Ale - Machine Learning Final Project .ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sunil Ale Final Project on Machine Learning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for numerical computing\n",
    "import numpy as np\n",
    "\n",
    "# for dataframes\n",
    "import pandas as pd\n",
    "\n",
    "# for easier visualization\n",
    "import seaborn as sns\n",
    "\n",
    "# for visualization and to display plots\n",
    "from matplotlib import pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "# import color maps\n",
    "from matplotlib.colors import ListedColormap\n",
    "\n",
    "# Ignore Warnings\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "from math import sqrt\n",
    "\n",
    "# to split train and test set\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# to perform hyperparameter tuning\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "\n",
    "from sklearn.linear_model import Ridge  # Linear Regression + L2 regularization\n",
    "from sklearn.linear_model import Lasso  # Linear Regression + L1 regularization\n",
    "from sklearn.svm import SVR # Support Vector Regressor\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.neighbors import KNeighborsRegressor \n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "\n",
    "# Evaluation Metrics\n",
    "from sklearn.metrics import mean_squared_error as mse\n",
    "from sklearn.metrics import r2_score as rs\n",
    "from sklearn.metrics import mean_absolute_error as mae\n",
    "\n",
    "#import xgboost\n",
    "import os\n",
    "mingw_path = 'C:\\\\Program Files\\\\mingw-w64\\\\x86_64-7.2.0-posix-seh-rt_v5-rev0\\\\mingw64\\\\bin'\n",
    "os.environ['PATH'] = mingw_path + ';' + os.environ['PATH']\n",
    "from xgboost import XGBRegressor\n",
    "from xgboost import plot_importance  # to plot feature importance\n",
    "\n",
    "# to save the final model on disk\n",
    "from sklearn.externals import joblib\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.set_printoptions(precision=2, suppress=True) #for printing floating point numbers upto  precision 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load real estate data from CSV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('BlackFriday 2.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(537577, 12)"
      ]
     },
     "execution_count": 147,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Columns of the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['User_ID', 'Product_ID', 'Gender', 'Age', 'Occupation', 'City_Category',\n",
       "       'Stay_In_Current_City_Years', 'Marital_Status', 'Product_Category_1',\n",
       "       'Product_Category_2', 'Product_Category_3', 'Purchase'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 148,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Display the first 5 rows to see example observations."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>User_ID</th>\n",
       "      <th>Product_ID</th>\n",
       "      <th>Gender</th>\n",
       "      <th>Age</th>\n",
       "      <th>Occupation</th>\n",
       "      <th>City_Category</th>\n",
       "      <th>Stay_In_Current_City_Years</th>\n",
       "      <th>Marital_Status</th>\n",
       "      <th>Product_Category_1</th>\n",
       "      <th>Product_Category_2</th>\n",
       "      <th>Product_Category_3</th>\n",
       "      <th>Purchase</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000001</td>\n",
       "      <td>P00069042</td>\n",
       "      <td>F</td>\n",
       "      <td>0-17</td>\n",
       "      <td>10</td>\n",
       "      <td>A</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>8370</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1000001</td>\n",
       "      <td>P00248942</td>\n",
       "      <td>F</td>\n",
       "      <td>0-17</td>\n",
       "      <td>10</td>\n",
       "      <td>A</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>6.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>15200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1000001</td>\n",
       "      <td>P00087842</td>\n",
       "      <td>F</td>\n",
       "      <td>0-17</td>\n",
       "      <td>10</td>\n",
       "      <td>A</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1422</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1000001</td>\n",
       "      <td>P00085442</td>\n",
       "      <td>F</td>\n",
       "      <td>0-17</td>\n",
       "      <td>10</td>\n",
       "      <td>A</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>14.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1057</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1000002</td>\n",
       "      <td>P00285442</td>\n",
       "      <td>M</td>\n",
       "      <td>55+</td>\n",
       "      <td>16</td>\n",
       "      <td>C</td>\n",
       "      <td>4+</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>7969</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   User_ID Product_ID Gender   Age  Occupation City_Category  \\\n",
       "0  1000001  P00069042      F  0-17          10             A   \n",
       "1  1000001  P00248942      F  0-17          10             A   \n",
       "2  1000001  P00087842      F  0-17          10             A   \n",
       "3  1000001  P00085442      F  0-17          10             A   \n",
       "4  1000002  P00285442      M   55+          16             C   \n",
       "\n",
       "  Stay_In_Current_City_Years  Marital_Status  Product_Category_1  \\\n",
       "0                          2               0                   3   \n",
       "1                          2               0                   1   \n",
       "2                          2               0                  12   \n",
       "3                          2               0                  12   \n",
       "4                         4+               0                   8   \n",
       "\n",
       "   Product_Category_2  Product_Category_3  Purchase  \n",
       "0                 NaN                 NaN      8370  \n",
       "1                 6.0                14.0     15200  \n",
       "2                 NaN                 NaN      1422  \n",
       "3                14.0                 NaN      1057  \n",
       "4                 NaN                 NaN      7969  "
      ]
     },
     "execution_count": 149,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.set_option('display.max_columns', 20) ## display max 20 columns\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Some feaures are numeric and some are categorical\n",
    "# Filtering the categorical features:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Product_ID                    object\n",
       "Gender                        object\n",
       "Age                           object\n",
       "City_Category                 object\n",
       "Stay_In_Current_City_Years    object\n",
       "dtype: object"
      ]
     },
     "execution_count": 150,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.dtypes[df.dtypes=='object']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Distributions of numeric features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x1152 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot histogram grid\n",
    "df.hist(figsize=(16,16), xrot=-45) ## Display the labels rotated by 45 degress\n",
    "\n",
    "# Clear the text \"residue\"\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Observations: We can make out quite a few observations:\n",
    "\n",
    "The items in Category 1 were sold the most in\n",
    "the numbers of 2 and 5. Only around 2,000 costumers bought\n",
    "around 7 items belonging to Category 1.\n",
    "\n",
    "Consider Product_Category_1, 2, 3:\n",
    "The items belonging to Product_Category_1 is the most\n",
    "popular category.\n",
    "\n",
    "Consider Purchase:\n",
    "The highest number (around 113,000) of people made\n",
    "purchases of \\\\$5000 to $7000 while only about \n",
    "2000 people bought around \\\\$23,000 worth of items from\n",
    "the store.\n",
    "\n",
    "Consider User_ID:\n",
