
       "      <td>3.564970</td>\n",
       "      <td>-5.178943</td>\n",
       "      <td>-5.159448</td>\n",
       "      <td>-2.580185</td>\n",
       "      <td>0.534896</td>\n",
       "      <td>...</td>\n",
       "      <td>1.137074</td>\n",
       "      <td>-0.716124</td>\n",
       "      <td>1.016299</td>\n",
       "      <td>3.055268</td>\n",
       "      <td>-0.415043</td>\n",
       "      <td>-0.712064</td>\n",
       "      <td>1.742104</td>\n",
       "      <td>0.337453</td>\n",
       "      <td>0.706408</td>\n",
       "      <td>6.483967</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72168</th>\n",
       "      <td>-5.815371</td>\n",
       "      <td>0.090122</td>\n",
       "      <td>-1.756488</td>\n",
       "      <td>-0.347117</td>\n",
       "      <td>-2.340362</td>\n",
       "      <td>-8.403755</td>\n",
       "      <td>4.236185</td>\n",
       "      <td>0.557626</td>\n",
       "      <td>-1.000335</td>\n",
       "      <td>-1.694324</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.059698</td>\n",
       "      <td>1.697802</td>\n",
       "      <td>0.728849</td>\n",
       "      <td>-0.816363</td>\n",
       "      <td>0.541905</td>\n",
       "      <td>0.731675</td>\n",
       "      <td>0.402328</td>\n",
       "      <td>-0.866669</td>\n",
       "      <td>0.615065</td>\n",
       "      <td>-0.552924</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>142535</th>\n",
       "      <td>-6.094946</td>\n",
       "      <td>-1.132857</td>\n",
       "      <td>-2.957879</td>\n",
       "      <td>0.130307</td>\n",
       "      <td>2.361835</td>\n",
       "      <td>-3.727978</td>\n",
       "      <td>1.167230</td>\n",
       "      <td>-3.244399</td>\n",
       "      <td>-0.304043</td>\n",
       "      <td>-0.437521</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.171043</td>\n",
       "      <td>-0.366000</td>\n",
       "      <td>-0.145396</td>\n",
       "      <td>0.473588</td>\n",
       "      <td>0.183927</td>\n",
       "      <td>-0.220939</td>\n",
       "      <td>0.381579</td>\n",
       "      <td>-0.700740</td>\n",
       "      <td>0.339453</td>\n",
       "      <td>0.065392</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 71 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                  0         1         2         3         4         5  \\\n",
       "user_id                                                                 \n",
       "19510     -5.910101 -1.817756 -3.132463 -0.107840  2.262797 -1.085222   \n",
       "147319    -6.179444 -2.108571 -3.644230  0.078246  4.048101 -1.165402   \n",
       "85028    175.175162 -2.025396 -0.748486  1.191982  5.387377  3.564970   \n",
       "72168     -5.815371  0.090122 -1.756488 -0.347117 -2.340362 -8.403755   \n",
       "142535    -6.094946 -1.132857 -2.957879  0.130307  2.361835 -3.727978   \n",
       "\n",
       "                6         7         8         9  ...        61        62  \\\n",
       "user_id                                          ...                       \n",
       "19510   -0.343117  1.695994  0.221188 -0.033039  ... -0.272325  0.211194   \n",
       "147319   1.540327 -0.342280 -0.159625 -0.925080  ...  0.326506  0.172667   \n",
       "85028   -5.178943 -5.159448 -2.580185  0.534896  ...  1.137074 -0.716124   \n",
       "72168    4.236185  0.557626 -1.000335 -1.694324  ... -1.059698  1.697802   \n",
       "142535   1.167230 -3.244399 -0.304043 -0.437521  ... -0.171043 -0.366000   \n",
       "\n",
       "               63        64        65        66        67        68        69  \\\n",
       "user_id                                                                         \n",
       "19510    0.998564  0.661929 -0.246326 -0.339113  0.429339  0.863432 -0.747858   \n",
       "147319  -1.231313 -0.543753  0.244803  0.049766 -0.890263  1.930898  0.833700   \n",
       "85028    1.016299  3.055268 -0.415043 -0.712064  1.742104  0.337453  0.706408   \n",
       "72168    0.728849 -0.816363  0.541905  0.731675  0.402328 -0.866669  0.615065   \n",
       "142535  -0.145396  0.473588  0.183927 -0.220939  0.381579 -0.700740  0.339453   \n",
       "\n",
       "               70  \n",
       "user_id            \n",
       "19510   -0.352087  \n",
       "147319   1.178497  \n",
       "85028    6.483967  \n",
       "72168   -0.552924  \n",
       "142535   0.065392  \n",
       "\n",
       "[5 rows x 71 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_path = \"../Outputs/customer_order_train_reduced.csv\"\n",
    "df = pd.read_csv(df_path, index_col='user_id')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "DBSCAN model is affected by two key parameters: \n",
    "1. clustering radius \n",
    "2. minimum sample size\n",
    "\n",
    "Below are several approaches we tried, \n",
    "to find out(narrow down) the appropriate parameters that fit into out model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Iterations\n",
    "### Iteration 1: Clustering Radius = 3 and Minimum Sample Size = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>62</th>\n",
       "      <th>63</th>\n",
       "      <th>64</th>\n",
       "      <th>65</th>\n",
       "      <th>66</th>\n",
       "      <th>67</th>\n",
       "      <th>68</th>\n",
       "      <th>69</th>\n",
       "      <th>70</th>\n",
       "      <th>cluster_db</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>user_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>19510</th>\n",
       "      <td>-5.910101</td>\n",
       "      <td>-1.817756</td>\n",
       "      <td>-3.132463</td>\n",
       "      <td>-0.107840</td>\n",
       "      <td>2.262797</td>\n",
       "      <td>-1.085222</td>\n",
       "      <td>-0.343117</td>\n",
       "      <td>1.695994</td>\n",
       "      <td>0.221188</td>\n",
       "      <td>-0.033039</td>\n",
       "      <td>...</td>\n",
       "      <td>0.211194</td>\n",
       "      <td>0.998564</td>\n",
       "      <td>0.661929</td>\n",
       "      <td>-0.246326</td>\n",
       "      <td>-0.339113</td>\n",
       "      <td>0.429339</td>\n",
       "      <td>0.863432</td>\n",
       "      <td>-0.747858</td>\n",
       "      <td>-0.352087</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111519</th>\n",
       "      <td>-4.806822</td>\n",
       "      <td>-1.446159</td>\n",
       "      <td>-1.935056</td>\n",
       "      <td>-0.490396</td>\n",
       "      <td>-0.156673</td>\n",
       "      <td>4.438962</td>\n",
       "      <td>-2.202389</td>\n",
       "      <td>-2.115206</td>\n",
       "      <td>-5.441359</td>\n",
       "      <td>1.365619</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.781457</td>\n",
       "      <td>1.774493</td>\n",
       "      <td>-0.213561</td>\n",
       "      <td>-0.979514</td>\n",
       "      <td>0.732921</td>\n",
       "      <td>-1.196980</td>\n",
       "      <td>0.024342</td>\n",
       "      <td>0.078431</td>\n",
       "      <td>0.667055</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>135017</th>\n",
       "      <td>212.989382</td>\n",
       "      <td>-3.405884</td>\n",
       "      <td>-1.692809</td>\n",
       "      <td>2.990067</td>\n",
       "      <td>8.951872</td>\n",
       "      <td>-8.676947</td>\n",
       "      <td>2.455723</td>\n",
       "      <td>7.668930</td>\n",
       "      <td>5.612855</td>\n",
       "      <td>2.470697</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.917756</td>\n",
       "      <td>-4.889316</td>\n",
       "      <td>2.396766</td>\n",
       "      <td>0.316241</td>\n",
       "      <td>-1.014786</td>\n",
       "      <td>5.129032</td>\n",
       "      <td>-1.290806</td>\n",
       "      <td>-0.365117</td>\n",
       "      <td>2.176869</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>74317</th>\n",
       "      <td>-5.126611</td>\n",
       "      <td>1.649594</td>\n",
       "      <td>-0.396849</td>\n",
       "      <td>-1.243026</td>\n",
       "      <td>-6.859559</td>\n",
       "      <td>1.254977</td>\n",
       "      <td>-5.574833</td>\n",
       "      <td>-2.597643</td>\n",
       "      <td>0.167228</td>\n",
