FinalProject_group031.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# COGS 108 - TRITON DRUNK"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Overview\n",
    "People love beers for different reasons. Some drink beers because beers contain two sleep-promoting \n",
    "B vitamins, making them terrific natural sleep aids while some just love the smell and taste of beer.\n",
    "Beers in the market vary in smell, taste, color, style, and ABV ( Alcohol by Volume). \n",
    "This project focus on how those properties influence their overall rating.\n",
    "\n",
    "We use the datasets from Beer Ratings - Predict the rating of a beer from user reviews (https://www.kaggle.com/c/beer-ratings/) to present the correlation between\n",
    "beer's properties and beer's overall rating. We also utilize python tool to draw some diagram so we can visualize the correlations.\n",
    "\n",
    "There is also an ethical problem involved in beer drinking. In many societies, drinking alcohol is forbidden because it causes serious problems, e.g. DUI (Driving under the influence), crime, etc. \n",
    "For that reason, in the second part of the project we will also focus on the correlation between alcohol and driving deaths.\n",
    "We will use data from Occupant and Alcohol-Impaired Driving Deaths in States (https://healthdata.gov/dataset/occupant-and-alcohol-impaired-driving-deaths-states-2005-2014)\n",
    "to present such correlation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Group Members\n",
    "* Toan Bui                  (XXXXXXXXX)\n",
    "* Hongru Yu                 (XXXXXXXXX)\n",
    "* Xiaokang (Paul) Lin       (XXXXXXXXX)\n",
    "* Haorong Chen              (XXXXXXXXX)\n",
    "* Niklas Sprute             (XXXXXXXXX)\n",
    "* Cong (Charlie) Li         (XXXXXXXXX)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Research Questions\n",
    "\n",
    "* Main Question: \n",
    "    * How do different properties of beer (i.e: taste, palate, aroma and appearance) affect its overall rating? In other words, which property contributes the most to the overall rating of a beer. Additionally, given user reviews, can we determine the sentiment of each review to predict the overall rating for a particular type of beer using NLP/Sentiment analysis?\n",
    "\n",
    "\n",
    "* Ethical Consideration:\n",
    "   *  Is there a correlation between the level of alcohol consumption and the number of DUI deaths within a particular state? In other words, if we were to assign to each state a 'State Alcohol-Addiction Index' (which signifies the level of alcohol dependency among the state population), then does a higher index correspond to a higher percentage of DUI deaths compared to all driving deaths? "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Background and Prior Work\n",
    "\n",
    "Whenever someone goes to a restaurant, they may struggle to order themselves the right type of beer. Because the process of making beer is very complex, every beer has its own unique taste. In order to analyze how different properties of beer shape its overall impression, an extensive beer survey has been conducted by consumer websites for advertisement purposes (https://www.kaggle.com/c/beer-ratings/). \n",
    "\n",
    "We decided to use this beer-ratings data to determine how each property (taste, palate, aroma and appearance) contributes to the overall rating of beer and how those correlations change among different beer types. At the end of our study, we aim to better inform consumers to choose the right beer or at least, when deciding which beer to drink, they can at least take into consideration specific characteristics of beers to guide them toward the best beer choice. \n",
    "\n",
    "However, due to our ethical considerations, the second part of our study aims to explore the possible correlation between alcohol consumption and DUI rates in different states. We obtained data from the Behavioral Risk Factor Surveillance System (BRFSS) which is the nation’s premier system of health-related telephone surveys that collect state data about U.S. residents regarding their health-related risk behaviors, chronic health conditions, and use of preventive services. \n",
    "\n",
    "There exists no prior research utilizing the specific beer-ratings dataset that was chosen for our analysis. \n",
    "\n",
    "The online survey data conducted by BRFSS will be used to produce an alcohol consumption rating on each state (which we called \"State Alcohol-Addiction Index\"). We will then explore the possible correlation between the consumption and DUI. With this part of our study, we aim to warn people about the danger of excessive drinking and what it can lead to afterwards.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Hypothesis\n",
    "\n",
    "People tend to enjoy a beverage for its taste instead of how it looks or smells. Take Coke for example: even though Coke has a dark color (instead of being colorful) and it also does not have much smell, people still enjoy drink Coke for its refreshing taste. \n",
    "\n",
    "Like Coke, beer is a type of beverage, and hence it is likely that people will also enjoy beer for its taste instead of its aroma or appearance. Hence, we predict that taste will be the feature that contributes the most to the overall rating of beer. \n",
    "\n",
    "To verify our hypothesis, we will take into consideration different properties of beer, from appearance, aroma, palate and taste to determine which feature correlates the most to the overall rating of beer. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dataset(s)\n",
    "\n",
    "* DATASET 1: Beer Ratings - Predict the rating of a beer from user reviews.\n",
    "    - Link: https://www.kaggle.com/c/beer-ratings/\n",
    "    \n",
    "    - For this dataset, we will use the data from appearance, aroma, overall, palate, taste to analyze the correlation between beer's characteristics and its overall rating. We will also use the review/text column to perform NLP analysis and predict beer rating from user's text reviews. \n",
    "    \n",
    "\n",
    "* DATASET 2: BRFSS Table of Alcohol Consumption\n",
    "    - Link: https://chronicdata.cdc.gov/Behavioral-Risk-Factors/BRFSS-Table-of-Alcohol-Consumption/dts9-xy2f\n",
    "    - For this dataset, we have the alcohol consumption data categorized by state, year and age. The data is obtained by random phone calls in the designated states asking questions to evaluate the degree of alcohol dependency of each individual, which is then categorized into 3 degrees (regular alcohol consumption, binge drinking and heavy drinking) The geographic location of the calls are also included. \n",
    "\n",
    "\n",
    "* DATASET 3: Occupant and Alcohol-Impaired Driving Deaths in States, 2005-2014\n",
    "\n",
    "    - Link: https://healthdata.gov/dataset/occupant-and-alcohol-impaired-driving-deaths-states-2005-2014\n",
    "    - This dataset contains data about all persons killed in crashes involving a driver with BAC >= .08 g/dL in different states. from 2005 to 2014\n",
    "    - We will use the Driving Deaths and Occupant Deaths columns to evaluate how DUI deaths contribute to total driving death (accidents)  \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Setup\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Imports - these provided for you. Do not import any other packages\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.metrics import mean_squared_error"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/khanhtoanbui/anaconda3/lib/python3.7/site-packages/IPython/core/interactiveshell.py:3020: DtypeWarning: Columns (13) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  interactivity=interactivity, compiler=compiler, result=result)\n"
     ]
    }
   ],
   "source": [
    "alc_consump_dat = pd.read_csv('data/BRFSS__Table_of_Alcohol_Consumption.csv')\n",
    "dui_dat = pd.read_csv('data/Occupant_and_Alcohol-Impaired_Driving_Deaths_in_States__2005-2014.csv')\n",
    "beer_dat = pd.read_csv('data/train.csv')\n",
    "beer_dat_NLP = pd.read_csv('data/train.csv')\n",
    "\n",
    "alc_consump_dat.drop(columns=['BreakoutID','LocationID','Break_Out_Category','GeoLocation','Locationdesc',\n",
    "                              'Class','ClassId','TopicId','Data_Value_Footnote_Symbol','DataSource',\n",
    "                              'Data_Value_Footnote','ResponseID', 'QuestionID', 'BreakOutCategoryID'], inplace=True)\n",
    "alc_consump_dat = alc_consump_dat.dropna()\n",
    "\n",
    "dui_dat.drop(columns = ['Location'],inplace=True)\n",
    "dui_dat = dui_dat.dropna()\n",
    "dui_dat.columns = dui_dat.columns.str.strip()\n",
    "\n",
    "beer_dat_NLP = beer_dat.drop(columns=['beer/ABV','beer/beerId','beer/brewerId','beer/name','index',\n",
    "                                      'user/ageInSeconds','user/gender','user/profileName','user/birthdayUnix',\n",
    "                                      'user/birthdayRaw','review/timeUnix', 'review/timeStruct']).dropna()\n",