    "On average, 50,000 Users were active shown by the\n",
    "number of their User_IDs."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Display summary statistics for the numerical features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {
    "scrolled": false
   },
   "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>User_ID</th>\n",
       "      <th>Occupation</th>\n",
       "      <th>Marital_Status</th>\n",
       "      <th>Product_Category_1</th>\n",
       "      <th>Product_Category_2</th>\n",
       "      <th>Product_Category_3</th>\n",
       "      <th>Purchase</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5.375770e+05</td>\n",
       "      <td>537577.00000</td>\n",
       "      <td>537577.000000</td>\n",
       "      <td>537577.000000</td>\n",
       "      <td>370591.000000</td>\n",
       "      <td>164278.000000</td>\n",
       "      <td>537577.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>1.002992e+06</td>\n",
       "      <td>8.08271</td>\n",
       "      <td>0.408797</td>\n",
       "      <td>5.295546</td>\n",
       "      <td>9.842144</td>\n",
       "      <td>12.669840</td>\n",
       "      <td>9333.859853</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1.714393e+03</td>\n",
       "      <td>6.52412</td>\n",
       "      <td>0.491612</td>\n",
       "      <td>3.750701</td>\n",
       "      <td>5.087259</td>\n",
       "      <td>4.124341</td>\n",
       "      <td>4981.022133</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>1.000001e+06</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>185.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>1.001495e+06</td>\n",
       "      <td>2.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>5866.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.003031e+06</td>\n",
       "      <td>7.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>14.000000</td>\n",
       "      <td>8062.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.004417e+06</td>\n",
       "      <td>14.00000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>12073.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.006040e+06</td>\n",
       "      <td>20.00000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>23961.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            User_ID    Occupation  Marital_Status  Product_Category_1  \\\n",
       "count  5.375770e+05  537577.00000   537577.000000       537577.000000   \n",
       "mean   1.002992e+06       8.08271        0.408797            5.295546   \n",
       "std    1.714393e+03       6.52412        0.491612            3.750701   \n",
       "min    1.000001e+06       0.00000        0.000000            1.000000   \n",
       "25%    1.001495e+06       2.00000        0.000000            1.000000   \n",
       "50%    1.003031e+06       7.00000        0.000000            5.000000   \n",
       "75%    1.004417e+06      14.00000        1.000000            8.000000   \n",
       "max    1.006040e+06      20.00000        1.000000           18.000000   \n",
       "\n",
       "       Product_Category_2  Product_Category_3       Purchase  \n",
       "count       370591.000000       164278.000000  537577.000000  \n",
       "mean             9.842144           12.669840    9333.859853  \n",
       "std              5.087259            4.124341    4981.022133  \n",
       "min              2.000000            3.000000     185.000000  \n",
       "25%              5.000000            9.000000    5866.000000  \n",
       "50%              9.000000           14.000000    8062.000000  \n",
       "75%             15.000000           16.000000   12073.000000  \n",
       "max             18.000000           18.000000   23961.000000  "
      ]
     },
     "execution_count": 152,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Obeservation:\n",
    "On average, a person bought $9333.86 worth of items."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Distributions of categorical features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Display summary statistics for categorical features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "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>Product_ID</th>\n",
       "      <th>Gender</th>\n",
       "      <th>Age</th>\n",
       "      <th>City_Category</th>\n",
       "      <th>Stay_In_Current_City_Years</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>537577</td>\n",
       "      <td>537577</td>\n",
       "      <td>537577</td>\n",
       "      <td>537577</td>\n",
       "      <td>537577</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unique</th>\n",
       "      <td>3623</td>\n",
       "      <td>2</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>top</th>\n",
       "      <td>P00265242</td>\n",
       "      <td>M</td>\n",
       "      <td>26-35</td>\n",
       "      <td>B</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>freq</th>\n",
       "      <td>1858</td>\n",
       "      <td>405380</td>\n",
       "      <td>214690</td>\n",
       "      <td>226493</td>\n",
       "      <td>189192</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Product_ID  Gender     Age City_Category Stay_In_Current_City_Years\n",
       "count      537577  537577  537577        537577                     537577\n",
       "unique       3623       2       7             3                          5\n",
       "top     P00265242       M   26-35             B                          1\n",
       "freq         1858  405380  214690        226493                     189192"
      ]
     },
     "execution_count": 154,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.describe(include=['object'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observation:\n",
    "- The most popular product was P00265242 with 1858 sales.\n",
    "- 189,192 of the coustumers had been living in the same for 1 year.\n",
    "- 214, 690 people of costumers were 26 to 35 years old.\n",
    "- Almost 80% of the costumers were male. 405380 out of 537577 costumers were male."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Bar plots for categorical Features\n",
    "Plot bar plot for the 'Age' feature."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f43b8dcc0>"
      ]
     },
     "execution_count": 155,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,8))\n",
    "sns.countplot(y='Age', data=df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Similarly Plot bar plot for the 'City_Category' feature."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f49da4a90>"
      ]
     },
     "execution_count": 156,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,8))\n",
    "sns.countplot(y='City_Category', data=df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Segmentations\n",
    "Segmentations are powerful ways to cut the data to observe the relationship between categorical features and numeric features.\n",
    "\n",
    "Segmenting the target variable by key categorical features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f49f2de80>"
      ]
     },
     "execution_count": 157,
     "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": [
    "sns.boxplot(y='Purchase', x='City_Category', data=df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observations\n",
    "In this case there are clearly some outliers in City A and B."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "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>User_ID</th>\n",
       "      <th>Product_ID</th>\n",
       "      <th>Gender</th>\n",
       "      <th>Age</th>\n",
       "      <th>Occupation</th>\n",
       "      <th>City_Category</th>\n",
       "      <th>Stay_In_Current_City_Years</th>\n",
       "      <th>Marital_Status</th>\n",
       "      <th>Product_Category_1</th>\n",
       "      <th>Product_Category_2</th>\n",
       "      <th>Product_Category_3</th>\n",
       "      <th>Purchase</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>503697</th>\n",
       "      <td>1005596</td>\n",
       "      <td>P00117642</td>\n",
       "      <td>M</td>\n",