       "      <td>-2.614968</td>\n",
       "      <td>...</td>\n",
       "      <td>2.262043</td>\n",
       "      <td>0.710900</td>\n",
       "      <td>-0.625739</td>\n",
       "      <td>3.306304</td>\n",
       "      <td>-0.824299</td>\n",
       "      <td>-2.177741</td>\n",
       "      <td>0.757316</td>\n",
       "      <td>0.510841</td>\n",
       "      <td>1.412737</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16200</th>\n",
       "      <td>-6.256468</td>\n",
       "      <td>4.613436</td>\n",
       "      <td>-0.221773</td>\n",
       "      <td>11.513154</td>\n",
       "      <td>5.271039</td>\n",
       "      <td>-0.205424</td>\n",
       "      <td>-0.983184</td>\n",
       "      <td>1.703804</td>\n",
       "      <td>0.399320</td>\n",
       "      <td>-0.257367</td>\n",
       "      <td>...</td>\n",
       "      <td>0.587844</td>\n",
       "      <td>0.569096</td>\n",
       "      <td>0.612419</td>\n",
       "      <td>-0.084372</td>\n",
       "      <td>-0.500902</td>\n",
       "      <td>0.366956</td>\n",
       "      <td>0.230599</td>\n",
       "      <td>-0.150323</td>\n",
       "      <td>0.659721</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>147232</th>\n",
       "      <td>-6.211017</td>\n",
       "      <td>-2.647336</td>\n",
       "      <td>-3.927965</td>\n",
       "      <td>0.299520</td>\n",
       "      <td>5.079107</td>\n",
       "      <td>-0.818853</td>\n",
       "      <td>0.247882</td>\n",
       "      <td>0.910282</td>\n",
       "      <td>-0.454900</td>\n",
       "      <td>-0.062453</td>\n",
       "      <td>...</td>\n",
       "      <td>0.302564</td>\n",
       "      <td>1.010004</td>\n",
       "      <td>0.426967</td>\n",
       "      <td>-0.627227</td>\n",
       "      <td>-0.138537</td>\n",
       "      <td>0.049120</td>\n",
       "      <td>0.413423</td>\n",
       "      <td>-0.212321</td>\n",
       "      <td>0.855941</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>84259</th>\n",
       "      <td>-6.293424</td>\n",
       "      <td>-3.881471</td>\n",
       "      <td>-2.361712</td>\n",
       "      <td>0.325190</td>\n",
       "      <td>6.061423</td>\n",
       "      <td>-0.750071</td>\n",
       "      <td>-0.031002</td>\n",
       "      <td>0.665127</td>\n",
       "      <td>-1.035822</td>\n",
       "      <td>-0.257883</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.237867</td>\n",
       "      <td>0.228055</td>\n",
       "      <td>-0.132464</td>\n",
       "      <td>0.097434</td>\n",
       "      <td>0.432506</td>\n",
       "      <td>0.155175</td>\n",
       "      <td>-0.069207</td>\n",
       "      <td>0.274245</td>\n",
       "      <td>0.052724</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7153</th>\n",
       "      <td>-6.224541</td>\n",
       "      <td>-3.347103</td>\n",
       "      <td>-3.154720</td>\n",
       "      <td>0.250362</td>\n",
       "      <td>5.496618</td>\n",
       "      <td>-0.551986</td>\n",
       "      <td>0.153831</td>\n",
       "      <td>0.937717</td>\n",
       "      <td>-1.007968</td>\n",
       "      <td>-0.216954</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.667452</td>\n",
       "      <td>0.401532</td>\n",
       "      <td>-0.296061</td>\n",
       "      <td>-0.069006</td>\n",
       "      <td>0.637682</td>\n",
       "      <td>0.253191</td>\n",
       "      <td>-0.067339</td>\n",
       "      <td>-0.139465</td>\n",
       "      <td>-0.392145</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8206</th>\n",
       "      <td>-6.273801</td>\n",
       "      <td>-2.865810</td>\n",
       "      <td>-4.001854</td>\n",
       "      <td>0.362304</td>\n",
       "      <td>5.385191</td>\n",
       "      <td>-1.620964</td>\n",
       "      <td>0.416888</td>\n",
       "      <td>2.010377</td>\n",
       "      <td>0.022642</td>\n",
       "      <td>-0.130830</td>\n",
       "      <td>...</td>\n",
       "      <td>0.239630</td>\n",
       "      <td>0.210548</td>\n",
       "      <td>0.502688</td>\n",
       "      <td>-0.475916</td>\n",
       "      <td>0.507586</td>\n",
       "      <td>0.299351</td>\n",
       "      <td>-0.406129</td>\n",
       "      <td>0.053151</td>\n",
       "      <td>-0.114358</td>\n",
       "      <td>17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90522</th>\n",
       "      <td>-6.233045</td>\n",
       "      <td>-2.730503</td>\n",
       "      <td>-3.888187</td>\n",
       "      <td>0.276768</td>\n",
       "      <td>4.878196</td>\n",
       "      <td>-2.020615</td>\n",
       "      <td>1.039537</td>\n",
       "      <td>1.719261</td>\n",
       "      <td>-0.519897</td>\n",
       "      <td>-0.346357</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.098922</td>\n",
       "      <td>0.389839</td>\n",
       "      <td>0.185612</td>\n",
       "      <td>-0.520321</td>\n",
       "      <td>0.057343</td>\n",
       "      <td>0.447906</td>\n",
       "      <td>0.055288</td>\n",
       "      <td>0.266153</td>\n",
       "      <td>-0.349917</td>\n",
       "      <td>17</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>20168 rows × 72 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                  0         1         2          3         4         5  \\\n",
       "user_id                                                                  \n",
       "19510     -5.910101 -1.817756 -3.132463  -0.107840  2.262797 -1.085222   \n",
       "111519    -4.806822 -1.446159 -1.935056  -0.490396 -0.156673  4.438962   \n",
       "135017   212.989382 -3.405884 -1.692809   2.990067  8.951872 -8.676947   \n",
       "74317     -5.126611  1.649594 -0.396849  -1.243026 -6.859559  1.254977   \n",
       "16200     -6.256468  4.613436 -0.221773  11.513154  5.271039 -0.205424   \n",
       "...             ...       ...       ...        ...       ...       ...   \n",
       "147232    -6.211017 -2.647336 -3.927965   0.299520  5.079107 -0.818853   \n",
       "84259     -6.293424 -3.881471 -2.361712   0.325190  6.061423 -0.750071   \n",
       "7153      -6.224541 -3.347103 -3.154720   0.250362  5.496618 -0.551986   \n",
       "8206      -6.273801 -2.865810 -4.001854   0.362304  5.385191 -1.620964   \n",
       "90522     -6.233045 -2.730503 -3.888187   0.276768  4.878196 -2.020615   \n",
       "\n",
       "                6         7         8         9  ...        62        63  \\\n",
       "user_id                                          ...                       \n",
       "19510   -0.343117  1.695994  0.221188 -0.033039  ...  0.211194  0.998564   \n",
       "111519  -2.202389 -2.115206 -5.441359  1.365619  ... -0.781457  1.774493   \n",
       "135017   2.455723  7.668930  5.612855  2.470697  ... -1.917756 -4.889316   \n",
       "74317   -5.574833 -2.597643  0.167228 -2.614968  ...  2.262043  0.710900   \n",
       "16200   -0.983184  1.703804  0.399320 -0.257367  ...  0.587844  0.569096   \n",
       "...           ...       ...       ...       ...  ...       ...       ...   \n",
       "147232   0.247882  0.910282 -0.454900 -0.062453  ...  0.302564  1.010004   \n",
       "84259   -0.031002  0.665127 -1.035822 -0.257883  ... -0.237867  0.228055   \n",
       "7153     0.153831  0.937717 -1.007968 -0.216954  ... -0.667452  0.401532   \n",
       "8206     0.416888  2.010377  0.022642 -0.130830  ...  0.239630  0.210548   \n",
       "90522    1.039537  1.719261 -0.519897 -0.346357  ... -0.098922  0.389839   \n",
       "\n",
       "               64        65        66        67        68        69        70  \\\n",
       "user_id                                                                         \n",
       "19510    0.661929 -0.246326 -0.339113  0.429339  0.863432 -0.747858 -0.352087   \n",
       "111519  -0.213561 -0.979514  0.732921 -1.196980  0.024342  0.078431  0.667055   \n",
       "135017   2.396766  0.316241 -1.014786  5.129032 -1.290806 -0.365117  2.176869   \n",
       "74317   -0.625739  3.306304 -0.824299 -2.177741  0.757316  0.510841  1.412737   \n",
       "16200    0.612419 -0.084372 -0.500902  0.366956  0.230599 -0.150323  0.659721   \n",
       "...           ...       ...       ...       ...       ...       ...       ...   \n",
