05_clustering.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\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": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Load the datasets\n",
    "current_path = os.getcwd()\n",
    "file = os.path.sep.join(['', 'datasets', 'mnist_data', 'mnist.pkl.gz'])\n",
    "\n",
    "f = gzip.open(current_path+file, 'rb')\n",
    "train_set, validation_set, test_set = pickle.load(f, encoding='latin1')\n",
    "f.close()\n",
    "\n",
    "X_train, y_train = train_set[0], train_set[1]\n",
    "X_validation, y_validation = validation_set[0], validation_set[1]\n",
    "X_test, y_test = test_set[0], test_set[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Create Pandas DataFrames from the datasets\n",
    "train_index = range(0,len(X_train))\n",
    "validation_index = range(len(X_train), \\\n",
    "                         len(X_train)+len(X_validation))\n",
    "test_index = range(len(X_train)+len(X_validation), \\\n",
    "                   len(X_train)+len(X_validation)+len(X_test))\n",
    "\n",
    "X_train = pd.DataFrame(data=X_train,index=train_index)\n",
    "y_train = pd.Series(data=y_train,index=train_index)\n",
    "\n",
    "X_validation = pd.DataFrame(data=X_validation,index=validation_index)\n",
    "y_validation = pd.Series(data=y_validation,index=validation_index)\n",
    "\n",
    "X_test = pd.DataFrame(data=X_test,index=test_index)\n",
    "y_test = pd.Series(data=y_test,index=test_index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Principal Component Analysis\n",
    "from sklearn.decomposition import PCA\n",
    "\n",
    "n_components = 784\n",
    "whiten = False\n",
    "random_state = 2018\n",
    "\n",
    "pca = PCA(n_components=n_components, whiten=whiten, \\\n",
    "          random_state=random_state)\n",
    "\n",
    "X_train_PCA = pca.fit_transform(X_train)\n",
    "X_train_PCA = pd.DataFrame(data=X_train_PCA, index=train_index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# K-means - Inertia as the number of clusters varies\n",
    "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_inertia = pd.DataFrame(data=[],index=range(2,21), \\\n",
    "                              columns=['inertia'])\n",
    "for n_clusters in range(2,21):\n",
    "    kmeans = KMeans(n_clusters=n_clusters, n_init=n_init, \\\n",
    "                max_iter=max_iter, tol=tol, random_state=random_state, \\\n",
    "                n_jobs=n_jobs)\n",
    "\n",
    "    cutoff = 99\n",
    "    kmeans.fit(X_train_PCA.loc[:,0:cutoff])\n",
    "    kMeans_inertia.loc[n_clusters] = kmeans.inertia_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2650427fdd8>"
      ]
     },
     "execution_count": 19,
     "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": [
    "kMeans_inertia.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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 = \\\n",
    "        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",
    "    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": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# K-means - Accuracy as the number of clusters varies\n",
    "\n",
    "n_clusters = 5\n",
    "n_init = 10\n",
    "max_iter = 300\n",
    "tol = 0.0001\n",
    "random_state = 2018\n",
    "n_jobs = 2\n",
    "\n",
    "kMeans_inertia = \\\n",
    "    pd.DataFrame(data=[],index=range(2,21),columns=['inertia'])\n",
    "overallAccuracy_kMeansDF = \\\n",
    "    pd.DataFrame(data=[],index=range(2,21),columns=['overallAccuracy'])\n",
    "\n",
    "for n_clusters in range(2,21):\n",
    "    kmeans = KMeans(n_clusters=n_clusters, n_init=n_init, \\\n",
    "                max_iter=max_iter, tol=tol, random_state=random_state, \\\n",
    "                n_jobs=n_jobs)\n",
    "\n",
    "    cutoff = 99\n",
    "    kmeans.fit(X_train_PCA.loc[:,0:cutoff])\n",
    "    kMeans_inertia.loc[n_clusters] = kmeans.inertia_\n",
    "    X_train_kmeansClustered = kmeans.predict(X_train_PCA.loc[:,0:cutoff])\n",
    "    X_train_kmeansClustered = \\\n",
    "        pd.DataFrame(data=X_train_kmeansClustered, index=X_train.index, \\\n",
    "                     columns=['cluster'])\n",
    "    \n",
    "    countByCluster_kMeans, countByLabel_kMeans, countMostFreq_kMeans, \\\n",
    "        accuracyDF_kMeans, overallAccuracy_kMeans, accuracyByLabel_kMeans \\\n",
    "        = analyzeCluster(X_train_kmeansClustered, y_train)\n",
    "    \n",
    "    overallAccuracy_kMeansDF.loc[n_clusters] = overallAccuracy_kMeans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x265043c9cf8>"
      ]
     },
     "execution_count": 26,
     "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": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0     0.636506\n",
       "1     0.928505\n",
       "2     0.848714\n",
       "3     0.521805\n",
       "4     0.714337\n",
       "5     0.950980\n",
       "6     0.893103\n",
       "7     0.919040\n",
       "8     0.404707\n",
       "9     0.500522\n",
       "10    0.381526\n",
       "11    0.587680\n",
       "12    0.463382\n",
       "13    0.958046\n",
