06_group_segmentation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Import libraries\n",
    "'''Main'''\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os, time, re\n",
    "import pickle, gzip\n",
    "\n",
    "'''Data Viz'''\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "color = sns.color_palette()\n",
    "import matplotlib as mpl\n",
    "\n",
    "%matplotlib inline\n",
    "\n",
    "'''Data Prep and Model Evaluation'''\n",
    "from sklearn import preprocessing as pp\n",
    "from sklearn.model_selection import train_test_split \n",
    "from sklearn.metrics import precision_recall_curve, average_precision_score\n",
    "from sklearn.metrics import roc_curve, auc, roc_auc_score\n",
    "\n",
    "'''Algorithms'''\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.cluster import KMeans\n",
    "import fastcluster\n",
    "from scipy.cluster.hierarchy import dendrogram, cophenet, fcluster\n",
    "from scipy.spatial.distance import pdist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ankur_\\Anaconda3\\envs\\unsupervisedLearning\\lib\\site-packages\\IPython\\core\\interactiveshell.py:2785: DtypeWarning: Columns (0,47) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  interactivity=interactivity, compiler=compiler, result=result)\n"
     ]
    }
   ],
   "source": [
    "# Load the data\n",
    "current_path = os.getcwd()\n",
    "file = os.path.sep.join(['', 'datasets', 'lending_club_data', 'LoanStats3a.csv'])\n",
    "data = pd.read_csv(current_path + file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Select columns to keep\n",
    "columnsToKeep = ['loan_amnt','funded_amnt','funded_amnt_inv','term', \\\n",
    "                 'int_rate','installment','grade','sub_grade', \\\n",
    "                 'emp_length','home_ownership','annual_inc', \\\n",
    "                 'verification_status','pymnt_plan','purpose', \\\n",
    "                 'addr_state','dti','delinq_2yrs','earliest_cr_line', \\\n",
    "                 'mths_since_last_delinq','mths_since_last_record', \\\n",
    "                 'open_acc','pub_rec','revol_bal','revol_util', \\\n",
    "                 'total_acc','initial_list_status','out_prncp', \\\n",
    "                 'out_prncp_inv','total_pymnt','total_pymnt_inv', \\\n",
    "                 'total_rec_prncp','total_rec_int','total_rec_late_fee', \\\n",
    "                 'recoveries','collection_recovery_fee','last_pymnt_d', \\\n",
    "                 'last_pymnt_amnt']\n",
    "\n",
    "data = data.loc[:,columnsToKeep]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(42542, 37)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>loan_amnt</th>\n",
       "      <th>funded_amnt</th>\n",
       "      <th>funded_amnt_inv</th>\n",
       "      <th>term</th>\n",
       "      <th>int_rate</th>\n",
       "      <th>installment</th>\n",
       "      <th>grade</th>\n",
       "      <th>sub_grade</th>\n",
       "      <th>emp_length</th>\n",
       "      <th>home_ownership</th>\n",
       "      <th>...</th>\n",
       "      <th>out_prncp_inv</th>\n",
       "      <th>total_pymnt</th>\n",
       "      <th>total_pymnt_inv</th>\n",
       "      <th>total_rec_prncp</th>\n",
       "      <th>total_rec_int</th>\n",
       "      <th>total_rec_late_fee</th>\n",
       "      <th>recoveries</th>\n",
       "      <th>collection_recovery_fee</th>\n",
       "      <th>last_pymnt_d</th>\n",
       "      <th>last_pymnt_amnt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5000.0</td>\n",
       "      <td>5000.0</td>\n",
       "      <td>4975.0</td>\n",
       "      <td>36 months</td>\n",
       "      <td>10.65%</td>\n",
       "      <td>162.87</td>\n",
       "      <td>B</td>\n",
       "      <td>B2</td>\n",
       "      <td>10+ years</td>\n",
       "      <td>RENT</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>5863.155187</td>\n",
       "      <td>5833.84</td>\n",
       "      <td>5000.00</td>\n",
       "      <td>863.16</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>Jan-15</td>\n",
       "      <td>171.62</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2500.0</td>\n",
       "      <td>2500.0</td>\n",
       "      <td>2500.0</td>\n",
       "      <td>60 months</td>\n",
       "      <td>15.27%</td>\n",
       "      <td>59.83</td>\n",
       "      <td>C</td>\n",
       "      <td>C4</td>\n",
       "      <td>&lt; 1 year</td>\n",
       "      <td>RENT</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1014.530000</td>\n",
       "      <td>1014.53</td>\n",
       "      <td>456.46</td>\n",
       "      <td>435.17</td>\n",
       "      <td>0.00</td>\n",
       "      <td>122.9</td>\n",
       "      <td>1.11</td>\n",
       "      <td>Apr-13</td>\n",
       "      <td>119.66</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2400.0</td>\n",
       "      <td>2400.0</td>\n",
       "      <td>2400.0</td>\n",
       "      <td>36 months</td>\n",
       "      <td>15.96%</td>\n",
       "      <td>84.33</td>\n",
       "      <td>C</td>\n",
       "      <td>C5</td>\n",
       "      <td>10+ years</td>\n",
       "      <td>RENT</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3005.666844</td>\n",
       "      <td>3005.67</td>\n",
       "      <td>2400.00</td>\n",