    "beer_dat = beer_dat_NLP.drop(columns=['review/text'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Analysis and Results\n",
    "\n",
    "## Part 1: Analyzing the Correlation between Beer Characteristics and its Overall Rating"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The beer-review dataset contains 95 unique types of beer\n",
      "\n",
      "However, we will only consider the 5 most-reviewed types of beer: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NUMBER OF REVIEWS</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>American Double / Imperial Stout</th>\n",
       "      <td>4493</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>American IPA</th>\n",
       "      <td>3048</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>American Double / Imperial IPA</th>\n",
       "      <td>2871</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Scotch Ale / Wee Heavy</th>\n",
       "      <td>2133</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Russian Imperial Stout</th>\n",
       "      <td>2030</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                   NUMBER OF REVIEWS\n",
       "American Double / Imperial Stout                4493\n",
       "American IPA                                    3048\n",
       "American Double / Imperial IPA                  2871\n",
       "Scotch Ale / Wee Heavy                          2133\n",
       "Russian Imperial Stout                          2030"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"The beer-review dataset contains \" + str(len(beer_dat['beer/style'].unique())) + \" unique types of beer\\n\")\n",
    "print(\"However, we will only consider the 5 most-reviewed types of beer: \")\n",
    "\n",
    "top_5_beer = beer_dat['beer/style'].value_counts().head(5).to_frame()\n",
    "top_5_beer.columns = [' NUMBER OF REVIEWS']\n",
    "top_5_beer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "to_contain = 'American Double / Imperial Stout|American IPA|American Double / Imperial IPA|Scotch Ale / Wee Heavy|Russian Imperial Stout'\n",
    "#|American Pale Ale (APA)|American Porter|Rauchbier|Rye Beer|Czech Pilsener'\n",
    "df_top_5 = beer_dat[beer_dat['beer/style'].str.contains(to_contain , na=False)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "df_top_5 = df_top_5.sort_values(by=['beer/style'])\n",
    "df_top_5 = df_top_5.reset_index()\n",
    "df_top_5.drop('index', axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "beer_styles = ['American Double / Imperial Stout', 'American IPA', 'American Double / Imperial IPA','Scotch Ale / Wee Heavy','Russian Imperial Stout']\n",
    "beer_ratings = [0,0,0,0,0]\n",
    "\n",
    "appearance_rating_dict = dict(zip(beer_styles,beer_ratings))\n",
    "palate_rating_dict = dict(zip(beer_styles,beer_ratings))\n",
    "aroma_rating_dict = dict(zip(beer_styles,beer_ratings))\n",
    "taste_rating_dict = dict(zip(beer_styles,beer_ratings))\n",
    "overall_rating_dict = dict(zip(beer_styles,beer_ratings))\n",
    "                 \n",
    "for style in beer_styles:\n",
    "    appearance_rating_dict[style] = round(df_top_5[df_top_5['beer/style'] == style]['review/appearance'].mean(),2)\n",
    "    palate_rating_dict[style] = round(df_top_5[df_top_5['beer/style'] == style]['review/palate'].mean(),2)\n",
    "    aroma_rating_dict[style] = round(df_top_5[df_top_5['beer/style'] == style]['review/aroma'].mean(),2)\n",
    "    taste_rating_dict[style] = round(df_top_5[df_top_5['beer/style'] == style]['review/taste'].mean(),2)\n",
    "    overall_rating_dict[style] = round(df_top_5[df_top_5['beer/style'] == style]['review/overall'].mean(),2)\n",
    "\n",
    "appearance_rating_df = pd.Series(appearance_rating_dict).to_frame()\n",
    "palate_rating_df = pd.Series(palate_rating_dict).to_frame()\n",
    "aroma_rating_df = pd.Series(aroma_rating_dict).to_frame()\n",
    "taste_rating_df = pd.Series(taste_rating_dict).to_frame()\n",
    "overall_rating_df = pd.Series(overall_rating_dict).to_frame()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Average Rating for each Individual Feature of the 5 Most-Reviewed Beer Styles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "An Overview of the Ratings for the 5 Most-reviewed Beer Types\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Appearance Rating</th>\n",
       "      <th>Palate Rating</th>\n",
       "      <th>Aroma Rating</th>\n",
       "      <th>Taste Rating</th>\n",
       "      <th>Overall Rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>American Double / Imperial Stout</th>\n",
       "      <td>4.32</td>\n",
       "      <td>4.35</td>\n",
       "      <td>4.40</td>\n",
       "      <td>4.48</td>\n",
       "      <td>4.33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>American IPA</th>\n",
       "      <td>3.97</td>\n",
       "      <td>3.92</td>\n",
       "      <td>3.97</td>\n",
       "      <td>4.01</td>\n",
       "      <td>4.04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>American Double / Imperial IPA</th>\n",
       "      <td>3.99</td>\n",
       "      <td>3.98</td>\n",
       "      <td>4.07</td>\n",
       "      <td>4.03</td>\n",
       "      <td>3.87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Scotch Ale / Wee Heavy</th>\n",
       "      <td>3.92</td>\n",
       "      <td>3.96</td>\n",
       "      <td>3.99</td>\n",
       "      <td>4.08</td>\n",
       "      <td>3.92</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Russian Imperial Stout</th>\n",
       "      <td>4.21</td>\n",
       "      <td>4.30</td>\n",
       "      <td>4.17</td>\n",
       "      <td>4.31</td>\n",
       "      <td>4.17</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  Appearance Rating  Palate Rating  \\\n",
       "American Double / Imperial Stout               4.32           4.35   \n",
       "American IPA                                   3.97           3.92   \n",
       "American Double / Imperial IPA                 3.99           3.98   \n",
       "Scotch Ale / Wee Heavy                         3.92           3.96   \n",
       "Russian Imperial Stout                         4.21           4.30   \n",
       "\n",
       "                                  Aroma Rating  Taste Rating  Overall Rating  \n",
       "American Double / Imperial Stout          4.40          4.48            4.33  \n",
       "American IPA                              3.97          4.01            4.04  \n",
       "American Double / Imperial IPA            4.07          4.03            3.87  \n",
       "Scotch Ale / Wee Heavy                    3.99          4.08            3.92  \n",
       "Russian Imperial Stout                    4.17          4.31            4.17  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "appearance_rating_df.columns = ['Appearance Rating']\n",
    "palate_rating_df.columns = ['Palate Rating']\n",
    "aroma_rating_df.columns = ['Aroma Rating']\n",
    "taste_rating_df.columns = ['Taste Rating']\n",
    "overall_rating_df.columns = ['Overall Rating']\n",
    "\n",
    "ratings_df = pd.concat([appearance_rating_df, palate_rating_df], axis=1, join='inner')\n",
    "ratings_df = pd.concat([ratings_df, aroma_rating_df], axis=1, join='inner')\n",
    "ratings_df = pd.concat([ratings_df, taste_rating_df], axis=1, join='inner')\n",
    "ratings_df = pd.concat([ratings_df, overall_rating_df], axis=1, join='inner')\n",
    "\n",
    "print(\"\\nAn Overview of the Ratings for the 5 Most-reviewed Beer Types\")\n",
    "\n",
    "#The dataframe that contains all the individual ratings\n",
    "ratings_df ## THIS CONTAINS THE AVERAGE RATINGS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ratings_df.plot.bar(title = 'Overview of the Ratings for the 5 Most-reviewed Beer Types',figsize= (12,8))\n",
    "labels = [3.5, 4.0, 4.5, 5.0]\n",
    "plt.yticks(labels)\n",
    "plt.ylim(3.5,5)\n",
    "plt.xticks(rotation = 35)\n",
    "plt.xlabel(\"Beer Type\")\n",
    "plt.ylabel(\"Rating\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observation:\n",
    "From the bar chart above, we can see that \"American Double/Imperial Stout\" is the beer type that receives the highest review in all categories (appearance, palate, aroma, taste and overall)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Bar Chart Representing the Distribution of Overall Beer Rating (of the 5 Most-Reviewed Beer Types)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_top_5['review/overall'].value_counts().sort_index().plot('bar',title = 'Overall Rating Distribution',figsize= (10,8))\n",
    "plt.xticks(rotation = 0)\n",
    "plt.xlabel(\"Overall Rating\")\n",
    "plt.ylabel(\"Count\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observation:\n",
    "From the bar chart above, we can see that people tend to give an overall rating of 4.0 for the 5 most-popular beer styles. This means that the most-reviewed beers also tend to receive a relatively high overall rating."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_top_5.columns = ['Style', 'Appearance', 'Aroma', 'OVERALL', 'Palate', 'Taste']\n",