       "14    0.870888\n",
       "15    0.942325\n",
       "16    0.791192\n",
       "17    0.843972\n",
       "18    0.455679\n",
       "19    0.926480\n",
       "dtype: float64"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracyByLabel_kMeans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# K-means - Accuracy as the number of components varies\n",
    "\n",
    "n_clusters = 20\n",
    "n_init = 10\n",
    "max_iter = 300\n",
    "tol = 0.0001\n",
    "random_state = 2018\n",
    "n_jobs = 2\n",
    "\n",
    "kMeans_inertia = pd.DataFrame(data=[],index=[9, 49, 99, 199, \\\n",
    "                    299, 399, 499, 599, 699, 783],columns=['inertia'])\n",
    "\n",
    "overallAccuracy_kMeansDF = pd.DataFrame(data=[],index=[9, 49, \\\n",
    "                    99, 199, 299, 399, 499, 599, 699, 783], \\\n",
    "                    columns=['overallAccuracy'])\n",
    "\n",
    "for cutoffNumber in [9, 49, 99, 199, 299, 399, 499, 599, 699, 783]:\n",
    "    kmeans = KMeans(n_clusters=n_clusters, n_init=n_init, \\\n",
    "                max_iter=max_iter, tol=tol, random_state=random_state, \\\n",
    "                n_jobs=n_jobs)\n",
    "\n",
    "    cutoff = cutoffNumber\n",
    "    kmeans.fit(X_train_PCA.loc[:,0:cutoff])\n",
    "    kMeans_inertia.loc[cutoff] = kmeans.inertia_\n",
    "    X_train_kmeansClustered = kmeans.predict(X_train_PCA.loc[:,0:cutoff])\n",
    "    X_train_kmeansClustered = pd.DataFrame(data=X_train_kmeansClustered, \\\n",
    "                                index=X_train.index, columns=['cluster'])\n",
    "    \n",
    "    countByCluster_kMeans, countByLabel_kMeans, countMostFreq_kMeans, \\\n",
    "        accuracyDF_kMeans, overallAccuracy_kMeans, accuracyByLabel_kMeans \\\n",
    "        = analyzeCluster(X_train_kmeansClustered, y_train)\n",
    "    \n",
    "    overallAccuracy_kMeansDF.loc[cutoff] = overallAccuracy_kMeans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x26504433d30>"
      ]
     },
     "execution_count": 28,
     "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": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# K-means - Accuracy as the number of components varies\n",
    "# On the original MNIST data (not PCA-reduced)\n",
    "\n",
    "n_clusters = 20\n",
    "n_init = 10\n",
    "max_iter = 300\n",
    "tol = 0.0001\n",
    "random_state = 2018\n",
    "n_jobs = 2\n",
    "\n",
    "kMeans_inertia = pd.DataFrame(data=[],index=[9, 49, 99, 199, \\\n",
    "                    299, 399, 499, 599, 699, 783],columns=['inertia'])\n",
    "\n",
    "overallAccuracy_kMeansDF = pd.DataFrame(data=[],index=[9, 49, \\\n",
    "                    99, 199, 299, 399, 499, 599, 699, 783], \\\n",
    "                    columns=['overallAccuracy'])\n",
    "\n",
    "for cutoffNumber in [9, 49, 99, 199, 299, 399, 499, 599, 699, 783]:\n",
    "    kmeans = KMeans(n_clusters=n_clusters, n_init=n_init, \\\n",
    "                max_iter=max_iter, tol=tol, random_state=random_state, \\\n",
    "                n_jobs=n_jobs)\n",
    "\n",
    "    cutoff = cutoffNumber\n",
    "    kmeans.fit(X_train.loc[:,0:cutoff])\n",
    "    kMeans_inertia.loc[cutoff] = kmeans.inertia_\n",
    "    X_train_kmeansClustered = kmeans.predict(X_train.loc[:,0:cutoff])\n",
    "    X_train_kmeansClustered = pd.DataFrame(data=X_train_kmeansClustered, \\\n",
    "                                index=X_train.index, columns=['cluster'])\n",
    "    \n",
    "    countByCluster_kMeans, countByLabel_kMeans, countMostFreq_kMeans, \\\n",
    "        accuracyDF_kMeans, overallAccuracy_kMeans, accuracyByLabel_kMeans \\\n",
    "        = analyzeCluster(X_train_kmeansClustered, y_train)\n",
    "    \n",
    "    overallAccuracy_kMeansDF.loc[cutoff] = overallAccuracy_kMeans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x26504483048>"
      ]
     },
     "execution_count": 31,
     "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": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Hierarchical clustering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import fastcluster\n",
    "from scipy.cluster.hierarchy import dendrogram, cophenet\n",
    "from scipy.spatial.distance import pdist\n",
    "\n",
    "cutoff = 99\n",
    "Z = fastcluster.linkage_vector(X_train_PCA.loc[:,0:cutoff], \\\n",
    "                               method='ward', metric='euclidean')\n",
    "Z_dataFrame = pd.DataFrame(data=Z, \\\n",
    "    columns=['clusterOne','clusterTwo','distance','newClusterSize'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>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>42194.0</td>\n",
       "      <td>43025.0</td>\n",
       "      <td>0.562682</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>28350.0</td>\n",
       "      <td>37674.0</td>\n",
       "      <td>0.590866</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>26696.0</td>\n",
       "      <td>44705.0</td>\n",
       "      <td>0.621506</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>12634.0</td>\n",
       "      <td>32823.0</td>\n",
       "      <td>0.627762</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>24707.0</td>\n",