       "      <td>605.67</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>Jun-14</td>\n",
       "      <td>649.91</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10000.0</td>\n",
       "      <td>10000.0</td>\n",
       "      <td>10000.0</td>\n",
       "      <td>36 months</td>\n",
       "      <td>13.49%</td>\n",
       "      <td>339.31</td>\n",
       "      <td>C</td>\n",
       "      <td>C1</td>\n",
       "      <td>10+ years</td>\n",
       "      <td>RENT</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>12231.890000</td>\n",
       "      <td>12231.89</td>\n",
       "      <td>10000.00</td>\n",
       "      <td>2214.92</td>\n",
       "      <td>16.97</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>Jan-15</td>\n",
       "      <td>357.48</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3000.0</td>\n",
       "      <td>3000.0</td>\n",
       "      <td>3000.0</td>\n",
       "      <td>60 months</td>\n",
       "      <td>12.69%</td>\n",
       "      <td>67.79</td>\n",
       "      <td>B</td>\n",
       "      <td>B5</td>\n",
       "      <td>1 year</td>\n",
       "      <td>RENT</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4066.908161</td>\n",
       "      <td>4066.91</td>\n",
       "      <td>3000.00</td>\n",
       "      <td>1066.91</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>Jan-17</td>\n",
       "      <td>67.30</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 37 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   loan_amnt  funded_amnt  funded_amnt_inv        term int_rate  installment  \\\n",
       "0     5000.0       5000.0           4975.0   36 months   10.65%       162.87   \n",
       "1     2500.0       2500.0           2500.0   60 months   15.27%        59.83   \n",
       "2     2400.0       2400.0           2400.0   36 months   15.96%        84.33   \n",
       "3    10000.0      10000.0          10000.0   36 months   13.49%       339.31   \n",
       "4     3000.0       3000.0           3000.0   60 months   12.69%        67.79   \n",
       "\n",
       "  grade sub_grade emp_length home_ownership       ...         out_prncp_inv  \\\n",
       "0     B        B2  10+ years           RENT       ...                   0.0   \n",
       "1     C        C4   < 1 year           RENT       ...                   0.0   \n",
       "2     C        C5  10+ years           RENT       ...                   0.0   \n",
       "3     C        C1  10+ years           RENT       ...                   0.0   \n",
       "4     B        B5     1 year           RENT       ...                   0.0   \n",
       "\n",
       "    total_pymnt total_pymnt_inv total_rec_prncp total_rec_int  \\\n",
       "0   5863.155187         5833.84         5000.00        863.16   \n",
       "1   1014.530000         1014.53          456.46        435.17   \n",
       "2   3005.666844         3005.67         2400.00        605.67   \n",
       "3  12231.890000        12231.89        10000.00       2214.92   \n",
       "4   4066.908161         4066.91         3000.00       1066.91   \n",
       "\n",
       "   total_rec_late_fee  recoveries collection_recovery_fee  last_pymnt_d  \\\n",
       "0                0.00         0.0                    0.00        Jan-15   \n",
       "1                0.00       122.9                    1.11        Apr-13   \n",
       "2                0.00         0.0                    0.00        Jun-14   \n",
       "3               16.97         0.0                    0.00        Jan-15   \n",
       "4                0.00         0.0                    0.00        Jan-17   \n",
       "\n",
       "   last_pymnt_amnt  \n",
       "0           171.62  \n",
       "1           119.66  \n",
       "2           649.91  \n",
       "3           357.48  \n",
       "4            67.30  \n",
       "\n",
       "[5 rows x 37 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Transform features from string to numeric\n",
    "for i in [\"term\",\"int_rate\",\"emp_length\",\"revol_util\"]:\n",
    "    data.loc[:,i] = \\\n",
    "        data.loc[:,i].apply(lambda x: re.sub(\"[^0-9]\", \"\", str(x)))\n",
    "    data.loc[:,i] = pd.to_numeric(data.loc[:,i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Determine which features are numerical\n",
    "numericalFeats = [x for x in data.columns if data[x].dtype != 'object']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "loan_amnt                      7\n",
       "funded_amnt                    7\n",
       "funded_amnt_inv                7\n",
       "term                           7\n",
       "int_rate                       7\n",
       "installment                    7\n",
       "emp_length                  1119\n",
       "annual_inc                    11\n",
       "dti                            7\n",
       "delinq_2yrs                   36\n",
       "mths_since_last_delinq     26933\n",
       "mths_since_last_record     38891\n",
       "open_acc                      36\n",
       "pub_rec                       36\n",
       "revol_bal                      7\n",
       "revol_util                    97\n",
       "total_acc                     36\n",
       "out_prncp                      7\n",
       "out_prncp_inv                  7\n",