    "df_top_5 = df_top_5[['Style', 'Appearance', 'Aroma','Palate', 'Taste', 'OVERALL', ]]\n",
    "data = df_top_5.sample(frac=1)  ## random shuffle the dataframe ( not in place)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Correlation Between Individual Beer Characteristics and the Overall Rating\n",
    "\n",
    "Correlation Matrix: the darker the cell signifies that two attributes are more correlated. We want to investigate how each atrribute is correlated with the overall rating. We found strong positive correlation between overall rating and taste, which supports our hypothesis. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmQAAAJCCAYAAABu5NuXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzs3XecVOX5v/Hr3l0QEKTvooKiiB01GrErYE1sMXaJJRaMxphojCb5GjWaiJoYe0NFI9ZgorHFhhLsLVEQREVBQWkBNFSBnef3xwzLLizFX2TOLnO9fc3LU54z85yzs2fv+ZxnDpFSQpIkSdkpy7oDkiRJpc6CTJIkKWMWZJIkSRmzIJMkScqYBZkkSVLGLMgkSZIyZkFWRBFxSESkiNg0676oLn82q15EVEfE2xHxbkQMjogWK2g/LiI6rKDNr7/ZXjZ+EdG+cJzfjohJEfFZrfmmX+N5ToyITquyrw1ZRHSOiL9HxIcR8VFEXBMRrSNiWkS0XqLtwxFxREScEBFTax3vtyNi84joGhFzC/OjIuKuiGiyxHNcU/hZldVadkJEXF9P31b4u6HGx4KsuI4GXgSOyuLFI6Iii9dtJJb5s4mI8uJ3Z7U0N6W0TUppS2A+8KNv4DktyJaQUppWOM7bADcDVy2aTynN/xpPdSJQkgVZRATwN+DhlFJ3YGOgJXA+8DTwvVptWwO7Ao8VFj1Q63hvk1IaVVj+UeFn0gPoDBxR6znKgEOA8cDuq3Tn1GBZkBVJRLQEdgFOovBHPyJ6RcSwiHio8Knp5kWfjiJiVkRcGRH/ioghEdGxsLxbRDwZEW9FxAuLEp2IODAiXouIf0fEsxFRVVh+UUQMiIingbsKn9ReKDzvvyJi51p9GRoRD0bE6Ii4p3BSIiK2j4iXI+KdiHg9IlpFRHlE/CEi3oiI4RFxarGP6TdlOT+b5yPiXmBEYdnZhXTn3Yj4WWFZ18Lxuq2w/J6I2CsiXip8su5ZaNezcAz/Xfj/JtnsbYPxArAR1KQLb0XEyIjoV1/j+tpExGVA80LqcE9h2Q8K79G3I+IWi+m6IuLRWsfx5MKyiogYFBEjCu/hMyPiSGAb4IHCsWxaOA/8s7D9PxadY1ZTfYB5KaU7AFJK1cBZ5IvU+6j7we0Q4MmU0pyVeeLCc70OrFtrcW/gXeAm8h8OVYpSSj6K8AB+ANxemH4Z2BboBcwDNgTKgWeAwwptEtC3MH0BcH1hegjQvTC9A/BcYbotEIXpk4ErC9MXAW8BzQvzLYBmhenuwJuF6V7Al+Q/uZUBr5D/1NcU+BjYvtBuLaAC6AecX1i2BvAmsEHWx/kb/tnMXrRPwHbkC7M1yX9SHgl8C+gKLCT/qbescKwHAgEcTP4Tds1xK0zvBfw16/3O4DjPKvy/Avg7cFphvl3h/83J/1FqX5gfB3RYQZtZtZ5/M+BRoElh/kbguKz3O+NjfhFwTq35RcexBTCqcN7YAfhHrTZtCv9/EdimML1G4Xdj0c+jLzAg6/1bhcftTPLJ4pLL/w1sBUyp9R58Eti/MH0CMBV4u9ajeeE88W6hTTPgeWCrWs97G3Bs4TzxWa338AkUzv1L9KPmd8PH6vPwElbxHA1cXZi+vzD/OPB6SuljgIi4j3wR9CCQAx4otL8b+FshydkZGFwIryB/ooR8IfVARKxNvogaW+u1H0kpzS1MNwGuj4htgGryUfwir6eUJhT68jb5k8iXwMSU0hsAKaX/FtbvA2wVEYcVtm1NvsCr/bqNxfJ+Nov2Z1fgoZTSbICI+BuwG/AIMDaltChFGwkMSSmliBhB/hhC/vj8OSK6ky+264wfKRHNC+8ryCdktxemz4yIQwrTXci/j6Ytse3KtNmTfOH8RuH3ozn5P5xa7KyIOKgw3RnoBowBNomIa4AnyF+SW9JmwBbAs4VjWw5MWPXdzUyQ/z1d1vJHgMMi4q/kk8Tax+yBlNIZdTbKH7Nuhfd/d+DBlNLwwrqmwHeBs1JKMyPiNWAf8ucglRALsiKIiPbkI/AtIyKRP5kl8ie/JX/pl/WPiybyCcwXKT8OYUnXAX9KKT0SEb3IfzJeZHat6bOAycDWheebV2vdV7Wmq8m/P5Z3YvpJSumpZfS3UVjBz6b2cYt6Nl+k9nHL1ZrPsfh37BLg+ZTSIRHRFRj6v/a9EZq75Hu38F7dC9gppTQnIoaSTxC+VptFTYE/p5R+9c13vfGLiL3Ij0/aMaU0NyJeJJ+WT4uIrYDvkE+GDiWfgNfZHBieUtqtqJ3Ozkjyx6FGRKxF/sPAR+QvW55P/rj8PaW0YCWe86OU0jaFD81DI+KglNIjwH7kP7CNKBRuLYA5WJCVHMeQFcdhwF0ppfVTSl1TSl3IJ0m7Aj0jYoPIjx07kvxlAsj/bBalT8cALxbSqbERcTjkB55GxNaFNq3JR90Axy+nL63JJ1458hH5isbYjAbWiYjtC6/ZKvJfDngKOC0K3xSKiI0jYs0VH4oGZ3k/m9qGAd+LiBaF/TyEfMqzsmr/fE74H/u8OmkNzCgUWpsCO37NNgti8bfVhpBPLSoBIqJdRKy/KjvfyLQGpheKsS2ARb/THckPdxgMXEj+kj3ATKBVYXoUsG6tMZFNC8+xuhoCtIiI46Dmiz1XAnem/Fix58knXT8mX5yttJTSROCXwKIPDkcDJxfOP12BDYB9YgXfQtbqx4KsOI4GHlpi2V/JF1qvAJeRHxcztla72cAWEfEW+QTn4sLyvsBJEfEO+U9xBxeWX0T+UuYLwH+W05cbgeMj4lXylytnL6ctKf+trCOB6wqv+Qz5dOI28ifpf0XEu8AtNM7EdXk/mxoppX8Bd5IfjPsacFtK6d9f43WuAPpHxEusuAguJU8CFRExnHyK+OrXbDMAGB4R96T8t9nOB54utH0GWHuV9r5xeZx8kfEO+XGprxWWdwGGFS6n3crib67eAdxWWJ7If3j5U2H7f5Mfe7ZaSikl8h+6Do+ID4EPyF9N+HVhfY78eaI9+Q9rtR0ZdW97sXM9L/Ew+Z/FHsC+1ErDCsMiXgQOLCw6ISIm1Hp0LiwfXmvZn76RHVemFg0CVwYKl2LOSSkdUM+6WSmllsXvlSRJKjYTMkmSpIyZkEmSJGXMhEySJCljFmSSJEkZsyCTJEnKmAXZaiyW8e8C6pvjMV71PMbF4XFe9TzGWh4LstWbv/yrnsd41fMYF4fHedXzGGuZLMgkSZIy1hjvrJ6JH8Vaje7+ILuxRqPq980zx2Xdha/tlmv+BLOmN5pjzLzl/sMMDdItV/aH/4xvPMe4xVpZ9+D/yy3XXQ1zvmw8x7kR3rLplmuvgtlfNJ6Or9lmef+G7zeu2H+vbk7/Ler+rYj3IVtJjamwaawaY0HW6DTCgqzRaaQFWaPj365Vz4KsqEzIJElS5kp9DFWp778kSVLmTMgkSVLmyqJBXUEsOhMySZKkjFmQSZIkZcxLlpIkKXOlnhCV+v5LkiRlzoRMkiRlrqy0x/SbkEmSJGXNhEySJGWu1BOiUt9/SZKkzJmQSZKkzHljWEmSJGXKhEySJGWu1BOiUt9/SZKkzJmQSZKkzHkfMkmSJGXKhEySJGWu1BOiUt9/SZKkzJmQSZKkzIX3IZMkSVKWLMgkSZIy5iVLSZKUuVJPiEp9/yVJkjJnQiZJkjLnjWElSZKUKRMySZKUuVJPiEp9/yVJkjJnQiZJkjJX5o1hJUmSlCUTMkmSlLlST4hKff8lSZIyZ0ImSZIy533IJEmSlCkTMkmSlLlST4hKff8lSZIyZ0ImSZIyV0ZpDyIzIZMkScqYBZkkSVLGvGQpSZIy520vJEmSlCkTMkmSlLlST4hKff8lSZIyZ0ImSZIy5xgySZIkZcqETJIkZc4bw0qSJClTJmSSJClzjiHTaufY22/giskf8ZsRr2bdldXGsJdfYd/vH8neBx/GgDvuWma7J599jk2224kRo94rYu8ar2Gvvs6+R53A3kccx4BB9y2z3ZPPD2OTXfZixHvvAzB81GgOPv5UDj7+VA46vh/P/PPFYnW50Rn20ivs+73D2Pug7zNg4J+XWn/f4L9y4OFHc/CRfTn6h6cw5qOPAZi/YAG/uvBiDjz8aA464hhee/OtYne90Rr20ivse8jh7H3QoQy4Y+ljvsiTzw5hk2138HwhYCULsog4JCJSRGy6qjuk/90rd97Ddft9P+turDaqq6u5+LIrue3aP/H4g/fx2FPPMObjsUu1mzV7NoPuH8zWW26RQS8bn+rqai6+8jpuu/JSHr/ndh579nnGjP1kqXazZs9h0OCH2Hrzxaef7ht25a+338jf/3wLt13ZnwuuuJqFC6uL2f1GIf/evYLbrr+Gx//6AI89+VRNwbXIgd/Zl0cH38ffH7iHk48/lv5/uhqAwX97GIBHB9/HHTdfz+V/uoZcLlf0fWhsqqurufjyP3DbdVfz+F/v57Enn2bMxx8v1W7W7NkMuu8vni9qKSvyo6FZ2T4dDbwIHLUK+7JMEeGl1a9hzAsvM2f6jKy7sdoYPnIU63fpTJfO69K0SRP232cvhgwdtlS7a24awMnH9WWNNZpm0MvGZ/h777N+53Xosu46+eO6Zy+GvPDSUu2uufVOTu57ZJ3j2rxZMyoqygH4av58osQvdSzL8HdH1n3v7rvPUu/dli1b1kzPnTuXKAysHvPxWHbsuT0A7du1o1WrlrxrkrNCw98dxfqdax/zves/X9x4CycffyxrrLFGBr1UQ7TCgiwiWgK7ACdRKMgioldEDIuIhyJiVETcHBFlhXWzIuLKiPhXRAyJiI6F5d0i4smIeCsiXliUtkXEgRHxWkT8OyKejYiqwvKLImJARDwN3BURXQvb/avw2LlWX4ZGxIMRMToi7onIn54jYvuIeDki3omI1yOiVUSUR8QfIuKNiBgeEad+84dVq5PJU6bSqaqyZr6qqpLJU6fWaTNq9PtMmjyF3rvvWuzuNVqTp/6HTpW1jmtlRyZPnVanzagPPmTSlCn03mXHpbZ/Z+R77N/3JA467hR++4uf1RRoWiz/3q2qma/vvQtwzwOD2evAQ/jDNddx/rk/B2DTjbszZOg/WbhwIeM/+4yRo0YzcdLkovW9sZo8dQqdOtU65pWVTJ5S3/lisueLJZRFcR8NzcokZN8DnkwpfQBMj4htC8t7Aj8HegDdgEXXyNYE/pVS2hb4J3BhYfkA4Ccppe2Ac4AbC8tfBHZMKX0LuB84t9ZrbwccnFI6BpgC7F143iOBa2u1+xbwM2BzYENgl4hoCjwA/DSltDWwFzCXfGH5ZUppe2B74JSI2KC+HY+IfhHxZkS8OYr5K3GotDpKKS21LGpFMrlcjv5/uobzzjqzmN1q9Oo/rounc7kc/a+9ifN+8qN6t996i814/J7befC2G7hl0H189ZW/o0tK1HOM62nX98jDefbRhzjnp2dw020DATj04APpVFXJoX2P59I/XMW3tt6K8nKL3hWp52299Pniyqs57+yfFrFXagxW5lLg0cDVhen7C/OPA6+nlD4GiIj7gF2BB4Ec+UII4G7gb4WUbWdgcK035qKctjPwQESsDTQFag/OeSSlNLcw3QS4PiK2AaqBjWu1ez2lNKHQl7eBrsCXwMSU0hsAKaX/FtbvA2wVEYcVtm0NdF/idSlsM4B8IcmPYq16fs1UCjpVVTJp8pSa+cmTp1DZoUPN/OzZc/hgzMcc1+90AKZOm85pZ53LTVddQY/NNyt6fxuLTpUdmTSl1nGdMpXKDu1r5mfPmcMHH4/juDPyic3U6dM57bwLuOnyi+mx2SY17bp1XZ/mzZrxwcdj6ywXdKqsZNLkxanW5MlTqOzYcZnt9993Hy669HIAKioq+PU5Z9esO+r4k+i6XpdV19nVRKfKSibVShInT5lCZcclzhcffcRxpyw6X0zjtJ+dw01X/7Hkzxelfh+y5RZkEdEe6ANsGREJKAcS8ETh/7Utq2BJ5JO4L1JK29Sz/jrgTymlRyKiF3BRrXWza02fBUwGti4837xa676qNV1Nfr9iGX0K8kndU8vor1RHj803Y9z48Yz/7HOqKjvy+NPPcuXvf1uzvlWrlrz23JM188f2O51zf/aTkj+5rkiPTTdh3ITPGP/5RKo6duDxIUO58sJf16xv1bIlrz3xt5r5Y884m3N/fCo9NtuE8Z9PZO3KSioqyvls0mTGfjqBddfulMVuNGg9tticcZ+OZ/xnn1FVWcnjTz3Nlf0vqdNm3Cef0nX99QAY+sJLrN8lX3TNnTuPRKJF8+a89OprlJeXs1G3DYu+D41Njy2WOF889QxXXrr4mOfPF0/XzB97ymmce9aZni+0woTsMOCulFLNOKuI+Cf5NKxn4VLfJ+QvIQ4oNCkrbHc/cAzwYkrpvxExNiIOTykNLozx2iql9A75hOqzwrbHL6cvrYEJKaVcRBxPvjhcntHAOhGxfUrpjYhoRf6S5VPAaRHxXEppQURsDHyWUpq93GdrRE66dyAb99qVlh3a03/8ezx64aW8PHBQ1t1qtCoqKrjg3J9z8hk/o7o6x6EHH0D3bhtyzU0D2HLzzdhzj92y7mKjVFFRzgVn/YSTz/5l/rgesB/dN+zKNbfeyZabbsyeu+28zG3fGv4utw66n4qKCsrKgovOOZN2bVoXsfeNQ0VFBRec9wtOPv1MqnM5Dj34QLp368Y1N96Sf+/22p27HxjMK6+9TkVFBWuttRaXX5IfZTJtxnROOv1MysrKqOrYkSt+99sVvJpg0TE/h5N/XDjmBx1YOF8Ujvkeu2fdRTVQUd84jpqVEUOBy1JKT9ZadiZwGjARmEp+DNkw4PRCsTQLuAr4LvnLhkemlKYWirebgLXJX368P6V0cUQcXGj/GfAqsH1KqVdEXATMSin9sfC63YG/AnOA58mnXC0Lqdo5KaUDCu2uB95MKd0ZEduTT+Caky/G9ips/zvgQPJp2VTgeymlL5d3oLxkuerdPHNc1l1Y/c1bbT53NFwt1sq6B6VhOX+79A1Zs01RryHevlbHov5QT/rv1AZ1jXS5BdkyN1qiCFpi3ayUUsult2rcLMhWPQuyIrAgW/UsyIrDgmzVsyArKu/vJUmSMtegqqMM/H8VZCmlocDQZaxb7dIxSZKkVcmETJIkZa4h3qy1mBriP+ckSZJUUkzIJElS5kr9xrAmZJIkSRkzIZMkSZlzDJkkSZIyZUImSZIyV+oJUanvvyRJUuZMyCRJUuZKfAiZCZkkSVLWTMgkSVLmyqK0MzITMkmSpIxZkEmSJGXMS5aSJClzpX3B0oRMkiQpcyZkkiQpcyZkkiRJypQJmSRJypwJmSRJkjJlQiZJkjIX3hhWkiRJWTIhkyRJmSvtfMyETJIkKXMmZJIkKXOlnhCV+v5LkiQtJSL2i4j3I2JMRPyynvXrR8SQiBgeEUMjonOtdcdHxIeFx/Er83oWZJIkKXMRxX0svy9RDtwAfAfYHDg6IjZfotkfgbtSSlsBFwP9C9u2Ay4EdgB6AhdGRNsV7b8FmSRJUl09gTEppY9TSvOB+4GDl2izOTCkMP18rfX7As+klKanlGYAzwD7regFLcgkSVLmotj/RfSLiDdrPfrV6s66wPha8xMKy2p7Bzi0MH0I0Coi2q/ktktxUL8kSSo5KaUBwIBlrK7vomZaYv4c4PqIOAEYBnwGLFzJbZdiQSZJklTXBKBLrfnOwOe1G6SUPge+DxARLYFDU0pfRsQEoNcS2w5d0Qt6yVKSJGUuivxYgTeA7hGxQUQ0BY4CHqnT34gOEbGojvoVMLAw/RSwT0S0LQzm36ewbLksyCRJkmpJKS0EziBfSL0H/CWlNDIiLo6IgwrNegHvR8QHQBXw+8K204FLyBd1bwAXF5YtV6S0wsuaAn4Ua3mgVrGbZ47Lugurv3mzs+7B6q/FWln3oDT4t2vVW7NNUf81o8far13UH+oB0yY2qH+tyYRMkiQpYw7qlyRJmStrUHlV8ZmQSZIkZcyETJIkZS5W5ruPqzETMkmSpIyZkEmSpMyVdj5mQiZJkpQ5EzJJkpS5KPGIzIRMkiQpYyZkkiQpcyUekJmQSZIkZc2ETJIkZa6sxDMyEzJJkqSMmZCtpJtnjsu6C6u9H7XqmnUXVns3ffh81l1Y7eUeHpB1F0pCbL5t1l1Y7ZXtfEhRX6+08zETMkmSpMxZkEmSJGXMS5aSJClz3hhWkiRJmTIhkyRJmSvxgMyETJIkKWsmZJIkKXNR4hmZCZkkSVLGTMgkSVLmyko7IDMhkyRJypoJmSRJylyJB2QmZJIkSVkzIZMkSZkzIZMkSVKmTMgkSVLmvA+ZJEmSMmVCJkmSMhelHZCZkEmSJGXNgkySJCljXrKUJEmZK/WEqNT3X5IkKXMmZJIkKXMlPqbfhEySJClrJmSSJClzUeL3vTAhkyRJypgJmSRJylxp52MmZJIkSZkzIZMkSZkzIZMkSVKmTMgkSVLm/JalJEmSMmVCJkmSMldW2gGZCZkkSVLWTMgkSVLmosQjMhMySZKkjFmQSZIkZcxLlpIkKXMlftcLEzJJkqSsmZBJkqTMmZBJkiQpUyZkkiQpc/7TSZIkScqUCZkkScpciQdkJmSSJElZMyGTJEmZcwyZJEmSMmVCJkmSMlfiAZkJmSRJUtZMyCRJUubKSjwisyBr5Ia9/Aq//+PV5KqrOfx7B9Hvh8fV2+7JZ5/jp+f9Hw8OGkiPzTcrci9XP8fefgM9DtiPmVOmckmPHbPuTqP1wpvv8PsBd5HL5Thsn970O+Kgets9+eJr/Kz/NQy++nf06L4hCxYu5Pxrb2XUmHFUV1dz8J67ceoRBxe5943DC2Mn0n/I21SnxGFbbcApOyz9+/+P0eO54eWRBLBpZRv+cED+Pd1v8DDemTiNbdftwE2H7lbknjceL4x4n0vvfZRcLnHY7ttzyv696qx/6MU3+cMD/6Cq7VoAHLPnThy+R0/e+/RzfnvXw8yaO4/ysjJOPaA3391h6wz2QA1BgyvIIuIQ4G/AZiml0Vn3pyGrrq7m4suu5I4br6GqqpLDjj2RPnvsxkYbblCn3azZsxl0/2C23nKLjHq6+nnlznsYev0ATrjrlqy70mhVV+e4+KY7GPi7X1HVoT2Hn3U+fXbclo3W61yn3aw5c7n7kafYepONapY9+eJrLFiwgEdvvJy5875i/9N+wf577Eznqo7F3o0GrTqX43fP/IvbjtiDqlbNOXLQs/Tutg4bdWhd02bcjJnc+tp73HNMH1o3a8q02fNq1v2w5ybMW1DNX975KIvuNwrVuRyXDPo7t59zElXtWnPExdfTe5vN2GjdqjrtvtNzK35zbN0PDc2aNuGyk4+ga6cOTJnxXw797XXs2mNj1mrRvJi70GCUeEDWIMeQHQ28CBy15IqIKC9+dxqu4SNHsX6XznTpvC5NmzRh/332YsjQYUu1u+amAZx8XF/WWKNpBr1cPY154WXmTJ+RdTcateEfjGG9darosnYVTZtU8N3dd2LIq28t1e7auwdz0mEH0LRpk5plQTBn3lcsrK5m3vz5NKmooGWJ/hFbnhETp7Ne25Z0adOSpuXlfGfT9XhuzOd12jz4zscc862NaN0sf35ov2azmnU7rV/Fmk0b3Of2BmX4x+NZr7I9XSrb07Sigu/23Jrn/j1qpbbdoFNHunbqAEBl27Vov9aaTP/v7FXZXTVgDaogi4iWwC7ASRQKsojoFRHPR8S9wIjCsrMj4t3C42eFZV0jYnRE3FZYfk9E7BURL0XEhxHRs9CuZ0S8HBH/Lvx/k2z29n83ecpUOlVV1sxXVVUyeerUOm1GjX6fSZOn0Hv3XYvdPWm5Jk+bwdod2tfMd+rQjsnTptdpM+qjcUycOo3ePbets3zfXXvSotka7PaD0+lzwpmc+P39adOqZVH63ZhMnjWXTq1a1Mx3atWcKbPm1mkzbsZMxk2fSd97hnDU3c/ywtiJxe5mozZlxn/p1G5x4ljVrjWTZ/x3qXZPv/UuB//man56w91MnPbFUuuHfzyeBQurWa+y3SrtrxquhvbR53vAkymlDyJiekQsOgv3BLZMKY2NiO2AHwI7AAG8FhH/BGYAGwGHA/2AN4BjgF2Bg4BfF55/NLB7SmlhROwFXAocWrQ9/AallJZaVvvGerlcjv5/uob+F/2mmN2SVk5971+WeP/eOoj+Z/1oqXYjPviIsrIyhg26gf/Omk3fcy9m5222pMvaVUu1LWVLH+GlVecSn8yYxZ1H9WbyzDkce9/z/P2H+7JWMxP1lZHqOcpLXnrrtc1m7L/DNjRtUsH9z7/Kr277C3ee169m/ZQv/st5tz5A/5OPoKysQeUkReWNYRuWo4H7C9P3F+YBXk8pjS1M7wo8lFKanVKaRX682aLRpmNTSiNSSjlgJDAk5auWEUDXQpvWwOCIeBe4CljmwKqI6BcRb0bEmwMG/vmb2cNvUKeqSiZNnlIzP3nyFCo7dKiZnz17Dh+M+Zjj+p1OnwMO4e0RIzntrHMZMeq9LLor1VHVoR0T/zOtZn7Sf6ZT2b5tzfzsufP48JPxHPfLS+jzwzN5Z/QYTr/4j4z48GMeG/oyu223NU0qKmjfpjXbbr4x744ZW9/LlLROLZszaeacmvlJM+dS2bLupd2qVs3p030dmpSX0blNS7q2a8UnM2YVu6uNVlXb1kya/mXN/OTpX1LZZq06bdq2XJOmTfL5x+F79GTkJ5/VrJs1dx4/uupOfvr9fdim23rF6bQapAZTkEVEe6APcFtEjAN+ARxJPgWrfVF9eSX0V7Wmc7XmcyxOAy8Bnk8pbQkcCDRjGVJKA1JK304pfbvficd/jb0pjh6bb8a48eMZ/9nnzF+wgMeffpY+eyz+JlSrVi157bknee6xh3jusYfYpscW3HTVFX7LUg1Cj4278clnk5gwaQrzFyzkiWGv0GeH7WrWt1qzBa/eN4Dn7riW5+64lq033YgbLziHHt03ZO2O7Xn1nZGklJgzbx7vjB7Dhp3XyXBvGqYt127HJzNmMeGLWcyvruYfoz+l90Z1j9Oe3dfl9U/zH+xmzPmKT2bMpEubNbPobqPUY4POfDJlGhOmTmf+woWKPH+FAAAgAElEQVQ88fo79P7W5nXaTPli8SXM5/49ig3Xzg81mb9wIT+5bhAH77It+22/VVH73RBFWXEfDU1DumR5GHBXSunURQsKlyKXHPw0DLgzIi4jX5wdAhz7NV6nNbDo48kJ/9+9bQAqKiq44Nyfc/IZP6O6OsehBx9A924bcs1NA9hy883Ycw+/pr6qnHTvQDbutSstO7Sn//j3ePTCS3l54KCsu9WoVJSX85vTTuCk31xGLpfj0L170X39zlw7aDBbdt+QPjtut8xtjzlgH3591c0cePq5pATf33t3NtnAdGFJFWVl/N9e23LKg8PI5RKH9NiA7h1ac92L77JFp7b02Whddu3aiZfHTuaAgU9SHsE5e2xNm+ZrAPCDe59j7PSZzFmwkN43Pcol+23Prht0ynivGpaK8nLO73sQJ185kFwux/d3+zbd163i2oeeZsuunenzrc25+5mXee7tUVSUl9F6zRb0P/lwAJ58fQRvfjCWL2bN4eEX819oufTkw9lsPT9clKKobxxSFiJiKHBZSunJWsvOBE4DPkopHVBr+dnAiYXZ21JKV0dEV+CxQvJFRNxZmH+w9rqI2An4MzAVeA44NqXUdYUdnDW9YRyo1diPWnXNugurvZs+fD7rLqz2ckMfzroLJSE233bFjfQ/Kdv5kKIO6hqzWfei/p3d6L0PG9SgtQZTkDV4FmSrnAXZqmdBtupZkBWHBdmqZ0FWXA3pkqUkSSpRJf4ly4YzqF+SJKlUmZBJkqTMeR8ySZIkZcqETJIkZa7EAzITMkmSpKyZkEmSpMyVlXhEZkImSZKUMRMySZKUuRIPyEzIJEmSsmZBJkmSlDEvWUqSpMx5Y1hJkiRlyoRMkiRlrsQDMhMySZKkrJmQSZKkzJmQSZIkKVMmZJIkKXNRVtoRmQmZJElSxkzIJElS5hxDJkmSpEyZkEmSpMyVlXhEZkImSZKUMRMySZKUuRIPyEzIJEmSsmZCJkmSMhclHpGZkEmSJGXMgkySJCljXrKUJEmZK/ErliZkkiRJWTMhkyRJmXNQvyRJkjJlQiZJkjJX4gGZCZkkSVLWTMgkSVLmHEMmSZKkTJmQSZKkzEWJR0QlvvuSJEnZMyGTJEmZcwyZJEmSMmVCJkmSsldmQiZJkqQMmZBJkqTsOYZMkiRJWbIgkyRJypiXLCVJUua87YUkSZIyZUImSZKy520vJEmSlCUTspU1b3bWPVjt3fTh81l3YbV3WvfeWXdhtXfDFcdm3YWSkEa8kXUXVn87H1Lc13MMmSRJkrJkQiZJkjIXjiGTJElSbRGxX0S8HxFjIuKX9ay/KiLeLjw+iIgvaq2rrrXukZV5PRMySZKUvQY0hiwiyoEbgL2BCcAbEfFISmnUojYppbNqtf8J8K1aTzE3pbTN13lNEzJJkqS6egJjUkofp5TmA/cDBy+n/dHAff/LC5qQSZKkzDWwMWTrAuNrzU8AdqivYUSsD2wAPFdrcbOIeBNYCFyWUnp4RS9oQSZJkkpORPQD+tVaNCClNGDR6no2Sct4qqOAB1NK1bWWrZdS+jwiNgSei4gRKaWPltcfCzJJkpS9Io8hKxRfA5axegLQpdZ8Z+DzZbQ9CvjxEs/9eeH/H0fEUPLjy5ZbkDmGTJIkqa43gO4RsUFENCVfdC31bcmI2ARoC7xSa1nbiFijMN0B2AUYteS2SzIhkyRJ2WtAY8hSSgsj4gzgKaAcGJhSGhkRFwNvppQWFWdHA/enlGpfztwMuCUicuSDr8tqfztzWSzIJEmSlpBSegJ4YollFywxf1E9270M9Pi6r+clS0mSpIyZkEmSpMxFA7oxbBZMyCRJkjJmQiZJkrLXgAb1Z8GETJIkKWMmZJIkKXuOIZMkSVKWTMgkSVLmosQjohLffUmSpOyZkEmSpOw5hkySJElZMiGTJEmZC+9DJkmSpCyZkEmSpOw5hkySJElZMiGTJEnZcwyZJEmSsmRBJkmSlDEvWUqSpMyFg/olSZKUJRMySZKUPQf1S5IkKUsmZJIkKXuOIZMkSVKWTMgkSVLm/JalJEmSMmVCJkmSsue3LCVJkpQlEzJJkpQ5x5BJkiQpUyZkkiQpe44hkyRJUpZMyCRJUvYcQyZJkqQsWZBJkiRlzEuWkiQpc+GgfkmSJGXJhEySJGWvxAf1W5A1QsNefZ3fX30juVyOww/8Dv2OPbredk8+P4yfnn8xD952Az0224Tho0bzm8uvAiCR+MmJx7H3HrsWs+uNygtvvsPvB9xFLpfjsH160++Ig+pt9+SLr/Gz/tcw+Orf0aP7hixYuJDzr72VUWPGUV1dzcF77sapRxxc5N43fsfefgM9DtiPmVOmckmPHbPuTuPVdXPK9jwMoow0/CXS688s1SQ22ZbY+bsApCkTSI/fuXhl02aUnfgb0ofvkIb8pUidblxeGDuJ/kPfoTqXOKzHBpzSc5Ol2vzj/Qnc8MooImDTDm34w/49Aej31xd5Z9J0tl2nPTcdskuxu64GpEEWZBFRDYwg37/3gONTSnOW034c8O2U0n+W0+bXKaVLv+m+Flt1dTUXX3kdd1x9OVWVHTns5B/TZ9ed2WiD9eu0mzV7DoMGP8TWm29as6z7hl356+03UlFRzpT/TOPg40+l9y47UVFRXuzdaPCqq3NcfNMdDPzdr6jq0J7DzzqfPjtuy0brda7Tbtacudz9yFNsvclGNcuefPE1FixYwKM3Xs7ceV+x/2m/YP89dqZzVcdi70aj9sqd9zD0+gGccNctWXel8YqgbO8jyP3lOpj5BWXHnkv6aARMm7S4TZuOxA77kLv3SvhqLrRoWfcpdj2ANP7DIne88ajOJX733NvcduiuVLVqwZH3PEfvbmuzUfu1atqMmzGTW18fzT1H9aJ1s6ZMmzOvZt0Pt9+YeQsW8pfhY7PofsPiGLIGaW5KaZuU0pbAfOBH38Bz/vobeI7MDX/vfdbvvA5d1l2Hpk2asP+evRjywktLtbvm1js5ue+RrLFG05plzZs1qym+vpo/v9TT4eUa/sEY1lunii5rV9G0SQXf3X0nhrz61lLtrr17MCcddgBNmzapWRYEc+Z9xcLqaubNn0+TigpatmhezO6vFsa88DJzps/IuhuN29pdYcZU+HIa5KpJo98iNtqqTpPYehfSv4flizGAObMWr6zqAi1awbjRxetzIzNi0nTWa7MmXdq0pGl5Gd/ZtDPPffR5nTYPjhjHMdt0o3Wz/Pm4fYtmNet2Wq+SNWudP1S6GmpBVtsLwEYAEfFwRLwVESMjol99jetrExGXAc0j4u2IuKew7AcR8Xph2S0R0ShioslT/0Onysqa+arKjkyeOq1Om1EffMikKVPovcvSl3neGfke+/c9iYOOO4Xf/uJnpmPLMHnaDNbu0L5mvlOHdkyeNr1Om1EfjWPi1Gn07rltneX77tqTFs3WYLcfnE6fE87kxO/vT5tWdVMHqShatiHNrFXUzvwCWrap0yTaVkK7SsqOOZuyvudA180XraGs1/dJ/3yoaN1tjCbPmkunVi1q5ju1bM6UmXPrtBk3YybjZsyi7/1DOere53lh7KQln0bk/3HxYj4amgZdkEVEBfAd8pcvAU5MKW0HfBs4MyLa17PZUm1SSr9kcerWNyI2A44EdkkpbQNUA33ref1+EfFmRLw54K57VsEefn0ppaWW1X5f5XI5+l97E+f9pP5QcestNuPxe27nwdtu4JZB9/HVV/NXVVcbt/qOM4sPdC6Xo/+tgzjv5B8s1W7EBx9RVlbGsEE38OzAq7njoScYP3HyKu2utPKWeG+XlRFtO5K7/2pyj91B2X7HwBrNiW/tTho7Ml/EaZmWPlOw1OD06lzikxmzuPPw3fnj/j254Jl/8d95nntVV4McQ0YhzSpMvwDcXpg+MyIOKUx3AboD05bYdmXa7AlsB7xRqJKbA1OW7ERKaQAwAID/jK/3967YOlV2ZNKUxV2dPGUqlbWSnNlz5vDBx+M47oyfAzB1+nROO+8Cbrr8Ynpstnigabeu69O8WTM++HhsneXKq+rQjon/Wfy2mfSf6VS2b1szP3vuPD78ZDzH/fISAP4z40tOv/iP3HjBOTw29GV2225rmlRU0L5Na7bdfGPeHTOWLmtXFX0/VOJmfUG0aru4aGjVBmZ9WadJmvkFfD4Ocrn8pc3pU6BtR1hnA6JzN2Kb3aHJGlBeDgu+Ig37e7H3okHr1LI5k2YuHuI8adZcKls2q9OmqlVztl67HU3Ky+jcek26tm3JJ1/MokendsXubsNW4mPIGmpBNreQXNWIiF7AXsBOKaU5ETEUaPZ12yxqCvw5pfSrb77rq1aPTTdh3ITPGP/5RKo6duDxIUO58sLFw+NatWzJa0/8rWb+2DPO5twfn0qPzTZh/OcTWbuykoqKcj6bNJmxn05g3bU7ZbEbDV6PjbvxyWeTmDBpCpXt2/HEsFf44y/OqFnfas0WvHrfgJr5Y395Ceee1Jce3Tfk1bff5dV3RnJQ712Z+9VXvDN6DMcf/J0sdkOlbuIn0LYSWreHmV8Qm25H7rE76zRJHw4nNtsORr4KzdfMt/9iGunxO2sKudhiR+i0nsVYPbbs1JZPvpjFhC9nU9myOf8YPYErvtuzTps9u63DE++P55AtujJj7ld8MmMWXVqvmVGP1VA11IKsPq2BGYVCa1Ogvu/BL6/NgohoklJaAAwB/h4RV6WUpkREO6BVSumTVb4X/6OKinIuOOsnnHz2L6muznHoAfvRfcOuXHPrnWy56cbsudvOy9z2reHvcuug+6moqKCsLLjonDNp16Z1EXvfeFSUl/Ob007gpN9cRi6X49C9e9F9/c5cO2gwW3bfkD47brfMbY85YB9+fdXNHHj6uaQE3997dzbZYL0i9n71cNK9A9m416607NCe/uPf49ELL+XlgYOy7lbjknLknv0LZYf9GMrKSCNegWkTiV32J036FD4aAeNGwQabUvbD8yHl8mPG5s3OuueNRkVZGf/XextO+euL5FLikC270r3DWlz30ki26NSWPt3WYdeuVbz8yWQOuPNpyiM4Z/cetGm+BgA/eGAoY6fPZM78hfQe8ASX7LMtu3Yt0Q/KDXBcVzFFfWOSshYRs1JKLZdYtgbwMLAu8D7QEbgopTR00W0vgJnLaXM5cBDwr8I4siOBX5EfR7cA+HFK6dVldqqBXLJcnaUvlrpqrG/Yad17Z92F1d4NVxybdRdKw1p+mFzVyk+9tKgV0sKfHFjUv7MV1z3aoCrABpmQLVmMFZZ9RX6Af33tu9aaXVab84Dzas0/ADzwP3VUkiR9M0o8IWvQ37KUJEkqBQ0yIZMkSSXGhEySJElZMiGTJEnZKyvtjKi0916SJKkBsCCTJEnKmJcsJUlS9hzUL0mSpCyZkEmSpOyZkEmSJClLJmSSJCl7JmSSJEnKkgmZJEnKnjeGlSRJUpZMyCRJUvYcQyZJkqQsmZBJkqTsmZBJkiQpSyZkkiQpeyZkkiRJypIJmSRJyp73IZMkSVKWLMgkSZIy5iVLSZKUPQf1S5IkKUsmZJIkKXsmZJIkScqSCZkkScqeCZkkSZKyZEImSZIyF94YVpIkSVkyIZMkSdlzDJkkSZKyZEImSZKyZ0ImSZKkLJmQSZKk7JmQSZIkKUsmZJIkKXveh0ySJElZsiCTJEnKmJcsJUlS9hzUL0mSpCyZkEmSpOyZkEmSJClLJmSSJCl7JmSSJEnKkgmZJEnKnjeGlSRJUpZMyFZWi7Wy7sFqL/fwgKy7sNq74Ypjs+7Cau/H5w7Kugsl4eq+22XdhdVe+alFfkHHkEmSJClLJmSSJCl7JmSSJEnKkgmZJEnKnt+ylCRJUpZMyCRJUvYcQyZJkqQsWZBJkiRlzEuWkiQpe16ylCRJUpZMyCRJUvZMyCRJkpQlEzJJkpQ9bwwrSZKkLJmQSZKk7DmGTJIkSVkyIZMkSdkzIZMkSVKWTMgkSVL2orQzotLee0mSpAbAhEySJGWvzDFkkiRJypAJmSRJyp5jyCRJkpQlCzJJkqSMeclSkiRlzxvDSpIkKUsmZJIkKXtlpZ0RlfbeS5Ik1SMi9ouI9yNiTET8chltjoiIURExMiLurbX8+Ij4sPA4fmVez4RMkiRlrwGNIYuIcuAGYG9gAvBGRDySUhpVq0134FfALimlGRFRWVjeDrgQ+DaQgLcK285Y3muakEmSJNXVExiTUvo4pTQfuB84eIk2pwA3LCq0UkpTCsv3BZ5JKU0vrHsG2G9FL2hCJkmSstewbgy7LjC+1vwEYIcl2mwMEBEvAeXARSmlJ5ex7borekELMkmSVHIioh/Qr9aiASmlAYtW17NJWmK+AugO9AI6Ay9ExJYrue1SLMgkSVL2ijyGrFB8DVjG6glAl1rznYHP62nzakppATA2It4nX6BNIF+k1d526Ir606DyQUmSpAbgDaB7RGwQEU2Bo4BHlmjzMNAbICI6kL+E+THwFLBPRLSNiLbAPoVly2VCJkmSsteA7kOWUloYEWeQL6TKgYEppZERcTHwZkrpERYXXqOAauAXKaVpABFxCfmiDuDilNL0Fb2mBZkkSdISUkpPAE8sseyCWtMJOLvwWHLbgcDAr/N6FmSSJCl7Deg+ZFloOPmgJElSiTIhkyRJ2WtY9yErutLee0mSpAbAgkySJCljXrKUJEnZK3NQvyRJkjJkQiZJkrLnoH5JkiRlyYRMkiRlzxvDSpIkKUsmZJIkKXuOIZMkSVKWTMgaoWEvvcLv/3AluVyOw793MP1OPL7O+vsG/5V7//IgZWVltGjRgkvO/xUbdduQ+QsWcOHv+vPuqPeICP7v3J+zw7e3y2gvGrYXxk6k/5C3qU6Jw7bagFN22GypNv8YPZ4bXh5JAJtWtuEPB+wIQL/Bw3hn4jS2XbcDNx26W5F73sh03ZyyPQ+DKCMNf4n0+jNLNYlNtiV2/i4AacoE0uN3Ll7ZtBllJ/6G9OE7pCF/KVKnVx/H3n4DPQ7Yj5lTpnJJjx2z7k6jVbbV9lQcewaUlVM99HGqH72vzvqKvqdTtvm38jNN1yDWastXpx5IrNeNJj88C5qvCblqFv79HnKvPZ/BHjQQJX4fsgZfkEVEe2BIYbYTUA1MLcz3TCnNX8nnORF4IqU06ZvvZfFUV1dz8WVXcMdN11NVVclhfY+nzx67sVG3DWvaHPidfTn68EMBGDJ0GP3/dDW333Atg//2MACPDr6PadOnc8oZP+PBu++krMygtLbqXI7fPfMvbjtiD6paNefIQc/Su9s6bNShdU2bcTNmcutr73HPMX1o3awp02bPq1n3w56bMG9BNX9556Msut94RFC29xHk/nIdzPyCsmPPJX00AqbV+hVt05HYYR9y914JX82FFi3rPsWuB5DGf1jkjq8+XrnzHoZeP4AT7rol6640XlFGxfE/ZcFlvyBNn0rTi28m99bLpM8/qWmy8J4ba6bL9z6E6No9PzP/Kxbc3J80+TNo0541fncLX414HebMLvJOqCFo8H+JU0rTUkrbpJS2AW4Grlo0v7LFWMGJ5Au6Rm34uyNZv0tnunRel6ZNmrD/vvswZOiwOm1atlz8R2vu3LkE+U8dYz4ey449twegfbt2tGrVkndHvVe8zjcSIyZOZ722LenSpiVNy8v5zqbr8dyYz+u0efCdjznmWxvRullTANqv2axm3U7rV7Fm0wb/WSd7a3eFGVPhy2mQqyaNfovYaKs6TWLrXUj/HpYvxgDmzFq8sqoLtGgF40YXr8+rmTEvvMyc6TOy7kajFt02JU3+nDR1IlQvpPrV5yjbbpdlti/fqQ+5V/IZQ5o0IV+MAXwxjfTlF0SrNsXodsMUUdxHA9PgC7LliYhHI+KtiBgZEScXllVExKCIGBER70bEmRFxJLAN8EBEvB0RTSNi+4j4Z2H7f0REVbZ7s3ImT5lKp6rFXa2qqmTy1KlLtbvngcHsdeAh/OGa6zj/3J8DsOnG3Rky9J8sXLiQ8Z99xshRo5k4aXLR+t5YTJ41l06tWtTMd2rVnCmz5tZpM27GTMZNn0nfe4Zw1N3P8sLYicXuZuPXsg1pZq1iYOYX0LLuH6NoWwntKik75mzK+p4DXTdftIayXt8n/fOhonVXqk+07UCaPqVmPk2fSrTtUH/j9lVE5drkRv576efZcFOoqCBN+byeDVUKGnVBBhyfUtoO2B44OyLaAtsBHVJKPVJKWwJ3pZQeAN4GjiwkbQFcAxxa2P5u4JIlnzwi+kXEmxHx5oCBdxZpl5YvkZZaVl+d3/fIw3n20Yc456dncNNtAwE49OAD6VRVyaF9j+fSP1zFt7beivLy8lXc48Zn6SO8tOpc4pMZs7jzqN788YAdueDJN/nvvK8T2Kp+Sxz9sjKibUdy919N7rE7KNvvGFijOfGt3UljR+aLOClL9SYt9Z9FynfqTfXr/4SUq7uiTTuanPYrFgy4HNLKnIFWU1FW3EcD09ivq5wVEQcVpjsD3YAxwCYRcQ3wBPB0PdttBmwBPBv5X6ZyYMKSjVJKA4ABAMz5skH8lnSqrGTS5MWp1uTJU6js2HGZ7fffdx8uuvRyACoqKvj1OWfXrDvq+JPoul6XVdfZRqpTy+ZMmjmnZn7SzLlUtmxep01Vq+ZsvU57mpSX0blNS7q2a8UnM2bRY+12xe5u4zXrC6JV28V/ulq1gVlf1mmSZn4Bn4+DXC5/aXP6FGjbEdbZgOjcjdhmd2iyBpSXw4KvSMP+Xuy9UIlL06cS7Spr5qNdR9KMafW2Ld+xDwv+fE3dhc1b0PSc/iwcPJD0kUNISlnDKxFXUkTsBewO7JhS2hoYDjRLKU0DtgJeBM4E6hutGsDwWmPReqSUvlOsvv8vemyxOeM+Hc/4zz5j/oIFPP7U0/TpVfebfOM++bRmeugLL7F+l3zRNXfuPObMzV96e+nV1ygvL6/zZQDlbbl2Oz6ZMYsJX8xifnU1/xj9Kb03WqdOmz27r8vrn+YvU8yY8xWfzJhJlzZrZtHdxmviJ9C2Elq3h7JyYtPtSGNG1GmSPhwO6xUGQDdfM9/+i2mkx+8kd8tvyA24gDT0IdLI1y3GlIn08Wii07pEx05QXkH5jn3I/evlpdrF2l2INVuRPhy5eGF5BU1+dgnVLzxN7vV/FrHXDVRZFPfRwDTmhKw1MD2lNDcitiB/2ZKI6AjMSykNjoix5L8IADATaFWYHgWsGxE9U0qvR0RToHtKaSQNXEVFBRec9wtOPv1MqnM5Dj34QLp368Y1N97Clptvxp69dufuBwbzymuvU1FRwVprrcXll1wIwLQZ0znp9DMpKyujqmNHrvjdbzPem4apoqyM/9trW055cBi5XOKQHhvQvUNrrnvxXbbo1JY+G63Lrl078fLYyRww8EnKIzhnj61p03wNAH5w73OMnT6TOQsW0vumR7lkv+3ZdYNG/32Sb17KkXv2L5Qd9mMoKyONeAWmTSR22Z806VP4aASMGwUbbErZD8+HlMuPGZvnN9C+KSfdO5CNe+1Kyw7t6T/+PR698FJeHjgo6241LrkcC/98LU3OvQLKyqj+5z9In42j4tAfkhv7fk1xVr5TH6pffa7OpmU79qJsk62IlmtRvvt+ACy45TLSp35DuxRFakTXqyPiImBWSumPEdEM+Dv5b06OBtYGfg3MAW4nn4Il4LyU0tMRcQT5cWJzgZ7AlsC15Iu0CuDKlNLAZb54A7lkuTqrvvePWXdh9TdjetY9WO39+FwLmmK4uq/3UFzVmt39fFFjpOq/XVPUv7Pl3/9pg4rJGlVClv5fe/ceXWV15nH8u5MQud9NogKDXLwgKCoirlIUXLXeLQrWS61a8VJri+1i6jjTOlNt1Vadar3TllIVrU6tqPVSHRGL1Yp4RVAcERBFCENA5CYkZ88fJ4QTCGBHcvYJ+X7Wylo579nnPc+bpWTnt593nxj/I+f7dcBXtzL0wAZe+wCQu3Pkq8DQHVmfJEn6fyrARvt8at5XL0mSVACaVEImSZJ2UgW4WWs+mZBJkiQlZkImSZLSs4dMkiRJKZmQSZKk9Apws9Z8MiGTJElKzIRMkiSlZw+ZJEmSUjIhkyRJ6bkPmSRJklIyIZMkSekVNe+MqHlfvSRJUgEwIZMkSenZQyZJkqSUnJBJkiQl5pKlJElKz41hJUmSlJIJmSRJSs+mfkmSJKVkQiZJktJzY1hJkiSlZEImSZLSs4dMkiRJKZmQSZKk9NyHTJIkSSmZkEmSpPTsIZMkSVJKJmSSJCk9e8gkSZKUkgmZJElKr8geMkmSJCXkhEySJCkxlywlSVJ6NvVLkiQpJRMySZKUnhvDSpIkKSUTMkmSlJ49ZJIkSUrJhEySJCUX7CGTJElSSiZkkiQpPXvIJEmSlJIJmSRJSs+ETJIkSSmZkEmSpPSKvMtSkiRJCZmQSZKk9Jp5D5kTss8rxtQV7PRCv4NSl7DTizNfTl3CTu/GMw9OXUKzcOmkV1KXsNO7457UFTQvzXs6KkmSVABMyCRJUnp+dJIkSZJSMiGTJEnpNfOm/uZ99ZIkSQXAhEySJKVnD5kkSZJSMiGTJEnp2UMmSZKklEzIJElSen64uCRJklIyIZMkSenZQyZJkqSUTMgkSVJ67kMmSZKklEzIJElSevaQSZIkKSUTMkmSlJ49ZJIkSUrJCZkkSVJiLllKkqT0bOqXJElSSiZkkiQpvaLmnRE176uXJEkqACZkkiQpueC2F5IkSUrJhEySJKXnXZaSJElKyYRMkiSlZw+ZJEmSUjIhkyRJ6dlDJkmSpJRMyCRJUnr2kEmSJCklEzJJkpSen2UpSZKklJyQSZIkJeaSpSRJSs+mfkmSJKVkQiZJktJzY1hJkiSlZEImSZLSs4dMkiRJKZmQSZKkAmBCJkmSpIRMyCRJUnr2kEmSJCklEzJJkpSeCZkkSaLurGoAAAw6SURBVJJSMiFr4v76txf52fX/SaYmw+iRJ3LBuWc3OO7J/36GsT/8V/54z0QG9Ns3z1U2PdNmzuHqex8lk4mMGnYI5x93RL3nH3p+Btfd/wTlndoDcMaRhzH68MG8/cEifnLXZFatXUdxUREXHj+cYw89IMEVNA3T5i3mmqlvUJOJjBqwJ+cP3nuLMU/M+ZBbX5xNCLBP145cd9xgAC548HneWFzFQbt34faRX8p36U1G0f6HUHLWJVBUTM3Ux6h59L56z5eceTFF/Q7MPijdhdC+E59deAKhR29anPt9aNUGMjVUPzyJzEvPJriCpu+s397KgOOP5tPKpVw1YEjqcgpY807IGnVCFkLoBtwK9CObxv0ZuAJ4H+gVY/wkZ+xk4F6gNXAd8FHOqc4A1gBvA3OAUmAGcF6McUPOOW4CRgHdY4yZ2mPnAINijJdsVtv82uP/u+OuOL9qamq48ufX8bvbbqa8vIxR3ziHEYd/mT69etUbt2r1au6+7wEO6L9fokqblppMhqvufpjfjjuP8s4dOPXKWxg+cF/67FFeb9wxg/fnx2edVO9Yy9IWXDvmVHpWdKVy+UpO+cnNDB2wF+1bt8rnJTQJNZnIT6e8zm9OGUp5u9Z8fdIUhvfejT5d2teNmb/8U349/R0mnXYEHVqWsmzNurrnzj1kL9ZtqOaBN+elKL9pCEWUnD2WDdf+M7FqKaVX3kHmlReIixbUDamedFvd98VfGUno2Tf7YP1nbLjjGuKSj6BjF3b56Z18NnM6rFmd54to+l6cOImpt4znnLvuTF2KClijLVmGEALwJ2ByjLEvsBfQFvgR8BTwtZyxHYChZCdsAPfHGAfmfM2uPT43xjgQGAB0A07NOUcRMBJYCAxrrOsqJG++NZt/6taN7t32oLRFC4776ld4Zupftxh30213Mubss9hll10SVNn0vPn+QnqUdaF7WRdKS0o4dvABTHlt9vZfCOxZsSs9K7oCUNapPV3at6Fqpb/AGjJzcRU9Orahe8e2lBYXccw+3Zgyd1G9MX+cOZ8zBvamQ8tSALq0bln33GE9ymhT2iKvNTc1ofc+xCWLiEs/hppqav4+haKDt54mFh82gsyLzwAQF3+YnYwBrFhG/GQFoV3HfJS903lv2gusqVqeuozCF0J+vwpMY/aQjQDWxRh/BxBjrAG+D3wLuA84LWfsSODJGOOaz3Pi2nNNB/bIOTwceAu4HTj9C1ffBCxZWklFxabUprysjCWVS+uNmf3OHBYvWcLwYUPzXV6TVbl8JRWdO9Q9Lu/cgSXLV24x7qlX3uKkH9/I2Fvv4eNlK7Z4/s33F7KhuoYeZZ0btd6masmqtVS0a133uKJtKyo/XVtvzPzlnzJ/+SrO/MNUTrv3WabNW5zvMpu00Kkrsaqy7nGsWkro1LXhwV3KCWW7kZn12pbn6bUPlJQQKxc18EJJO0JjTsj2A17JPRBjXAl8AMwHDg4hdKl96jSyk7SNvh5CeD3nq956TwihJXAo8GTO4dNrz/EQcHwIYaf/0znGLY+FnFl/JpPhmhtu5LIfjM1jVU1fZMsf7OZ/TB0xcF+eue4yHr7qUg7r14fLf/NAvecrV6zksl/fz8/OG01RkffONKSB/3y3+EHXZCILlq9i4uhhXH/cYK54+lVWrlufl/p2Cg2mAA3+5Ck+bDg105+DbLfHJh070+Lbl7Nh/M8b/kdH0g7RmL8pAg3/n7/x+CPAqBBCV2Ag2WXMjTZfstz4Z3PvEMLrwDLggxjjmwAhhFLgWLLLoyuBl4CjvvAFhHBBCGFGCGHG+AkTv+jpdriKsjIWL15S93hJZSVlu27663f16jW8O3cu3zz/YkYc9zVen/kW3750HDNnv52i3CajvFMHFlfVtTeypOoTyjq2rzemU9s2lLbItmCOPnwwsxZsanlctXYdF/1yImNPPoqBvXvkp+gmqKJtKxZ/uikUX7xqLWVtW9YbU96uFSP67EaL4iK6dWhDz05tWbBiVb5LbbJi1VJC57K6x6HzrsTlyxocWzxkBDUvTql/sFVrSsddQ/V/TSDO9d8NNbKQ56/tlRPC0SGEOSGE90II/7KNcaNCCDGEMKj2cc8QwtqcUOmOz3P5jTkhmwUMyj0QQmgPdAfmsmnZchTwcG5z/jZs7CHrAwwJIZxYe/xooAMws7ZZfyg7YNkyxjg+xjgoxjjogm+d80VPt8MN2G9f5i9cyMKPFrF+wwYe+8vTjDh8U/tcu3ZteWnKU0x5bDJTHpvMwAH9uf3G673LcjsG7NmNBZXL+HBpFeurq3l8+hsMP7BfvTGVKzYtYU55bTa9dsv+0ltfXc13b76bk750EEcfsn9e625q+ld0YsGKVXz4yWrW12R44p0PGd5r93pjjuy9O9MXZpfhl6/9jAXLV9G9Q5sU5TZJ8f13CBV7EHatgOISioeMIPPqC1uMC7t1J7RpR/yfWZsOFpfQ4tKrqJn2FJnpz+Wxaim9EEIx2ZsSjyF7Y+LpIYR+DYxrB3yPbBCUa25OqHTR53nPxrzL8hng2hDCN2OMd9Ve3A3AxBjjmhDCs8Dvge8A3/1HThxj/Lh2tno52aTtdGBMjPE+gBBCG2BeCKH1Nk7T5JWUlHDFZeMY853vUZPJcMqJJ9C3dy9uuv1O+vfblyMPbxb3NuxwJcXF/OjMExlzwwQymQwnf3kQffco51cPPUX/nt0YcWA/7nn6Baa8PpuS4iI6tGnNNWNGA/Dk9JnMeHceK1atYfLz2RX7q8eMZt8eu2/rLZulkqIi/m34QM5/8HkyMTKyf0/6dm3PzX+bxX4VnRjRe3eG9iznhQVLOH7iUxSHwLhhA+jYKntzyjfun8q8qk9Zs76a4eMf56qjDmJoz4rEV1VgMhmqf/8rWvzwF1BURM1zTxA/mk/JKeeSmTenbnJWfNgIav5ePx0rGnIERXvvT2jbnuJhRwOw4c5riR/MzftlNHXn3TuBvY4YStuuXbhm4ds8+u9X88KEu1OXVYAKqtF+MPBejPF9gBDCH4CTgM3v8LoK+AUw7ou+YYiN2BMQQugO3AbsQzaNexwYF2P8rPb5m4DRQLfNtqnYfNuLi4FFwJ9jjP1rxwXgdbIz04eAnrXLlRvf+0/A/UAr4BYgt+t6CPA82e0zNjZMPBBj/MFWL2b1CpsnGlnmDfc4amxx5supS9jpbZj2YuoSmoVLJ72y/UH6Qu6IK/M6Q4qL3s3r79miPfa+ELgg59D4GON4yC5DAkfHGMfUPj4LODR3C60QwoHAj2KMp4QQppKd38wIIfQku0r4LrCydsy07dXTqPuQxRgXAids4/mxwNjNjk0EJm7lJf1zxkVg446bW9zGFmM8OedhQ+frubW6JElSnuV5K4raydf4rTy9zTtiarfa+iVwTgPjPgZ6xBiXhRAOBiaHEPbLDY0a4u1fkiRJ9X1Itud9o25kV+o2akc2JJpa27s+BHgkhDAoxvhZjHEZQIzxFbJ983tt7w396CRJkpReYW3W+jLQN4SwJ9kWqtPIfmoQALWfNFS3rcFmS5a7AlUxxpoQQi+gL9lPKNomJ2SSJEk5YozVIYRLgL8AxcCEGOOsEMKVwIwY4yPbePkw4MoQQjVQA1wUY6za3ns6IZMkSQWgoBIyYoyPk70ZMffYFVsZe0TO9w8CD/6j72cPmSRJUmImZJIkKb3C6iHLOxMySZKkxEzIJElSATAhkyRJUkImZJIkKT17yCRJkpSSEzJJkqTEXLKUJEnpuWQpSZKklEzIJElSATAhkyRJUkImZJIkKblgD5kkSZJSMiGTJEnpmZBJkiQpJRMySZJUAEzIJEmSlJAJmSRJSs8eMkmSJKVkQiZJktIzIZMkSVJKJmSSJKkAmJBJkiQpISdkkiRJiblkKUmS0rOpX5IkSSmZkEmSpPSad0BmQiZJkpSaCZkkSSoAzTsiMyGTJElKzIRMkiSl512WkiRJSsmETJIkpWdCJkmSpJRMyCRJUgEwIZMkSVJCJmSSJCk9e8gkSZKUkgmZJElKz4RMkiRJKTkhkyRJSswlS0mSVABcspQkSVJCJmSSJCm9Zt7UH2KMqWuQJElq1lyylCRJSswJmSRJUmJOyCRJkhJzQiZJkpSYEzJJkqTEnJBJkiQl5oRMkiQpMSdkkiRJiTkhkyRJSswJmSRJUmL/B16qjAFCLQQGAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Using Pearson Correlation\n",
    "plt.figure(figsize=(10,10))\n",
    "cor = data.corr()\n",
    "ax = sns.heatmap(cor, annot=True, cmap=plt.cm.Reds)\n",
    "ax.xaxis.set_ticks_position('top')\n",
    "loc, labels = plt.yticks()\n",
    "ax.set_yticklabels(labels, rotation=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### OBSERVATION:\n",
    "\n",
    "* Based on the correlation matrix above, we can see that taste correlates the most (corr = 0.72) with the overall beer rating.\n",
