Kmeans.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# KMeans"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import the packages and data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from IPython.display import Image\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "matplotlib.rcParams['figure.figsize'] = [12, 12]\n",
    "pd.set_option(\"precision\", 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv('Data/customer_data.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca_data = pd.read_csv('Data/pca_customer_data.csv')\n",
    "#data = table.to_pandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(810341, 7)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(810341, 16)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_data.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Edit PCA df "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>Unnamed: 0</th>\n",
       "      <th>PC1</th>\n",
       "      <th>PC2</th>\n",
       "      <th>PC3</th>\n",
       "      <th>PC4</th>\n",
       "      <th>PC5</th>\n",
       "      <th>PC6</th>\n",
       "      <th>PC7</th>\n",
       "      <th>PC8</th>\n",
       "      <th>PC9</th>\n",
       "      <th>PC10</th>\n",
       "      <th>PC11</th>\n",
       "      <th>PC12</th>\n",
       "      <th>PC13</th>\n",
       "      <th>PC14</th>\n",
       "      <th>UniqueID</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>-0.36368</td>\n",
       "      <td>-0.07125</td>\n",
       "      <td>0.02035</td>\n",
       "      <td>-0.03988</td>\n",
       "      <td>0.26995</td>\n",
       "      <td>-0.14028</td>\n",
       "      <td>0.76859</td>\n",
       "      <td>0.41567</td>\n",
       "      <td>-0.24630</td>\n",
       "      <td>-0.00124</td>\n",
       "      <td>0.22165</td>\n",
       "      <td>-0.38622</td>\n",
       "      <td>-0.35088</td>\n",
       "      <td>0.76266</td>\n",
       "      <td>5962ec84553dcb84e508e2fc45789278</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>-0.36494</td>\n",
       "      <td>-0.19491</td>\n",
       "      <td>0.02874</td>\n",
       "      <td>-0.20343</td>\n",
       "      <td>0.11951</td>\n",
       "      <td>0.63265</td>\n",
       "      <td>1.10307</td>\n",
       "      <td>0.02585</td>\n",
       "      <td>-0.00788</td>\n",
       "      <td>-0.10726</td>\n",
       "      <td>-0.13155</td>\n",
       "      <td>-0.14486</td>\n",
       "      <td>-0.01571</td>\n",
       "      <td>-0.01835</td>\n",
       "      <td>547c0d52bbfe1b93b9411d7b55f3bd82</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0      PC1      PC2      PC3      PC4      PC5      PC6      PC7  \\\n",
       "0           0 -0.36368 -0.07125  0.02035 -0.03988  0.26995 -0.14028  0.76859   \n",
       "1           1 -0.36494 -0.19491  0.02874 -0.20343  0.11951  0.63265  1.10307   \n",
       "\n",
       "       PC8      PC9     PC10     PC11     PC12     PC13     PC14  \\\n",
       "0  0.41567 -0.24630 -0.00124  0.22165 -0.38622 -0.35088  0.76266   \n",
       "1  0.02585 -0.00788 -0.10726 -0.13155 -0.14486 -0.01571 -0.01835   \n",
       "\n",
       "                           UniqueID  \n",
       "0  5962ec84553dcb84e508e2fc45789278  \n",
       "1  547c0d52bbfe1b93b9411d7b55f3bd82  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_data.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "del pca_data[\"Unnamed: 0\"]\n",
    "pca_data = pca_data.drop(columns=\"UniqueID\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>PC1</th>\n",
       "      <th>PC2</th>\n",
       "      <th>PC3</th>\n",
       "      <th>PC4</th>\n",
       "      <th>PC5</th>\n",
       "      <th>PC6</th>\n",
       "      <th>PC7</th>\n",
       "      <th>PC8</th>\n",
       "      <th>PC9</th>\n",
       "      <th>PC10</th>\n",
       "      <th>PC11</th>\n",
       "      <th>PC12</th>\n",
       "      <th>PC13</th>\n",
       "      <th>PC14</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.36368</td>\n",
       "      <td>-0.07125</td>\n",
       "      <td>0.02035</td>\n",
       "      <td>-0.03988</td>\n",
       "      <td>0.26995</td>\n",
       "      <td>-0.14028</td>\n",
       "      <td>0.76859</td>\n",