       "total_pymnt                    7\n",
       "total_pymnt_inv                7\n",
       "total_rec_prncp                7\n",
       "total_rec_int                  7\n",
       "total_rec_late_fee             7\n",
       "recoveries                     7\n",
       "collection_recovery_fee        7\n",
       "last_pymnt_amnt                7\n",
       "dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display NaNs by feature\n",
    "nanCounter = np.isnan(data.loc[:,numericalFeats]).sum()\n",
    "nanCounter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Impute NaNs with mean \n",
    "fillWithMean = ['loan_amnt','funded_amnt','funded_amnt_inv','term', \\\n",
    "                'int_rate','installment','emp_length','annual_inc',\\\n",
    "                'dti','open_acc','revol_bal','revol_util','total_acc',\\\n",
    "                'out_prncp','out_prncp_inv','total_pymnt', \\\n",
    "                'total_pymnt_inv','total_rec_prncp','total_rec_int', \\\n",
    "                'last_pymnt_amnt']\n",
    "\n",
    "# Impute NaNs with zero\n",
    "fillWithZero = ['delinq_2yrs','mths_since_last_delinq', \\\n",
    "                'mths_since_last_record','pub_rec','total_rec_late_fee', \\\n",
    "                'recoveries','collection_recovery_fee']\n",
    "\n",
    "# Perform imputation\n",
    "im = pp.Imputer(strategy='mean')   \n",
    "data.loc[:,fillWithMean] = im.fit_transform(data[fillWithMean])\n",
    "\n",
    "data.loc[:,fillWithZero] = data.loc[:,fillWithZero].fillna(value=0,axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "loan_amnt                  0\n",
       "funded_amnt                0\n",
       "funded_amnt_inv            0\n",
       "term                       0\n",
       "int_rate                   0\n",
       "installment                0\n",
       "emp_length                 0\n",
       "annual_inc                 0\n",
       "dti                        0\n",
       "delinq_2yrs                0\n",
       "mths_since_last_delinq     0\n",
       "mths_since_last_record     0\n",
       "open_acc                   0\n",
       "pub_rec                    0\n",
       "revol_bal                  0\n",
       "revol_util                 0\n",
       "total_acc                  0\n",
       "out_prncp                  0\n",
       "out_prncp_inv              0\n",
       "total_pymnt                0\n",
       "total_pymnt_inv            0\n",
       "total_rec_prncp            0\n",
       "total_rec_int              0\n",
       "total_rec_late_fee         0\n",
       "recoveries                 0\n",
       "collection_recovery_fee    0\n",
       "last_pymnt_amnt            0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check for NaNs one last time\n",
    "nanCounter = np.isnan(data.loc[:,numericalFeats]).sum()\n",
    "nanCounter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Feature engineering\n",
    "data['installmentOverLoanAmnt'] = data.installment/data.loan_amnt\n",
    "data['loanAmntOverIncome'] = data.loan_amnt/data.annual_inc\n",
    "data['revol_balOverIncome'] = data.revol_bal/data.annual_inc\n",
    "data['totalPymntOverIncome'] = data.total_pymnt/data.annual_inc\n",
    "data['totalPymntInvOverIncome'] = data.total_pymnt_inv/data.annual_inc\n",
    "data['totalRecPrncpOverIncome'] = data.total_rec_prncp/data.annual_inc\n",
    "data['totalRecIncOverIncome'] = data.total_rec_int/data.annual_inc\n",
    "\n",
    "newFeats = ['installmentOverLoanAmnt','loanAmntOverIncome', \\\n",
    "            'revol_balOverIncome','totalPymntOverIncome', \\\n",
    "           'totalPymntInvOverIncome','totalRecPrncpOverIncome', \\\n",
    "            'totalRecIncOverIncome']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Select features for training\n",
    "numericalPlusNewFeats = numericalFeats+newFeats\n",
    "X_train = data.loc[:,numericalPlusNewFeats]\n",
    "\n",
    "# Scale data\n",
    "sX = pp.StandardScaler()\n",
    "X_train.loc[:,:] = sX.fit_transform(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['loan_amnt', 'funded_amnt', 'funded_amnt_inv', 'term', 'int_rate',\n",
       "       'installment', 'emp_length', 'annual_inc', 'dti', 'delinq_2yrs',\n",
       "       'mths_since_last_delinq', 'mths_since_last_record', 'open_acc',\n",
       "       'pub_rec', 'revol_bal', 'revol_util', 'total_acc', 'out_prncp',\n",
       "       'out_prncp_inv', 'total_pymnt', 'total_pymnt_inv', 'total_rec_prncp',\n",
       "       'total_rec_int', 'total_rec_late_fee', 'recoveries',\n",
       "       'collection_recovery_fee', 'last_pymnt_amnt', 'installmentOverLoanAmnt',\n",
       "       'loanAmntOverIncome', 'revol_balOverIncome', 'totalPymntOverIncome',\n",
       "       'totalPymntInvOverIncome', 'totalRecPrncpOverIncome',\n",
       "       'totalRecIncOverIncome'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['B', 'C', 'A', 'E', 'F', 'D', 'G', nan], dtype=object)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Designate labels for evaluation\n",
    "labels = data.grade\n",
    "labels.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Fill missing labels\n",
    "labels = labels.fillna(value=\"Z\")\n",
    "\n",
    "# Convert labels to numerical values\n",