       "      <td>36-45</td>\n",
       "      <td>12</td>\n",
       "      <td>B</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>16.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>292083</th>\n",
       "      <td>1003045</td>\n",
       "      <td>P00052842</td>\n",
       "      <td>M</td>\n",
       "      <td>46-50</td>\n",
       "      <td>1</td>\n",
       "      <td>B</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>15.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>437804</th>\n",
       "      <td>1001387</td>\n",
       "      <td>P00086242</td>\n",
       "      <td>F</td>\n",
       "      <td>51-55</td>\n",
       "      <td>13</td>\n",
       "      <td>B</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23959</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>416883</th>\n",
       "      <td>1004117</td>\n",
       "      <td>P00161842</td>\n",
       "      <td>M</td>\n",
       "      <td>18-25</td>\n",
       "      <td>4</td>\n",
       "      <td>B</td>\n",
       "      <td>4+</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>13.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>23958</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>449656</th>\n",
       "      <td>1003301</td>\n",
       "      <td>P00086242</td>\n",
       "      <td>F</td>\n",
       "      <td>26-35</td>\n",
       "      <td>2</td>\n",
       "      <td>B</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23955</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        User_ID Product_ID Gender    Age  Occupation City_Category  \\\n",
       "503697  1005596  P00117642      M  36-45          12             B   \n",
       "292083  1003045  P00052842      M  46-50           1             B   \n",
       "437804  1001387  P00086242      F  51-55          13             B   \n",
       "416883  1004117  P00161842      M  18-25           4             B   \n",
       "449656  1003301  P00086242      F  26-35           2             B   \n",
       "\n",
       "       Stay_In_Current_City_Years  Marital_Status  Product_Category_1  \\\n",
       "503697                          1               0                  10   \n",
       "292083                          2               1                  10   \n",
       "437804                          1               1                  10   \n",
       "416883                         4+               0                  10   \n",
       "449656                          3               0                  10   \n",
       "\n",
       "        Product_Category_2  Product_Category_3  Purchase  \n",
       "503697                16.0                 NaN     23960  \n",
       "292083                15.0                 NaN     23960  \n",
       "437804                 NaN                 NaN     23959  \n",
       "416883                13.0                16.0     23958  \n",
       "449656                 NaN                 NaN     23955  "
      ]
     },
     "execution_count": 158,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['City_Category']=='B'].sort_values(['Purchase'], ascending = False).head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observation: \n",
    "On average, for all categories of cities, the same amount of purchases, which is of around \\\\$7000, were made on average. However,\n",
    "75th quartile of purchase made in City categoried C was slightly higher than that of the others."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "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>User_ID</th>\n",
       "      <th>Occupation</th>\n",
       "      <th>Marital_Status</th>\n",
       "      <th>Product_Category_1</th>\n",
       "      <th>Product_Category_2</th>\n",
       "      <th>Product_Category_3</th>\n",
       "      <th>Purchase</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>City_Category</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A</th>\n",
       "      <td>1.002901e+06</td>\n",
       "      <td>7.857306</td>\n",
       "      <td>0.381988</td>\n",
       "      <td>5.437036</td>\n",
       "      <td>9.945026</td>\n",
       "      <td>12.682530</td>\n",
       "      <td>8958.011014</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>B</th>\n",
       "      <td>1.003033e+06</td>\n",
       "      <td>7.973244</td>\n",
       "      <td>0.407430</td>\n",
       "      <td>5.300420</td>\n",
       "      <td>9.825763</td>\n",
       "      <td>12.674811</td>\n",
       "      <td>9198.657848</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C</th>\n",
       "      <td>1.003015e+06</td>\n",
       "      <td>8.427538</td>\n",
       "      <td>0.433955</td>\n",
       "      <td>5.165964</td>\n",
       "      <td>9.779649</td>\n",
       "      <td>12.654322</td>\n",
       "      <td>9844.441855</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    User_ID  Occupation  Marital_Status  Product_Category_1  \\\n",
       "City_Category                                                                 \n",
       "A              1.002901e+06    7.857306        0.381988            5.437036   \n",
       "B              1.003033e+06    7.973244        0.407430            5.300420   \n",
       "C              1.003015e+06    8.427538        0.433955            5.165964   \n",
       "\n",
       "               Product_Category_2  Product_Category_3     Purchase  \n",
       "City_Category                                                       \n",
       "A                        9.945026           12.682530  8958.011014  \n",
       "B                        9.825763           12.674811  9198.657848  \n",
       "C                        9.779649           12.654322  9844.441855  "
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.groupby('City_Category').mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observations :\n",
    "\n",
    "City of category C makes the most amount of sales per\n",
    "person which is worth \\\\$9844.44 on average. It around \\\\$600\n",
    "more than that of City Category B.\n",
    "\n",
    "However, City Category A is making more sales per person\n",
    "for all categories of the products.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f3e38cb00>"
      ]
     },
     "execution_count": 160,
     "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": [
    "sns.boxplot(y='Purchase', x='Gender', data=df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Observation\n",
    "\n",
    "There are clearly some outliers in both Genders. Both genders have made on average around \\\\$8000 worth of purchases."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Segment by Gender and display the means and standard deviations within each class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"2\" halign=\"left\">User_ID</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Occupation</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Marital_Status</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Product_Category_1</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Product_Category_2</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Product_Category_3</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Purchase</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Gender</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>F</th>\n",
       "      <td>1.003088e+06</td>\n",
       "      <td>1774.236455</td>\n",
       "      <td>6.742672</td>\n",
       "      <td>6.242116</td>\n",
       "      <td>0.417733</td>\n",
       "      <td>0.493188</td>\n",
       "      <td>5.595445</td>\n",
       "      <td>3.476495</td>\n",
       "      <td>10.007969</td>\n",
       "      <td>4.872295</td>\n",
       "      <td>12.452318</td>\n",
       "      <td>4.126782</td>\n",
       "      <td>8809.761349</td>\n",
       "      <td>4721.440055</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>M</th>\n",
       "      <td>1.002961e+06</td>\n",
       "      <td>1693.251916</td>\n",
       "      <td>8.519705</td>\n",
       "      <td>6.554518</td>\n",
       "      <td>0.405883</td>\n",