       "      <td>43151.0</td>\n",
       "      <td>0.637668</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>20465.0</td>\n",
       "      <td>24483.0</td>\n",
       "      <td>0.662557</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>466.0</td>\n",
       "      <td>42098.0</td>\n",
       "      <td>0.664189</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>46542.0</td>\n",
       "      <td>49961.0</td>\n",
       "      <td>0.665520</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2301.0</td>\n",
       "      <td>5732.0</td>\n",
       "      <td>0.671215</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>37564.0</td>\n",
       "      <td>47668.0</td>\n",
       "      <td>0.675121</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>3375.0</td>\n",
       "      <td>26243.0</td>\n",
       "      <td>0.685797</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>15722.0</td>\n",
       "      <td>30368.0</td>\n",
       "      <td>0.686356</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>21247.0</td>\n",
       "      <td>21575.0</td>\n",
       "      <td>0.694412</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>14900.0</td>\n",
       "      <td>42486.0</td>\n",
       "      <td>0.696769</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>30100.0</td>\n",
       "      <td>41908.0</td>\n",
       "      <td>0.699261</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>12040.0</td>\n",
       "      <td>13254.0</td>\n",
       "      <td>0.701134</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>10508.0</td>\n",
       "      <td>25434.0</td>\n",
       "      <td>0.708872</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>30695.0</td>\n",
       "      <td>30757.0</td>\n",
       "      <td>0.710023</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>31019.0</td>\n",
       "      <td>31033.0</td>\n",
       "      <td>0.712052</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>36264.0</td>\n",
       "      <td>37285.0</td>\n",
       "      <td>0.713130</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    clusterOne  clusterTwo  distance  newClusterSize\n",
       "0      42194.0     43025.0  0.562682             2.0\n",
       "1      28350.0     37674.0  0.590866             2.0\n",
       "2      26696.0     44705.0  0.621506             2.0\n",
       "3      12634.0     32823.0  0.627762             2.0\n",
       "4      24707.0     43151.0  0.637668             2.0\n",
       "5      20465.0     24483.0  0.662557             2.0\n",
       "6        466.0     42098.0  0.664189             2.0\n",
       "7      46542.0     49961.0  0.665520             2.0\n",
       "8       2301.0      5732.0  0.671215             2.0\n",
       "9      37564.0     47668.0  0.675121             2.0\n",
       "10      3375.0     26243.0  0.685797             2.0\n",
       "11     15722.0     30368.0  0.686356             2.0\n",
       "12     21247.0     21575.0  0.694412             2.0\n",
       "13     14900.0     42486.0  0.696769             2.0\n",
       "14     30100.0     41908.0  0.699261             2.0\n",
       "15     12040.0     13254.0  0.701134             2.0\n",
       "16     10508.0     25434.0  0.708872             2.0\n",
       "17     30695.0     30757.0  0.710023             2.0\n",
       "18     31019.0     31033.0  0.712052             2.0\n",
       "19     36264.0     37285.0  0.713130             2.0"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Z_dataFrame.iloc[:20]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>clusterOne</th>\n",
       "      <th>clusterTwo</th>\n",
       "      <th>distance</th>\n",
       "      <th>newClusterSize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>49980</th>\n",
       "      <td>99965.0</td>\n",
       "      <td>99972.0</td>\n",
       "      <td>161.106998</td>\n",
       "      <td>5197.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49981</th>\n",
       "      <td>99932.0</td>\n",
       "      <td>99980.0</td>\n",
       "      <td>172.070003</td>\n",
       "      <td>6505.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49982</th>\n",
       "      <td>99945.0</td>\n",
       "      <td>99960.0</td>\n",
       "      <td>182.840860</td>\n",
       "      <td>3245.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49983</th>\n",
       "      <td>99964.0</td>\n",
       "      <td>99976.0</td>\n",
       "      <td>184.475761</td>\n",
       "      <td>3683.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49984</th>\n",
       "      <td>99974.0</td>\n",
       "      <td>99979.0</td>\n",
       "      <td>185.027847</td>\n",
       "      <td>7744.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49985</th>\n",
       "      <td>99940.0</td>\n",
       "      <td>99975.0</td>\n",
       "      <td>185.345207</td>\n",
       "      <td>5596.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49986</th>\n",
       "      <td>99957.0</td>\n",
       "      <td>99967.0</td>\n",
       "      <td>211.854714</td>\n",