    "* This supports our initial hypothesis, which predicted that taste is the most important feature of a beer.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predicting Overall Beer Rating from Taste, Appearance, Aroma and Palate Ratings  \n",
    "\n",
    "* In this section, we are splitting the rating data into train and test set (with a 80:20 ratio)\n",
    "* Given the ratings for the 4 features (taste, appearance, aroma and palate), we will use linear regression to predict the overall rating for each beer style."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pick_feature(datum): \n",
    "    feat=[]\n",
    "    feat.append( datum['Taste'])\n",
    "    feat.append( datum['Appearance'])\n",
    "    feat.append( datum['Aroma'])\n",
    "    feat.append( datum['Palate']) \n",
    "    ##feat.append(datum['review/overall'])\n",
    "    return feat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = []\n",
    "y = []\n",
    "for index, row in data.iterrows():\n",
    "    x.append(pick_feature(row))\n",
    "    y.append(row['OVERALL'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "#SPLITTING INTO TRAIN/TEST SET\n",
    "train_size = int(len(x)*float(0.8))\n",
    "test_size =int(len(x)*float(0.2))\n",
    "\n",
    "trainX = x[ :train_size]\n",
    "trainY = y[ :train_size]\n",
    "testX = x[-test_size:]\n",
    "testY = y[-test_size:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/khanhtoanbui/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:2: FutureWarning: `rcond` parameter will change to the default of machine precision times ``max(M, N)`` where M and N are the input matrix dimensions.\n",
      "To use the future default and silence this warning we advise to pass `rcond=None`, to keep using the old, explicitly pass `rcond=-1`.\n",
      "  \n"
     ]
    }
   ],
   "source": [
    "## train a linear regression model\n",
    "theta1,residuals1,rank1,s1 = np.linalg.lstsq(trainX ,trainY)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "trainX = np.matrix(trainX)  # convert list to matrix\n",
    "theta1 = theta1.reshape(4,1) #transpose\n",
    "testX = np.matrix(testX)\n",
    "expY = testX*theta1 # experiment value\n",
    "actY = np.matrix(testY)\n",
    "expY=expY.reshape(expY.shape[1],expY.shape[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Mean Squared Error on the Prediction\n",
    "\n",
    "* We predicted the overall rating for each beer, given their individual ratings on each of the 4 features (taste, appearance, aroma and palate). \n",
    "* The MSE for our linear regression prediction model is below 0.17, which is respectable given that the rating ranges from 0.0 to 5.0."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.15719546579470547"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_MSE = mean_squared_error(expY, actY)\n",
    "test_MSE"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sentiment/NLP Analysis From User Review Text\n",
    "\n",
    "* Model Definition:\n",
    "    - TFIDF Vectorizer\n",
    "        - Exclude >0.9, <0.001 most frequent document-freq words\n",
    "        - Binary mode\n",
    "        - 1-gram + 2-gram\n",
    "    - Logistic Regressor\n",
    "        - L1 norm\n",
    "\n",
    "* Output:\n",
    "    - Performance of log regression\n",
    "    - Top 150 most positive / most negative coefficient n-grams"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import metrics, preprocessing\n",
    "from sklearn.linear_model import LogisticRegression, LogisticRegressionCV\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "\n",
    "import re\n",
    "from nltk.stem.snowball import SnowballStemmer\n",
    "class CustomTokenizer(object):\n",
    "    def __init__(self):\n",
    "        self.stemmer = SnowballStemmer(\"english\")\n",
    "        self.tokenize = re.compile(r\"(?u)\\b\\w\\w+\\b\")\n",
    "    def __call__(self, articles):\n",
    "        return [self.stemmer.stem(t) for t in self.tokenize.findall(articles)]\n",
    "\n",
    "def define_model():\n",
    "    # [Sentences/Terms, ...] <-> [Feature vectors, ...]\n",
    "    X_vectorizer = TfidfVectorizer(\n",
    "        binary=True,\n",
    "        max_df=0.9,\n",
    "        min_df=0.001,\n",
    "        ngram_range=(1, 2),\n",
    "        #tokenizer=CustomTokenizer() #takes a while\n",
    "    )\n",
    "\n",
    "    # {\"NEGATIVE\", POSITIVE\"} <-> {0, 1}\n",
    "    y_encoder = preprocessing.LabelEncoder()\n",
    "\n",
    "    # [Feature vectors, ...] -> [{0,1}, ...]\n",
    "    clf = LogisticRegression(\n",
    "        penalty='l1',   # Intuitively more likely for certain words to signal\n",
    "                        # positive/negative rating. L1 regularization causes\n",
    "                        # the model coeffs to tend towards sparsity.\n",
    "        solver='liblinear', # liblinear works well for L1-reg on small dataset\n",
    "        multi_class='ovr',\n",
    "        random_state=None,\n",
    "        #cv=5,\n",
    "        #n_jobs=3,\n",
    "        max_iter=10000\n",
    "    )\n",
    "\n",
    "    return X_vectorizer, y_encoder, clf\n",
    "\n",
    "def evaluate(clf, data):\n",
    "    y_pred = clf.predict(data.X)\n",
    "    acc = metrics.accuracy_score(data.y, y_pred)\n",
    "    print(f\"Accuracy on {data.name}  is: {acc}\\n\")\n",
    "    return acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Data:\n",
    "    def __init__(self, name, raw_X, raw_y, X, y):\n",
    "        self.name = name\n",
    "        self.raw_X = raw_X\n",
    "        self.raw_y = raw_y\n",
    "        self.X = X\n",
    "        self.y = y\n",
    "\n",
    "def read_files(X_vectorizer, y_encoder, train_fname, dev_fname):\n",
    "    def read_labeled(fname):\n",
    "        with open(fname, 'r') as file:\n",
    "            raw_X = []\n",
    "            raw_y = []\n",
    "            for line in file:\n",
    "                label, text = line.strip().split('\\t')\n",
    "                raw_X.append(text)\n",
    "                raw_y.append(label)\n",
    "        return raw_X, raw_y\n",
    "\n",
    "    fnf = \"%s set file not found. File name must end in \\\"%s\\\"\"\n",
    "    if train_fname is None:\n",
    "        raise Exception(fnf % ('train', 'train.tsv'))\n",
    "    if dev_fname is None:\n",
    "        raise Exception(fnf % ('dev', 'dev.tsv'))\n",
    "    train_raw_X, train_raw_y = read_labeled(train_fname)\n",
    "    dev_raw_X, dev_raw_y = read_labeled(dev_fname)\n",
    "\n",
    "    train_X = X_vectorizer.fit_transform(train_raw_X)\n",
    "    train_y = y_encoder.fit_transform(train_raw_y)\n",
    "\n",
    "    dev_X   = X_vectorizer.transform(dev_raw_X)\n",
    "    dev_y   = y_encoder.transform(dev_raw_y)\n",
    "\n",
    "    train_data = Data('train', train_raw_X, train_raw_y, train_X, train_y)\n",
    "    dev_data = Data('dev', dev_raw_X, dev_raw_y, dev_X, dev_y)\n",
    "\n",
    "    return train_data, dev_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_size = int(0.95 * len(beer_dat_NLP))\n",
    "train_dat = beer_dat_NLP[:train_size]\n",
    "with open('data/train.tsv', 'w') as train:\n",
    "    for score, text in zip(train_dat['review/overall'], train_dat['review/text']):\n",
    "        label = ('POSITIVE', 'NEGATIVE')[score < 4]\n",
    "        text = text.replace('\\t', ' ')\n",
    "        train.write(f\"{label}\\t{text}\\n\")\n",
    "dev_dat = beer_dat_NLP[train_size:]\n",
    "with open('data/dev.tsv', 'w') as dev:\n",
    "    for score, text in zip(dev_dat['review/overall'], dev_dat['review/text']):\n",
    "        label = ('POSITIVE', 'NEGATIVE')[score < 4]\n",
    "        text = text.replace('\\t', ' ')\n",
    "        dev.write(f\"{label}\\t{text}\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on train  is: 0.8296785062473677\n",
      "\n",
      "Accuracy on dev  is: 0.8138666666666666\n",
      "\n",
      "150 most positive grams:\n",
      "['nice', 'indeed', 'this is', 'full', 'good', 'up', 'solid', 'very nice', 'with this', 'oil', 'black', 'chocolate', 'molasses', 'tasty', 'wonderful', 'refreshing', 'oak', 'very solid', 'nicely', 'balanced', 'quaffable', 'creamy', 'well', 'right', 'rich', 'rounded', 'very well', 'today', 'lovely', 'easy', 'love', 'damn', 'delicious', 'drinkable', 'ipa', 'perfect', 'fine', 'smooth', 'very smooth', 'coffee', 'complex', 'kbs', 'well done', 'yet', 'founders', 'pleasant', 'session', 'great', 'of flavor', 'very pleasant', 'very drinkable', 'not too', 'not overpowering', 'very easy', 'citrus', 'crisp', 'pleasantly', 'but not', 'easy to', 'wonderfully', 'really nice', 'is good', 'disappoint', 'beautiful', 'amazing', 'helles', 'mocha', 'the coffee', 'bad thing', 'fresh', 'quite good', 'easily', 'nicely done', 'summer', 'sessionable', 'perfectly', 'raspberries', 'raspberry', 'is great', 'great beer', 'and slight', 'grapefruit', 'silky', '16', 'flavorful', 'smoky', 'brown head', 'pineapple', 'fantastic', 'my favorite', 'coffee flavors', 'is high', 'awesome', 'dangerous', 'up there', 'the rye', 'could drink', 'this all', 'all night', 'the best', 'and well', 'year', 'wait', 'low abv', 'highly', 'espresso', 'goodness', 'than the', 'let it', 'must try', 'growler', 'maple', 'scent of', 'decent lacing', 'very refreshing', 'case', 'job', 'lighter', 'really good', 'quite drinkable', 'well worth', 'very good', 'excellent', 'enjoyable', 'very enjoyable', 'release', 'incredible', 'exceptional', 'recommend', 'and one', 'already', 'really enjoyed', 'treat', 'recommended', 'winner', 'beautifully', 'very clean', 'an imperial', 'very balanced', 'just enough', 'german', 'cbs', 'highly recommend', 'superb', 'thick with', 'extremely drinkable', 'yummy', 'highly drinkable', 'perfection', 'notch']\n",
