02 PCA.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Overview\n",
    "\n",
    "This notebook is for fitting PCA model on the general German population data to reduce the dimension of the data and understand what the important components mean. We will apply the fitted PCA model to the customers data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load data and packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import libraries here; add more as necessary\n",
    "import os\n",
    "import datetime\n",
    "import re\n",
    "import pickle \n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.decomposition import PCA\n",
    "pd.set_option('max_colwidth', 200)\n",
    "pd.set_option('display.max_columns', 100)\n",
    "pd.set_option('max_rows', 400)\n",
    "\n",
    "# magic word for producing visualizations in notebook\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv(\"azdias_checkpoint_0307.csv\").set_index(\"LNR\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "attribute_values = pd.read_csv(\"attribute_values_cleaned_checkpoint.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "customers = pd.read_csv(\"customers_transformed_0308.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "attribute_descriptions = attribute_values[attribute_values['Description'].notna()][['Attribute','Description']]\\\n",
    "                         .groupby('Attribute').first().reset_index()\n",
    "attribute_descriptions['Attribute1'] = attribute_descriptions['Attribute']\\\n",
    "                                       .apply(lambda x: \"_\".join(x.split(\"_\")[:-1]))\n",
    "attribute_descriptions['Attribute2'] = attribute_descriptions['Attribute']\\\n",
    "                                       .apply(lambda x: x.split(\"_\")[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def attribute_to_descriptions(x):\n",
    "    \n",
    "    \"\"\"\n",
    "    Create a helper dataframe to help translate column names of azdias to the actual meanings so that it's \n",
    "    easier for us to understand the meanings of the components later\n",
    "    \"\"\"\n",
    "    \n",
    "    try:\n",
    "        \n",
    "        if x in list(attribute_descriptions['Attribute']):\n",
    "            result = attribute_descriptions[attribute_descriptions['Attribute'] == x]['Description'].values[0]\n",
    "        \n",
    "        elif \"_\".join(x.split(\"_\")[:-1]) in list(attribute_descriptions['Attribute']):\n",
    "            y = \"_\".join(x.split(\"_\")[:-1])\n",
    "            result = attribute_descriptions[attribute_descriptions['Attribute'] == y]['Description'].values[0]\n",
    "        \n",
    "        elif x in list(attribute_descriptions['Attribute1']):\n",
    "            result = attribute_descriptions[attribute_descriptions['Attribute1'] == x]['Description'].values[0]\n",
    "                       \n",
    "        elif \"_\".join(x.split(\"_\")[:-1]) in list(attribute_descriptions['Attribute1']):\n",
    "            z = \"_\".join(x.split(\"_\")[:-1])\n",
    "            result = attribute_descriptions[attribute_descriptions['Attribute1'] == z]['Description'].values[0]\n",
    "        \n",
    "        elif \"_\".join(x.split(\"_\")[:-2]) in list(attribute_descriptions['Attribute1']):\n",
    "            w = \"_\".join(x.split(\"_\")[:-2])\n",
    "            result = attribute_descriptions[attribute_descriptions['Attribute1'] == w]['Description'].values[0]\n",
    "       \n",
    "        elif x.split(\"_\")[0] in list(attribute_descriptions['Attribute2']):\n",
    "            u = x.split(\"_\")[0]\n",
    "            result = attribute_descriptions[attribute_descriptions['Attribute2'] == u]['Description'].values[0]\n",
    "        \n",
    "        return result\n",
    "    \n",
    "    except:\n",
    "\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Perform Dimensionality Reduction (PCA) on German general population data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca = PCA(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [],
   "source": [
    "general_pca = pca.fit_transform(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [],
   "source": [
    "def scree_plot(pca):\n",
    "    '''\n",
    "    Creates a scree plot associated with the principal components \n",
    "    \n",
    "    INPUT: pca - the result of instantian of PCA in scikit learn\n",
    "            \n",
    "    OUTPUT:\n",
    "            None\n",
    "    '''\n",
    "    num_components=len(pca.explained_variance_ratio_)\n",
    "    ind = np.arange(num_components)\n",
    "    vals = pca.explained_variance_ratio_\n",
    " \n",
    "    plt.figure(figsize=(20, 6))\n",
    "    ax = plt.subplot(111)\n",
    "    cumvals = np.cumsum(vals)\n",
    "    ax.bar(ind, vals)\n",
    "    ax.plot(ind, cumvals)\n",
    "    for i in range(num_components):\n",
    "        ax.annotate(r\"%s%%\" % ((str(vals[i]*100)[:4])), (ind[i]+0.2, vals[i]), va=\"bottom\", ha=\"center\", fontsize=12)\n",
    " \n",
    "    ax.xaxis.set_tick_params(width=0)\n",
    "    ax.yaxis.set_tick_params(width=2, length=12)\n",
    " \n",
    "    ax.set_xlabel(\"Principal Component\")\n",
    "    ax.set_ylabel(\"Variance Explained (%)\")\n",
    "    plt.title('Explained Variance Per Principal Component')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "scree_plot(pca)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6982385273555503"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Investigate the variance accounted for by each principal component.\n",
    "sum(pca.explained_variance_ratio_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The first 100 components can explain 70% of the variance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca_components = pd.DataFrame(pca.components_, columns=data.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca_components.columns = [attribute_to_descriptions(i) for i in list(pca_components.columns)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def display_important_features_for_pc(index):\n",