       "      <td>5957.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49987</th>\n",
       "      <td>99938.0</td>\n",
       "      <td>99983.0</td>\n",
       "      <td>215.494857</td>\n",
       "      <td>4846.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49988</th>\n",
       "      <td>99978.0</td>\n",
       "      <td>99984.0</td>\n",
       "      <td>216.760365</td>\n",
       "      <td>11072.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49989</th>\n",
       "      <td>99970.0</td>\n",
       "      <td>99973.0</td>\n",
       "      <td>217.355871</td>\n",
       "      <td>4899.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49990</th>\n",
       "      <td>99969.0</td>\n",
       "      <td>99986.0</td>\n",
       "      <td>225.468298</td>\n",
       "      <td>8270.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49991</th>\n",
       "      <td>99981.0</td>\n",
       "      <td>99982.0</td>\n",
       "      <td>238.845135</td>\n",
       "      <td>9750.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49992</th>\n",
       "      <td>99968.0</td>\n",
       "      <td>99977.0</td>\n",
       "      <td>266.146782</td>\n",
       "      <td>5567.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49993</th>\n",
       "      <td>99985.0</td>\n",
       "      <td>99989.0</td>\n",
       "      <td>270.929453</td>\n",
       "      <td>10495.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49994</th>\n",
       "      <td>99990.0</td>\n",
       "      <td>99991.0</td>\n",
       "      <td>346.840948</td>\n",
       "      <td>18020.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49995</th>\n",
       "      <td>99988.0</td>\n",
       "      <td>99993.0</td>\n",
       "      <td>394.365194</td>\n",
       "      <td>21567.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49996</th>\n",
       "      <td>99987.0</td>\n",
       "      <td>99995.0</td>\n",
       "      <td>425.142387</td>\n",
       "      <td>26413.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49997</th>\n",
       "      <td>99992.0</td>\n",
       "      <td>99994.0</td>\n",
       "      <td>440.148301</td>\n",
       "      <td>23587.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49998</th>\n",
       "      <td>99996.0</td>\n",
       "      <td>99997.0</td>\n",
       "      <td>494.383866</td>\n",
       "      <td>50000.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       clusterOne  clusterTwo    distance  newClusterSize\n",
       "49980     99965.0     99972.0  161.106998          5197.0\n",
       "49981     99932.0     99980.0  172.070003          6505.0\n",
       "49982     99945.0     99960.0  182.840860          3245.0\n",
       "49983     99964.0     99976.0  184.475761          3683.0\n",
       "49984     99974.0     99979.0  185.027847          7744.0\n",
       "49985     99940.0     99975.0  185.345207          5596.0\n",
       "49986     99957.0     99967.0  211.854714          5957.0\n",
       "49987     99938.0     99983.0  215.494857          4846.0\n",
       "49988     99978.0     99984.0  216.760365         11072.0\n",
       "49989     99970.0     99973.0  217.355871          4899.0\n",
       "49990     99969.0     99986.0  225.468298          8270.0\n",
       "49991     99981.0     99982.0  238.845135          9750.0\n",
       "49992     99968.0     99977.0  266.146782          5567.0\n",
       "49993     99985.0     99989.0  270.929453         10495.0\n",
       "49994     99990.0     99991.0  346.840948         18020.0\n",
       "49995     99988.0     99993.0  394.365194         21567.0\n",
       "49996     99987.0     99995.0  425.142387         26413.0\n",
       "49997     99992.0     99994.0  440.148301         23587.0\n",
       "49998     99996.0     99997.0  494.383866         50000.0"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Z_dataFrame.iloc[49980:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from scipy.cluster.hierarchy import fcluster\n",
    "\n",
    "distance_threshold = 160\n",
    "clusters = fcluster(Z, distance_threshold, criterion='distance')\n",
    "X_train_hierClustered = \\\n",
    "    pd.DataFrame(data=clusters,index=X_train_PCA.index,columns=['cluster'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of distinct clusters:  20\n"
     ]
    }
   ],
   "source": [
    "print(\"Number of distinct clusters: \", \\\n",
    "      len(X_train_hierClustered['cluster'].unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Overall accuracy from hierarchical clustering:  0.76882\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": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy by cluster for hierarchical clustering\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0     0.987962\n",
       "1     0.983727\n",
       "2     0.988998\n",
       "3     0.597356\n",
       "4     0.678642\n",
       "5     0.442478\n",
       "6     0.950033\n",
       "7     0.829060\n",
       "8     0.976062\n",
       "9     0.986141\n",
       "10    0.990183\n",
       "11    0.992183\n",
       "12    0.971033\n",
       "13    0.554273\n",
