07-performance-evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Performance Evaluation\n",
    "\n",
    "In this notebook we compare recomendations based on the embedding similarity match with recomendations based on user correlation. The later is just a reference. We don't have the \"correct\" similars to measure performance properly. In an e-commerce, for example, we compare algorithms using A/B testing with business metrics like CTR and conversion. The better the business metric, the better the recommendation approach."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from plotnine import *\n",
    "\n",
    "from src.evaluation import spearman_corr, precision_recall"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Ranking Correlation\n",
    "\n",
    "In the Similarity Match notebook we have computed the cosine similarity between pairs of movies. Now, we will compare the user correlation ranking with the cosine similarity ranking based from the embedding features. In addition, we will check if the embedding features carries any type of popularity bias.\n",
    "\n",
    "The Spearman Ranking Correlation Coefficient will give us a value between -1 (exactly opposite order) and +1 (same order). Like the classical Pearson's Correlation Coefficient, values above 0.3 can be seen as an evidence of correlation.\n",
    "\n",
    "For each target movie we will compute the correlation between the user similarity ranking and the cosine similarity ranking. Then, we will evaluate the distribution of the correlation measures."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "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>num_votes</th>\n",
       "      <th>num_likes</th>\n",
       "      <th>avg_rating</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>68410</td>\n",
       "      <td>50939.0</td>\n",
       "      <td>3.886508</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>27138</td>\n",
       "      <td>12345.0</td>\n",
       "      <td>3.246499</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15579</td>\n",
       "      <td>5861.0</td>\n",
       "      <td>3.173792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2989</td>\n",
       "      <td>848.0</td>\n",
       "      <td>2.874540</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>15466</td>\n",
       "      <td>5064.0</td>\n",
       "      <td>3.077169</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    num_votes  num_likes  avg_rating\n",
       "id                                  \n",
       "1       68410    50939.0    3.886508\n",
       "2       27138    12345.0    3.246499\n",
       "3       15579     5861.0    3.173792\n",
       "4        2989      848.0    2.874540\n",
       "5       15466     5064.0    3.077169"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ratings = pd.read_csv('data/movie_ratings.csv')\n",
    "ratings['like'] = np.where(ratings['rating'] >= 3.5, 1.0, 0.0)\n",
    "ratings = ratings.groupby('id').agg(\n",
    "    num_votes=('user', 'count'),\n",
    "    num_likes=('like', 'sum'),\n",
    "    avg_rating=('rating', 'mean')\n",
    ")\n",
    "ratings.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ranking with Text Embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "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>id_right</th>\n",
       "      <th>similarity</th>\n",
       "      <th>id_left</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>4886</td>\n",
       "      <td>0.433568</td>\n",
       "      <td>2</td>\n",
       "      <td>0.235485</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>110</td>\n",
       "      <td>0.401254</td>\n",
       "      <td>2</td>\n",
       "      <td>0.242349</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4306</td>\n",
       "      <td>0.387088</td>\n",
       "      <td>2</td>\n",
       "      <td>0.240394</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1097</td>\n",
       "      <td>0.383907</td>\n",
       "      <td>2</td>\n",
       "      <td>0.233132</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1291</td>\n",
       "      <td>0.352362</td>\n",
       "      <td>2</td>\n",
       "      <td>0.232228</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   id_right  similarity  id_left    target\n",
       "0      4886    0.433568        2  0.235485\n",
       "1       110    0.401254        2  0.242349\n",
       "2      4306    0.387088        2  0.240394\n",
       "3      1097    0.383907        2  0.233132\n",
       "4      1291    0.352362        2  0.232228"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_ranking = pd.read_csv('output/movie_ranking_tfidf.csv')\n",
    "tfidf_ranking.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "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>corr</th>\n",
       "      <th>num_votes</th>\n",
       "      <th>num_likes</th>\n",
       "      <th>avg_rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.127273</td>\n",
       "      <td>27138</td>\n",
       "      <td>12345.0</td>\n",
       "      <td>3.246499</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>-0.515152</td>\n",
       "      <td>25699</td>\n",
       "      <td>14913.0</td>\n",
       "      <td>3.568368</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>-0.054545</td>\n",
       "      <td>35900</td>\n",
       "      <td>21916.0</td>\n",
       "      <td>3.613858</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>-0.187879</td>\n",
       "      <td>62129</td>\n",
       "      <td>55297.0</td>\n",
       "      <td>4.292038</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>-0.163636</td>\n",