       "      <td>0.491063</td>\n",
       "      <td>5.197748</td>\n",
       "      <td>3.830816</td>\n",
       "      <td>9.789072</td>\n",
       "      <td>5.153046</td>\n",
       "      <td>12.732924</td>\n",
       "      <td>4.121502</td>\n",
       "      <td>9504.771713</td>\n",
       "      <td>5051.057256</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             User_ID              Occupation           Marital_Status  \\\n",
       "                mean          std       mean       std           mean   \n",
       "Gender                                                                  \n",
       "F       1.003088e+06  1774.236455   6.742672  6.242116       0.417733   \n",
       "M       1.002961e+06  1693.251916   8.519705  6.554518       0.405883   \n",
       "\n",
       "                 Product_Category_1           Product_Category_2            \\\n",
       "             std               mean       std               mean       std   \n",
       "Gender                                                                       \n",
       "F       0.493188           5.595445  3.476495          10.007969  4.872295   \n",
       "M       0.491063           5.197748  3.830816           9.789072  5.153046   \n",
       "\n",
       "       Product_Category_3               Purchase               \n",
       "                     mean       std         mean          std  \n",
       "Gender                                                         \n",
       "F               12.452318  4.126782  8809.761349  4721.440055  \n",
       "M               12.732924  4.121502  9504.771713  5051.057256  "
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.groupby('Gender').agg([np.mean, np.std])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Correlations\n",
    "Finally, let's take a look at the relationships between numeric features and other numeric features.\n",
    "Correlation is a value between -1 and 1 that represents how closely values for two separate features.\n",
    "Positive correlation means that as one feature increases, the other increases.\n",
    "Negative correlation means that as one feature increases, the other decreases.\n",
    "Correlations near -1 or 1 indicate a strong relationship.\n",
    "Those closer to 0 indicate a weak relationship.\n",
    "0 indicates no relationship."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "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>User_ID</th>\n",
       "      <th>Occupation</th>\n",
       "      <th>Marital_Status</th>\n",
       "      <th>Product_Category_1</th>\n",
       "      <th>Product_Category_2</th>\n",
       "      <th>Product_Category_3</th>\n",
       "      <th>Purchase</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>User_ID</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.023024</td>\n",
       "      <td>0.018732</td>\n",
       "      <td>0.003687</td>\n",
       "      <td>0.001471</td>\n",
       "      <td>0.004045</td>\n",
       "      <td>0.005389</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Occupation</th>\n",
       "      <td>-0.023024</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.024691</td>\n",
       "      <td>-0.008114</td>\n",
       "      <td>-0.000031</td>\n",
       "      <td>0.013452</td>\n",
       "      <td>0.021104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Marital_Status</th>\n",
       "      <td>0.018732</td>\n",
       "      <td>0.024691</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.020546</td>\n",
       "      <td>0.015116</td>\n",
       "      <td>0.019452</td>\n",
       "      <td>0.000129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Product_Category_1</th>\n",
       "      <td>0.003687</td>\n",
       "      <td>-0.008114</td>\n",
       "      <td>0.020546</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.540423</td>\n",
       "      <td>0.229490</td>\n",
       "      <td>-0.314125</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Product_Category_2</th>\n",
       "      <td>0.001471</td>\n",
       "      <td>-0.000031</td>\n",
       "      <td>0.015116</td>\n",
       "      <td>0.540423</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.543544</td>\n",
       "      <td>-0.209973</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Product_Category_3</th>\n",
       "      <td>0.004045</td>\n",
       "      <td>0.013452</td>\n",
       "      <td>0.019452</td>\n",
       "      <td>0.229490</td>\n",
       "      <td>0.543544</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.022257</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Purchase</th>\n",
       "      <td>0.005389</td>\n",
       "      <td>0.021104</td>\n",
       "      <td>0.000129</td>\n",
       "      <td>-0.314125</td>\n",
       "      <td>-0.209973</td>\n",
       "      <td>-0.022257</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     User_ID  Occupation  Marital_Status  Product_Category_1  \\\n",
       "User_ID             1.000000   -0.023024        0.018732            0.003687   \n",
       "Occupation         -0.023024    1.000000        0.024691           -0.008114   \n",
       "Marital_Status      0.018732    0.024691        1.000000            0.020546   \n",
       "Product_Category_1  0.003687   -0.008114        0.020546            1.000000   \n",
       "Product_Category_2  0.001471   -0.000031        0.015116            0.540423   \n",
       "Product_Category_3  0.004045    0.013452        0.019452            0.229490   \n",
       "Purchase            0.005389    0.021104        0.000129           -0.314125   \n",
       "\n",
       "                    Product_Category_2  Product_Category_3  Purchase  \n",
       "User_ID                       0.001471            0.004045  0.005389  \n",
       "Occupation                   -0.000031            0.013452  0.021104  \n",
       "Marital_Status                0.015116            0.019452  0.000129  \n",
       "Product_Category_1            0.540423            0.229490 -0.314125  \n",
       "Product_Category_2            1.000000            0.543544 -0.209973  \n",
       "Product_Category_3            0.543544            1.000000 -0.022257  \n",
       "Purchase                     -0.209973           -0.022257  1.000000  "
      ]
     },
     "execution_count": 163,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f3d991630>"
      ]
     },
     "execution_count": 164,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(20,20))\n",
    "sns.heatmap(df.corr())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAoYAAAKcCAYAAAB1x3uEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzs3X28VWWd///XFsw7vBk1UEEFUj+WhuCA1tg3tZJCJezndKONeTeomVJZv3SkCRl/85umacbuqCTvG00dv5XkSGqlk1kWKqIpfJLC9ChfAu8QRFHc3z/WhZ4IDgc47MU55/V8PHiw97XXutbn2udwzpvrWmvtRrPZRJIkSdqs7gIkSZK0aTAYSpIkCTAYSpIkqTAYSpIkCTAYSpIkqTAYSpIkCYC+dRcgSZKkdRMRlwFHA3/KzP1L247AdcBg4FHgQ5n5TEQ0gK8CRwIvACdl5n2r69cZQ0mSpO7nCuB9q7SdB/w0M/cGflqeA4wB9i5/TgO+taZODYaSJEndTGb+HHh6leZxwJXl8ZXAMe3ar8rMZmbeDewQEbuurl+XknsOP8JGktTbNFp6sEbrftfus0+cTjW7t9LUzJy6lt0GZOZ8gMycHxH9S/tA4PF227WVtvmrdmAwlCRJ2sSUELi2INhZqwvQqw25LiVLkiT1DAtWLhGXv/9U2tuA3dttNwh4cnUdGAwlSZJ6hmnAieXxicCN7do/FhGNiHgb8NzKJedVuZQsSZLUzUTE94DDgJ0jog2YBHwRuD4iTgUeAz5YNr+Z6lY1c6luV3PymvptNJtes9BD+IWUJPU2Pfbik2aztWNbyaVkSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFQZDSZIkAQZDSZIkFX3rLmBTFRGDgZsyc/92bRcASzLzyxvrWBFxGHAj8Adga2AB8KXMvKkrjylJkrQqZwxbKCI6G8TvzMwRmRnABOAbEfHujViaJEmSM4brIyImAGcArwAPZ+ZHImIb4OvAW6ne1wsy88aIOAk4CtgS2AZ417ocKzPvj4h/As4Cftp1o5AkSfpzzhiun/OAEZk5jCogAkwEfpaZo4DDgX8rYRHg7cCJmblOobCd+4B9N6RgSZKktTEYrlmzg/YHgKsj4u+oZg0BRgPnRcT9wB1UM4R7lNduy8ynN6CWxgbsK0mS1CkuJa/ZU8BfrdK2IzCPamn4ncD7gX+MiP2owtuxmZntd4iIg4GlG1jLCGD2BvYhSZLUIWcM1yAzlwDzV170ERE7Au8DfgHsnpm3A58DdgD6AbcAZ0dEo2w/oivqiIhhwD8CU7qiP0mSpDVxxrBjHwOmRMS/l+eTgceA2yNie6pZwosy89mIuBD4CvBACYePAkev53H/V0TMpLpdzZ+ACZnphSeSJGmjajSbazqVTt2MX0hJUm/T0nPwG43W/a5tNuu5vsClZEmSJAEuJbdURLwV+O4qzS9l5sF11CNJktSeS8k9h19ISVJv41JyF3MpWZIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSQD0rbsAdZ3G5EbdJdSiOalZdwmSJPUIzhhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSp6Ft3AZIkSeq8iAjgunZNQ4EvADsA44GFpf38zLx5XfpuNJvNLilStWs2JjfqrqEWzUl+D0tSL9XSX3yNBi37hdNsdm5sEdEHeAI4GDgZWJKZX17f47qULEmS1H29G/h9Zv6xKzozGEqSJHVfHwG+1+75WRHxQERcFhF/ta6deY6hJEnSJiYiTgNOa9c0NTOnrrLNG4D3A/9Qmr4FXAg0y9//DpyyLsc1GEqSJG1iSgicupbNxgD3ZeaCss+ClS9ExHeAm9b1uC4lS5IkdU/H0W4ZOSJ2bffaB4DfrmuHzhhKkiR1MxGxNXAEcHq75i9FxHCqpeRHV3mtU7xdTc/h7WokSb1Nr79dTVdzKVmSJEmAwVCSJEmFwVCSJEmAwVBd4Pi3Hs+sM2Yx64xZ3HXKXQwbMKzukiRJ0nrocVclR8QgYArwFqrgexPw/2bm8prqOQb4XWY+XJ7/E/DzzPxJHfVsDPOemcehVxzKsy8+y/v2eh9Tj57K2y59W91lSZKkddSjZgwjogF8H/hhZu4N7AP0A/65xrKOoQqpAGTmF3pSKAT4VduvePbFZwG4u+1uBm03qOaKJEnS+uhRt6uJiHcDkzLzne3atgPmAXsAk4H3Ut3f5zuZ+fWIGAV8FdgGeInqw6iPBUZm5lmlj5uAL2fmHRGxBLgYOBx4BvhIZi6MiPFUH13zBmAucAIwnGrG8rny51jgH4GbMvOGUu+XqWZuZwAfz8yXIuJR4EpgLLA58MHMnLOW4W8St6v5zNs/w74778v4H41v2TG9XY0k9VrerqaL9agZQ2A/4N72DZm5GHgM+HtgCDAiM4cBV5fPGLwO+GRmHgC8B1i2lmNsQ/XxMwcC/wNMKu3fz8xRpZ/ZwKmZ+UtgGtVS9vDM/P3KTiJiS+AK4MOZ+VaqcPjxdsdZVI7xLeCz6/g+1OKwwYdx6ohTOfcn59ZdiiRJWg89LRg2YLVpvgG8E/h2Zr4CkJlPAwHMz8wZpW3xytc78CpVmAT4T+Ad5fH+EXFnRDwIfJQqpHYkgHmZ+bvy/MpS40rfL3/fCwxeS18td+aoM5l5+kxmnj6TXfvtylv7v5VLxl7CuGvH8fSyp+suT5IkrYeeFgwfAka2byhLybuz+tC4piD5Cn/+3mzZwTFX7n8FcFaZ/Zu8ln1WHrsjL5W/V7AJXiT0zRnfZMTFIxhx8Qj6btaX73/4+5zwgxN45OlH6i5NkiStp54WDH8KbB0RHwOIiD7Av1OFtluBMyKib3ltR2AOsFs5z5CI2La8/igwPCI2i4jdgYPaHWMz4G/L4+OBX5TH2wLzI2JzqhnDlZ4vr61qDjA4IvYqz0+gWprudr5w6BfYaaud+OZR32Tm6TOZMX5G3SVJkqT1sMnNRG2IzGxGxAeAb0bEP1KFuJuB86lm3vYBHoiIl6kuPvlGRHwY+HpEbEV1fuF7gLuoLlh5EPgtcF+7wywF9ouIe6kuKPlwaf9H4NfAH8t+K8PgtcB3ImICrwdKMvPFiDgZ+K8SRmcA3+7SN6RFxv9ofEsvNpEkSRtHj7oquRUiYklm9qu7jtXYJK5KroNXJUtSr+VVyV2spy0lS5IkaT0ZDNfRJjpbKEmStMEMhpIkSQIMhpIkSSoMhpIkSQIMhpIkSSoMhpIkSQIMhpIkSSoMhpIkSQIMhpIkSSoMhpIkSQIMhpIkSSoMhpIkSQIMhpIkSSoMhpIkSQKgb90FSJIkdQdNGi09Wh2cMZQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRIAjWazWXcN6hp+ISVJvU2jtUdrtO53bbPZ2rEVzhhKkiQJgL51F6CuVMt/LjYBK/8D19vHL0nShnHGUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSYXBUJIkSYDBUJIkSUXfugtQ9zZ/fl8+97ldWLSoD5ttBh/60HOceOKzdZfVEr157JKknqnRbDbrrkFdowmNlh/0T3/qw8KFfdlvv5dYsqTBscfuyZQpT7LXXstbWMXK7+HWjn/TGDu8Pn5J6nVa+4O/0WjdD9xms/W/1NlEl5IjohkR3233vG9ELIyIm9axn90i4obyeHhEHNmJfQ7r6DgRMSAiboqIWRHxcETcXNoHR8Txnei/U9t1F/37r2C//V4CoF+/JkOHLmfBgt4xEd2bxy5J6pk2yWAILAX2j4ityvMjgCfWpYOI6JuZT2bm35am4cBag2En/BNwW2YekJlvAc4r7YOBzgS+zm7X7bS19WX27C044IAX6y6l5Xrz2CVJPcemPL0xHTgKuAE4Dvge8L8AIuIg4CvAVsAy4OTMzIg4qeyzJbBNRJwC3AQcSBXotoqIdwD/AsxbXR+dqGtX4NaVTzLzgfLwi8CbI+J+4ErgB8B3gW3K62dl5i9Xs90zwMjMPKuM7Sbgy8CdwKXASKq1wssy86JOvXM1WLq0wYQJu3H++Qvp1+/Vustpqd48dklSz7KpzhgCXAt8JCK2BIYBv2732hzgnZk5AvgC8P+3e+3twImZ+a6VDZm5vGx3XWYOz8zr1tJHR6YAl0bE7RExMSJ2K+3nAXeW/i8C/gQckZkHAh8GvraG7dZkODAwM/fPzLcCl3eyvpZ7+WWYMGE3xo5dzOjRS+oup6V689glST3PJjtjmJkPRMRgqtnCm1d5eXvgyojYm2o2bfN2r92WmU934hAd9dFRXbdExFDgfcAYYGZE7L+aTTcHvhERw4EVwD6d6b+dPwBDI+LrwH/TbpZyU9JswsSJuzB06HJOPrl3XZHbm8cuSeqZNuUZQ4BpVMuq31ul/ULg9szcHxhLtXS80tJO9t1RHx3KzKcz85rMPAGYAbxzNZt9GlgAHEC1HPyGNXT3Cn/+ddiyHOOZsu8dwCeASzpbXyvde++W3Hjjdtx999aMG7cH48btwf/8zzZr37EH6M1jlyT1TJvsjGFxGfBcZj4YEYe1a9+e1y9GOamTfT0PbLuBfRAR7wLuzswXImJb4E3AY8Crq+m/LTNfjYgTgT5rqONR4MyI2AwYCBxUjrMzsDwz/3dE/B64orM1ttLIkS+S+bu6y6hFbx67JKln2qSDYWa2AV9dzUtfoloGPgf4WSe7ux04r1z08S/r2QfAX1MtEa+c6bskM2dExObAKxExiyrEfRP43xHxwXLslTOZD6yy3VeoLoR5EPgtcF/ZbiBweQmMAP+wDjVKkiStM29w3XPUcoPrTUM9N7jedPhvWFKv5Q2uu9imfo6hJEmSWmSTXkquU0ScDHxylea7MvMTddQjSZK0sbmU3HO4lNzrxy9JvY5LyV3MGUNJkqRuJiIepbrTyQrglcwcGRE7AtdRffzuo8CHyu3vOs1zDCVJkrqnw8snqY0sz88DfpqZewM/Lc/XicFQkiSpZxgHXFkeXwkcs64duJQsSZK0iYmI04DT2jVNzcyp7Z43gVsjoglcXF4bkJnzATJzfkT0X9fjGgwlSZI2MSXoTe1gk0My88kS/m6LiDldcVyXkiVJkrqZzHyy/P0n4AdUH6m7ICJ2BSh//2ld+zUYSpIkdSMRsU1EbLvyMTCa6mN1pwEnls1OBG5c1769j2HP4X0Me/34JanX6ZX3MYyIoVSzhFCdFnhNZv5zROwEXA/sATwGfDAzn16XwxoMew6DYa8fvyT1Or0yGG5MLiVLkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJMBhKkiSpMBhKkiQJgL51F6Cu1Ky7gJr19vFLkrRhnDGUJEkS4Ixhj9KY3Ki7hFo0J1Uzhb19/NA7x+9MsSR1HWcMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMJUmSBBgMJUmSVBgMtcHe+6b3MucTc3jk7Ec495Bz6y6npWKn4Jen/JIXJ77IZ97+mbrLaan58/tywgmDGDNmT446ak+uvHKHukuSJG2gvnUXoO5ts8ZmTDlyCkd89wjaFrcxY/wMpuU0Zi+aXXdpLfH0sqeZ8OMJHLPvMXWX0nJ9+jQ577yF7LffSyxZ0uDYY/fkkENeYK+9ltddmiRpPa01GEbECuDBsu1s4MTMfGF9DhYRJwEjM/Os9dz31sx8soNtNgcuBI4FXgJeACZl5vQO9vkUMHV9x7QxRMS+wOXAgcDEzPxyzSWt0UEDD2Lu03OZ9+w8AK596FrG7TuO2b/oHcFw4QsLWfjCQo7a+6i6S2m5/v1X0L//CgD69WsydOhyFizoazCUpG6sM0vJyzJzeGbuDywHzmj/YkQ0IqIVS9InAbutZZsLgV2B/Uu9Y4Ft17LPp4CtN7i6DkTEus7MPg1MADbZQLjSwG0H8vjix1973ra4jYHbDqyxItWhra0vs2dvwQEHvFh3KZKkDbCugeVOYFhEDAamA7cDbweOiYi/Ac4HGsB/Z+a5ABFxMvAPwHzgd1QzeUTEFcBNmXlDeb4kM/uVx58DTgBeLce5BxgJXB0Ry4C3Z+ay9oVFxNbAeGBIZr4EkJkLgOvL698CRgFbATdk5qSImEAVNm+PiEWZeXhEjAYmA1sAvwdOzswlEXEk8B/AIuA+YGhmHh0ROwKXAUOpZihPy8wHIuKC0vdgYFFE7A6cnZn3l3ruAj6emQ+s+iZn5p+AP0XEJj8N1Wg0/qKtSbOGSlSXpUsbTJiwG+efv5B+/V6tuxxJ0gbo9ExfmfUaQ7WsDBDAVZk5AngZ+FfgXcBwYFREHBMRu1KFrEOAI4C3dOI4Y4BjgIMz8wDgSyU83gN8tMxeLlvNrnsBj2Xm4jV0PTEzRwLDgEMjYlhmfg14Eji8hMKdgc8D78nMA8sxz4mILYGLgTGZ+Q7gje36nQzMzMxhVMH4qnav/TUwLjOPBy6hmvUkIvYBtlhdKOxu2ha3sft2u7/2fNB2g3jy+TWu9vcIZ446k5mnz2Tm6TPZtd+udZdTq5dfhgkTdmPs2MWMHr2k7nIkaeNqtvBPTToTDLeKiPupQtJjwKWl/Y+ZeXd5PAq4IzMXZuYrwNXAO4GD27UvB67rxPHeA1y+8py/zHy688Pp0Ici4j5gJrAfqw+pbyvtd5UxnwjsCewL/CEz55Xtvtdun3cA3y21/gzYKSK2L69Naxdi/ws4upwHeQpwRReNq1YznpjB3jvtzeAdBrP5Zpvzkf0+wrScVndZG9U3Z3yTERePYMTFI5i/ZH7d5dSm2YSJE3dh6NDlnHzys3WXI0nqAp1ZSl6WmcPbN0QEwNJ2TX+5nvi6NeXeVyjBNCIawBva9bU+WXkusEdEbJuZz69S7xDgs8CozHymLGNvuZo+GsBtmXncKvuP6OC4qxv7yvpfe48y84WIuA0YB3yIamm821vRXMFZN5/FLX93C30afbjs/st4eOHDdZfVMgO2GcA9p93Ddltsx6vNV/nU2z7FW6a8heeXP7/2nbu5e+/dkhtv3I599nmJceP2AOCcc57i0EOXrmVPSdKmqqtuV/Nr4KtlKfYZ4Djg68BvSvtOwGLgg8Csss+jVEut11OFpc1L+63AFyLimhKmdiyzhs/TwYUkZdtLga9FxOmZubwsZb+bavl7KfBcRAygWhK/o+y6st9FwN3AlIjYKzPnlvMWBwFzgKERMTgzHwU+3O7QPwc+ClwYEYcBizJzcQnPq7oE+BFwZxfOhNZu+tzpTP/GGi/87tEWLF3A7hftvvYNe6CRI18k83d1lyFJ6kJdcjVxZs6nusDkdqrgd19m3ljaLwB+BfyE6qKNlb5Dda7fb6iWnJeWvn4MTAPuKcu5ny3bXwF8OyLuj4it1lDK54GFwMMR8Vvgh8DCzJxFtYT8ENWFIne122cqMD0ibs/MhVTnAX4vIh6gCor7luXgM4EfR8QvgAXAc2X/C4CRZfsvUi0/r+l9upcqIF++pm0AImKXiGgDzgE+HxFtEbFdR/tIkiRtqEaz6RWknRER/crVyQ1gCvBIZl60jn3sRjVTuW9mdvXlm83G5I5W9Huu5qTqe7i3j7/jMzp6Mn+GSb1Yi3/wNVr4A6dZyw91PxKv88aXGcyHgO2prlLutIj4GNWS+8SNEAolSZI2WLecMYyIHwBDVmk+NzNvqaOe9VXu8fjJVZrvysxPrEd3zhj28vE7YyipF3LGsIt1y89KzswP1F1DV8jMy1nL+YaSJEmt4lKyJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSCoOhJEmSAIOhJEmSir51FyBJkqTOi4jdgauAXYBXgamZ+dWIuAAYDywsm56fmTevS98GQ0mSpO7lFeAzmXlfRGwL3BsRt5XXLsrML69vxwZDSZKkbiQz5wPzy+PnI2I2MLAr+m40m82u6Ef18wspSeptGi0+XMt+10bsczpwWrumqZk59S+3i8HAz4H9gXOAk4DFwD1Us4rPrMtxDYY9h19ISVJv02ODITTXOraI6Af8D/DPmfn9iBgALKLKBBcCu2bmKetyVINhz9FsTG7xv49NRHNS9T3c28ff8p+Pm4wy/kYvHL8/v6VeGwwjYnPgJuCWzPyP1bw+GLgpM/dfl6N6uxpJkqRuJCIawKXA7PahMCJ2bbfZB4DfrmvfXnwiSZLUvRwCnAA8GBH3l7bzgeMiYjjVUsqjwOnr2rHBUJIkqRvJzF+w+mX0dbpn4eq4lCxJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYChJkqTCYChJkiTAYKgu8N43vZc5n5jDI2c/wrmHnFt3OS331fd9lUfOfoRZZ8xixC4j6i6nZebP78sJJwxizJg9OeqoPbnyyh3qLqk15s2DBx6AmTNhxow/f+0zn4FmE3baqZ7aJGkD9a27AHVvmzU2Y8qRUzjiu0fQtriNGeNnMC2nMXvR7LpLa4kxe41h7x33Zu+v783BAw/mW0d9i7dd+ra6y2qJPn2anHfeQvbb7yWWLGlw7LF7csghL7DXXsvrLm3jO/xweOqpP28bNAiOOAL++Md6apKkLrDWYBgRK4AHy7azgRMz84X1OVhEnASMzMyz1nPfWzPzyQ622Ry4EDgWeAl4AZiUmdM72OdTwNT1HdPGEBEfBVZOvS0BPp6Zs2osaY0OGngQc5+ey7xn5wFw7UPXMm7fccz+Re8IhuP2HcdVD1wFwK+f+DU7bLkDu/Tbhf+z5P/UXNnG17//Cvr3XwFAv35Nhg5dzoIFfXtHMFydiy6Cz30Obryx7kokab11Zil5WWYOz8z9geXAGe1fjIhGRLRiSfokYLe1bHMhsCuwf6l3LLDtWvb5FLD1BlfXgYhY15nZecChmTmMakxTu76qrjFw24E8vvjx1563LW5j4LYDa6yotQZuO5DHn+u941+pra0vs2dvwQEHvFh3KRtfswm33gr33APjx1dtY8fCE09US8yS1I2ta2C5ExgWEYOB6cDtwNuBYyLib4DzgQbw35l5LkBEnAz8AzAf+B3VTB4RcQVwU2beUJ4vycx+5fHngBOAV8tx7gFGAldHxDLg7Zm5rH1hEbE1MB4YkpkvAWTmAuD68vq3gFHAVsANmTkpIiZQhc3bI2JRZh4eEaOBycAWwO+BkzNzSUQcCfwHsAi4DxiamUdHxI7AZcBQqhnK0zLzgYi4oPQ9GFgUEbsDZ2fm/aWeu6hmAv/iN0lm/rLd07uBQWv9ytSk0Wj8RVuTZg2V1KNB7x4/wNKlDSZM2I3zz19Iv36v1l3OxnfIITB/PrzxjXDbbTBnDkycCKNH112ZJG2wTs/0lVmvMVTLygABXJWZI4CXgX8F3gUMB0ZFxDERsStVyDoEOAJ4SyeOMwY4Bjg4Mw8AvlTC4z3AR8vs5bLV7LoX8FhmLl5D1xMzcyQwDDg0IoZl5teAJ4HDSyjcGfg88J7MPLAc85yI2BK4GBiTme8A3tiu38nAzDK7dz5wVbvX/hoYl5nHA5dQzXoSEfsAW6wuFK7GqVTheJPUtriN3bfb/bXng7YbxJPPr3G1v0c4c9SZzDx9JjNPn8mTS55k9+171/jbe/llmDBhN8aOXczo0UvqLqc15s+v/l64EH7wAzj0UBgyBGbNqi5MGTQI7rsPBgyot05JWg+dCYZbRcT9VCHpMeBWkMz6AAAgAElEQVTS0v7HzLy7PB4F3JGZCzPzFeBq4J3Awe3alwPXdeJ47wEuX3nOX2Y+3fnhdOhDEXEfMBPYj9WH1LeV9rvKmE8E9gT2Bf6QmfPKdt9rt887gO+WWn8G7BQR25fXprULsf8FHF3OgzwFuGJtBUfE4VTBcJO91HfGEzPYe6e9GbzDYDbfbHM+st9HmJbT6i5ro/rmjG8y4uIRjLh4BD+c80M+NuxjABw88GCee+m5XnF+IVQrqhMn7sLQocs5+eRn6y6nNbbeGvr1e/3x6NHVlckDBlThcMgQaGuDAw+EBQvqrVWS1kNnlpKXZebw9g0RAbC0XdNfrqe9bk3raq9QgmlENIA3tOtrfdbi5gJ7RMS2mfn8KvUOAT4LjMrMZ8oy9par6aMB3JaZx62yf0f3IFnd2FfW/9p7lJkvRMRtwDjgQ1RL42sUEcOoZhnHZOZTHW1bpxXNFZx181nc8ne30KfRh8vuv4yHFz5cd1ktc/MjN3Pk3kcy9+y5vPDyC5x848l1l9Qy9967JTfeuB377PMS48btAcA55zzFoYcuXcue3diAAdUsIUDfvnDNNXDLLfXWJEldqKtuV/Nr4KtlKfYZ4Djg68BvSvtOwGLgg8DKq2sfpVpqvZ4qLG1e2m8FvhAR15QwtWOZNXyeDi4kKdteCnwtIk7PzOVlKfvdVMvfS4HnImIA1ZL4HWXXlf0uojqfb0pE7JWZc8t5i4OAOcDQiBicmY8CH2536J8DHwUujIjDgEWZubiE51VdAvwIuLOjmdCI2AP4PnBCZv5uTdttKqbPnc70b2yyq90b3Vk3r/NF9j3CyJEv0g2+PbvWvHkwfHjH2wwZ0ppaJGkj6JKriTNzPtUFJrdTBb/7MvPG0n4B8CvgJ1QXbaz0Hapz/X5DteS8tPT1Y2AacE9Zzv1s2f4K4NsRcX9EbLWGUj4PLAQejojfAj8EFpZbvcwEHqK6UOSudvtMBaZHxO2ZuZDqPMDvRcQDVEFx37IcfCbw44j4BbAAeK7sfwEwsmz/Rarl5zW9T/dSBeTL17RN8QVgJ+CbZbz3rGV7SZKkDdZoNnvXFZTrKyL6lauTG8AU4JHMvGgd+9iNaqZy38zs6ss3m43JHa3o91zNSdX3cG8ff8dndPRkZfyruUK+x/Pnt9Tif/iNFv6ja9byQ82PxOu88WUG8yFge6qrlDstIj5GteQ+cSOEQkmSpA3WLWcMI+IHwKon8pybmd3qLPByj8dPrtJ8V2Z+Yj26c8awl4/fGcNeOP5u+PNb6mLOGHaxbvlZyZn5gbpr6AqZeTlrP99QkiSpJVxKliRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUmEwlCRJEmAwlCRJUtFoNpt116Cu4RdSktTbNFp8uBb+rm22eGwVZwwlSZIEQN+6C1DXaUyu5T8XtWtOqv4D19vH3/L/OG8yevP4y9gbvXHsgCteUpdzxlCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJElF37oLkCRJ6g4ak1t3rOak1h2rPWcMJUmSBDhjKEmS1O1ExPuArwJ9gEsy84td0a8zhpIkSd1IRPQBpgBjgLcAx0XEW7qib4OhJElS93IQMDcz/5CZy4FrgXFd0bHBUJIkqXsZCDze7nlbadtgnmMoSZK0iYmI04DT2jVNzcyp5XFjNbs0u+K4BkNJkqRNTAmBU9fwchuwe7vng4Anu+K4BkNJkqTuZQawd0QMAZ4APgIc3xUde46hJElSN5KZrwBnAbcAs4HrM/OhrujbGUNJkqRuJjNvBm7u6n6dMZQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMNQGe++b3sucT8zhkbMf4dxDzq27nJa69P2XsuCzC3jw4w/WXUrLzZ/flxNOGMSYMXty1FF7cuWVO9Rd0ka3pjF/5Ss7MXbsnowbtwennDKQBQv61FzpRjRvHjzwAMycCTNm/Plrn/kMNJuw00711CZpgxkMtUE2a2zGlCOnMObqMbxlyls4bv/jePPOb667rJa54v4reN9/vq/uMmrRp0+T885byPTpf+S66x7jmmt2YO7cN9Rd1ka1pjH//d8/w49+9EduvPExDjtsKVOm9PBgdPjhMGIEjBr1etugQXDEEfDHP9ZXl6QNttbb1UTECuDBsu1s4MTMfGF9DhYRJwEjM/Os9dz31sxc4529I2Jz4ELgWOAl4AVgUmZO72CfT1F9zMx6jWljiIhxVON4FXgF+FRm/qLeqlbvoIEHMffpucx7dh4A1z50LeP2HcfsX8yuubLWuPOxO9lz+z3rLqMW/fuvoH//FQD069dk6NDlLFjQl732Wl5zZRtPZ8a8bFmDxuo+rKqnu+gi+Nzn4MYb665E0gbozIzhsswcnpn7A8uBM9q/GBGNiGjFzONJwG5r2eZCYFdg/1LvWGDbtezzKWDrDa6uAxGxrveL/ClwQGYOB04BLun6qrrGwG0H8vji1z/Hu21xGwO37ZLP8VY30tbWl9mzt+CAA16su5SWWXXMF120E4ceOoQf/Wg7PvnJp2qubiNqNuHWW+Gee2D8+Kpt7Fh44olqiVlSt7augeVOYFhEDAamA7cDbweOiYi/Ac6n+mDn/87McwEi4mTgH4D5wO+oZvKIiCuAmzLzhvJ8SWb2K48/B5xANWM2HbgHGAlcHRHLgLdn5rL2hUXE1sB4YEhmvgSQmQuA68vr3wJGAVsBN2TmpIiYQBU2b4+IRZl5eESMBiYDWwC/B07OzCURcSTwH8Ai4D5gaGYeHRE7ApcBQ6lmKE/LzAci4oLS92BgUUTsDpydmfeXeu4CPp6Zf/GTNDOXtHu6DV30wdgbQ2M1UyPNTbdcbQRLlzaYMGE3zj9/If36vVp3OS2xujF/+tNP8elPP8XFF/8V//mfOzBhQg8Nh4ccAvPnwxvfCLfdBnPmwMSJMHp03ZVJ6gKdnukrs15jqJaVAQK4KjNHAC8D/wq8CxgOjIqIYyJiV6qQdQhwBPCWThxnDHAMcHBmHgB8qYTHe4CPltnLZavZdS/gscxcvIauJ2bmSGAYcGhEDMvMr1F96PThJRTuDHweeE9mHliOeU5EbAlcDIzJzHcAb2zX72RgZmYOowrGV7V77a+BcZl5PNWs30lljPsAW6wuFLZ7Hz4QEXOA/6aaNdwktS1uY/ftXv8c70HbDeLJ57vkc7zVDbz8MkyYsBtjxy5m9Ogla9+hB1jbmI8++nluvbVfDZW1yPz51d8LF8IPfgCHHgpDhsCsWdWFKYMGwX33wYAB9dYpab10JhhuFRH3U4Wkx4BLS/sfM/Pu8ngUcEdmLiyf33c18E7g4Hbty4HrOnG89wCXrzznLzOf7vxwOvShiLgPmAnsx+pD6ttK+11lzCcCewL7An/IzHllu++12+cdwHdLrT8DdoqI7ctr09qF2P8Cji7nQZ4CXNFRsZn5g8zclyokX7guA22lGU/MYO+d9mbwDoPZfLPN+ch+H2FaTqu7LLVAswkTJ+7C0KHLOfnkZ+supyXWNOZHH938tcc/+1k/hg7toedZbr019Ov3+uPRo6srkwcMqMLhkCHQ1gYHHggLFtRbq6T10pml5GXlXLfXRATA0nZNHZ1qvaZ1xVcowTQiGsDKyxkbHezTkbnAHhGxbWY+v0q9Q4DPAqMy85myjL3lavpoALdl5nGr7D+ig+Oubuwr63/tPcrMFyLiNmAc8CGqpfG1ysyfR8SbImLnzFzUmX1aaUVzBWfdfBa3/N0t9Gn04bL7L+PhhQ/XXVbLXPP/XMNhgw9j56135vFPP86kOyZx2czL6i6rJe69d0tuvHE79tnnJcaN2wOAc855ikMPXbqWPbuvNY35hhu2Y968N9BowMCBLzN58p9qrnQjGTCgmiUE6NsXrrkGbrml3pokdal1PcdwTX4NfLUsxT4DHAd8HfhNad8JWAx8EJhV9nmUaqn1eqqwtPK/3LcCX4iIa0qY2rHMGj5PBxeSlG0vBb4WEadn5vKylP1uquXvpcBzETGAakn8jrLryn4XAXcDUyJir8ycW85bHATMAYZGxODMfBT4cLtD/xz4KHBhRBwGLMrMxSU8r+oS4EfAnR3NhEbEXsDvM7MZEQdSheZN9oSl6XOnM/0ba7zwu0c7/vvH111CbUaOfJHM39VdRkutacw9OQz/mXnzYPjwjrcZMqQ1tUjaKLrkauLMnE91gcntVMHvvsy8sbRfAPwK+AnVRRsrfYfqXL/fUC05Ly19/RiYBtxTlnM/W7a/Avh2RNwfEVutoZTPAwuBhyPit8APgYWZOYtqCfkhqgtF7mq3z1RgekTcnpkLqc4D/F5EPEAVFPcty8FnAj+OiF8AC4Dnyv4XACPL9l+kWn5e0/t0L1VAvnxN2xTHAr8t458CfDgzvaJDkiRtVI1m07zRGRHRr1yd3KAKa49k5kXr2MduVDOV+2ZmV1++2WxM7o03T4PmpOp7uLePv+MzOnqy3jz+MvZeeeNEqpM+1du19Ju/MbnRsm+65qRmLf+w/eSTzhtfZvAeAranukq50yLiY1RL7hM3QiiUJEnaYF11jmFLRcQPgFVPZDk3MzfaWdBldnCdZghX2f8q/vxWNivv8fjJVTa9KzM/sb7HkSRJWl/dMhhm5gfqrqErZOblrP18Q0mSpJZwKVmSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJEmFwVCSJEmAwVCSJElF37oLUNdpTmrWXUKtevv4wfH3Ws1ePHZJXcoZQ0mSJAHOGPYojcmNukuoxeszhb1z/Ctnynr71783jr83jx0cv6sk2hicMZQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMJQkSRJgMJQkSVJhMNQGufT9l7Lgswt48OMP1l1Ky82f35cTThjEmDF7ctRRe3LllTvUXVLLvfdN72XOJ+bwyNmPcO4h59ZdTku8P97PrDNmMfP0mcwYP4NDdj/ktdemf3Q6z5z7DD867kc1VrjxHP/W45l1xixmnTGLu065i2EDhr32Wm/7WdDReyF1Z33rLqAuEbECeJDqPZgNnJiZL2xgn4OBmzJz/w2vsHu44v4r+MZvvsFVH7iq7lJark+fJuedt5D99nuJJUsaHHvsnhxyyAvstdfyuktric0amzHlyCkc8d0jaFvcxozxM5iW05i9aHbdpW1UP/3DT5mW0wB4a/+3cv0Hr+fNU94MwL/98t/YevOtOf2vT6+zxI1m3jPzOPSKQ3n2xWd5317vY+rRU3nbpW8Det/Pgo7eC6k7680zhssyc3gJccuBMzq7Y0T02kC9qjsfu5Onlz1ddxm16N9/Bfvt9xIA/fo1GTp0OQsW9J5vjYMGHsTcp+cy79l5vPzqy1z70LWM23dc3WVtdEtfXvra423esA3NZvO15z+b9zOef+n5OspqiV+1/YpnX3wWgLvb7mbQdoNee623/Szo6L2QurPe81usY3cCw1ad8YuIzwL9MvOCiLgD+CVwCDAtIv4T+DYwtPTxceBJoE9EfAf4G+AJYFxmLouI8cBpwBuAucAJmflCRHwQmASsAJ7LzHdGRB/gi8BhwBbAlMy8eGO/CVp/bW19mT17Cw444MW6S2mZgdsO5PHFj7/2vG1xGwcPPLjGilrnmH2P4V/e/S/036Y/R11zVN3l1OLUEacyfe70usvYJPheqCfpzTOGwGuzf2OolpXXZofMPDQz/x34GvA/mXkAcCDwUNlmb6ogtx/wLHBsaf9+Zo4q288GTi3tXwDeW9rfX9pOpQqJo4BRwPiIGLJBA9VGs3RpgwkTduP88xfSr9+rdZfTMo1G4y/amjRXs2XP88M5P+TNU97MMdcew4WHX1h3OS132ODDOHXEqZz7k95xXmlHfC/U0/TmGcOtIuL+8vhO4FJgt7Xsc127x+8CPgaQmSuA5yLir4B5mbmy33uBweXx/hHx/wE7AP2AW0r7XcAVEXE98P3SNppqBvNvy/PtqQLnvHUaoTa6l1+GCRN2Y+zYxYwevaTuclqqbXEbu2+3+2vPB203iCeff7LGijaeM0edyfgDxwNw5NVHMn/JfKBaPn3TX72JnbbaiaeWPVVniRvNqmPfeeuduWTsJYy5ekyvWjoG3wv1Dr05GC7LzOHtGyLiFf58FnXLVfZZytq91O7xCmCr8vgK4JjMnBURJ1EtE5OZZ0TEwcBRwP0RMRxoAGdn5i1ok9VswsSJuzB06P9t786jLCvrc49/i2ZqhmZSLyqjDI8XEJmjUZm9BhmvuARFQDAoiEDkLldERAZZ0SDEIAQMBBAHTBxQJJEpyKwgNCBN1OciRC7ghMo8d1P3j72LPjRFt3Wo2i+93+ezVi3O2ftU+/yqyqrfefd+3/dp9tvvwdJxOnfjfTeyzkrrsMbya3Dfw/ex5/p78r7z31c61pQ47cbTOO3G0wBYa4W1nju+8cobs/i0xXvbFMLza191xqqcv8f57P3dvbnjT3cUTta9fC2iBjU3huP5HfAqSSsBjwI7ARe/yGsvp7mv8B/bewKXXsC/vSzwG0mLAXvR3H+IpLVs3wDcIGlnYFWa0cSDJP3Q9jOS1gXus/3nNKadOu9d57H1GlvziqVewT0fu4ejrzyas285u3SsTsycuSQXXDCDddd9il13XQ2Aww//I1tt9bL7Nk2JOaNz+OgPPsol77+EaSPTOPvWs/nZ/T8rHWvK7b7e7uyz4T488+wzPPHME+zx7T2eO3f1B67m9a94Pcssvgz3fOwePvj9D3LpnZcWTDu5Pr3Vp1lp+kqctmPTHM1+djabn7k5UN/vgvl9LSIWZmkMB7RN2HHADTSXbX8xn5cfBpwh6YM0I4MHAb+Zz+uPav/du2nuZ1y2Pf55SevQjBJeDvwUuI3mEvTNkkaA+4HdhixrSvV1hOjPsdlmT2L/39Ixirrolxdx0al13XR/wnUncMJ1J4x7bssvb9lxmm4dcOEBHHDhAeOeq+13wfy+FhELs5HBpRZioTY6cuwLJwPUYPTosZ/hOuunnfBR+/e/xvprrh1S/9zffVXr9Js/cuxIZ1/00aNHi/xgVz8rOSIiIiIaaQwjIiIiAkhjGBERERGtNIYRERERAWRWckRERESvSPo8sDPwNHAnsJ/tB9utf38OuH3p9bYPHPzcNIYRERER/XIZcITt2ZL+HjgCGNu38c55N/gYlMYwIiIiokdsD66sfz3w7hd77bzSGEZERES8zEj6EPChgUNn2D5jiH9qf+DfBp6vKekW4GHgU7avGXxxGsOIiIiIl5m2CXzRRlDSfwIrj3PqSNsXtK85EpgNfL099xtgNdt/lLQp8D1J69t+eOyT0xhGRERELGRsbz+/85L2BXYCtrM92n7OU8BT7eOZku4E1gVuGvu8LFcTERER0SOS/opmsskuth8fOP5KSdPax68D1gHuGvzcjBhGRERE9MupwBLAZZJg7rI0WwLHSZoNzAEOtP2nwU9MYxgRERHRI7bXfpHj3wG+M7/PzaXkiIiIiADSGEZEREREK41hRERERABpDCMiIiKilcYwIiIiIoA0hhERERHRSmMYEREREUAaw4iIiIhopTGMiIiICCCNYURERES00hhGREREBJDGMCIiIiJaaQwjIiIiAkhjGBERERGtNIYRERERAaQxjIiIiIhWGsOIiIiIANIYRkREREQrjWFEREREAGkMIyIiIqI1Mjo6WjpDTI58IyMiojYjnf6PHTvS2d/a0aNHO61tTBrDiIiIiAByKTkiIiIiWmkMIyIiIgJIYxgRERERrTSGEREREQGkMYyIiIiIVhrDiIiIiADSGEZEREREK41hRERERABpDCMiJo2kZUpniG5JWrF0hhIkrS5p+/bxdEnLls4Uk2PR0gFi4SRpG+AQQO2hnwOn2r6yWKgOSVoX+DiwOgP/P7K9bbFQHZN0GHAO8AjwL8DGwCdsX1o0WFk/A1YrHWIqSXoDcCbwWuAi4G9tP9Ce+4ntLUrmm0qS3kLzs/4ssD9wPLCWpMWA99j+ccl8XZF0APAhYEVgLWAV4EvAdiVzxeRIYxgTJmlH4FTgOOBYmr0qNwHOlvRR2z8oma8j36L5RXgmMKdwllL2t32ypHcArwT2o2kUe90YSjr8RU6NADWMGJ4OHANcD/w1cK2kXWzfCSxWMlgHvgC8h+b7/B/AbravlbQJcArwlpLhOnQwsAVwA4DtOyS9qmykmCxpDGMYH6f5hfjTgWO3SrqJ5pdjDY3hbNunlw5R2NgG7+8EzrH9U0lFNn3v2N8Bnwdmj3OuhttzlrF9cfv4REkzgYsl7Q2MFszVhcVszwKQdL/tawFs3yxpetlonXrK9tNSc8FI0qL0/3tfjTSGMYyV52kKAbB9m6T/USJQARdK+gjwXeCpsYO2/1QuUudmSroUWBM4or3H6NnCmbpwM/A92zPnPSHprwvk6dqIpOVsPwRg+wpJuwPfobm02GeDjf8R85xbvMsghV0l6ZPAdElvBz4CXFg4U0ySGt7dxuR7bMhzfbIvzcjpj4CZ7cdNRRN174PAJ4DNbT9O84dxv7KROrEfcPeLnNusyyCF/D3wPwcP2L6N5v6y84sk6s5RkpYCsP29sYOS1gK+UixV9z4B3A/MAj5Mc5XoU0UTxaQZGR3N6G9MjKQHgavHOTUCvNX2Ch1HigIkbTnecdvj/WxUR9Iptg8pnaOUmuuvqfZ2VvYq7ZuD6IFcSo5h7Dqfcyd2lqKgdhbiQcBYc3Ql8M+2nykWqnsfH3i8JM3N6DOBamZmL0AtExFeTM3197p2SVcCu9D0ELcC90u6yvaLTcyKhUgaw5gw21eVzvAycDrNDMzT2ud7t8dquMcMANs7Dz6XtCpwQqE4EdGd5Ww/3N5Te47toyVlxLAn0hjGhEmaxXxmoNnesMM4pWxu+40Dz38o6QUTcipzL7BB6RARMeUWlfRqmqV7jiwdJiZXGsMYxk6lA7wMzJG0Vrt2G5JeR2XrGUo6hblvEBYBNgJqb44H1bB0z/zUXH/faz8OuAS41vaN7e+/OwpnikmSxjAmzPaLzch8Hkk/tv3mqc5TyMeBKyTdRfNHYHXqmJE7aHAW9mzgG7avKxWma5I2sH37fF5ycmdhCqi5/pprB7D9LZpF/see3wXsXi5RTKbMSo4pI+kW2xuXzjFVJC1BsyXgCPAL208t4FN6RdJhtk9e0LG+knQtzRI9XwbOs/1g2UTdqrn+mmsHkLQkzXJV69NMPAPA9v7FQsWkyTqGMZV6965D0rbtf98F7AisTbNX6I7tsZrsO86xD3QdohTbbwX2AlYFbpJ0XrvYbxVqrr/m2ltfBVYG3gFcRbNX8iNFE8WkyYhhTBlJN9vepHSOySTp2HYG3jnjnB6t4R2zpPcC7wPeClwzcGpZYI7t7YsEK0TSNGA34IvAwzQjyJ+03ffFnoG666+19rGrQZJus71hu3zXJbazVFUP5B7DmEq9uwHb9tHtw+Ns//fgOUlrFohUwo+A3wCvAE4aOP4IUM2SFZI2pLmvdEfgMmDnds/c1wA/pue7gNRcf821t8bWa31Q0gbAb4E1ysWJyZTGMIbSvlO+ZAGjQ3t3laeA7wDzjoZ+G9i0QJZOtZOP7gb6OrHoz3UqcCbNCNETYwdt/1pSDduD1Vx/zbUDnCFpBeAo4PvAMsCny0aKyZJLyTE0Sd8H9rb9UOksXZH0epobrk/g+Tt/zAA+bnv9IsEKkPQm4BSafXMXB6YBj9meUTRYB9o3Rl+xvVfpLCXUXH/NtUcdMmIYL8WTwCxJlwGPjR20fWi5SFNONOs4Lg8M7vzxCHBAkUTlnArsSbNsxWbAPjSTcXrP9hxJK0la3PbTpfN0reb6a659TLsiw+40l4+f6yNsH1cqU0yeNIbxUvxH+1EN2xcAF0h6s+0fl85Tmu1fSppmew5wjqQflc7UobuB69qR88E3Rv9QLlKnaq6/5toBLgAeotkbvaplumqQxjCGZvtcSdOB1Wy7dJ6O3SLpYOpex+txSYsDt0o6gWZCytKFM3Xp1+3HIjQzsmtTc/011w6wiu2/Kh0ipkbuMYyhSdoZOBFY3Paakjaima27S+FoU07St4Bf0CzbchzNmmY/t31Y0WAdkrQ68Dua+ws/BiwH/NPYNoG1kLQszVJFj5bOUkLN9ddau6QzgFNszyqdJSZfGsMYmqSZwLbAlWM7nEiaZfsNZZNNvazjlZ1P2mU6vgqs2B76A7CP7f8ql6o7Nddfa+2SZtFsXLAosA5wF82l5BGaBnnDgvFikuRScrwUs20/JGnwWC3vNLKOV7PzybxN4Ng/FA0AAA2iSURBVAfGOdZXZwCH274CQNLWNEuY/GXJUB2quf5aa9+pdICYemkM46W4XdL7gGmS1gEOpVn8uAbjreN1VNlI3RjY+WTN9ub7MTOAP5ZJVcTSY40BgO0rJdV0j2XN9VdZe7uG6dhSVf9l+5H2+bLAejSTcmIhl8YwXopDgCNpLiV8A7gE+EzRRB2x/S/tw6uA15XMUkB2PmncJekomkuKAO8H/ns+r++bmuuvuXaA03n+Av+PjXMsFlJpDGNoth+naQyPbBd9Xdr2k4VjdULSSsAxwFtoLp9fA3zGdu9HzObd+aT9WmwJPGp7dslsHdsfOJZm+7MR4GqabdJqUXP9NdcOMGL7uduGbD8rKf1ET2TySQxN0nnAgcAcmvWslgP+wfbniwbrQLuo99XA19pDewFbL2CLwF6Q9O/AJ2zfLunVwM3ATcBawBm2/7FowIiYUpLOB66kGSUE+Aiwje3dioWKSZMOP16K9Ww/LGkv4AfA39I0iL1vDIEVbQ9eNj9eUi2/FNe0fXv7eD/gMtv7tPcZXQdU0RhKupAXTrZ6iKZJ/ue+j57XXH/NtbcOBL4IfIrm63A58KGiiWLSpDGMl2KxdpmW3YBTbT8jqZYh6Csk7Ql8s33+burZBeaZgcfb0czGxPYjkp4tE6mIu4BX0txfC7AHzbqO69J8TfYulKsrNddfbe3tbUN72d6zdJaYGmkM46X4Es0N17cBV7cLHj9cNlJnPgwcztxLyYsAj0k6nGY9rxnFkk29eyQdAtxLc7P5xQDtLjiLlQzWsY1tbznw/EJJV9veUlKv17Nr1Vx/tbW3e0XvCnyhdJaYGmkMY8La5mfMF2guJbwfuBbYpkiojtmucRusMR+k2e1le2AP2w+2x98EnFMsVfdeKWk12/8PQNJqNDO1AZ4uF6szNddfc+3Q7BN9KvBvPH+v6JvLRYrJksYwhjFeU7Q6zQzlY4B/7TRNAZK2HO+47au7ztI127+nucdo3uNXAM+t7SbpFNuHdJmtY/8HuFbSnTQzU9cEPtKuZ3du0WTdqLn+mmuHuQt5HzdwbJRmJ6xYyGVWckwaSSsC/2m792tZtTefj1kS2AKYWdOWeAsi6ea+/yxIWgJ4PU1z8IsKJh08T83111x79FtGDGPS2P6TpJHSObpge+fB55JWBU4oFCcKkLQUzX2mq9s+QNI6kmT730tn60LN9ddcO4CkT4933PZx4x2PhcsipQNEf0jaFnigdI5C7gU2KB0iOnUOzf1kb26f3wscXy5O52quv+baobmvcOxjDrAD9e0V31sZMYwJkzSLF67htSLwa2Cf7hN1T9IpzP0aLAJsBPy0XKKXpb6PHq9le49272hsP1HLiHmr5vprrh3bg1thIulEmj3jowfSGMYwdprn+SjwR9uPjffinrpp4PFs4Bu2rysV5mXq5NIBptjT7RI9owCS1qLZN7wWNddfc+3jWYr69ozvrTSGMWHtXrm1+zbwpO050Cz6Kmmpdv/oXnuRXR+eY3uX9r9f7ipTIcfQrOG4qqSv0+ybXdN+ucdQb/3HUG/t8141mkaz2HfuL+yJzEqOGIKk64HtbT/aPl8GuNT2X87/Mxd+kraa33nbV3WVpTRJK9Gs3zgCXG/7D4Ujdarm+iuvffWBp7OB39meXSpPTK40hhFDkHSr7Y0WdCz6S9Lltrdb0LG+qrn+WmuXtCTNGqZrA7OAs9IQ9k8uJUcM5zFJm4yt9C9pU+CJwpk6JWkd4LPAejRrOQJgu9f3GrV/HJcCXiFpBeZOspkBvKZYsI7UXH/NtbfOpdkr/RqamcjrAYcVTRSTLo1hxHD+BviWpF+3z18N7FEwTwnnAEfTbIu4Dc09VjXMzPwwzff/NcBM5tb8MPBPpUJ1qOb6a64dYD3bbwCQdBbwk8J5YgrkUnLEkCQtBoi5Ox88UzhSpyTNtL2ppFkDfyyusf220tm6IOkQ26eUzlFKzfXXWvu8uxnVsLtRjdIYRgxB0sHA120/2D5fAXiv7dPKJuuOpOuAt9HM0P4hcB/wOdsqGqxDkjbghZfSv1IuUbdqrr/G2iXNoVnUGpo3xNOBx9vHo7ZnlMoWkyeXkiOGc4Dt5y4d2X5A0gFANY0hzSW1pYBDgc/QXE6uYoFzAElHA1vTNAc/oLnn6lqg183BmJrrr7V229NKZ4iply3xIoazyOBOB5KmAYsXzFPCGrYftX2v7f1s7w6sVjpUh94NbAf81vZ+wBuBJcpG6lTN9ddce/RcGsOI4VwKfFPSdu0e0f9Ks+BtTY74M4/11RO2nwVmS5oB/J66dn+ouf6aa4+ey6XkiOEcBRxAs6bXCE2jeFbRRB2RtAPwTuC1kr44cGoGzWK3tbhJ0vLAmTQzVB+lrlmaNddfc+3Rc5l8EjEBkhYF/o5maZZ7aJrCVWmWbvlkDTOTJb0R2IhmC6xPD5x6BLjC9gNFghUkaQ1ghu3bSmcpoeb6a649+imNYcQESPoCsCzwMduPtMeWBU6iubxUzWKvkhatcdcDSe8AlrX97XmO7wX83vZlZZJ1o+b6a6496pFLyRETsxOwru3n3lHZfkTSQcAvqGAXAEnftP0e4BZJL3hnaXvDArG6dCyw8zjHLwe+C/S9Oai5/pprj0pk8knExIwONoVjbM8Bahl+H2t+d6L5IznvR98tZfv+eQ/a/i2wdIE8Xau5/pprj0qkMYyYmJ9JesFafZLeTzNi2Hu2f9Muz3OW7bvn/SidrwNLtveaPk+7E870Anm6VnP9NdcelUhjGDExBwMHS7pS0kmSTpR0Fc0izwcVztaZdoT0cUnLlc5SwPnAmZKeGyFqH3+pPdd3Nddfc+1RidxjGDEBtu8D/qJdu3B9mlnJF9m+vGyyIp4EZkm6jLnbZGH70HKROvEp4HjgbkljI6Sr0SxXdFSxVN2puf6aa49KZFZyRAxF0r7jHbd9btdZSpA0HVi7ffpL20/Mc/7tfZ6lWnP9Ndce/ZfGMCJiCki62fYmpXOUUnP9NdceC79cSo6IoUhaB/gssB6w5Nhx29karDGy4Jf0Ws3111x7LOQy+SQihnUOcDrNNnjbAF8Bvlo00ctL7Zdjaq6/5tpjIZfGMCKGNb2ddDPSLlVzDLBt4UwREfESpDGMiGE9KWkR4A5JH5X0v4FXlQ7VFUlLLODYr7pL072a66+59ui/NIYRMay/AZaiWcNxU2BvYNyZyj314/kds/2uDrOUUHP9NdcePZfJJxExFNs3tg8fBfYrmaVLklYGXgtMl7QxcycazKBplHut5vprrj3qkcYwIiZE0vfnd972Ll1lKeQdwAeAVYCTmNscPAx8slCmLtVcf821RyWyjmFETIik+4F7gG8ANzDP0hy2ryqRq2uSdrf9ndI5Sqm5/pprj/7LPYYRMVEr04yObACcDLwd+IPtq2ppClubSlp+7ImkFSQdXzJQx2quv+bao+fSGEbEhNieY/ti2/sCbwJ+CVwp6ZDC0bq2g+0Hx57YfgB4Z8E8Xau5/pprj55LYxgREyZpCUnvAr4GHAx8ETi/bKrOTRtcoqTdP/cFy5j0WM3111x79Fwmn0TEhEg6l+Yy8kXAsbZvLxyplK8Bl0s6h2ani/2Bc8tG6lTN9ddce/RcJp9ExIRIehZ4rH06+AtkBBi1PaP7VGVI2gHYjqb2S21fUjhSp2quv+bao9/SGEZEREQEkEvJERFDkfQIc0dMFwcWAx6rZcS05vprrj36L41hRMQQbC87+FzSbsAWheJ0rub6a649+i+zkiMiJoHt7wHbls5RSs3111x79E9GDCMihtAu1zNmEWAznj8Zp9dqrr/m2qP/0hhGRAxn54HHs4FfAbuWiVJEzfXXXHv0XGYlR0RERASQEcOIiAmRdArzuWxo+9AO43Su5vprrj3qkcknERETcxMwE1gS2AS4o/3YCJhTMFdXaq6/5tqjErmUHBExBElXAP/L9jPt88VodsDYpmyybtRcf821R/9lxDAiYjivAQbXs1umPVaLmuuvufboudxjGBExnM8Bt7SjRwBbAceUi9O5muuvufbouVxKjogYkqSVgb+gmZDwE9u/LRypUzXXX3Pt0W8ZMYyIGN4WwNvax6PAhQWzlFBz/TXXHj2WEcOIiCFI+hywOfD19tB7gZtsH1EuVXdqrr/m2qP/MmIYETGcdwIb2X4WQNK5wC1ALc1BzfXXXHv0XGYlR0QMb/mBx8sVS1FOzfXXXHv0WEYMIyKG81nmzkwdAbakrhGjmuuvufboudxjGBExQZJGgFWA2TT3mo0AN9QyM7Xm+muuPeqQxjAiYgiSZtretHSOUmquv+bao/9yj2FExHCul7R56RAF1Vx/zbVHz2XEMCJiCJJ+Bgj4FfAYzSXFUdsblszVlZrrr7n26L9MPomIGM4OpQMUVnP9NdcePZfGMCJiAiQtCRwIrA3MAs6yPbtsqu7UXH/NtUc9co9hRMTEnAtsRtMY7ACcVDZO52quv+baoxIZMYyImJj1bL8BQNJZwE8K5+lazfXXXHtUIiOGERET88zYg0ovI9Zcf821RyUyKzkiYgIkzaGZiQrNbNTpwOPMnZk6o1S2LtRcf821Rz3SGEZEREQEkEvJEREREdFKYxgRERERQBrDiIiIiGilMYyIiIgIII1hRERERLT+P0MUGBJMUHuMAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "mask=np.zeros_like(df.corr())\n",
    "mask[np.triu_indices_from(mask)] = True\n",
    "plt.figure(figsize=(10,10))\n",
    "with sns.axes_style(\"white\"):\n",
    "    ax = sns.heatmap(df.corr()*100, mask=mask, fmt='.0f', annot=True, lw=1, cmap=ListedColormap(['green', 'yellow', 'red','blue']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Dropping the duplicates (De-duplication)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(537577, 12)\n"
     ]
    }
   ],
   "source": [
    "df = df.drop_duplicates()\n",
    "print( df.shape )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {},
   "outputs": [],
   "source": [
    "# It looks like we didn't have any duplicates in our original dataset. \n",
    "# Even so, it's a good idea to check this as an easy first step for cleaning your dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fix structural errors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The basement feature has some nan values, to handle them:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([nan, 14., 17.,  5.,  4., 16., 15.,  8.,  9., 13.,  6., 12.,  3.,\n",
       "       18., 11., 10.])"