       "147232   0.426967 -0.627227 -0.138537  0.049120  0.413423 -0.212321  0.855941   \n",
       "84259   -0.132464  0.097434  0.432506  0.155175 -0.069207  0.274245  0.052724   \n",
       "7153    -0.296061 -0.069006  0.637682  0.253191 -0.067339 -0.139465 -0.392145   \n",
       "8206     0.502688 -0.475916  0.507586  0.299351 -0.406129  0.053151 -0.114358   \n",
       "90522    0.185612 -0.520321  0.057343  0.447906  0.055288  0.266153 -0.349917   \n",
       "\n",
       "         cluster_db  \n",
       "user_id              \n",
       "19510            -1  \n",
       "111519           -1  \n",
       "135017           -1  \n",
       "74317            -1  \n",
       "16200            -1  \n",
       "...             ...  \n",
       "147232           15  \n",
       "84259            16  \n",
       "7153             16  \n",
       "8206             17  \n",
       "90522            17  \n",
       "\n",
       "[20168 rows x 72 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = df.values\n",
    "# For the first tiral\n",
    "# set radius as 3 and minimum sample size as 2\n",
    "db = DBSCAN(eps=3, min_samples=2).fit(X)\n",
    " \n",
    "labels = db.labels_ \n",
    "df['cluster_db'] = labels  # add the clustering result back to the dataframe\n",
    "df.sort_values('cluster_db')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since DBSCAN is differentiate clusters based on their density and depend on the parameters we set up, some objects may be considered as noise(outlier) which will not been included in any cluster."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ideally, we want every single datapoint been grouped into cluster, but in most cases noise is always exist. So the first step for us is to reduce the nosie to certain level (in our case, we chose noise level = 0.009 as threshold), in other words, make sure the parameters we picked has a noise level no more than 0.009"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In addition, in order to compare with K-mean result, we would like to see the grouping result when cluser number euqal to 5."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>61</th>\n",
       "      <th>62</th>\n",
       "      <th>63</th>\n",
       "      <th>64</th>\n",
       "      <th>65</th>\n",
       "      <th>66</th>\n",
       "      <th>67</th>\n",
       "      <th>68</th>\n",
       "      <th>69</th>\n",
       "      <th>70</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cluster_db</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>-1</th>\n",
       "      <td>0.067217</td>\n",
       "      <td>0.049193</td>\n",
       "      <td>0.006158</td>\n",
       "      <td>-0.003273</td>\n",
       "      <td>-0.063361</td>\n",
       "      <td>0.010248</td>\n",
       "      <td>-0.006080</td>\n",
       "      <td>-0.011418</td>\n",
       "      <td>0.002808</td>\n",
       "      <td>0.002833</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000982</td>\n",
       "      <td>0.001888</td>\n",
       "      <td>-0.003702</td>\n",
       "      <td>0.001829</td>\n",
       "      <td>0.001197</td>\n",
       "      <td>-0.001128</td>\n",
       "      <td>-0.000786</td>\n",
       "      <td>0.000542</td>\n",
       "      <td>0.000156</td>\n",
       "      <td>0.002775</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-6.895333</td>\n",
       "      <td>-29.013606</td>\n",
       "      <td>49.087990</td>\n",
       "      <td>0.075192</td>\n",
       "      <td>10.541499</td>\n",
       "      <td>0.414168</td>\n",
       "      <td>1.693615</td>\n",
       "      <td>0.775856</td>\n",
       "      <td>0.010336</td>\n",
       "      <td>-0.009800</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.111120</td>\n",
       "      <td>0.115260</td>\n",
       "      <td>0.549128</td>\n",
       "      <td>-0.379190</td>\n",
       "      <td>0.367533</td>\n",
       "      <td>0.369147</td>\n",
       "      <td>0.320080</td>\n",
       "      <td>-0.380862</td>\n",
       "      <td>0.025982</td>\n",
       "      <td>0.017502</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-6.256492</td>\n",
       "      <td>-2.989861</td>\n",
       "      <td>-3.875127</td>\n",
       "      <td>0.324630</td>\n",
       "      <td>5.633935</td>\n",
       "      <td>-1.106020</td>\n",
       "      <td>0.480189</td>\n",
       "      <td>1.138911</td>\n",
       "      <td>-0.210023</td>\n",
       "      <td>-0.345828</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.077205</td>\n",
       "      <td>-0.195910</td>\n",
       "      <td>0.331077</td>\n",
       "      <td>-0.147238</td>\n",
       "      <td>-0.147858</td>\n",
       "      <td>0.064920</td>\n",
       "      <td>0.060119</td>\n",
       "      <td>-0.001149</td>\n",
       "      <td>-0.017002</td>\n",
       "      <td>-0.303921</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-6.298746</td>\n",
       "      <td>-2.642974</td>\n",
       "      <td>-3.767719</td>\n",
       "      <td>0.298217</td>\n",
       "      <td>4.998708</td>\n",
       "      <td>-2.780339</td>\n",
       "      <td>0.559280</td>\n",
       "      <td>2.503800</td>\n",
       "      <td>1.182191</td>\n",
       "      <td>0.507989</td>\n",
       "      <td>...</td>\n",
       "      <td>0.011739</td>\n",
       "      <td>-0.516446</td>\n",
       "      <td>0.951742</td>\n",
       "      <td>-1.093328</td>\n",
       "      <td>-0.284629</td>\n",
       "      <td>0.946349</td>\n",
       "      <td>-0.452101</td>\n",
       "      <td>-0.507880</td>\n",
       "      <td>0.180149</td>\n",
       "      <td>-0.622664</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-6.275318</td>\n",
       "      <td>-2.885413</td>\n",
       "      <td>-4.109611</td>\n",
       "      <td>0.308644</td>\n",
       "      <td>5.768160</td>\n",
       "      <td>-0.503303</td>\n",
       "      <td>-0.376914</td>\n",
       "      <td>0.183829</td>\n",
       "      <td>-1.052025</td>\n",
       "      <td>-0.182080</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.029521</td>\n",
       "      <td>0.010474</td>\n",
       "      <td>0.166469</td>\n",
       "      <td>-0.381934</td>\n",
       "      <td>-0.350124</td>\n",
       "      <td>0.159750</td>\n",
       "      <td>-0.004079</td>\n",
       "      <td>-0.143524</td>\n",
       "      <td>-0.093563</td>\n",
       "      <td>-0.650821</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-6.225647</td>\n",
       "      <td>-2.890878</td>\n",
       "      <td>-4.024569</td>\n",
       "      <td>0.295333</td>\n",
       "      <td>5.452970</td>\n",
       "      <td>-0.697997</td>\n",
       "      <td>-0.223623</td>\n",
       "      <td>0.838530</td>\n",
       "      <td>-1.223010</td>\n",
       "      <td>-0.169858</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.205555</td>\n",
       "      <td>-0.297950</td>\n",
       "      <td>0.374490</td>\n",
       "      <td>-0.181045</td>\n",
       "      <td>-0.189673</td>\n",
       "      <td>0.317169</td>\n",
       "      <td>-0.103736</td>\n",
       "      <td>0.069058</td>\n",
       "      <td>0.030504</td>\n",
       "      <td>-0.215669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>-6.211767</td>\n",
       "      <td>-2.830695</td>\n",
       "      <td>-4.046308</td>\n",
       "      <td>0.236430</td>\n",
       "      <td>5.359576</td>\n",
       "      <td>-0.151066</td>\n",
       "      <td>0.226879</td>\n",
       "      <td>0.497887</td>\n",
       "      <td>-1.280133</td>\n",
       "      <td>-0.028194</td>\n",
       "      <td>...</td>\n",
       "      <td>0.301486</td>\n",
       "      <td>0.161067</td>\n",
       "      <td>0.374133</td>\n",
       "      <td>0.501784</td>\n",
       "      <td>-0.794218</td>\n",
       "      <td>-0.741958</td>\n",
       "      <td>0.559697</td>\n",
       "      <td>-0.093948</td>\n",
       "      <td>0.176557</td>\n",
       "      <td>0.846894</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>-6.599431</td>\n",
       "      <td>-14.641259</td>\n",
       "      <td>19.477480</td>\n",
       "      <td>0.284402</td>\n",