    "lbl = pp.LabelEncoder()\n",
    "lbl.fit(list(labels.values))\n",
    "labels = pd.Series(data=lbl.transform(labels.values), name=\"grade\")\n",
    "\n",
    "# Store as y_train\n",
    "y_train = labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "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>grade</th>\n",
       "      <th>grade</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>4</td>\n",
       "      <td>E</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>5</td>\n",
       "      <td>F</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>3</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>3</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42512</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42513</th>\n",
       "      <td>4</td>\n",
       "      <td>E</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42514</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42515</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42516</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42517</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42518</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42519</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42520</th>\n",
       "      <td>4</td>\n",
       "      <td>E</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42521</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42522</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42523</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42524</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42525</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42526</th>\n",
       "      <td>3</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42527</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42528</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42529</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42530</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42531</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42532</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42533</th>\n",
       "      <td>2</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42534</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42535</th>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42536</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42537</th>\n",
       "      <td>0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42538</th>\n",
       "      <td>7</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42539</th>\n",
       "      <td>7</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42540</th>\n",
       "      <td>7</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42541</th>\n",
       "      <td>7</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>42542 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      grade grade\n",
       "0         1     B\n",
       "1         2     C\n",
       "2         2     C\n",
       "3         2     C\n",
       "4         1     B\n",
       "5         0     A\n",
       "6         2     C\n",
       "7         4     E\n",
       "8         5     F\n",
       "9         1     B\n",
       "10        2     C\n",
       "11        1     B\n",
       "12        2     C\n",
       "13        1     B\n",
       "14        1     B\n",
       "15        3     D\n",
       "16        2     C\n",
       "17        0     A\n",
       "18        1     B\n",
       "19        0     A\n",
       "20        2     C\n",
       "21        1     B\n",
       "22        1     B\n",
       "23        1     B\n",
       "24        1     B\n",
       "25        1     B\n",
       "26        2     C\n",
       "27        3     D\n",
       "28        1     B\n",
       "29        1     B\n",
       "...     ...   ...\n",
       "42512     1     B\n",
       "42513     4     E\n",
       "42514     0     A\n",
       "42515     2     C\n",
       "42516     1     B\n",
       "42517     1     B\n",
       "42518     2     C\n",
       "42519     2     C\n",
       "42520     4     E\n",
       "42521     0     A\n",
       "42522     0     A\n",
       "42523     0     A\n",
       "42524     0     A\n",
       "42525     0     A\n",
       "42526     3     D\n",
       "42527     2     C\n",
       "42528     2     C\n",
       "42529     1     B\n",
       "42530     1     B\n",
       "42531     1     B\n",
       "42532     1     B\n",
       "42533     2     C\n",
       "42534     1     B\n",
       "42535     1     B\n",
       "42536     0     A\n",
       "42537     0     A\n",
       "42538     7   NaN\n",
       "42539     7   NaN\n",
       "42540     7   NaN\n",
       "42541     7   NaN\n",
       "\n",
       "[42542 rows x 2 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labelsOriginalVSNew = pd.concat([labels, data.grade],axis=1)\n",
    "labelsOriginalVSNew"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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>int_rate</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>grade</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0.0</th>\n",
       "      <td>734.270844</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.0</th>\n",