      "\n",
      "150 most negative grams:\n",
      "['odd', 'aftertaste', 'bottle', 'but', 'weird', 'not', 'bad', 'very sweet', 'bad but', 'skunk', 'couldn', 'to finish', 'bland', 'lacking', 'for me', 'average', 'interesting', 'though', 'sour', 'awful', 'novelty', 'lacks', 'fades quickly', 'but little', 'unfortunately', 'nothing', 'perfect pint', 'some roasted', 'not very', 'ok', 'cloying', 'beer but', 'too', 'too much', 'watery', 'faint', 'astringent', 'hot', 'alcohol and', 'not sure', 'just not', 'try', 'decent', 'brews', 'else', 'good but', 'tastes more', 'kind', 'water', 'least', 'detracts', 'sipper', 'not great', 'harsh', 'flat', 'extract', 'very dry', 'clarity', 'high carbonation', 'drink but', 'not something', 'more than', 'than one', 'for my', 'my taste', 'wouldn', 'just too', 'drinkable but', 'not session', 'more hops', 'disappointing', 'brew but', 'strange', 'metallic', 'little too', 'boring', 'not really', 'colour', 'caramel with', 'bit too', 'needs', 'hard time', 'revisit', 'don think', 'worth try', 'and alcohol', 'not my', 'imo', 'bubbles that', 'fairly easy', 'cheap', 'hoping', 'terrible', 'not one', 'not impressed', 'decent enough', 'corn', 'weak', 'cut', 'euro', 'alright', 'okay', 'with pretty', 'unbalanced', 'medicinal', 'sweet for', 'tasty but', 'rough', 'dull', 'putting', 'dr', 'trafalgar', 'but one', 'not good', 'difficult', 'try but', 'fusel', 'common', 'unpleasant', 'move', 'way too', 'great summer', 'worse', 'tough', 'just didn', 'would not', 'fairly drinkable', 'and done', 'potential', 'disappointment', 'horrible', 'light beer', 'that disappears', 'cardboard', 'unremarkable', 'great but', 'that great', 'worst', 'the worst', 'unless', 'just don', 'infected', 'attempt', 'meh', 'thin bodied', 'are too', 'suffers', 'drinkability suffers', 'drain', 'sub']\n"
     ]
    }
   ],
   "source": [
    "X_vectorizer, y_encoder, clf = define_model()\n",
    "\n",
    "train_data, dev_data = read_files(X_vectorizer, y_encoder, \"data/train.tsv\", \"data/dev.tsv\")\n",
    "\n",
    "clf.fit(train_data.X, train_data.y)\n",
    "\n",
    "evaluate(clf, train_data)\n",
    "evaluate(clf, dev_data)\n",
    "\n",
    "N = 150\n",
    "print(f\"{N} most positive grams:\")\n",
    "w = []\n",
    "indices = np.argpartition(clf.coef_, -N)[0][-N:]\n",
    "for k, v in X_vectorizer.vocabulary_.items():\n",
    "    for i in indices:\n",
    "        if v == i:\n",
    "            w.append(k)\n",
    "print(w)\n",
    "\n",
    "print(f\"\\n{N} most negative grams:\")\n",
    "w = []\n",
    "indices = np.argpartition(-clf.coef_, -N)[0][-N:]\n",
    "for k, v in X_vectorizer.vocabulary_.items():\n",
    "    for i in indices:\n",
    "        if v == i:\n",
    "            w.append(k)\n",
    "print(w)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observation\n",
    "* The Accuracy of our prediction model is reasonably high at 83%.\n",
    "* some of interesting positive traits are \"chocolate\", \"molasses\", \"balanced\", \"rounded\", \"decent lacing\", \"maple\", \"german\".\n",
    "* some of interesting negative traits are \"bland\", \"water\", \"sour\", \"strange\", \"fade quickly\", \"cheap\"."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Part 2: ETHICS (How Alcohol Consumption Contributes to Driving Deaths)\n",
    "\n",
    "In many societies, drinking alcohol is forbidden because it causes serious problems, e.g. DUI (Driving under the influence), crime, etc. \n",
    "\n",
    "For that reason, in this part of the project, we will also focus on the correlation between alcohol consumption and driving deaths.\n",
    "\n",
    "We will use data from Occupant and Alcohol-Impaired Driving Deaths in States (https://healthdata.gov/dataset/occupant-and-alcohol-impaired-driving-deaths-states-2005-2014) as well as BRFSS Table of Alcohol Consumption (https://chronicdata.cdc.gov/Behavioral-Risk-Factors/BRFSS-Table-of-Alcohol-Consumption/dts9-xy2f) to present such a correlation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Only keeping years before 2014 since our DUI data is from 2005-2014\n",
    "# The overall evaluation is selected and we only care about people who response yes in the survey data\n",
    "alc_dat_overall = alc_consump_dat[alc_consump_dat['Year'] <= 2014]\n",
    "alc_dat_overall = alc_dat_overall[alc_dat_overall['Break_Out'] == 'Overall']\n",
    "alc_dat_overall = alc_dat_overall[alc_dat_overall['Response'] == 'Yes']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "alc_dat_overall.drop(columns = ['Question','Response', 'Confidence_limit_Low', 'Confidence_limit_High', \n",
    "                                'Display_order', 'Data_value_unit','Data_value_type', \n",
    "                                'Break_Out', 'Sample_Size'], inplace = True)\n",
    "alc_dat_overall.reset_index(drop = True, inplace = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### We will assign weights for different types of alcohol consumption based on addiction level:\n",
    "* Regular Alcohol Consumption is the least serious type of addiction, with an addiction index of 1\n",
    "* Binge Drinking has an addiction index of 2\n",
    "* Heavy Drinking has an addiction index of 3 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "#create a function that quantifies drinking level\n",
    "def quantify_topic(drinking_level):\n",
    "    return_val = 0\n",
    "    if 'Alcohol Consumption' in drinking_level:\n",
    "        return_val = 1\n",
    "    elif 'Binge Drinking' in drinking_level:\n",
    "        return_val = 2\n",
    "    elif 'Heavy Drinking' in drinking_level:\n",
    "        return_val = 3\n",
    "    return return_val"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "alc_dat_overall['Topic'] =alc_dat_overall['Topic'].apply(quantify_topic)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "alc_dat_overall['multi'] = alc_dat_overall['Topic'] * alc_dat_overall['Data_value']\n",
    "alc_dat_overall.sort_values('Locationabbr')\n",
    "print()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# generate an alcholic rating list of states\n",
    "result = {}\n",
    "for loc in alc_dat_overall['Locationabbr'].unique():\n",
    "    df = alc_dat_overall[alc_dat_overall['Locationabbr']==loc].copy()\n",
    "    df['val'] = df['Topic']*df['Data_value']\n",
    "    average = int(df['val'].sum(axis=0) / 4)\n",
    "    result.setdefault(loc, average)\n",
    "final = pd.DataFrame(columns=['State', 'Alcoholic Rating'])\n",
    "final['State'] = alc_dat_overall['Locationabbr'].unique()\n",
    "final['Alcoholic Rating'] = final['State'].map(result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "#calculating percentage death caused by DUI\n",
    "hold = []\n",
    "i = 0\n",
    "while i < dui_dat.shape[0]:\n",
    "    hold.append(dui_dat.iloc[i]['Alcohol-Impaired Driving Deaths']/dui_dat.iloc[i]['Occupant Deaths']*100)\n",
    "    i+=1\n",
    "dui_dat['% Death Caused by DUI'] = hold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "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>State</th>\n",
       "      <th>Alcohol-Impaired Driving Deaths</th>\n",
       "      <th>Occupant Deaths</th>\n",
       "      <th>% Death Caused by DUI</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>HI</td>\n",
       "      <td>453</td>\n",
       "      <td>587</td>\n",
       "      <td>77.17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>DC</td>\n",
       "      <td>96</td>\n",
       "      <td>130</td>\n",
       "      <td>73.85</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>RI</td>\n",
       "      <td>265</td>\n",
       "      <td>429</td>\n",
       "      <td>61.77</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>CT</td>\n",
       "      <td>1049</td>\n",
       "      <td>1806</td>\n",
       "      <td>58.08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>DE</td>\n",
       "      <td>430</td>\n",
       "      <td>767</td>\n",
       "      <td>56.06</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  State  Alcohol-Impaired Driving Deaths  Occupant Deaths  \\\n",
       "0    HI                              453              587   \n",
       "1    DC                               96              130   \n",
       "2    RI                              265              429   \n",
       "3    CT                             1049             1806   \n",