       "      <td>0.41567</td>\n",
       "      <td>-0.24630</td>\n",
       "      <td>-0.00124</td>\n",
       "      <td>0.22167</td>\n",
       "      <td>-0.38623</td>\n",
       "      <td>-0.35095</td>\n",
       "      <td>0.76248</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.36494</td>\n",
       "      <td>-0.19491</td>\n",
       "      <td>0.02874</td>\n",
       "      <td>-0.20343</td>\n",
       "      <td>0.11951</td>\n",
       "      <td>0.63265</td>\n",
       "      <td>1.10307</td>\n",
       "      <td>0.02585</td>\n",
       "      <td>-0.00788</td>\n",
       "      <td>-0.10726</td>\n",
       "      <td>-0.13154</td>\n",
       "      <td>-0.14486</td>\n",
       "      <td>-0.01572</td>\n",
       "      <td>-0.01838</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.36073</td>\n",
       "      <td>-0.13856</td>\n",
       "      <td>0.00033</td>\n",
       "      <td>-0.07730</td>\n",
       "      <td>0.06696</td>\n",
       "      <td>-0.05990</td>\n",
       "      <td>0.01344</td>\n",
       "      <td>0.14090</td>\n",
       "      <td>-0.16061</td>\n",
       "      <td>0.09449</td>\n",
       "      <td>0.45173</td>\n",
       "      <td>0.00763</td>\n",
       "      <td>-0.33430</td>\n",
       "      <td>0.78235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.37153</td>\n",
       "      <td>0.09313</td>\n",
       "      <td>0.70203</td>\n",
       "      <td>-0.20990</td>\n",
       "      <td>-0.37500</td>\n",
       "      <td>0.01482</td>\n",
       "      <td>-0.19426</td>\n",
       "      <td>0.18343</td>\n",
       "      <td>-0.24443</td>\n",
       "      <td>0.09521</td>\n",
       "      <td>0.50377</td>\n",
       "      <td>-0.39686</td>\n",
       "      <td>0.68769</td>\n",
       "      <td>-0.16564</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.35632</td>\n",
       "      <td>-0.14719</td>\n",
       "      <td>0.03584</td>\n",
       "      <td>-0.27739</td>\n",
       "      <td>0.76592</td>\n",
       "      <td>0.11624</td>\n",
       "      <td>-0.36774</td>\n",
       "      <td>-0.05292</td>\n",
       "      <td>0.00815</td>\n",
       "      <td>0.19761</td>\n",
       "      <td>0.27032</td>\n",
       "      <td>-0.33024</td>\n",
       "      <td>-0.31666</td>\n",
       "      <td>0.79632</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       PC1      PC2      PC3      PC4      PC5      PC6      PC7      PC8  \\\n",
       "0 -0.36368 -0.07125  0.02035 -0.03988  0.26995 -0.14028  0.76859  0.41567   \n",
       "1 -0.36494 -0.19491  0.02874 -0.20343  0.11951  0.63265  1.10307  0.02585   \n",
       "2 -0.36073 -0.13856  0.00033 -0.07730  0.06696 -0.05990  0.01344  0.14090   \n",
       "3 -0.37153  0.09313  0.70203 -0.20990 -0.37500  0.01482 -0.19426  0.18343   \n",
       "4 -0.35632 -0.14719  0.03584 -0.27739  0.76592  0.11624 -0.36774 -0.05292   \n",
       "\n",
       "       PC9     PC10     PC11     PC12     PC13     PC14  \n",
       "0 -0.24630 -0.00124  0.22167 -0.38623 -0.35095  0.76248  \n",
       "1 -0.00788 -0.10726 -0.13154 -0.14486 -0.01572 -0.01838  \n",
       "2 -0.16061  0.09449  0.45173  0.00763 -0.33430  0.78235  \n",
       "3 -0.24443  0.09521  0.50377 -0.39686  0.68769 -0.16564  \n",
       "4  0.00815  0.19761  0.27032 -0.33024 -0.31666  0.79632  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Edit Customer df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [],
   "source": [
    "#data = data[0:1000]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>Unnamed: 0</th>\n",
       "      <th>UniqueID</th>\n",
       "      <th>FrequencyofVisits</th>\n",
       "      <th>Age</th>\n",
       "      <th>Sex</th>\n",
       "      <th>Payer</th>\n",
       "      <th>Facility</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>5962ec84553dcb84e508e2fc45789278</td>\n",
       "      <td>12</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ADVANCECARE</td>\n",
       "      <td>CCB</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>547c0d52bbfe1b93b9411d7b55f3bd82</td>\n",
       "      <td>3</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ADVANCECARE</td>\n",