       "      <td>1101.420857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2.0</th>\n",
       "      <td>1349.988902</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3.0</th>\n",
       "      <td>1557.714927</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4.0</th>\n",
       "      <td>1737.676783</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5.0</th>\n",
       "      <td>1926.530361</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6.0</th>\n",
       "      <td>2045.125000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7.0</th>\n",
       "      <td>1216.501563</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          int_rate\n",
       "grade             \n",
       "0.0     734.270844\n",
       "1.0    1101.420857\n",
       "2.0    1349.988902\n",
       "3.0    1557.714927\n",
       "4.0    1737.676783\n",
       "5.0    1926.530361\n",
       "6.0    2045.125000\n",
       "7.0    1216.501563"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Compare loan grades with interest rates\n",
    "interestAndGrade = pd.DataFrame(data=[data.int_rate,labels])\n",
    "interestAndGrade = interestAndGrade.T\n",
    "\n",
    "interestAndGrade.groupby(\"grade\").mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Goodness of the Clusters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def analyzeCluster(clusterDF, labelsDF):\n",
    "    countByCluster = \\\n",
    "        pd.DataFrame(data=clusterDF['cluster'].value_counts())\n",
    "    countByCluster.reset_index(inplace=True,drop=False)\n",
    "    countByCluster.columns = ['cluster','clusterCount']\n",
    "        \n",
    "    preds = pd.concat([labelsDF,clusterDF], axis=1)\n",
    "    preds.columns = ['trueLabel','cluster']\n",
    "    \n",
    "    countByLabel = pd.DataFrame(data=preds.groupby('trueLabel').count())\n",
    "        \n",
    "    countMostFreq = pd.DataFrame(data=preds.groupby('cluster').agg( \\\n",
    "        lambda x:x.value_counts().iloc[0]))\n",
    "    countMostFreq.reset_index(inplace=True,drop=False)\n",
    "    countMostFreq.columns = ['cluster','countMostFrequent']\n",
    "    \n",
    "    accuracyDF = countMostFreq.merge(countByCluster, \\\n",
    "        left_on=\"cluster\",right_on=\"cluster\")\n",
    "    \n",
    "    overallAccuracy = accuracyDF.countMostFrequent.sum()/ \\\n",
    "        accuracyDF.clusterCount.sum()\n",
    "    \n",
    "    accuracyByLabel = accuracyDF.countMostFrequent/ \\\n",
    "        accuracyDF.clusterCount\n",
    "    \n",
    "    return countByCluster, countByLabel, countMostFreq, \\\n",
    "        accuracyDF, overallAccuracy, accuracyByLabel"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# K-means Application"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "\n",
    "n_clusters = 10\n",
    "n_init = 10\n",
    "max_iter = 300\n",
    "tol = 0.0001\n",
    "random_state = 2018\n",
    "n_jobs = 2\n",
    "\n",
    "kmeans = KMeans(n_clusters=n_clusters, n_init=n_init, \\\n",
    "                max_iter=max_iter, tol=tol, \\\n",
    "                random_state=random_state, n_jobs=n_jobs)\n",
    "\n",
    "kMeans_inertia = pd.DataFrame(data=[],index=range(10,31), \\\n",
    "                              columns=['inertia'])\n",
    "\n",
    "overallAccuracy_kMeansDF = pd.DataFrame(data=[], \\\n",
    "    index=range(10,31),columns=['overallAccuracy'])\n",
    "\n",
    "for n_clusters in range(10,31):\n",
    "    kmeans = KMeans(n_clusters=n_clusters, n_init=n_init, \\\n",
    "                    max_iter=max_iter, tol=tol, \\\n",
    "                    random_state=random_state, n_jobs=n_jobs)\n",
    "\n",
    "    kmeans.fit(X_train)\n",
    "    kMeans_inertia.loc[n_clusters] = kmeans.inertia_\n",
    "    X_train_kmeansClustered = kmeans.predict(X_train)\n",
    "    X_train_kmeansClustered = pd.DataFrame(data= \\\n",
    "        X_train_kmeansClustered, index=X_train.index, \\\n",
    "        columns=['cluster'])\n",
    "    \n",
    "    countByCluster_kMeans, countByLabel_kMeans, \\\n",
    "    countMostFreq_kMeans, accuracyDF_kMeans, \\\n",
    "    overallAccuracy_kMeans, accuracyByLabel_kMeans = \\\n",
    "    analyzeCluster(X_train_kmeansClustered, y_train)\n",
    "    \n",
    "    overallAccuracy_kMeansDF.loc[n_clusters] = \\\n",
    "        overallAccuracy_kMeans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2352ac07860>"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "overallAccuracy_kMeansDF.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0     0.326633\n",
       "1     0.258993\n",
       "2     0.292240\n",
       "3     0.234242\n",
       "4     0.388794\n",
       "5     0.325654\n",
       "6     0.303797\n",
       "7     0.762116\n",
       "8     0.222222\n",
       "9     0.391381\n",
       "10    0.292910\n",
       "11    0.317533\n",
       "12    0.206897\n",
       "13    0.312709\n",
       "14    0.345233\n",
       "15    0.682208\n",
       "16    0.327250\n",
       "17    0.366605\n",
       "18    0.234783\n",
       "19    0.288757\n",
       "20    0.500000\n",
       "21    0.375466\n",
       "22    0.332203\n",
       "23    0.252252\n",
       "24    0.338509\n",
       "25    0.232000\n",
       "26    0.464418\n",
       "27    0.261583\n",
       "28    0.376327\n",