    "    return pd.concat([pca_components.loc[index,:].sort_values(ascending=False).head(3),\n",
    "                      pca_components.loc[index,:].sort_values(ascending=False).tail(3)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Explain the components"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### First component\n",
    "\n",
    "Low mobility, 1-2 family houses in the PLZ8, low share of AUDI within the PLZ8."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "moving patterns                                         0.133800\n",
       "number of 1-2 family houses in the PLZ8                 0.128520\n",
       "share of AUDI within the PLZ8                           0.128020\n",
       "CAMEO classification 2015 - detailled classification   -0.124806\n",
       "share of AUDI within the PLZ8                          -0.127276\n",
       "number of 6-10 family houses in the PLZ8               -0.127580\n",
       "Name: 0, dtype: float64"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "display_important_features_for_pc(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Second component\n",
    "\n",
    "High online affinity and high recent transaction activity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "online affinity                                                        0.145429\n",
       "transaction activity TOTAL POOL in the last 24 months                  0.140497\n",
       "transaction activity TOTAL POOL in the last 12 months                  0.132982\n",
       "customer journey typology                                             -0.130357\n",
       "actuality of the last transaction for the segment mail-order ONLINE   -0.131349\n",
       "actuality of the last transaction with the complete file ONLINE       -0.136164\n",
       "Name: 1, dtype: float64"
      ]
     },
     "execution_count": 109,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "display_important_features_for_pc(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Third component\n",
    "\n",
    "High share of BMW & Mercedes Benz, high share of upper middle class cars and upper class cars (BMW5er, BMW7er etc.), low share of small and very small cars (Ford Fiesta, Ford Ka etc.) in the PLZ8, low share of cars with max speed between 140 and 210 km/h within the PLZ8, low numbers of cars with 5 seats in the PLZ8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "share of BMW & Mercedes Benz within the PLZ8                                   0.178220\n",
       "share of upper middle class cars and upper class cars (BMW5er, BMW7er etc.)    0.153275\n",
       "share of MERCEDES within the PLZ8                                              0.151223\n",
       "share of small and very small cars (Ford Fiesta, Ford Ka etc.) in the PLZ8    -0.126484\n",
       "share of cars with max speed between 140 and 210 km/h within the PLZ8         -0.127418\n",
       "number of cars with 5 seats in the PLZ8                                       -0.144333\n",
       "Name: 2, dtype: float64"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "display_important_features_for_pc(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Forth component\n",
    "\n",
    "low \"financial typology: be prepared\", older age,  highly likely to be money saver or investor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "financial typology: be prepared                 0.152177\n",
       "age classification through prename analysis     0.141429\n",
       "customer journey typology                       0.112580\n",
       "financial typology: investor                   -0.163427\n",
       "financial typology: unremarkable               -0.164740\n",
       "financial typology: money saver                -0.168190\n",
       "Name: 3, dtype: float64"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "display_important_features_for_pc(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Extract important features\n",
    "\n",
    "We will extract the top 2 features (that can be found explanations) for the top 20 components. We will use the output dataframe later."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [],
   "source": [
    "#names of important features\n",
    "im_features = []  \n",
    "#meanings of important features\n",
    "im_meanings = []\n",
    "#original feature names\n",
    "im_original_features = []\n",
    "#component_list\n",
    "com_list = []\n",
    "\n",
    "pca_components_pd = pd.DataFrame(pca.components_, columns=data.columns)\n",
    "\n",
    "for i in range(20):\n",
    "    #get the list of the features ranked by values in the component\n",
    "    c = list(pca_components_pd.iloc[i].sort_values(ascending=False).index)\n",
    "    j = 0\n",
    "    for f in c:\n",
    "        d = attribute_to_descriptions(f)\n",
    "        if d != f and j != 2:\n",
    "            j += 1\n",
    "            im_features.append(f)\n",
    "            im_meanings.append(d)\n",
    "            im_original_features.append(attribute_values[attribute_values['Description']==d]['Attribute'].values[0])\n",
    "            com_list.append(i)\n",
    "            \n",
    "important_features_pd = pd.DataFrame({'feature' : im_features,\n",
    "                                                         'description' : im_meanings,\n",
    "                                                         'original_feature' : im_original_features,\n",
    "                                                         'component' : com_list})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "metadata": {},
   "outputs": [],
   "source": [
    "attribute_values_1 = attribute_values\\\n",
    ".groupby('Attribute')['value__meaning'].apply(list).reset_index()\\\n",
    ".merge(attribute_values[['Attribute','Description']][attribute_values['Description'].notna()], on=\"Attribute\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "metadata": {},