       "      <td>8.367922</td>\n",
       "      <td>-0.540982</td>\n",
       "      <td>0.971633</td>\n",
       "      <td>0.562056</td>\n",
       "      <td>-0.180984</td>\n",
       "      <td>-0.087323</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.335899</td>\n",
       "      <td>0.076976</td>\n",
       "      <td>0.458008</td>\n",
       "      <td>-0.212925</td>\n",
       "      <td>0.250406</td>\n",
       "      <td>-0.004220</td>\n",
       "      <td>-0.186271</td>\n",
       "      <td>-0.178820</td>\n",
       "      <td>0.140160</td>\n",
       "      <td>-0.095515</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>-6.869163</td>\n",
       "      <td>-26.808639</td>\n",
       "      <td>44.255382</td>\n",
       "      <td>0.133914</td>\n",
       "      <td>10.606468</td>\n",
       "      <td>0.355194</td>\n",
       "      <td>1.757410</td>\n",
       "      <td>0.172339</td>\n",
       "      <td>-0.565913</td>\n",
       "      <td>0.535145</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.148232</td>\n",
       "      <td>0.023886</td>\n",
       "      <td>0.383691</td>\n",
       "      <td>-0.493321</td>\n",
       "      <td>0.376593</td>\n",
       "      <td>0.254687</td>\n",
       "      <td>0.039829</td>\n",
       "      <td>-0.507704</td>\n",
       "      <td>-0.013724</td>\n",
       "      <td>0.132308</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>-6.954996</td>\n",
       "      <td>-30.314563</td>\n",
       "      <td>51.413216</td>\n",
       "      <td>0.095895</td>\n",
       "      <td>11.296909</td>\n",
       "      <td>0.657250</td>\n",
       "      <td>2.001597</td>\n",
       "      <td>-0.119141</td>\n",
       "      <td>-0.667306</td>\n",
       "      <td>0.764632</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.243244</td>\n",
       "      <td>0.103723</td>\n",
       "      <td>0.222393</td>\n",
       "      <td>-0.350282</td>\n",
       "      <td>0.695888</td>\n",
       "      <td>0.204880</td>\n",
       "      <td>0.209158</td>\n",
       "      <td>-0.585362</td>\n",
       "      <td>-0.256143</td>\n",
       "      <td>0.343968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>-6.221319</td>\n",
       "      <td>-3.038135</td>\n",
       "      <td>-3.714809</td>\n",
       "      <td>0.066835</td>\n",
       "      <td>4.839335</td>\n",
       "      <td>3.157141</td>\n",
       "      <td>6.182710</td>\n",
       "      <td>1.323994</td>\n",
       "      <td>2.052513</td>\n",
       "      <td>-0.769645</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.216993</td>\n",
       "      <td>-0.217262</td>\n",
       "      <td>0.459316</td>\n",
       "      <td>0.026191</td>\n",
       "      <td>-0.007094</td>\n",
       "      <td>-0.129000</td>\n",
       "      <td>-0.378193</td>\n",
       "      <td>0.012471</td>\n",
       "      <td>-0.127241</td>\n",
       "      <td>-0.560938</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>-6.243159</td>\n",
       "      <td>-2.755719</td>\n",
       "      <td>-3.883527</td>\n",
       "      <td>0.321883</td>\n",
       "      <td>5.121779</td>\n",
       "      <td>-1.408617</td>\n",
       "      <td>0.485838</td>\n",
       "      <td>1.714001</td>\n",
       "      <td>0.129109</td>\n",
       "      <td>-0.249605</td>\n",
       "      <td>...</td>\n",
       "      <td>0.074967</td>\n",
       "      <td>0.346838</td>\n",
       "      <td>0.710125</td>\n",
       "      <td>0.341874</td>\n",
       "      <td>0.283295</td>\n",
       "      <td>0.636307</td>\n",
       "      <td>2.452808</td>\n",
       "      <td>-0.012269</td>\n",
       "      <td>0.373567</td>\n",
       "      <td>-0.046459</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>-6.316858</td>\n",
       "      <td>-2.940206</td>\n",
       "      <td>-3.997706</td>\n",
       "      <td>0.379632</td>\n",
       "      <td>5.793638</td>\n",
       "      <td>-1.855474</td>\n",
       "      <td>0.360271</td>\n",
       "      <td>1.198003</td>\n",
       "      <td>0.047999</td>\n",
       "      <td>-0.251526</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.184925</td>\n",
       "      <td>-0.032836</td>\n",
       "      <td>-0.341299</td>\n",
       "      <td>-0.388026</td>\n",
       "      <td>-0.213138</td>\n",
       "      <td>-0.035716</td>\n",
       "      <td>0.558747</td>\n",
       "      <td>0.082680</td>\n",
       "      <td>-0.404211</td>\n",
       "      <td>-0.469070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>-6.155640</td>\n",
       "      <td>-2.972040</td>\n",
       "      <td>-3.508241</td>\n",
       "      <td>0.207440</td>\n",
       "      <td>4.900823</td>\n",
       "      <td>-0.217872</td>\n",
       "      <td>-0.075857</td>\n",
       "      <td>0.813035</td>\n",
       "      <td>-2.013936</td>\n",
       "      <td>0.446348</td>\n",
       "      <td>...</td>\n",
       "      <td>0.114634</td>\n",
       "      <td>-0.334970</td>\n",
       "      <td>0.459600</td>\n",
       "      <td>0.017832</td>\n",
       "      <td>-0.206095</td>\n",
       "      <td>0.157694</td>\n",
       "      <td>-0.082625</td>\n",
       "      <td>-0.047039</td>\n",
       "      <td>-0.098649</td>\n",
       "      <td>-0.420766</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>-6.294608</td>\n",
       "      <td>-2.910474</td>\n",
       "      <td>-4.076164</td>\n",
       "      <td>0.379485</td>\n",
       "      <td>5.736282</td>\n",
       "      <td>-1.521189</td>\n",
       "      <td>0.389787</td>\n",
       "      <td>1.246243</td>\n",
       "      <td>-0.048998</td>\n",
       "      <td>-0.240415</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.198550</td>\n",
       "      <td>-0.499003</td>\n",
       "      <td>0.191417</td>\n",
       "      <td>-0.922431</td>\n",
       "      <td>-0.086788</td>\n",
       "      <td>0.695874</td>\n",
       "      <td>-0.160089</td>\n",
       "      <td>-0.560474</td>\n",
       "      <td>0.020206</td>\n",
       "      <td>-0.444084</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>-6.204212</td>\n",
       "      <td>-2.820351</td>\n",
       "      <td>-3.955055</td>\n",
       "      <td>0.279818</td>\n",
       "      <td>5.221431</td>\n",
       "      <td>-0.928959</td>\n",
       "      <td>0.024450</td>\n",
       "      <td>1.341518</td>\n",
       "      <td>-0.455063</td>\n",
       "      <td>-0.431003</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.249298</td>\n",
       "      <td>-0.714295</td>\n",
       "      <td>0.644013</td>\n",
       "      <td>-0.787790</td>\n",
       "      <td>0.086342</td>\n",
       "      <td>0.934733</td>\n",
       "      <td>-0.406814</td>\n",
       "      <td>-0.486253</td>\n",
       "      <td>-0.286250</td>\n",
       "      <td>-0.064451</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>-6.251613</td>\n",
       "      <td>-2.744909</td>\n",
       "      <td>-3.968714</td>\n",
       "      <td>0.334472</td>\n",
       "      <td>5.330407</td>\n",
       "      <td>-0.981616</td>\n",
       "      <td>0.310586</td>\n",
       "      <td>0.968836</td>\n",
       "      <td>-0.343817</td>\n",
       "      <td>-0.127372</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.337373</td>\n",
       "      <td>-0.018574</td>\n",
       "      <td>0.879097</td>\n",
       "      <td>0.323645</td>\n",
       "      <td>-0.104845</td>\n",
       "      <td>-0.161734</td>\n",
       "      <td>0.094969</td>\n",
       "      <td>-0.057930</td>\n",
       "      <td>-0.027212</td>\n",
       "      <td>0.537397</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>-6.258983</td>\n",
       "      <td>-3.614287</td>\n",
       "      <td>-2.758216</td>\n",
       "      <td>0.287776</td>\n",
       "      <td>5.779020</td>\n",
       "      <td>-0.651029</td>\n",
       "      <td>0.061414</td>\n",
       "      <td>0.801422</td>\n",
       "      <td>-1.021895</td>\n",
       "      <td>-0.237418</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.049929</td>\n",
       "      <td>-0.452659</td>\n",
       "      <td>0.314794</td>\n",
       "      <td>-0.214262</td>\n",