       "29    0.269129\n",
       "dtype: float64"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracyByLabel_kMeans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Hierarchical Clustering Application"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import fastcluster\n",
    "from scipy.cluster.hierarchy import dendrogram\n",
    "from scipy.cluster.hierarchy import cophenet\n",
    "from scipy.spatial.distance import pdist\n",
    "\n",
    "Z = fastcluster.linkage_vector(X_train, method='ward', \\\n",
    "                               metric='euclidean')\n",
    "\n",
    "Z_dataFrame = pd.DataFrame(data=Z,columns=['clusterOne', \\\n",
    "                'clusterTwo','distance','newClusterSize'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>clusterOne</th>\n",
       "      <th>clusterTwo</th>\n",
       "      <th>distance</th>\n",
       "      <th>newClusterSize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>39786.0</td>\n",
       "      <td>39787.0</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>39788.0</td>\n",
       "      <td>42542.0</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>42538.0</td>\n",
       "      <td>42539.0</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>42540.0</td>\n",
       "      <td>42544.0</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>42541.0</td>\n",
       "      <td>42545.0</td>\n",
       "      <td>3.399350e-17</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>42543.0</td>\n",
       "      <td>42546.0</td>\n",
       "      <td>5.139334e-17</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>33251.0</td>\n",
       "      <td>33261.0</td>\n",
       "      <td>1.561313e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>42512.0</td>\n",
       "      <td>42535.0</td>\n",
       "      <td>3.342654e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>42219.0</td>\n",
       "      <td>42316.0</td>\n",
       "      <td>3.368231e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>6112.0</td>\n",
       "      <td>21928.0</td>\n",
       "      <td>3.384368e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>33248.0</td>\n",
       "      <td>33275.0</td>\n",
       "      <td>3.583819e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>33253.0</td>\n",
       "      <td>33265.0</td>\n",
       "      <td>3.595331e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>33258.0</td>\n",
       "      <td>42552.0</td>\n",
       "      <td>3.719377e-01</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>20430.0</td>\n",
       "      <td>23299.0</td>\n",
       "      <td>3.757307e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>5455.0</td>\n",
       "      <td>32845.0</td>\n",
       "      <td>3.828709e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>28615.0</td>\n",
       "      <td>30306.0</td>\n",
       "      <td>3.900294e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>9056.0</td>\n",
       "      <td>9769.0</td>\n",
       "      <td>3.967378e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>11162.0</td>\n",
       "      <td>13857.0</td>\n",
       "      <td>3.991124e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>33270.0</td>\n",
       "      <td>42548.0</td>\n",
       "      <td>3.995620e-01</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>17422.0</td>\n",
       "      <td>17986.0</td>\n",
       "      <td>4.061704e-01</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    clusterOne  clusterTwo      distance  newClusterSize\n",
       "0      39786.0     39787.0  0.000000e+00             2.0\n",
       "1      39788.0     42542.0  0.000000e+00             3.0\n",
       "2      42538.0     42539.0  0.000000e+00             2.0\n",
       "3      42540.0     42544.0  0.000000e+00             3.0\n",
       "4      42541.0     42545.0  3.399350e-17             4.0\n",
       "5      42543.0     42546.0  5.139334e-17             7.0\n",
       "6      33251.0     33261.0  1.561313e-01             2.0\n",
       "7      42512.0     42535.0  3.342654e-01             2.0\n",
       "8      42219.0     42316.0  3.368231e-01             2.0\n",
       "9       6112.0     21928.0  3.384368e-01             2.0\n",
       "10     33248.0     33275.0  3.583819e-01             2.0\n",
       "11     33253.0     33265.0  3.595331e-01             2.0\n",
       "12     33258.0     42552.0  3.719377e-01             3.0\n",
       "13     20430.0     23299.0  3.757307e-01             2.0\n",
       "14      5455.0     32845.0  3.828709e-01             2.0\n",
       "15     28615.0     30306.0  3.900294e-01             2.0\n",
       "16      9056.0      9769.0  3.967378e-01             2.0\n",
       "17     11162.0     13857.0  3.991124e-01             2.0\n",
       "18     33270.0     42548.0  3.995620e-01             3.0\n",
       "19     17422.0     17986.0  4.061704e-01             2.0"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Z_dataFrame[:20]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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>clusterOne</th>\n",
       "      <th>clusterTwo</th>\n",
       "      <th>distance</th>\n",
       "      <th>newClusterSize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>42521</th>\n",