       "14    0.553617\n",
       "15    0.720183\n",
       "16    0.538891\n",
       "17    0.484590\n",
       "18    0.957732\n",
       "19    0.977310\n",
       "dtype: float64"
      ]
     },
     "execution_count": 15,
     "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": [
    "# DBSCAN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.242"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.cluster import DBSCAN\n",
    "\n",
    "eps = 3\n",
    "min_samples = 5\n",
    "leaf_size = 30\n",
    "n_jobs = 4\n",
    "\n",
    "db = DBSCAN(eps=eps, min_samples=min_samples, leaf_size=leaf_size, \n",
    "            n_jobs=n_jobs)\n",
    "\n",
    "cutoff = 99\n",
    "X_train_PCA_dbscanClustered = db.fit_predict(X_train_PCA.loc[:,0:cutoff])\n",
    "X_train_PCA_dbscanClustered = \\\n",
    "    pd.DataFrame(data=X_train_PCA_dbscanClustered, index=X_train.index, \\\n",
    "                 columns=['cluster'])\n",
    "\n",
    "countByCluster_dbscan, countByLabel_dbscan, countMostFreq_dbscan, \\\n",
    "    accuracyDF_dbscan, overallAccuracy_dbscan, accuracyByLabel_dbscan \\\n",
    "    = analyzeCluster(X_train_PCA_dbscanClustered, y_train)\n",
    "\n",
    "overallAccuracy_dbscan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Overall accuracy from DBSCAN:  0.242\n"
     ]
    }
   ],
   "source": [
    "print(\"Overall accuracy from DBSCAN: \",overallAccuracy_dbscan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster results for DBSCAN\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>39575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>8885</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>8</td>\n",
       "      <td>720</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5</td>\n",
       "      <td>92</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>18</td>\n",
       "      <td>51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>38</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>41</td>\n",
       "      <td>22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>39</td>\n",
       "      <td>22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>4</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>20</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>10</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>2</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>33</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>25</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>66</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>12</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>28</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>36</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>40</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>23</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>37</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>61</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>51</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>56</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>24</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>63</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>30</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>35</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>74</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>58</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>87</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72</th>\n",
       "      <td>22</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73</th>\n",
       "      <td>21</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>74</th>\n",
       "      <td>52</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>47</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>76</th>\n",
       "      <td>80</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>48</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>79</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>77</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>14</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81</th>\n",
       "      <td>78</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>82</th>\n",
       "      <td>46</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>83</th>\n",
       "      <td>82</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>84</th>\n",
       "      <td>72</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>75</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>45</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87</th>\n",
       "      <td>92</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>60</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>57</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>89</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91</th>\n",
       "      <td>50</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>55</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>93</th>\n",
       "      <td>86</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94</th>\n",