       "      <td>HCP</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0                          UniqueID  FrequencyofVisits   Age  \\\n",
       "0           0  5962ec84553dcb84e508e2fc45789278                 12  37.0   \n",
       "1           1  547c0d52bbfe1b93b9411d7b55f3bd82                  3  37.0   \n",
       "\n",
       "        Sex        Payer Facility  \n",
       "0  Feminino  ADVANCECARE      CCB  \n",
       "1  Feminino  ADVANCECARE      HCP  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "del data[\"Unnamed: 0\"]\n",
    "data = data.drop(columns=\"UniqueID\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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>FrequencyofVisits</th>\n",
       "      <th>Age</th>\n",
       "      <th>Sex</th>\n",
       "      <th>Payer</th>\n",
       "      <th>Facility</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>12</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ADVANCECARE</td>\n",
       "      <td>CCB</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ADVANCECARE</td>\n",
       "      <td>HCP</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   FrequencyofVisits   Age       Sex        Payer Facility\n",
       "0                 12  37.0  Feminino  ADVANCECARE      CCB\n",
       "1                  3  37.0  Feminino  ADVANCECARE      HCP"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Working with K-means"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loading the data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PC1     float64\n",
       "PC2     float64\n",
       "PC3     float64\n",
       "PC4     float64\n",
       "PC5     float64\n",
       "PC6     float64\n",
       "PC7     float64\n",
       "PC8     float64\n",
       "PC9     float64\n",
       "PC10    float64\n",
       "PC11    float64\n",
       "PC12    float64\n",
       "PC13    float64\n",
       "PC14    float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_data.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_data.isnull().values.any()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### K-means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.cluster import KMeans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_processed= pca_data.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KMeans(algorithm='auto', copy_x=True, init='k-means++', max_iter=300,\n",
       "    n_clusters=3, n_init=10, n_jobs=None, precompute_distances='auto',\n",
       "    random_state=42, tol=0.0001, verbose=0)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimator_kmeans = KMeans(random_state=42, n_clusters=3)\n",
    "estimator_kmeans.fit(data_processed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_id = estimator_kmeans.labels_\n",
    "#cluster_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "centroids = estimator_kmeans.cluster_centers_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3, 14)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "centroids.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1034150.2068234462"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimator_kmeans.inertia_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import euclidean_distances"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "centroid_distances = euclidean_distances(centroids)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3, 3)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "centroid_distances.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(810341,)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimator_kmeans.labels_.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Add cluster label to both df's\n",
    "data['cluster_id'] = cluster_id\n",
    "data_processed['cluster_id'] = cluster_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "numerical_data = data.select_dtypes(np.number)\n",
    "categorical_data = data.select_dtypes([object, \"category\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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>FrequencyofVisits</th>\n",
       "      <th>Age</th>\n",
       "      <th>Sex</th>\n",
       "      <th>Payer</th>\n",
       "      <th>Facility</th>\n",