       "      <td>0.014214</td>\n",
       "      <td>0.535094</td>\n",
       "      <td>0.204183</td>\n",
       "      <td>-0.068273</td>\n",
       "      <td>0.067390</td>\n",
       "      <td>-0.169710</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>-6.253423</td>\n",
       "      <td>-2.798157</td>\n",
       "      <td>-3.945020</td>\n",
       "      <td>0.319536</td>\n",
       "      <td>5.131693</td>\n",
       "      <td>-1.820789</td>\n",
       "      <td>0.728213</td>\n",
       "      <td>1.864819</td>\n",
       "      <td>-0.248627</td>\n",
       "      <td>-0.238593</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.327797</td>\n",
       "      <td>0.070354</td>\n",
       "      <td>0.300193</td>\n",
       "      <td>0.344150</td>\n",
       "      <td>-0.498119</td>\n",
       "      <td>0.282464</td>\n",
       "      <td>0.373628</td>\n",
       "      <td>-0.175421</td>\n",
       "      <td>0.159652</td>\n",
       "      <td>-0.232138</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>19 rows × 71 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                   0          1          2         3          4         5  \\\n",
       "cluster_db                                                                  \n",
       "-1          0.067217   0.049193   0.006158 -0.003273  -0.063361  0.010248   \n",
       " 0         -6.895333 -29.013606  49.087990  0.075192  10.541499  0.414168   \n",
       " 1         -6.256492  -2.989861  -3.875127  0.324630   5.633935 -1.106020   \n",
       " 2         -6.298746  -2.642974  -3.767719  0.298217   4.998708 -2.780339   \n",
       " 3         -6.275318  -2.885413  -4.109611  0.308644   5.768160 -0.503303   \n",
       " 4         -6.225647  -2.890878  -4.024569  0.295333   5.452970 -0.697997   \n",
       " 5         -6.211767  -2.830695  -4.046308  0.236430   5.359576 -0.151066   \n",
       " 6         -6.599431 -14.641259  19.477480  0.284402   8.367922 -0.540982   \n",
       " 7         -6.869163 -26.808639  44.255382  0.133914  10.606468  0.355194   \n",
       " 8         -6.954996 -30.314563  51.413216  0.095895  11.296909  0.657250   \n",
       " 9         -6.221319  -3.038135  -3.714809  0.066835   4.839335  3.157141   \n",
       " 10        -6.243159  -2.755719  -3.883527  0.321883   5.121779 -1.408617   \n",
       " 11        -6.316858  -2.940206  -3.997706  0.379632   5.793638 -1.855474   \n",
       " 12        -6.155640  -2.972040  -3.508241  0.207440   4.900823 -0.217872   \n",
       " 13        -6.294608  -2.910474  -4.076164  0.379485   5.736282 -1.521189   \n",
       " 14        -6.204212  -2.820351  -3.955055  0.279818   5.221431 -0.928959   \n",
       " 15        -6.251613  -2.744909  -3.968714  0.334472   5.330407 -0.981616   \n",
       " 16        -6.258983  -3.614287  -2.758216  0.287776   5.779020 -0.651029   \n",
       " 17        -6.253423  -2.798157  -3.945020  0.319536   5.131693 -1.820789   \n",
       "\n",
       "                   6         7         8         9  ...        61        62  \\\n",
       "cluster_db                                          ...                       \n",
       "-1         -0.006080 -0.011418  0.002808  0.002833  ...  0.000982  0.001888   \n",
       " 0          1.693615  0.775856  0.010336 -0.009800  ... -0.111120  0.115260   \n",
       " 1          0.480189  1.138911 -0.210023 -0.345828  ... -0.077205 -0.195910   \n",
       " 2          0.559280  2.503800  1.182191  0.507989  ...  0.011739 -0.516446   \n",
       " 3         -0.376914  0.183829 -1.052025 -0.182080  ... -0.029521  0.010474   \n",
       " 4         -0.223623  0.838530 -1.223010 -0.169858  ... -0.205555 -0.297950   \n",
       " 5          0.226879  0.497887 -1.280133 -0.028194  ...  0.301486  0.161067   \n",
       " 6          0.971633  0.562056 -0.180984 -0.087323  ... -0.335899  0.076976   \n",
       " 7          1.757410  0.172339 -0.565913  0.535145  ... -0.148232  0.023886   \n",
       " 8          2.001597 -0.119141 -0.667306  0.764632  ... -0.243244  0.103723   \n",
       " 9          6.182710  1.323994  2.052513 -0.769645  ... -0.216993 -0.217262   \n",
       " 10         0.485838  1.714001  0.129109 -0.249605  ...  0.074967  0.346838   \n",
       " 11         0.360271  1.198003  0.047999 -0.251526  ... -0.184925 -0.032836   \n",
       " 12        -0.075857  0.813035 -2.013936  0.446348  ...  0.114634 -0.334970   \n",
       " 13         0.389787  1.246243 -0.048998 -0.240415  ... -0.198550 -0.499003   \n",
       " 14         0.024450  1.341518 -0.455063 -0.431003  ... -0.249298 -0.714295   \n",
       " 15         0.310586  0.968836 -0.343817 -0.127372  ... -0.337373 -0.018574   \n",
       " 16         0.061414  0.801422 -1.021895 -0.237418  ... -0.049929 -0.452659   \n",
       " 17         0.728213  1.864819 -0.248627 -0.238593  ... -0.327797  0.070354   \n",
       "\n",
       "                  63        64        65        66        67        68  \\\n",
       "cluster_db                                                               \n",
       "-1         -0.003702  0.001829  0.001197 -0.001128 -0.000786  0.000542   \n",
       " 0          0.549128 -0.379190  0.367533  0.369147  0.320080 -0.380862   \n",
       " 1          0.331077 -0.147238 -0.147858  0.064920  0.060119 -0.001149   \n",
       " 2          0.951742 -1.093328 -0.284629  0.946349 -0.452101 -0.507880   \n",
       " 3          0.166469 -0.381934 -0.350124  0.159750 -0.004079 -0.143524   \n",
       " 4          0.374490 -0.181045 -0.189673  0.317169 -0.103736  0.069058   \n",
       " 5          0.374133  0.501784 -0.794218 -0.741958  0.559697 -0.093948   \n",
       " 6          0.458008 -0.212925  0.250406 -0.004220 -0.186271 -0.178820   \n",
       " 7          0.383691 -0.493321  0.376593  0.254687  0.039829 -0.507704   \n",
       " 8          0.222393 -0.350282  0.695888  0.204880  0.209158 -0.585362   \n",
       " 9          0.459316  0.026191 -0.007094 -0.129000 -0.378193  0.012471   \n",
       " 10         0.710125  0.341874  0.283295  0.636307  2.452808 -0.012269   \n",
       " 11        -0.341299 -0.388026 -0.213138 -0.035716  0.558747  0.082680   \n",
       " 12         0.459600  0.017832 -0.206095  0.157694 -0.082625 -0.047039   \n",
       " 13         0.191417 -0.922431 -0.086788  0.695874 -0.160089 -0.560474   \n",
       " 14         0.644013 -0.787790  0.086342  0.934733 -0.406814 -0.486253   \n",
       " 15         0.879097  0.323645 -0.104845 -0.161734  0.094969 -0.057930   \n",
       " 16         0.314794 -0.214262  0.014214  0.535094  0.204183 -0.068273   \n",
       " 17         0.300193  0.344150 -0.498119  0.282464  0.373628 -0.175421   \n",
       "\n",
       "                  69        70  \n",
       "cluster_db                      \n",
       "-1          0.000156  0.002775  \n",
       " 0          0.025982  0.017502  \n",
       " 1         -0.017002 -0.303921  \n",
       " 2          0.180149 -0.622664  \n",
       " 3         -0.093563 -0.650821  \n",
       " 4          0.030504 -0.215669  \n",
       " 5          0.176557  0.846894  \n",
       " 6          0.140160 -0.095515  \n",
       " 7         -0.013724  0.132308  \n",
       " 8         -0.256143  0.343968  \n",
       " 9         -0.127241 -0.560938  \n",
       " 10         0.373567 -0.046459  \n",
       " 11        -0.404211 -0.469070  \n",
       " 12        -0.098649 -0.420766  \n",
       " 13         0.020206 -0.444084  \n",
       " 14        -0.286250 -0.064451  \n",
       " 15        -0.027212  0.537397  \n",
       " 16         0.067390 -0.169710  \n",
       " 17         0.159652 -0.232138  \n",
       "\n",
       "[19 rows x 71 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.groupby('cluster_db').mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.989438714795716"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# calculate the noise level (in DBSCAN model noise are cluster as \"-1\")\n",
    "raito = len(labels[labels[:] == -1]) / len(labels)\n",
    "raito"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see than the nosie level is about 0.99, which mean about 99% of the datapoints were not grouped into any cluster. \n",