       "      <td>23348</td>\n",
       "      <td>14857.0</td>\n",
       "      <td>3.746595</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        corr  num_votes  num_likes  avg_rating\n",
       "2  -0.127273      27138    12345.0    3.246499\n",
       "21 -0.515152      25699    14913.0    3.568368\n",
       "34 -0.054545      35900    21916.0    3.613858\n",
       "50 -0.187879      62129    55297.0    4.292038\n",
       "62 -0.163636      23348    14857.0    3.746595"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_corr = spearman_corr(tfidf_ranking, 'similarity', 'target', 'id_left')\n",
    "tfidf_corr = tfidf_corr.join(ratings, how='inner')\n",
    "tfidf_corr.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:ylabel='Frequency'>"
      ]
     },
     "execution_count": 43,
     "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": [
    "tfidf_corr['corr'].plot.hist(bins=30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For the greatest part of the target movies we have a positive rankling correlation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    382.000000\n",
       "mean       0.122926\n",
       "std        0.369945\n",
       "min       -0.733333\n",
       "25%       -0.151515\n",
       "50%        0.151515\n",
       "75%        0.406061\n",
       "max        0.915152\n",
       "Name: corr, dtype: float64"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_corr['corr'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='corr', ylabel='num_votes'>"
      ]
     },
     "execution_count": 45,
     "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": [
    "tfidf_corr.plot.scatter(x='corr', y='num_votes')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From this scatter plot we can see that there is no popularity bias in the model performance."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ranking with Content Embedding (Text + Crew)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "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>id_right</th>\n",
       "      <th>similarity</th>\n",
       "      <th>id_left</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6377</td>\n",
       "      <td>0.382239</td>\n",
       "      <td>2</td>\n",
       "      <td>0.229354</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4306</td>\n",
       "      <td>0.336810</td>\n",
       "      <td>2</td>\n",
       "      <td>0.240394</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>364</td>\n",
       "      <td>0.310683</td>\n",
       "      <td>2</td>\n",
       "      <td>0.335812</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1097</td>\n",
       "      <td>0.279052</td>\n",
       "      <td>2</td>\n",
       "      <td>0.233132</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4886</td>\n",
       "      <td>0.272782</td>\n",
       "      <td>2</td>\n",
       "      <td>0.235485</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   id_right  similarity  id_left    target\n",
       "0      6377    0.382239        2  0.229354\n",
       "1      4306    0.336810        2  0.240394\n",
       "2       364    0.310683        2  0.335812\n",
       "3      1097    0.279052        2  0.233132\n",
       "4      4886    0.272782        2  0.235485"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_crew_ranking = pd.read_csv('output/movie_ranking_tfidf_crew.csv')\n",
    "tfidf_crew_ranking.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "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>corr</th>\n",
       "      <th>num_votes</th>\n",
       "      <th>num_likes</th>\n",
       "      <th>avg_rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.066667</td>\n",
       "      <td>27138</td>\n",
       "      <td>12345.0</td>\n",
       "      <td>3.246499</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>0.272727</td>\n",
       "      <td>25699</td>\n",
       "      <td>14913.0</td>\n",
       "      <td>3.568368</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>-0.066667</td>\n",
       "      <td>35900</td>\n",
       "      <td>21916.0</td>\n",
       "      <td>3.613858</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>-0.248485</td>\n",
       "      <td>62129</td>\n",
       "      <td>55297.0</td>\n",
       "      <td>4.292038</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>-0.212121</td>\n",
       "      <td>23348</td>\n",
       "      <td>14857.0</td>\n",
       "      <td>3.746595</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        corr  num_votes  num_likes  avg_rating\n",
       "2  -0.066667      27138    12345.0    3.246499\n",
       "21  0.272727      25699    14913.0    3.568368\n",
       "34 -0.066667      35900    21916.0    3.613858\n",
       "50 -0.248485      62129    55297.0    4.292038\n",
       "62 -0.212121      23348    14857.0    3.746595"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_crew_corr = spearman_corr(tfidf_crew_ranking, 'similarity', 'target', 'id_left')\n",
    "tfidf_crew_corr = tfidf_crew_corr.join(ratings, how='inner')\n",
    "tfidf_crew_corr.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:ylabel='Frequency'>"
      ]
     },
     "execution_count": 48,
     "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": [
    "tfidf_crew_corr['corr'].plot.hist(bins=30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that the number of movies with correlation above 0.25 decreased."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    382.000000\n",
       "mean       0.091607\n",
       "std        0.361220\n",
       "min       -0.781818\n",
       "25%       -0.175758\n",
       "50%        0.090909\n",