       "      <td>85038.0</td>\n",
       "      <td>85043.0</td>\n",
       "      <td>132.715723</td>\n",
       "      <td>3969.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42522</th>\n",
       "      <td>85051.0</td>\n",
       "      <td>85052.0</td>\n",
       "      <td>141.386569</td>\n",
       "      <td>2899.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42523</th>\n",
       "      <td>85026.0</td>\n",
       "      <td>85027.0</td>\n",
       "      <td>146.976703</td>\n",
       "      <td>2351.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42524</th>\n",
       "      <td>85048.0</td>\n",
       "      <td>85049.0</td>\n",
       "      <td>152.660192</td>\n",
       "      <td>5691.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42525</th>\n",
       "      <td>85036.0</td>\n",
       "      <td>85059.0</td>\n",
       "      <td>153.512281</td>\n",
       "      <td>5956.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42526</th>\n",
       "      <td>85033.0</td>\n",
       "      <td>85044.0</td>\n",
       "      <td>160.825959</td>\n",
       "      <td>2203.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42527</th>\n",
       "      <td>85055.0</td>\n",
       "      <td>85061.0</td>\n",
       "      <td>163.701428</td>\n",
       "      <td>668.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42528</th>\n",
       "      <td>85062.0</td>\n",
       "      <td>85066.0</td>\n",
       "      <td>168.199295</td>\n",
       "      <td>6897.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42529</th>\n",
       "      <td>85054.0</td>\n",
       "      <td>85060.0</td>\n",
       "      <td>168.924039</td>\n",
       "      <td>9414.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42530</th>\n",
       "      <td>85028.0</td>\n",
       "      <td>85064.0</td>\n",
       "      <td>185.215769</td>\n",
       "      <td>3118.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42531</th>\n",
       "      <td>85067.0</td>\n",
       "      <td>85071.0</td>\n",
       "      <td>187.832588</td>\n",
       "      <td>15370.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42532</th>\n",
       "      <td>85056.0</td>\n",
       "      <td>85073.0</td>\n",
       "      <td>203.212147</td>\n",
       "      <td>17995.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42533</th>\n",
       "      <td>85057.0</td>\n",
       "      <td>85063.0</td>\n",
       "      <td>205.285993</td>\n",
       "      <td>9221.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42534</th>\n",
       "      <td>85068.0</td>\n",
       "      <td>85072.0</td>\n",
       "      <td>207.902660</td>\n",
       "      <td>5321.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42535</th>\n",
       "      <td>85069.0</td>\n",
       "      <td>85075.0</td>\n",
       "      <td>236.754581</td>\n",
       "      <td>9889.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42536</th>\n",
       "      <td>85070.0</td>\n",
       "      <td>85077.0</td>\n",
       "      <td>298.587755</td>\n",
       "      <td>16786.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42537</th>\n",
       "      <td>85058.0</td>\n",
       "      <td>85078.0</td>\n",
       "      <td>309.946867</td>\n",
       "      <td>16875.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42538</th>\n",
       "      <td>85074.0</td>\n",
       "      <td>85079.0</td>\n",
       "      <td>375.698458</td>\n",
       "      <td>34870.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42539</th>\n",
       "      <td>85065.0</td>\n",
       "      <td>85080.0</td>\n",
       "      <td>400.711547</td>\n",
       "      <td>37221.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42540</th>\n",
       "      <td>85076.0</td>\n",
       "      <td>85081.0</td>\n",
       "      <td>644.047472</td>\n",
       "      <td>42542.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       clusterOne  clusterTwo    distance  newClusterSize\n",
       "42521     85038.0     85043.0  132.715723          3969.0\n",
       "42522     85051.0     85052.0  141.386569          2899.0\n",
       "42523     85026.0     85027.0  146.976703          2351.0\n",
       "42524     85048.0     85049.0  152.660192          5691.0\n",
       "42525     85036.0     85059.0  153.512281          5956.0\n",
       "42526     85033.0     85044.0  160.825959          2203.0\n",
       "42527     85055.0     85061.0  163.701428           668.0\n",
       "42528     85062.0     85066.0  168.199295          6897.0\n",
       "42529     85054.0     85060.0  168.924039          9414.0\n",
       "42530     85028.0     85064.0  185.215769          3118.0\n",
       "42531     85067.0     85071.0  187.832588         15370.0\n",
       "42532     85056.0     85073.0  203.212147         17995.0\n",
       "42533     85057.0     85063.0  205.285993          9221.0\n",
       "42534     85068.0     85072.0  207.902660          5321.0\n",
       "42535     85069.0     85075.0  236.754581          9889.0\n",
       "42536     85070.0     85077.0  298.587755         16786.0\n",
       "42537     85058.0     85078.0  309.946867         16875.0\n",
       "42538     85074.0     85079.0  375.698458         34870.0\n",