    "\n",
    "We kept trying..and noticed that noise level decrease when we incread the radius."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Iterating by testing with cluster radius of 5 to 105"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8784708449028164\n",
      "number of clusters: 75\n",
      "coefficient: -0.474\n",
      "0.26913923046410154\n",
      "number of clusters: 33\n",
      "coefficient: 0.018\n",
      "0.10551368504561681\n",
      "number of clusters: 24\n",
      "coefficient: 0.382\n",
      "0.04789765965886553\n",
      "number of clusters: 15\n",
      "coefficient: 0.384\n",
      "0.02360174533915113\n",
      "number of clusters: 6\n",
      "coefficient: 0.559\n",
      "0.012048790162633875\n",
      "number of clusters: 11\n",
      "coefficient: 0.633\n"
     ]
    }
   ],
   "source": [
    "raito_list_test=[]\n",
    "for x in range(5, 65, 10):\n",
    "    X_test = df.values\n",
    "    db_x = DBSCAN(eps=x, min_samples=2).fit(X_test)\n",
    "    labels = db_x.labels_ \n",
    "    \n",
    "    # calculate the noise level\n",
    "    raito_test = len(labels[labels[:] == -1]) / len(labels)\n",
    "    raito_list_test.append(raito_test)\n",
    "    \n",
    "    # get the number of clusters\n",
    "    n_clusters_ = len(set(labels)) - (1 if -1 in labels else 0)\n",
    "    \n",
    "    print(raito_test)\n",
    "    print('number of clusters: %d' % n_clusters_)\n",
    "    print(\"coefficient: %0.3f\" % metrics.silhouette_score(X, labels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.006346687822292741\n",
      "number of clusters: 6\n",
      "coefficient: 0.699\n",
      "0.0034212614042046807\n",
      "number of clusters: 3\n",
      "coefficient: 0.722\n",
      "0.0020825069416898056\n",
      "number of clusters: 2\n",
      "coefficient: 0.836\n",
      "0.0010412534708449028\n",
      "number of clusters: 1\n",
      "coefficient: 0.873\n",
      "0.0008429194763982546\n",
      "number of clusters: 1\n",
      "coefficient: 0.879\n"
     ]
    }
   ],
   "source": [
    "for x in range(65, 115, 10):\n",
    "    X_test = df.values\n",
    "    db_x = DBSCAN(eps=x, min_samples=2).fit(X_test)\n",
    "    labels = db_x.labels_ \n",
    "    \n",
    "    # calculate the noise level\n",
    "    raito_test = len(labels[labels[:] == -1]) / len(labels)\n",
    "    raito_list_test.append(raito_test)\n",
    "    \n",
    "    # get the number of clusters\n",
    "    n_clusters_ = len(set(labels)) - (1 if -1 in labels else 0)\n",
    "    \n",
    "    print(raito_test)\n",
    "    print('number of clusters: %d' % n_clusters_)\n",
    "    print(\"coefficient: %0.3f\" % metrics.silhouette_score(X, labels))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "it looks like when radius reaches 75, the cluster number would be less than 3, which does not meet criteria mentioned above. Also, the noise level is higher than our threshold when radius is below 55. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Increasing Minimum Sample Size from 2 to 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8835283617612059\n",
      "number of clusters: 24\n",
      "coefficient: -0.438\n",
      "0.27151923839746134\n",
      "number of clusters: 9\n",
      "coefficient: 0.153\n",
      "0.10739785799285997\n",
      "number of clusters: 5\n",
      "coefficient: 0.444\n",
      "0.0485918286394288\n",
      "number of clusters: 8\n",
      "coefficient: 0.487\n",
      "0.023998413328044428\n",
      "number of clusters: 2\n",
      "coefficient: 0.706\n",
      "0.012842126140420468\n",
      "number of clusters: 3\n",
      "coefficient: 0.722\n",
      "0.006644188813962713\n",
      "number of clusters: 3\n",
      "coefficient: 0.723\n",
      "0.003520428401428005\n",
      "number of clusters: 2\n",
      "coefficient: 0.722\n",
      "0.00218167393891313\n",
      "number of clusters: 1\n",
      "coefficient: 0.843\n",
      "0.0010412534708449028\n",
      "number of clusters: 1\n",
      "coefficient: 0.873\n",
      "0.0008429194763982546\n",
      "number of clusters: 1\n",
      "coefficient: 0.879\n"
     ]
    }
   ],
   "source": [
    "# we increased the minimum sample size from 2 to 3\n",
    "raito_list_test_2=[]\n",
    "for x in range(5, 115, 10):\n",
    "    X_test = df.values\n",
    "    db_x = DBSCAN(eps=x, min_samples=3).fit(X_test)\n",
    "    labels = db_x.labels_ \n",
    "    \n",
    "    # calculate the noise level\n",
    "    raito_test = len(labels[labels[:] == -1]) / len(labels)\n",
    "    raito_list_test_2.append(raito_test)\n",
    "    \n",
    "    # get the number of clusters\n",
    "    n_clusters_ = len(set(labels)) - (1 if -1 in labels else 0)\n",
    "    \n",
    "    print(raito_test)\n",
    "    print('number of clusters: %d' % n_clusters_)\n",
    "    print(\"coefficient: %0.3f\" % metrics.silhouette_score(X, labels))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It seems no radius range meets our criteria(noise level<0.009 & cluster number = 5)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. Back to Minimum Sample Size of 2 and Testing Radius Between 55 and 75"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.012048790162633875\n",
      "number of clusters: 11\n",
      "coefficient: 0.633\n",
      "0.011305037683458944\n",
      "number of clusters: 11\n",
      "coefficient: 0.593\n",
      "0.010858786195953986\n",
      "number of clusters: 10\n",
      "coefficient: 0.639\n",
      "0.010263784212614042\n",
      "number of clusters: 9\n",
      "coefficient: 0.639\n",
      "0.00971836572788576\n",
      "number of clusters: 9\n",
      "coefficient: 0.635\n",
      "0.009222530741769139\n",
      "number of clusters: 8\n",
      "coefficient: 0.688\n",
      "0.008478778262594208\n",
      "number of clusters: 8\n",
      "coefficient: 0.688\n",
      "0.00803252677508925\n",
      "number of clusters: 8\n",
      "coefficient: 0.688\n",
      "0.007387941293137644\n",
      "number of clusters: 7\n",
      "coefficient: 0.687\n",
      "0.006991273304244348\n",
      "number of clusters: 6\n",
      "coefficient: 0.689\n",
      "0.006346687822292741\n",
      "number of clusters: 6\n",
      "coefficient: 0.699\n",
      "0.005999603332011107\n",
      "number of clusters: 6\n",
      "coefficient: 0.699\n",
      "0.005454184847282825\n",
      "number of clusters: 5\n",
      "coefficient: 0.699\n",
      "0.005255850852836176\n",
      "number of clusters: 4\n",
      "coefficient: 0.722\n",
      "0.004908766362554542\n",
      "number of clusters: 5\n",
      "coefficient: 0.722\n",
      "0.0046608488694962315\n",
      "number of clusters: 5\n",
      "coefficient: 0.722\n",
      "0.0043137643792145975\n",
      "number of clusters: 5\n",
      "coefficient: 0.661\n",
      "0.004065846886156287\n",
      "number of clusters: 5\n",
      "coefficient: 0.661\n",
      "0.0039170963903213014\n",
      "number of clusters: 3\n",
      "coefficient: 0.722\n",
      "0.003669178897262991\n",
      "number of clusters: 3\n",
      "coefficient: 0.722\n"
     ]
    }
   ],
   "source": [
    "# head back to min_samples=2 and test test radius range from 55 to 75\n",
    "raito_list_test_3=[]\n",
    "for x in range(55, 75):\n",
    "    X_test = df.values\n",
    "    db_x = DBSCAN(eps=x, min_samples=2).fit(X_test)\n",
    "    labels = db_x.labels_ \n",
    "    \n",
    "    # calculate the noise level\n",
    "    raito_test = len(labels[labels[:] == -1]) / len(labels)\n",
    "    raito_list_test_3.append(raito_test)\n",
    "    \n",
    "    # get the number of clusters\n",
    "    n_clusters_ = len(set(labels)) - (1 if -1 in labels else 0)\n",
    "    \n",
    "    print(raito_test)\n",
    "    print('number of clusters: %d' % n_clusters_)\n",
    "    print(\"coefficient: %0.3f\" % metrics.silhouette_score(X, labels))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "we found that when radius equals to 67, 70, 71, 72, the clustering results meets our criteria"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5. Clustering Radius = 67 and Minimum Sample Size = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cluster -1: 110\n",