      ]
     },
     "execution_count": 170,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Product_Category_3.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5.295546498455105"
      ]
     },
     "execution_count": 173,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Product_Category_1.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Generating mean for all Product categories:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean_cat1 = round(df.Product_Category_1.mean())\n",
    "mean_cat2 = round(df.Product_Category_2.mean())\n",
    "mean_cat3 = round(df.Product_Category_3.mean())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Replacing NaNs in all product categories with their means:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.Product_Category_2.fillna(mean_cat1, inplace=True)\n",
    "df.Product_Category_1.fillna(mean_cat2, inplace=True)\n",
    "df.Product_Category_3.fillna(mean_cat3, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Typos and capitalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f3d4cb470>"
      ]
     },
     "execution_count": 178,
     "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": [
    "# Class distributions for 'roof'\n",
    "sns.countplot(y='City_Category', data=df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Mislabeled classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Finally, we'll check for classes that are labeled as separate classes when they should really be the same.\n",
    "\n",
    "# e.g. If 'N/A' and 'Not Applicable' appear as two separate classes, we should combine them.\n",
    "# let's plot the class distributions for 'exterior_walls':"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f3d9918d0>"
      ]
     },
     "execution_count": 180,
     "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": [
    "sns.countplot(y='Gender', data=df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Removing Outliers\n",
    "Outliers can cause problems with certain types of models.\n",
    "Boxplots are a nice way to detect outliers\n",
    "Let's start with a box plot of your target variable, since that's what you're actually trying to predict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f43460c50>"
      ]
     },
     "execution_count": 181,
     "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": [
    "sns.boxplot(df.Purchase)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Interpretation\n",
    "\n",
    "The two vertical bars on the ends are the minimum and maximum values. The box in the middle is has data ranging from\n",
    "25th percentile to 75th percentile. The vertical bar in the middle is the median. Half of all observations fall in the box.\n",
    "Since there are some outliers to the end of the boxplot, they have to be removed from the data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Removing the outliers\n",
    "Purchases of more than \\\\$21000 seem to have outliers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f43438be0>"
      ]
     },
     "execution_count": 182,
     "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": [
    "# setting the filter for Purchase to 21500, and decreasing from there\n",
    "dg = df[df['Purchase']<21307]\n",
    "# there is no out lier below 21307\n",
    "sns.boxplot(dg.Purchase)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12f433b9c18>"
      ]
     },
     "execution_count": 183,
     "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": [
    "## Checking outliers in size of the house\n",
    "sns.boxplot(dg.User_ID)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Label missing categorical data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Product_ID                    0\n",
       "Gender                        0\n",
       "Age                           0\n",
       "City_Category                 0\n",
       "Stay_In_Current_City_Years    0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 184,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display number of missing values by categorical feature\n",
    "dg.select_dtypes(include=['object']).isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Interpretation\n",
    "No values in any of the categorical data seem to be misssing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Flag and fill missing numeric data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "User_ID               0\n",
       "Occupation            0\n",
       "Marital_Status        0\n",
       "Product_Category_1    0\n",
       "Product_Category_2    0\n",
       "Product_Category_3    0\n",
       "Purchase              0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 185,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display number of missing values by numeric feature\n",
    "dg.select_dtypes(exclude=['object']).isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observation\n",
    "Well, it looks like we don't have any numerical features with missing values in this dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Before we move on to the next module, let's save the new dataframe we worked hard to clean."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This makes sure we don't have to re-do all the cleaning after clossing the session"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save cleaned dataframe to new file\n",
    "dg.to_csv(r'cleaneddf.csv', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Encode dummy variables (One Hot Encoding)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Machine learning algorithms cannot directly handle categorical features. Specifically, they cannot handle text values.\n",
    "Therefore, we need to create dummy variables for our categorical features.\n",
    "Dummy variables are a set of binary (0 or 1) features that each represent a single class from a categorical feature."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "dh = dg.drop(['User_ID'], axis = 1) #Remove an unimportant categorial column for machine learning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a new dataframe with dummy variables for for our categorical features.\n",
    "dh = pd.get_dummies(dh, columns=['Occupation', 'Marital_Status'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save analytical base table\n",
    "dh.to_csv(r'analyticaldf.csv', index=None)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Machine Learning Models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "dh = pd.read_csv(\"analyticaldf.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(534777, 32)"
      ]
     },
     "execution_count": 198,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dh.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train and Test Splits"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Separate your dataframe into separate objects for the target variable (y) \n",
    "and the input features (X) and perform the train and test split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create separate object for target variable\n",
    "y = dh.Purchase\n",
    "# Create separate object for input features\n",
    "X = dh.drop('Purchase', axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Split X and y into train and test sets: 80-20\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=1234)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's confirm we have the right number of observations in each subset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(427821, 31) (106956, 31) (427821,) (106956,)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape, X_test.shape, y_train.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data standardization\n",
    "In Data Standardization we perform zero mean centring and unit scaling; i.e. we make the mean of all the features as zero and the standard deviation as 1. hus we use mean and standard deviation of each feature. It is very important to save the mean and standard deviation for each of the feature from the training set, because we use the same mean and standard deviation in the test set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_mean = X_train.mean()\n",
    "train_std = X_train.std()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Standardize the train data set\n",
    "X_train = (X_train - train_mean) / train_std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Check for mean and std dev.\n",
    "X_train.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Note: We use train_mean and train_std_dev to standardize test data set\n",
    "X_test = (X_test - train_mean) / train_std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Check for mean and std dev. - not exactly 0 and 1\n",
    "X_test.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model 1 - Baseline Model\n",
    "In this model, for every test data point, we will simply predict the average of the train labels as the output.\n",
    "We will use this simple model to perform hypothesis testing for other complex models."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Train results\n",
    "y_train_pred = np.ones(y_train.shape[0])*y_train.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Test results\n",
    "y_pred = np.ones(y_test.shape[0])*y_train.mean()\n",
    "from sklearn.metrics import r2_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Train Results for Baseline Model:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_train.values, y_train_pred)))\n",
    "print(\"R-squared: \", r2_score(y_train.values, y_train_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_train.values, y_train_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Results for Baseline Model:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_test, y_pred)))\n",
    "print(\"R-squared: \", r2_score(y_test, y_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model-2 Ridge Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuned_params = {'alpha': [0.0001, 0.001, 0.01, 0.1, 1, 10, 100, 1000, 10000, 100000]}\n",
    "model = GridSearchCV(Ridge(), tuned_params, scoring = 'neg_mean_absolute_error', cv=10, n_jobs=-1)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Train results\n",
    "y_train_pred = model.predict(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Test results\n",
    "y_pred = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Train Results for Ridge Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_train.values, y_train_pred)))\n",
    "print(\"R-squared: \", r2_score(y_train.values, y_train_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_train.values, y_train_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Test Results for Ridge Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_test, y_pred)))\n",
    "print(\"R-squared: \", r2_score(y_test, y_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature Importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Building the model again with the best hyperparameters\n",
    "model = Ridge(alpha=100)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "indices = np.argsort(-abs(model.coef_))\n",
    "print(\"The features in order of importance are:\")\n",
    "print(50*'-')\n",
    "for feature in X.columns[indices]:\n",
    "    print(feature)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model-3 Support Vector Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuned_params = {'C': [0.0001, 0.001, 0.01, 0.1, 1, 10, 100, 1000, 10000, 100000], 'gamma': [0.0001, 0.001, 0.01, 0.1, 1, 10, 100, 1000, 10000, 100000]}\n",
    "model = GridSearchCV(SVR(), tuned_params, scoring = 'neg_mean_absolute_error', cv=5, n_jobs=-1)\n",
    "model.fit(X_train, y_train)\n",
    "## This takes around 20 minutes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Building the model again with the best hyperparameters\n",
    "model = SVR(C=100000, gamma=0.01)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Train results\n",
    "y_train_pred = model.predict(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Test results\n",
    "y_pred = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Train Results for Support Vector Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_train.values, y_train_pred)))\n",
    "print(\"R-squared: \", r2_score(y_train.values, y_train_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_train.values, y_train_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Test Results for Support Vector Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_test, y_pred)))\n",
    "print(\"R-squared: \", r2_score(y_test, y_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model-4 Random Forest Regression\n",
    "Reference for random search on random forest\n",
    "https://towardsdatascience.com/hyperparameter-tuning-the-random-forest-in-python-using-scikit-learn-28d2aa77dd74"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuned_params = {'n_estimators': [100, 200, 300, 400, 500], 'min_samples_split': [2, 5, 10], 'min_samples_leaf': [1, 2, 4]}\n",
    "model = RandomizedSearchCV(RandomForestRegressor(), tuned_params, n_iter=20, scoring = 'neg_mean_absolute_error', cv=5, n_jobs=-1)\n",
    "model.fit(X_train, y_train)\n",
    "## This takes around 15 minutes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Train results\n",
    "y_train_pred = model.predict(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Test results\n",
    "y_pred = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Train Results for Random Forest Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_train.values, y_train_pred)))\n",
    "print(\"R-squared: \", r2_score(y_train.values, y_train_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_train.values, y_train_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Test Results for Random Forest Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_test, y_pred)))\n",
    "print(\"R-squared: \", r2_score(y_test, y_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature Importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Building the model again with the best hyperparameters\n",
    "model = RandomForestRegressor(n_estimators=200, min_samples_split=10, min_samples_leaf=2)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "indices = np.argsort(-model.feature_importances_)\n",
    "print(\"The features in order of importance are:\")\n",
    "print(50*'-')\n",
    "for feature in X.columns[indices]:\n",
    "    print(feature)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model-5 XGBoost Regression\n",
    "Reference for random search on xgboost\n",
    "https://gist.github.com/wrwr/3f6b66bf4ee01bf48be965f60d14454d"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuned_params = {'max_depth': [1, 2, 3, 4, 5], 'learning_rate': [0.01, 0.05, 0.1], 'n_estimators': [100, 200, 300, 400, 500], 'reg_lambda': [0.001, 0.1, 1.0, 10.0, 100.0]}\n",
    "model = RandomizedSearchCV(XGBRegressor(), tuned_params, n_iter=20, scoring = 'neg_mean_absolute_error', cv=5, n_jobs=-1)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Train results\n",
    "y_train_pred = model.predict(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Test results\n",
    "y_pred = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Train Results for XGBoost Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_train.values, y_train_pred)))\n",
    "print(\"R-squared: \", rs(y_train.values, y_train_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_train.values, y_train_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Test Results for XGBoost Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_test, y_pred)))\n",
    "print(\"R-squared: \", r2_score(y_test, y_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature Importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Building the model again with the best hyperparameters\n",
    "model = XGBRegressor(max_depth=2,learning_rate=0.05,n_estimators=400, reg_lambda=0.001)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Function to include figsize parameter\n",
    "## Reference: https://stackoverflow.com/questions/40081888/xgboost-plot-importance-figure-size\n",
    "def my_plot_importance(booster, figsize, **kwargs): \n",
    "    from matplotlib import pyplot as plt\n",
    "    from xgboost import plot_importance\n",
    "    fig, ax = plt.subplots(1,1,figsize=figsize)\n",
    "    return plot_importance(booster=booster, ax=ax, **kwargs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "my_plot_importance(model, (10,10))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model-6 Lasso Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuned_params = {'alpha': [0.0001, 0.001, 0.01, 0.1, 1, 10, 100, 1000, 10000, 100000]}\n",
    "model = GridSearchCV(Lasso(), tuned_params, scoring = 'neg_mean_absolute_error', cv=20, n_jobs=-1)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Train results\n",
    "y_train_pred = model.predict(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Test results\n",
    "y_pred = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Train Results for Lasso Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_train.values, y_train_pred)))\n",
    "print(\"R-squared: \", rs(y_train.values, y_train_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_train.values, y_train_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Test Results for Lasso Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_test, y_pred)))\n",
    "print(\"R-squared: \", rs(y_test, y_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature Importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Building the model again with the best hyperparameters\n",
    "model = Lasso(alpha=1000)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "indices = np.argsort(-abs(model.coef_))\n",
    "print(\"The features in order of importance are:\")\n",
    "print(50*'-')\n",
    "for feature in X.columns[indices]:\n",
    "    print(feature)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model-7 Descision Tree Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuned_params = {'min_samples_split': [2, 3, 4, 5, 7], 'min_samples_leaf': [1, 2, 3, 4, 6], 'max_depth': [2, 3, 4, 5, 6, 7]}\n",
    "model = RandomizedSearchCV(DecisionTreeRegressor(), tuned_params, n_iter=20, scoring = 'neg_mean_absolute_error', cv=10, n_jobs=-1)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Train results\n",
    "y_train_pred = model.predict(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Test results\n",
    "y_pred = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Train Results for Decision Tree Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_train.values, y_train_pred)))\n",
    "print(\"R-squared: \", rs(y_train.values, y_train_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_train.values, y_train_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Test Results for Decision Tree Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_test, y_pred)))\n",
    "print(\"R-squared: \", rs(y_test, y_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model-8 KN Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# creating odd list of K for KNN\n",
    "neighbors = list(range(1,50,2))\n",
    "# empty list that will hold cv scores\n",
    "cv_scores = []\n",
    "\n",
    "# perform 10-fold cross validation\n",
    "for k in neighbors:\n",
    "    knn = KNeighborsRegressor(n_neighbors=k)\n",
    "    scores = cross_val_score(knn, X_train, y_train, cv=10, scoring='neg_mean_absolute_error')\n",
    "    cv_scores.append(scores.mean())\n",
    "\n",
    "# changing to misclassification error\n",
    "MSE = [1 - x for x in cv_scores]\n",
    "\n",
    "# determining best k\n",
    "optimal_k = neighbors[MSE.index(min(MSE))]\n",
    "print('\\nThe optimal number of neighbors is %d.' % optimal_k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = KNeighborsRegressor(n_neighbors = optimal_k)\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Train results\n",
    "y_train_pred = model.predict(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Predict Test results\n",
    "y_pred = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Train Results for KN Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_train.values, y_train_pred)))\n",
    "print(\"R-squared: \", rs(y_train.values, y_train_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_train.values, y_train_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Test Results for KN Regression:\")\n",
    "print(\"*******************************\")\n",
    "print(\"Root mean squared error: \", sqrt(mse(y_test, y_pred)))\n",
    "print(\"R-squared: \", rs(y_test, y_pred))\n",
    "print(\"Mean Absolute Error: \", mae(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Comparing all models and Saving the winning model to disk"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "win_model = RandomForestRegressor(n_estimators=200, min_samples_split=10, min_samples_leaf=2)\n",
    "win_model.fit(X_train, y_train)\n",
    "with open('rfr_real_estate.pkl', 'wb') as pickle_file:\n",
    "       joblib.dump(win_model, 'rfr_real_estate.pkl')"
   ]
  }
 ],
 "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
}