       "4    DE                              430              767   \n",
       "\n",
       "   % Death Caused by DUI  \n",
       "0                  77.17  \n",
       "1                  73.85  \n",
       "2                  61.77  \n",
       "3                  58.08  \n",
       "4                  56.06  "
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dui_dat['% Death Caused by DUI'] = dui_dat['% Death Caused by DUI'].round(2)\n",
    "dui_dat.sort_values(by = ['% Death Caused by DUI'], inplace = True, ascending=False)\n",
    "dui_dat.reset_index(drop = True, inplace = True)\n",
    "dui_dat.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# a bar graph to show general information about %Death caused by DUI\n",
    "\n",
    "dui_dat.plot('State', '% Death Caused by DUI', kind = 'bar', figsize = (15, 10))\n",
    "plt.title(\"% Death Caused by DUI by States\")\n",
    "plt.xlabel(\"States\")\n",
    "plt.ylabel(\"Percentage\")\n",
    "plt.legend(\"\")\n",
    "\n",
    "labels = [20, 30,40,50, 60,70,80]\n",
    "plt.yticks(labels)\n",
    "plt.ylim(20,80)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observation:\n",
    "* From the plot above, we can see that Rhode Island is the state with the highest percentage of driving death due to DUI (61.77%)\n",
    "\n",
    "\n",
    "* Even though it may seems that Hawaii and DC are the states with the highest percentage of driving death due to DUI, it turns out that these two states are outliers. This is pointed out in the box plots below. We believe this is due to the fact that Hawaii and DC are small states with low population, meaning that there will be less traffic. Hence, people are less likely to be involved in traffic accidents unless they are driving under influence.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### BOX PLOTS SHOWING THE SPREAD OF THE DATA "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a23cb89e8>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# check if there are outliers\n",
    "dui_dat.boxplot(column = ['% Death Caused by DUI'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "    It appears that 2 states/regions are outliers in our dataset. They are Hawaii and DC."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a241f60f0>"
      ]
     },
     "execution_count": 32,
     "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": [
    "# remove outliers\n",
    "dui_dat.drop([0, 1], inplace = True)\n",
    "dui_dat.reset_index(drop = True)\n",
    "dui_dat.boxplot(column = ['% Death Caused by DUI'])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### OBSERVATION: \n",
    "Out of 49 observations, the mean is 43% with lower quartile 38% and upper quartile at 49%."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Bar Chart Representing the Percentage of DUI Deaths by States (With Outliers Removed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "dui_dat.plot('State', '% Death Caused by DUI', kind = 'bar', figsize = (15, 10))\n",
    "plt.title(\"% Death Caused by DUI by States\")\n",
    "plt.xlabel(\"States\")\n",
    "plt.ylabel(\"Percentage\")\n",
    "plt.legend(\"\")\n",
    "\n",
    "labels = [20, 30,40,50, 60, 70]\n",
    "plt.yticks(labels)\n",
    "plt.ylim(20,70)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Observation: \n",
    "\n",
    "* After removing the outliers (Hawaii and DC), it is clear that Rhode Island is the state with the highest percentage of driving death due to DUI (61.77%)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    49.00\n",
       "mean     43.44\n",
       "std       7.75\n",
       "min      24.23\n",
       "25%      37.74\n",
       "50%      42.15\n",
       "75%      49.34\n",
       "max      61.77\n",
       "Name: % Death Caused by DUI, dtype: float64"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dui_dat['% Death Caused by DUI'].describe().round(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "merged_final = final.merge(dui_dat, how='inner', on=['State'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "merged_final.plot.scatter(x='Alcoholic Rating',y='% Death Caused by DUI', figsize=(12,8))\n",
    "sns.regplot(x='Alcoholic Rating',y='% Death Caused by DUI', data=merged_final, fit_reg=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### OBSERVATION: \n",
    "* We define \"State Alcohol-Addiction Index\" as higher index means having more people with alcohol addiction\n",
    "* There exists a positive correlation between State Alcohol-Addiction Index and % death caused by DUI"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Ethics and Privacy\n",
    "\n",
    "\n",
    "### Maintain Data Anonymously:\n",
    "\n",
    "\n",
    "- Data Collection:\n",
    "    The DUI and Alcohol Consumption data were both collected by BRFSS, a national phone survey program certified by the CDC. Random sampling was used to collect data. We have scrambled out geographic of the phone locations. The alcohol ratings data is made publicly available by Kaggle, a trusted data science website. The online ID for each user review is dropped. The specific collection method is unknown but it is trusted.\n",
    "    \n",
    "\n",
    "- Data Storage: Since all data used is publicly available, there is no concern of protect and secure the data. After the project submission, we plan to scramble all data immediately.\n",
    "\n",
    "\n",
    "- For analysis and modeling, we follow the basic guidelines of fairness, explainability and bias. The code is attached for specific examinations\n",
    "\n",
    "     * The beer-rating data is taken from Kaggle.com. The dataset contains private information such as username, age and birthday. However, we followed the safety-harbor method to safely remove such information to protect user privacy. We expect little bias in this dataset since it is simply an online survey of beer ratings. However, it is possible that non-drinkers are likely to give a lower rating. We minimized such bias by only focusing on the 5 most-popular beer types, thus ignoring outliers. \n",
    "     \n",
    "     * The BRFSS: Table of Alcohol Consumption dataset contains sensitive information such as gender, age, income group and GeoLocation. Again, we followed the safety-harbor method to safely remove such information to protect user privacy. We expect no bias in this dataset since it covers all gender, age and income groups among all states in America. \n",
    "     \n",
    "     * The Occupant and Alcohol-Impaired Driving Deaths in States, 2005-2014 dataset contains GeoLocation information, which we removed to protect user privacy. We expect no bias in this dataset since it covers DUI data from all states in America.\n",
    " \n",
    "\n",
    "### Ethical Considerations:\n",
    "This study will not be used to advocate alcohol usage among underage students\n",
    "Although the result may show preference and guidance of reviewing beers, any consequence of binge drinking is not the intention of our study.\n",
    "To prevent the negative consequences that the study will bring, we included an additional part where we explore the correlation between alcohol consumption and DUI rates\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the first part of our study about beer ratings, we found out that the most popular beer styles among American drinkers are \"American Double/Imperial Stout\", \"American IPA\", \"American Double/Imperial IPA\", \"Scotch Ale/Wee Heavy\", \"Russian Imperial Stout\". And among all beer characteristics, \"taste\" and \"palate\" are the ones that have most impact on a person's overall rating of beer. This suggests that beer consumers mainly value how beer tastes rather than fancy packaging on the visual level.\n",
    "\n",
    "We created two prediction models for different purposes: \n",
    "* In first model, we are able to predict a user's subjective overall rating from their ratings in \"taste\", \"palate\", \"aroma\" and \"appearance\". And the result is reasonably accurate with MSE less than 1.7 (each individual rating ranges from 0.0 to 5.0). \n",
    "* In the second model, we were able to extract some of the positive and negative traits of a good beer as presented earlier. These are the traits the beer manufacturers can consider and put more emphasis on in future production of beer.\n",
    "\n",
    "In the second part of our study about DUI, we have a solid conclusion that driving under influences are very deadly, as it makes up for 43% of total driving deaths on average. And, as the percentage of people consuming alcohols increases, the percentage of death under DUI also increases, which indicates a strong positive correlation. This implies that the region with stronger drinking culture are more likely to ignore DUI laws, which often results in more deaths due to DUI. We believe that this study about DUI will help the society recognize the seriousness of driving under influence as it is the most direct cause of driving deaths."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}