      "cluster 0: 20048\n",
      "cluster 1: 3\n",
      "cluster 2: 3\n",
      "cluster 3: 2\n",
      "cluster 4: 2\n"
     ]
    }
   ],
   "source": [
    "# radius 67\n",
    "db = DBSCAN(eps=67, min_samples=2).fit(X)\n",
    " \n",
    "labels = db.labels_ \n",
    "df['cluster_db'] = labels  # add the clustering result back to the dataframe\n",
    "df.sort_values('cluster_db')\n",
    "\n",
    "\n",
    "# print our the number of observations per cluster\n",
    "for c in np.unique(df.cluster_db):\n",
    "        print('cluster {}: {}'.format(c, np.sum(df.cluster_db==c)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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": [
    "n_clusters_ = len(set(labels)) - (1 if -1 in labels else 0)\n",
    "for i in range(n_clusters_):\n",
    "    one_cluster = X[labels == i]\n",
    "    plt.plot(one_cluster[:,0],one_cluster[:,1],'o')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6. Clustering Radius = 70 and Minimum Sample Size = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cluster -1: 94\n",
      "cluster 0: 20064\n",
      "cluster 1: 2\n",
      "cluster 2: 3\n",
      "cluster 3: 3\n",
      "cluster 4: 2\n"
     ]
    }
   ],
   "source": [
    "# radius 70\n",
    "db = DBSCAN(eps=70, min_samples=2).fit(X)\n",
    " \n",
    "labels = db.labels_ \n",
    "df['cluster_db'] = labels  # add the clustering result back to the dataframe\n",
    "df.sort_values('cluster_db')\n",
    "\n",
    "# print our the number of observations per cluster\n",
    "for c in np.unique(df.cluster_db):\n",
    "        print('cluster {}: {}'.format(c, np.sum(df.cluster_db==c)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7. Clustering Radius = 72 and Minimum Sample Size = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cluster -1: 82\n",
      "cluster 0: 20076\n",
      "cluster 1: 2\n",
      "cluster 2: 3\n",
      "cluster 3: 3\n",
      "cluster 4: 2\n"
     ]
    }
   ],
   "source": [
    "# radius 72\n",
    "db = DBSCAN(eps=72, min_samples=2).fit(X)\n",
    " \n",
    "labels = db.labels_ \n",
    "df['cluster_db'] = labels  # add the clustering result back to the dataframe\n",
    "df.sort_values('cluster_db')\n",
    "\n",
    "# print our the number of observations per cluster\n",
    "for c in np.unique(df.cluster_db):\n",
    "        print('cluster {}: {}'.format(c, np.sum(df.cluster_db==c)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Over the 'qualified' ones we chose, we could see that the majority of the datapoints are in one cluster."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 8. Maybe force group to have at least 50 people?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cluster -1: 10304\n",
      "cluster 0: 9864\n"
     ]
    }
   ],
   "source": [
    "db = DBSCAN(eps=10, min_samples=50).fit(X)\n",
    " \n",
    "labels = db.labels_ \n",
    "df['cluster_db'] = labels  # add the clustering result back to the dataframe\n",
    "df.sort_values('cluster_db')\n",
    "\n",
    "# print our the number of observations per cluster\n",
    "for c in np.unique(df.cluster_db):\n",
    "        print('cluster {}: {}'.format(c, np.sum(df.cluster_db==c)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It doesn't seem to produce more than 2 clusters by setting high minimum sample size\n",
    "\n",
    "Thus, we think DBSCAN may not be the appropriate method to cluster our dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7a. Let's check out the purchase pattern for each cluster for radius = 72 and sample size = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "cust_train = pd.read_csv('../Outputs/customer_order_train.csv', index_col='user_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>air fresheners candles</th>\n",
       "      <th>asian foods</th>\n",
       "      <th>baby accessories</th>\n",
       "      <th>baby bath body care</th>\n",
       "      <th>baby food formula</th>\n",
       "      <th>bakery desserts</th>\n",
       "      <th>baking ingredients</th>\n",
       "      <th>baking supplies decor</th>\n",
       "      <th>beauty</th>\n",
       "      <th>beers coolers</th>\n",
       "      <th>...</th>\n",
       "      <th>meat seafood</th>\n",
       "      <th>missing_y</th>\n",
       "      <th>other_y</th>\n",
       "      <th>pantry</th>\n",
       "      <th>personal care</th>\n",
       "      <th>pets</th>\n",
       "      <th>produce</th>\n",
       "      <th>snacks</th>\n",
       "      <th>organic</th>\n",
       "      <th>customer type</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>user_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>19510</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>15</td>\n",
       "      <td>0</td>\n",
       "      <td>76</td>\n",
       "      <td>14</td>\n",
       "      <td>152</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>147319</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>18</td>\n",
       "      <td>6</td>\n",
       "      <td>23</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85028</th>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>126</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>47</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>173</td>\n",
       "      <td>49</td>\n",
       "      <td>68</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72168</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>13</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>43</td>\n",
       "      <td>31</td>\n",
       "      <td>93</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>142535</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13607</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>16</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>398</td>\n",
       "      <td>67</td>\n",
       "      <td>519</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117498</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>25</td>\n",
       "      <td>6</td>\n",
       "      <td>60</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164881</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>33</td>\n",
       "      <td>6</td>\n",
       "      <td>11</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1918</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>43</td>\n",
       "      <td>14</td>\n",
       "      <td>101</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34069</th>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>18</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>61</td>\n",
       "      <td>28</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>20168 rows × 158 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         air fresheners candles  asian foods  baby accessories  \\\n",
       "user_id                                                          \n",
       "19510                         0            0                 0   \n",
       "147319                        0            1                 0   \n",
       "85028                         0            8                 0   \n",
       "72168                         0            0                 0   \n",
       "142535                        0            1                 0   \n",
       "...                         ...          ...               ...   \n",
       "13607                         0            0                 0   \n",
       "117498                        0            0                 0   \n",
       "164881                        0            1                 0   \n",
       "1918                          0            1                 0   \n",
       "34069                         0            2                 0   \n",
       "\n",
       "         baby bath body care  baby food formula  bakery desserts  \\\n",
       "user_id                                                            \n",
       "19510                      0                  0                0   \n",
       "147319                     0                  0                1   \n",
       "85028                      3                126                0   \n",
       "72168                      0                  0                0   \n",