       "      <th>cluster_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>12</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ADVANCECARE</td>\n",
       "      <td>CCB</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ADVANCECARE</td>\n",
       "      <td>HCP</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ENTIDADES COM DESCONTO</td>\n",
       "      <td>CCB</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>16</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>MÉDIS</td>\n",
       "      <td>HCS</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>PARTICULARES</td>\n",
       "      <td>CCB</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   FrequencyofVisits   Age       Sex                   Payer Facility  \\\n",
       "0                 12  37.0  Feminino             ADVANCECARE      CCB   \n",
       "1                  3  37.0  Feminino             ADVANCECARE      HCP   \n",
       "2                  2  37.0  Feminino  ENTIDADES COM DESCONTO      CCB   \n",
       "3                 16  37.0  Feminino                   MÉDIS      HCS   \n",
       "4                  1  37.0  Feminino            PARTICULARES      CCB   \n",
       "\n",
       "   cluster_id  \n",
       "0           2  \n",
       "1           2  \n",
       "2           2  \n",
       "3           2  \n",
       "4           2  "
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cluster_summary(cluster_id):\n",
    "    cluster = data[data.cluster_id==cluster_id]\n",
    "    cluster_summary = cluster[categorical_data.columns].mode().to_dict(orient=\"records\")[0]\n",
    "    cluster_summary.update(cluster.median().to_dict())\n",
    "    return cluster_summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'Sex': 'Feminino', 'Payer': 'MÉDIS', 'Facility': 'HCD'}]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#cluster = data[data.cluster_id==2]\n",
    "#cluster[categorical_data.columns].mode().to_dict(orient=\"records\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.2436452524992418"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Distance between centroid of cluster 0 and 2 \n",
    "#centroid_distances[0,2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cluster_comparison(*cluster_ids):\n",
    "    summaries = []\n",
    "    for cluster_id in cluster_ids:\n",
    "        summaries.append(cluster_summary(cluster_id))\n",
    "    return pd.DataFrame(summaries).set_index(\"cluster_id\").T"
   ]
  },
  {
   "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>cluster_id</th>\n",
       "      <th>0.0</th>\n",
       "      <th>1.0</th>\n",
       "      <th>2.0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Age</th>\n",
       "      <td>53</td>\n",
       "      <td>54</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Facility</th>\n",
       "      <td>HCP</td>\n",
       "      <td>HCD</td>\n",
       "      <td>HCD</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>FrequencyofVisits</th>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Payer</th>\n",
       "      <td>ADSE</td>\n",
       "      <td>MULTICARE</td>\n",
       "      <td>MÉDIS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sex</th>\n",
       "      <td>Feminino</td>\n",
       "      <td>Masculino</td>\n",
       "      <td>Feminino</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "cluster_id              0.0        1.0       2.0\n",
       "Age                      53         54        52\n",
       "Facility                HCP        HCD       HCD\n",
       "FrequencyofVisits         4          4         4\n",
       "Payer                  ADSE  MULTICARE     MÉDIS\n",
       "Sex                Feminino  Masculino  Feminino"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cluster_comparison(0,1,2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [],
   "source": [
    "def kmeans_cluster(df, n_clusters=2):\n",
    "    model = KMeans(n_clusters=n_clusters, random_state=42)\n",
    "    clusters = model.fit_predict(df)\n",
    "    cluster_results = df.copy()\n",
    "    cluster_results['Cluster'] = clusters\n",
    "    return cluster_results\n",
    "\n",
    "def graph_summary_clustering(results):\n",
    "    cluster_size = results.groupby(['Cluster']).size().reset_index()\n",