       "      <td>85</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>43</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>83</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>11</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98</th>\n",
       "      <td>54</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>96</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>100 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    cluster  clusterCount\n",
       "0        -1         39575\n",
       "1         0          8885\n",
       "2         8           720\n",
       "3         5            92\n",
       "4        18            51\n",
       "5        38            38\n",
       "6        41            22\n",
       "7        39            22\n",
       "8         4            16\n",
       "9        20            16\n",
       "10       10            16\n",
       "11        2            15\n",
       "12       33            13\n",
       "13       25            12\n",
       "14       66            12\n",
       "15       12            12\n",
       "16       28            11\n",
       "17       36            11\n",
       "18       40            11\n",
       "19        9            10\n",
       "20       23            10\n",
       "21       37             8\n",
       "22       61             8\n",
       "23       51             8\n",
       "24       56             8\n",
       "25       24             8\n",
       "26       63             8\n",
       "27       30             8\n",
       "28       35             8\n",
       "29       74             7\n",
       "..      ...           ...\n",
       "70       58             5\n",
       "71       87             5\n",
       "72       22             5\n",
       "73       21             5\n",
       "74       52             5\n",
       "75       47             5\n",
       "76       80             5\n",
       "77       48             5\n",
       "78       79             5\n",
       "79       77             5\n",
       "80       14             5\n",
       "81       78             5\n",
       "82       46             5\n",
       "83       82             4\n",
       "84       72             4\n",
       "85       75             4\n",
       "86       45             4\n",
       "87       92             4\n",
       "88       60             4\n",
       "89       57             4\n",
       "90       89             4\n",
       "91       50             4\n",
       "92       55             4\n",
       "93       86             4\n",
       "94       85             4\n",
       "95       43             4\n",
       "96       83             4\n",
       "97       11             4\n",
       "98       54             3\n",
       "99       96             3\n",
       "\n",
       "[100 rows x 2 columns]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Cluster results for DBSCAN\")\n",
    "countByCluster_dbscan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# HDBSCAN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import hdbscan\n",
    "\n",
    "min_cluster_size = 30\n",
    "min_samples = None\n",
    "alpha = 1.0\n",
    "cluster_selection_method = 'eom'\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",
    "cutoff = 10\n",
    "X_train_PCA_hdbscanClustered = \\\n",
    "    hdb.fit_predict(X_train_PCA.loc[:,0:cutoff])\n",
    "\n",
    "X_train_PCA_hdbscanClustered = \\\n",
    "    pd.DataFrame(data=X_train_PCA_hdbscanClustered, \\\n",
    "    index=X_train.index, columns=['cluster'])\n",
    "\n",
    "countByCluster_hdbscan, countByLabel_hdbscan, \\\n",
    "    countMostFreq_hdbscan, accuracyDF_hdbscan, \\\n",
    "    overallAccuracy_hdbscan, accuracyByLabel_hdbscan \\\n",
    "    = analyzeCluster(X_train_PCA_hdbscanClustered, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Overall accuracy from HDBSCAN:  0.24696\n"
     ]
    }
   ],
   "source": [
    "print(\"Overall accuracy from HDBSCAN: \",overallAccuracy_hdbscan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "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>42570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>5140</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7</td>\n",
       "      <td>942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6</td>\n",
       "      <td>295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>3</td>\n",
       "      <td>252</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1</td>\n",
       "      <td>119</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>5</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   cluster  clusterCount\n",
       "0       -1         42570\n",
       "1        4          5140\n",
       "2        7           942\n",
       "3        0           605\n",
       "4        6           295\n",
       "5        3           252\n",
       "6        1           119\n",
       "7        5            45\n",
       "8        2            32"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Cluster results for HDBSCAN\")\n",
    "countByCluster_hdbscan"
   ]
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}