   "outputs": [],
   "source": [
    "important_features_pd = important_features_pd\\\n",
    ".merge(attribute_values_1[['Attribute', 'value__meaning']].rename({'Attribute':'original_feature'}, axis=1), how=\"left\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "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>feature</th>\n",
       "      <th>description</th>\n",
       "      <th>original_feature</th>\n",
       "      <th>component</th>\n",
       "      <th>value__meaning</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>MOBI_REGIO</td>\n",
       "      <td>moving patterns</td>\n",
       "      <td>MOBI_REGIO</td>\n",
       "      <td>0</td>\n",
       "      <td>[1__very high mobility, 2__high mobility, 3__middle mobility, 4__low mobility, 5__very low mobility , 6__none]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PLZ8_ANTG1</td>\n",
       "      <td>number of 1-2 family houses in the PLZ8</td>\n",
       "      <td>PLZ8_ANTG1</td>\n",
       "      <td>0</td>\n",
       "      <td>[-1__unknown, 0__none, 1__low share, 2__average share, 3__high share, 4__very high share]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ONLINE_AFFINITAET</td>\n",
       "      <td>online affinity</td>\n",
       "      <td>ONLINE_AFFINITAET</td>\n",
       "      <td>1</td>\n",
       "      <td>[0__none, 1__low, 2__middle, 3__high, 4__very high, 5__highest  ]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>D19_GESAMT_ANZ_24</td>\n",
       "      <td>transaction activity TOTAL POOL in the last 24 months</td>\n",
       "      <td>D19_GESAMT_ANZ_24</td>\n",
       "      <td>1</td>\n",
       "      <td>[0__no transactions known, 1__very low activity, 2__low activity, 3__slightly increased activity, 4__increased activity, 5__high activity, 6__very high activity]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>KBA13_HERST_BMW_BENZ</td>\n",
       "      <td>share of BMW &amp; Mercedes Benz within the PLZ8</td>\n",
       "      <td>KBA13_HERST_BMW_BENZ</td>\n",
       "      <td>2</td>\n",
       "      <td>[-1__unknown, 0__none, 1__very low, 2__low, 3__average, 4__high, 5__very high]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                feature  \\\n",
       "0            MOBI_REGIO   \n",
       "1            PLZ8_ANTG1   \n",
       "2     ONLINE_AFFINITAET   \n",
       "3     D19_GESAMT_ANZ_24   \n",
       "4  KBA13_HERST_BMW_BENZ   \n",
       "\n",
       "                                              description  \\\n",
       "0                                         moving patterns   \n",
       "1                 number of 1-2 family houses in the PLZ8   \n",
       "2                                         online affinity   \n",
       "3  transaction activity TOTAL POOL in the last 24 months    \n",
       "4            share of BMW & Mercedes Benz within the PLZ8   \n",
       "\n",
       "       original_feature  component  \\\n",
       "0            MOBI_REGIO          0   \n",
       "1            PLZ8_ANTG1          0   \n",
       "2     ONLINE_AFFINITAET          1   \n",
       "3     D19_GESAMT_ANZ_24          1   \n",
       "4  KBA13_HERST_BMW_BENZ          2   \n",
       "\n",
       "                                                                                                                                                      value__meaning  \n",
       "0                                                     [1__very high mobility, 2__high mobility, 3__middle mobility, 4__low mobility, 5__very low mobility , 6__none]  \n",
       "1                                                                          [-1__unknown, 0__none, 1__low share, 2__average share, 3__high share, 4__very high share]  \n",
       "2                                                                                                  [0__none, 1__low, 2__middle, 3__high, 4__very high, 5__highest  ]  \n",
       "3  [0__no transactions known, 1__very low activity, 2__low activity, 3__slightly increased activity, 4__increased activity, 5__high activity, 6__very high activity]  \n",
       "4                                                                                     [-1__unknown, 0__none, 1__very low, 2__low, 3__average, 4__high, 5__very high]  "
      ]
     },
     "execution_count": 228,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "important_features_pd.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Perform the PCA transformation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca = PCA(100).fit(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "customers_pca = pca.transform(customers)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Checkpoint"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.DataFrame(pca.components_, columns=data.columns).to_csv(\"pca_components_0308.csv\", index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.DataFrame(general_pca).to_csv(\"general_pca_0308.csv\", index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.DataFrame(customers_pca).to_csv(\"customers_pca_0308.csv\", index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"pca.p\", 'wb') as pickle_file:\n",
    "    pickle.dump(pca, pickle_file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"top_20_features.p\", 'wb') as pickle_file:\n",
    "    pickle.dump(top_20_features, pickle_file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [],
   "source": [
    "data[top_20_features].to_csv(\"general_important_features.csv\", index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [],
   "source": [
    "customers[top_20_features].to_csv(\"customer_important_features.csv\", index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "metadata": {},
   "outputs": [],
   "source": [
    "important_features_pd.to_csv(\"important_features.csv\", index=False)"
   ]
  }
 ],
 "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}