    "    cluster_size.columns = ['Cluster', 'Count']\n",
    "    cluster_means = results.groupby(['Cluster'], as_index=False).mean()\n",
    "    cluster_summary = pd.merge(cluster_size, cluster_means, on='Cluster')\n",
    "    cluster_summary = cluster_summary.drop([\"Count\"], axis=1).set_index(\"Cluster\")\n",
    "    return cluster_summary[sorted(cluster_summary.columns)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ideal number of clusters?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a function that evaluates the intertia and percentage variance explained for a value of K\n",
    "def evaluate_k_kmeans(k, scoring, **kwargs):\n",
    "    scoring_measure = None\n",
    "    if scoring==\"inertia\":\n",
    "        scoring_measure = inertia_measure\n",
    "    elif scoring==\"variance\":\n",
    "        scoring_measure = variance_measure\n",
    "        \n",
    "    estimator_kmeans = KMeans(random_state=42, n_clusters=k)\n",
    "    estimator_kmeans.fit(data_processed)\n",
    "    return scoring_measure(estimator_kmeans, **kwargs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%time\n",
    "range_k = [1,2,3,4,5,6]\n",
    "results_k = {}\n",
    "for k in range_k:\n",
    "    results_k[k] = evaluate_k_kmeans_inertia(k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_k"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x1008 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax1 = plt.subplots()\n",
    "\n",
    "ax1.plot(\n",
    "    [c[0] for c in results_k.items()],\n",
    "    [c[1][0] for c in results_k.items()], label=\"inertia\", color=\"red\")\n",
    "ax1.set_ylabel('inertia', color=\"red\")\n",
    "\n",
    "\n",
    "ax2 = ax1.twinx()\n",
    "ax2.plot(\n",
    "    [c[0] for c in results_k.items()],\n",
    "    [c[1][1] for c in results_k.items()], label=\"percentage variance explained\", color=\"blue\")\n",
    "ax2.set_ylabel('percentage variance', color='blue')\n",
    "\n",
    "plt.xlabel(\"K\")\n",
    "plt.legend()\n",
    "plt.title(\"Percentage variance / Inertia by. K\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sse = {}\n",
    "for k in range(1, 10):\n",
    "   kmeans = KMeans(n_clusters=k).fit(data_processed)\n",
    "   #print(data[\"clusters\"])\n",
    "   sse[k] = kmeans.inertia_ # Inertia: Sum of distances of samples to their closest cluster center\n",
    "plt.figure()\n",
    "plt.plot(list(sse.keys()), list(sse.values()))\n",
    "plt.xlabel(\"Number of clusters\")\n",
    "plt.ylabel(\"SSE\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Update the number of clusters "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KMeans(algorithm='auto', copy_x=True, init='k-means++', max_iter=300,\n",
       "    n_clusters=5, n_init=10, n_jobs=None, precompute_distances='auto',\n",
       "    random_state=None, tol=0.0001, verbose=0)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clusterer = KMeans(n_clusters=5)\n",
    "clusterer.fit(data_processed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_summary = graph_summary_clustering(cluster_results);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "matplotlib.rcParams['figure.figsize'] = [14, 14]\n",
    "import seaborn as sns\n",
    "sns.heatmap(cluster_summary.transpose(), annot=True);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluate the clusters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2    238402\n",
       "1    156346\n",
       "3    142714\n",
       "4    136866\n",
       "0    136013\n",
       "dtype: int64"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series(clusterer.labels_).value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_id = clusterer.labels_\n",
    "data[\"cluster_id\"] = cluster_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "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>FrequencyofVisits</th>\n",
       "      <th>Age</th>\n",
       "      <th>Sex</th>\n",
       "      <th>Payer</th>\n",
       "      <th>Facility</th>\n",
       "      <th>cluster_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>12</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ADVANCECARE</td>\n",
       "      <td>CCB</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ADVANCECARE</td>\n",