       "142535                     0                  0                0   \n",
       "...                      ...                ...              ...   \n",
       "13607                      0                 19                0   \n",
       "117498                     0                  0                0   \n",
       "164881                     0                  0                0   \n",
       "1918                       0                  0                0   \n",
       "34069                      0                  0                0   \n",
       "\n",
       "         baking ingredients  baking supplies decor  beauty  beers coolers  \\\n",
       "user_id                                                                     \n",
       "19510                     0                      0       0              0   \n",
       "147319                    1                      2       0              0   \n",
       "85028                     1                      0       0              0   \n",
       "72168                     4                      0       0              0   \n",
       "142535                    1                      0       0              0   \n",
       "...                     ...                    ...     ...            ...   \n",
       "13607                     1                      0       0              0   \n",
       "117498                    0                      0       0              0   \n",
       "164881                    6                      0       0              0   \n",
       "1918                      0                      0       0              0   \n",
       "34069                     5                      1       0              0   \n",
       "\n",
       "         ...  meat seafood  missing_y  other_y  pantry  personal care  pets  \\\n",
       "user_id  ...                                                                  \n",
       "19510    ...             1          0        0       5             15     0   \n",
       "147319   ...             3          1        0       7              0     0   \n",
       "85028    ...            12          0        1      47              0     0   \n",
       "72168    ...             2          0        0      13              3     0   \n",
       "142535   ...             6          0        0       8              4     0   \n",
       "...      ...           ...        ...      ...     ...            ...   ...   \n",
       "13607    ...            16          1        0       9              3     0   \n",
       "117498   ...             1          9        0      11              3     1   \n",
       "164881   ...             1          0        0      12              1     0   \n",
       "1918     ...             2          0        0       3              0     0   \n",
       "34069    ...             6          0        0      18              8     0   \n",
       "\n",
       "         produce  snacks  organic  customer type  \n",
       "user_id                                           \n",
       "19510         76      14      152              0  \n",
       "147319        18       6       23              0  \n",
       "85028        173      49       68              0  \n",
       "72168         43      31       93              0  \n",
       "142535        12       5        1              0  \n",
       "...          ...     ...      ...            ...  \n",
       "13607        398      67      519              0  \n",
       "117498        25       6       60              0  \n",
       "164881        33       6       11              0  \n",
       "1918          43      14      101              0  \n",
       "34069         61      28        4              0  \n",
       "\n",
       "[20168 rows x 158 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cust_train['customer type'] = db.labels_\n",
    "cust_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# function to plot purchase pattern heat map\n",
    "# `order_history`: order history data (unscaled, non-dimension-reduction)\n",
    "# `plot_threshold`: percent of purchase for a product category to not display in heat map (0.02 = 2%)\n",
    "#                   [reduce number of categories to plot]\n",
    "def customer_group_heatmap(order_history, plot_threshold=0.02):\n",
    "    # 1) convert number of purchase to percent\n",
    "    # aisle categories\n",
    "    order_history.iloc[:,0:134] = order_history.iloc[:,0:134].div(order_history.Total, axis=0)\n",
    "    # department categories\n",
    "    order_history.iloc[:,135:156] = order_history.iloc[:,135:156].div(order_history.Total, axis=0)\n",
    "    # organic product\n",
    "    order_history.iloc[:,156] = order_history.iloc[:,156].div(order_history.Total, axis=0)\n",
    "    \n",
    "    # 2) average by customer group\n",
    "    cg_order_history = order_history.groupby('customer type').mean()\n",
    "\n",
    "    cg_aisle = cg_order_history.iloc[:,0:134]\n",
    "    cg_dept = cg_order_history.iloc[:,135:157]\n",
    "    cg_total = cg_order_history.iloc[:,[134]].round(0)\n",
    "    \n",
    "    # 3) threshold for a product category to display in heatmap\n",
    "    cg_aisle = cg_aisle.loc[:,(cg_aisle>plot_threshold).any()]\n",
    "    cg_dept = cg_dept.loc[:,(cg_dept>plot_threshold).any()]\n",
    "    \n",
    "    # 4) Number of customers per group\n",
    "    cg_count = order_history.iloc[:,-2:].groupby('customer type').count()\n",
    "    \n",
    "    # 5) plot heat map\n",
    "    fig = plt.figure(figsize=(10,12))\n",
    "    gridspec.GridSpec(10,1)\n",
    "    \n",
    "    # Map 1: Number of Customer in each cluster\n",
    "    plt.subplot2grid((10,1), (0,0))\n",
    "    plt.title('Number of Customers')\n",
    "    sns.heatmap(cg_count.transpose(), yticklabels=False, cmap=\"Greys\", annot=True, fmt='d', linewidths=0.2)\n",
    "    plt.xlabel(\"\")\n",
    "    \n",
    "    # Map 2: Avearage number of products purchased\n",
    "    plt.subplot2grid((10,1), (1,0))\n",
    "    plt.title('Average # of Items Purchased by Customer Type')\n",
    "    sns.heatmap(cg_total.transpose(), cmap=\"Reds\", annot=True, fmt='0g', linewidths=0.2)\n",
    "    plt.xlabel(\"\")\n",
    "\n",
    "    # Map 3: Average percent of products by aisle bought by customer cluster\n",
    "    plt.subplot2grid((10,1), (2,0), rowspan=4)\n",
    "    plt.title('Percent of Items Purchased by Customer Type (Aisle)')\n",
    "    sns.heatmap(cg_aisle.transpose(), cmap=\"Blues\", linewidths=0.1)\n",
    "    plt.xlabel(\"\")\n",
    "    \n",
    "    # Map 4: Average percent of products by department bought by customer cluster\n",
    "    plt.subplot2grid((10,1), (6,0), rowspan=4)\n",
    "    plt.title('Percent of Items Purchased by Customer Type (Department)')\n",
    "    sns.heatmap(cg_dept.transpose(), cmap=\"Greens\", linewidths=0.1)\n",
    "\n",
    "    plt.tight_layout()\n",
    "    plt.savefig(f'../Outputs/DBSCAN_Images/dbscan_72_2.png')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x864 with 8 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "customer_group_heatmap(cust_train.copy(), 0.04)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Groups are different from K-Means and less distinguished\n",
    "\n",
    "| # | Name              | Description                                                        |\n",
    "|:-:| ----------------- | ------------------------------------------------------------------ |\n",
    "| -1| Outlier           | 82 customers in this group, seems to buy a bit of everything       |\n",
    "| 0 | The Majority      | Most customers in this group, buying mostly fresh fruits and veges.|\n",
    "| 1 | College Folks     | Bought a lot of frozen meals and energy bars                       | \n",
    "| 2 | Cat Owner Group 1 | Half of their purchase are in cat food care, followed by beverages |\n",
    "| 3 | Organi-tarian     | Bought mostly organic produce                                      |\n",
    "| 4 | Cat Owner Group 2 | Half of their purchase are in cat food care, followed by dairy eggs|"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}