       "42539     85065.0     85080.0  400.711547         37221.0\n",
       "42540     85076.0     85081.0  644.047472         42542.0"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Z_dataFrame[42521:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from scipy.cluster.hierarchy import fcluster\n",
    "\n",
    "distance_threshold = 100\n",
    "clusters = fcluster(Z, distance_threshold, criterion='distance')\n",
    "X_train_hierClustered = pd.DataFrame(data=clusters, \\\n",
    "    index=X_train.index,columns=['cluster'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of distinct clusters:  32\n"
     ]
    }
   ],
   "source": [
    "print(\"Number of distinct clusters: \", \\\n",
    "      len(X_train_hierClustered['cluster'].unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Overall accuracy from hierarchical clustering:  0.3651685393258427\n"
     ]
    }
   ],
   "source": [
    "countByCluster_hierClust, countByLabel_hierClust, \\\n",
    "    countMostFreq_hierClust, accuracyDF_hierClust, \\\n",
    "    overallAccuracy_hierClust, accuracyByLabel_hierClust = \\\n",
    "    analyzeCluster(X_train_hierClustered, y_train)\n",
    "\n",
    "print(\"Overall accuracy from hierarchical clustering: \", \\\n",
    "      overallAccuracy_hierClust)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy by cluster for hierarchical clustering\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0     0.304124\n",
       "1     0.219001\n",
       "2     0.228311\n",
       "3     0.379722\n",
       "4     0.240064\n",
       "5     0.272011\n",
       "6     0.314560\n",
       "7     0.263930\n",
       "8     0.246138\n",
       "9     0.318942\n",
       "10    0.302752\n",
       "11    0.269772\n",
       "12    0.335717\n",
       "13    0.330403\n",
       "14    0.346320\n",
       "15    0.440141\n",
       "16    0.744155\n",
       "17    0.502227\n",
       "18    0.294118\n",
       "19    0.236111\n",
       "20    0.254727\n",
       "21    0.241042\n",
       "22    0.317979\n",
       "23    0.308771\n",
       "24    0.284314\n",
       "25    0.243243\n",
       "26    0.500000\n",
       "27    0.289157\n",
       "28    0.365283\n",
       "29    0.479693\n",
       "30    0.393559\n",
       "31    0.340875\n",
       "dtype: float64"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Accuracy by cluster for hierarchical clustering\")\n",
    "accuracyByLabel_hierClust"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# HDBSCAN Application"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import hdbscan\n",
    "\n",
    "min_cluster_size = 20\n",
    "min_samples = 20\n",
    "alpha = 1.0\n",
    "cluster_selection_method = 'leaf'\n",
    "\n",
    "hdb = hdbscan.HDBSCAN(min_cluster_size=min_cluster_size, \\\n",
    "    min_samples=min_samples, alpha=alpha, \\\n",
    "    cluster_selection_method=cluster_selection_method)\n",
    "\n",
    "X_train_hdbscanClustered = hdb.fit_predict(X_train)\n",
    "X_train_hdbscanClustered = pd.DataFrame(data= \\\n",
    "    X_train_hdbscanClustered, index=X_train.index, \\\n",
    "    columns=['cluster'])\n",
    "\n",
    "countByCluster_hdbscan, countByLabel_hdbscan, \\\n",
    "    countMostFreq_hdbscan, accuracyDF_hdbscan, \\\n",
    "    overallAccuracy_hdbscan, accuracyByLabel_hdbscan = \\\n",
    "    analyzeCluster(X_train_hdbscanClustered, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Overall accuracy from HDBSCAN:  0.3246203751586667\n"
     ]
    }
   ],
   "source": [
    "print(\"Overall accuracy from HDBSCAN: \", overallAccuracy_hdbscan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster results for HDBSCAN\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>cluster</th>\n",
       "      <th>clusterCount</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-1</td>\n",
       "      <td>32708</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7</td>\n",
       "      <td>4070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>3668</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>1096</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>773</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>120</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>3</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>5</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   cluster  clusterCount\n",
       "0       -1         32708\n",
       "1        7          4070\n",
       "2        2          3668\n",
       "3        1          1096\n",
       "4        4           773\n",
       "5        0           120\n",
       "6        6            49\n",
       "7        3            38\n",
       "8        5            20"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Cluster results for HDBSCAN\")\n",
    "countByCluster_hdbscan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    0.284487\n",
       "1    0.341667\n",
       "2    0.414234\n",
       "3    0.332061\n",
       "4    0.552632\n",
       "5    0.438551\n",
       "6    0.400000\n",
       "7    0.408163\n",
       "8    0.590663\n",
       "dtype: float64"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracyByLabel_hdbscan"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}