       "      <td>HCP</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>ENTIDADES COM DESCONTO</td>\n",
       "      <td>CCB</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>16</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>MÉDIS</td>\n",
       "      <td>HCS</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>PARTICULARES</td>\n",
       "      <td>CCB</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   FrequencyofVisits   Age       Sex                   Payer Facility  \\\n",
       "0                 12  37.0  Feminino             ADVANCECARE      CCB   \n",
       "1                  3  37.0  Feminino             ADVANCECARE      HCP   \n",
       "2                  2  37.0  Feminino  ENTIDADES COM DESCONTO      CCB   \n",
       "3                 16  37.0  Feminino                   MÉDIS      HCS   \n",
       "4                  1  37.0  Feminino            PARTICULARES      CCB   \n",
       "\n",
       "   cluster_id  \n",
       "0           1  \n",
       "1           1  \n",
       "2           1  \n",
       "3           1  \n",
       "4           1  "
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([2, 4, 1, 3, 0])"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.cluster_id.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "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>cluster_id</th>\n",
       "      <th>0.0</th>\n",
       "      <th>1.0</th>\n",
       "      <th>2.0</th>\n",
       "      <th>3.0</th>\n",
       "      <th>4.0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Age</th>\n",
       "      <td>55</td>\n",
       "      <td>53</td>\n",
       "      <td>52</td>\n",
       "      <td>53</td>\n",
       "      <td>53</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Facility</th>\n",
       "      <td>HCD</td>\n",
       "      <td>HCP</td>\n",
       "      <td>HCD</td>\n",
       "      <td>HCD</td>\n",
       "      <td>HCD</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>FrequencyofVisits</th>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Payer</th>\n",
       "      <td>PARTICULARES</td>\n",
       "      <td>ADSE</td>\n",
       "      <td>PARTICULARES</td>\n",
       "      <td>MULTICARE</td>\n",
       "      <td>MÉDIS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sex</th>\n",
       "      <td>Masculino</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>Feminino</td>\n",
       "      <td>Feminino</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "cluster_id                  0.0       1.0           2.0        3.0       4.0\n",
       "Age                          55        53            52         53        53\n",
       "Facility                    HCD       HCP           HCD        HCD       HCD\n",
       "FrequencyofVisits             3         4             3          4         5\n",
       "Payer              PARTICULARES      ADSE  PARTICULARES  MULTICARE     MÉDIS\n",
       "Sex                   Masculino  Feminino      Feminino   Feminino  Feminino"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cluster_comparison(0,1,2,3,4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1386.3767842926277"
      ]
     },
     "execution_count": 200,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimator_kmeans.inertia_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MiniBatchKMeans is faster and has a higher inertia."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Internal Evaluation metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.metrics import silhouette_score, calinski_harabaz_score "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.2579394917267587"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cross_val_score(clusterer, data_processed, y=None, \n",
    "                scoring=silhouette_score_cv, cv=3).mean() #a small value means it is well matched"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "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>PC1</th>\n",
       "      <th>PC2</th>\n",
       "      <th>PC3</th>\n",
       "      <th>PC4</th>\n",
       "      <th>PC5</th>\n",
       "      <th>PC6</th>\n",
       "      <th>PC7</th>\n",
       "      <th>PC8</th>\n",
       "      <th>PC9</th>\n",
       "      <th>PC10</th>\n",
       "      <th>PC11</th>\n",
       "      <th>PC12</th>\n",
       "      <th>PC13</th>\n",
       "      <th>PC14</th>\n",
       "      <th>cluster_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.36368</td>\n",
       "      <td>-0.07125</td>\n",
       "      <td>0.02035</td>\n",
       "      <td>-0.03988</td>\n",
       "      <td>0.26995</td>\n",
       "      <td>-0.14028</td>\n",
       "      <td>0.76859</td>\n",
       "      <td>0.41567</td>\n",
       "      <td>-0.24630</td>\n",
       "      <td>-0.00124</td>\n",
       "      <td>0.22165</td>\n",
       "      <td>-0.38622</td>\n",
       "      <td>-0.35088</td>\n",
       "      <td>0.76266</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.36494</td>\n",
       "      <td>-0.19491</td>\n",
       "      <td>0.02874</td>\n",
       "      <td>-0.20343</td>\n",
       "      <td>0.11951</td>\n",
       "      <td>0.63265</td>\n",
       "      <td>1.10307</td>\n",
       "      <td>0.02585</td>\n",
       "      <td>-0.00788</td>\n",
       "      <td>-0.10726</td>\n",
       "      <td>-0.13155</td>\n",
       "      <td>-0.14486</td>\n",
       "      <td>-0.01571</td>\n",
       "      <td>-0.01835</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.36073</td>\n",
       "      <td>-0.13856</td>\n",
       "      <td>0.00033</td>\n",
       "      <td>-0.07730</td>\n",
       "      <td>0.06696</td>\n",
       "      <td>-0.05990</td>\n",
       "      <td>0.01344</td>\n",
       "      <td>0.14090</td>\n",
       "      <td>-0.16061</td>\n",
       "      <td>0.09449</td>\n",
       "      <td>0.45172</td>\n",
       "      <td>0.00763</td>\n",
       "      <td>-0.33428</td>\n",
       "      <td>0.78242</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.37153</td>\n",
       "      <td>0.09313</td>\n",
       "      <td>0.70203</td>\n",
       "      <td>-0.20990</td>\n",
       "      <td>-0.37500</td>\n",
       "      <td>0.01482</td>\n",
       "      <td>-0.19426</td>\n",
       "      <td>0.18343</td>\n",
       "      <td>-0.24443</td>\n",
       "      <td>0.09521</td>\n",
       "      <td>0.50376</td>\n",
       "      <td>-0.39685</td>\n",
       "      <td>0.68774</td>\n",
       "      <td>-0.16549</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.35632</td>\n",
       "      <td>-0.14719</td>\n",
       "      <td>0.03584</td>\n",
       "      <td>-0.27739</td>\n",
       "      <td>0.76592</td>\n",
       "      <td>0.11624</td>\n",
       "      <td>-0.36774</td>\n",
       "      <td>-0.05292</td>\n",
       "      <td>0.00815</td>\n",
       "      <td>0.19761</td>\n",
       "      <td>0.27030</td>\n",
       "      <td>-0.33023</td>\n",
       "      <td>-0.31660</td>\n",
       "      <td>0.79647</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       PC1      PC2      PC3      PC4      PC5      PC6      PC7      PC8  \\\n",
       "0 -0.36368 -0.07125  0.02035 -0.03988  0.26995 -0.14028  0.76859  0.41567   \n",
       "1 -0.36494 -0.19491  0.02874 -0.20343  0.11951  0.63265  1.10307  0.02585   \n",
       "2 -0.36073 -0.13856  0.00033 -0.07730  0.06696 -0.05990  0.01344  0.14090   \n",
       "3 -0.37153  0.09313  0.70203 -0.20990 -0.37500  0.01482 -0.19426  0.18343   \n",
       "4 -0.35632 -0.14719  0.03584 -0.27739  0.76592  0.11624 -0.36774 -0.05292   \n",
       "\n",
       "       PC9     PC10     PC11     PC12     PC13     PC14  cluster_id  \n",
       "0 -0.24630 -0.00124  0.22165 -0.38622 -0.35088  0.76266           2  \n",
       "1 -0.00788 -0.10726 -0.13155 -0.14486 -0.01571 -0.01835           2  \n",
       "2 -0.16061  0.09449  0.45172  0.00763 -0.33428  0.78242           2  \n",
       "3 -0.24443  0.09521  0.50376 -0.39685  0.68774 -0.16549           2  \n",
       "4  0.00815  0.19761  0.27030 -0.33023 -0.31660  0.79647           2  "
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_processed.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "309234.2572875792"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "calinski_harabaz_score(data_processed, cluster_id)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}