       "75%        0.369697\n",
       "max        0.890909\n",
       "Name: corr, dtype: float64"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_crew_corr['corr'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='corr', ylabel='num_votes'>"
      ]
     },
     "execution_count": 50,
     "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": [
    "tfidf_crew_corr.plot.scatter(x='corr', y='num_votes')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again, popularity bias is not observed."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ranking with Content Embedding (Text + Crew + Cast)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "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>id_right</th>\n",
       "      <th>similarity</th>\n",
       "      <th>id_left</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>4306</td>\n",
       "      <td>0.384327</td>\n",
       "      <td>2</td>\n",
       "      <td>0.240394</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2716</td>\n",
       "      <td>0.223461</td>\n",
       "      <td>2</td>\n",
       "      <td>0.222825</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>364</td>\n",
       "      <td>0.164086</td>\n",
       "      <td>2</td>\n",
       "      <td>0.335812</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>736</td>\n",
       "      <td>0.158615</td>\n",
       "      <td>2</td>\n",
       "      <td>0.256185</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2011</td>\n",
       "      <td>0.147242</td>\n",
       "      <td>2</td>\n",
       "      <td>0.248665</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   id_right  similarity  id_left    target\n",
       "0      4306    0.384327        2  0.240394\n",
       "1      2716    0.223461        2  0.222825\n",
       "2       364    0.164086        2  0.335812\n",
       "3       736    0.158615        2  0.256185\n",
       "4      2011    0.147242        2  0.248665"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_crew_cast_ranking = pd.read_csv('output/movie_ranking_tfidf_crew_cast.csv')\n",
    "tfidf_crew_cast_ranking.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "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>corr</th>\n",
       "      <th>num_votes</th>\n",
       "      <th>num_likes</th>\n",
       "      <th>avg_rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.175758</td>\n",
       "      <td>27138</td>\n",
       "      <td>12345.0</td>\n",
       "      <td>3.246499</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>0.454545</td>\n",
       "      <td>25699</td>\n",
       "      <td>14913.0</td>\n",
       "      <td>3.568368</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>-0.248485</td>\n",
       "      <td>35900</td>\n",
       "      <td>21916.0</td>\n",
       "      <td>3.613858</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>0.248485</td>\n",
       "      <td>62129</td>\n",
       "      <td>55297.0</td>\n",
       "      <td>4.292038</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>-0.406061</td>\n",
       "      <td>23348</td>\n",
       "      <td>14857.0</td>\n",
       "      <td>3.746595</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        corr  num_votes  num_likes  avg_rating\n",
       "2  -0.175758      27138    12345.0    3.246499\n",
       "21  0.454545      25699    14913.0    3.568368\n",
       "34 -0.248485      35900    21916.0    3.613858\n",
       "50  0.248485      62129    55297.0    4.292038\n",
       "62 -0.406061      23348    14857.0    3.746595"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_crew_cast_corr = spearman_corr(tfidf_crew_cast_ranking, 'similarity', 'target', 'id_left')\n",
    "tfidf_crew_cast_corr = tfidf_crew_cast_corr.join(ratings, how='inner')\n",
    "tfidf_crew_cast_corr.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:ylabel='Frequency'>"
      ]
     },
     "execution_count": 53,
     "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": [
    "tfidf_crew_cast_corr['corr'].plot.hist(bins=30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that the number of movies with negative correlation increased."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    382.000000\n",
       "mean       0.019959\n",
       "std        0.344919\n",
       "min       -0.818182\n",
       "25%       -0.236364\n",
       "50%        0.012121\n",
       "75%        0.281818\n",
       "max        0.793939\n",
       "Name: corr, dtype: float64"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_crew_cast_corr['corr'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='corr', ylabel='num_votes'>"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZgAAAEGCAYAAABYV4NmAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/d3fzzAAAACXBIWXMAAAsTAAALEwEAmpwYAABCwElEQVR4nO2de5wcdZXov6d7HnmREIaH5GVgA2rCmihR0CgXia4sIHivAVlEWBdkdVFQ3q5XEbzeK+JjxaC7LLqCuGKIrkSQvcsmcDUI0UETSABhBJZMgjxCCElI5tF97h9VNVNdXdVd3V3VXdNzvp/PJDO/rsep6qrf+f3O6yeqimEYhmEkTa7VAhiGYRjtiSkYwzAMIxVMwRiGYRipYArGMAzDSAVTMIZhGEYqdLRagCyx//7769y5c1sthmEYxpjiwQcffFFVDwi2m4LxMXfuXHp7e1sthmEYxphCRP4rrD1VE5mIfE9EnheRjb62/UTkbhF5wv1/uu+zz4hIn4j8QUTe62s/UkQedj+7TkTEbe8WkR+77etEZK5vn7PdczwhImeneZ2GYRhGOWn7YL4PHB9ouwJYraqHAavdvxGR+cDpwAJ3n2+LSN7d5zvAecBh7o93zHOA7ao6D/gGcI17rP2AK4GjgLcCV/oVmWEYhpE+qSoYVf0l8FKg+RTgJvf3m4D3+9pvVdUBVX0K6APeKiIHA1NV9X51yg7cHNjHO9ZKYKk7u3kvcLeqvqSq24G7KVd0hmEYRoq0IorsIFV9FsD9/0C3fSaw2bddv9s20/092F6yj6oOAzuAngrHKkNEzhORXhHpfeGFFxq4LMMwDMNPlsKUJaRNK7TXu09po+oNqrpYVRcfcEBZEIRhGIZRJ61QMM+5Zi/c/5932/uB2b7tZgFb3fZZIe0l+4hIBzANxyQXdSwjBbbtGmDD5pfZtmug1aIYhpEhWqFgVgFeVNfZwO2+9tPdyLBDcJz5v3HNaDtF5GjXv3JWYB/vWMuANa6f5v8CfyEi013n/l+4bUbC3L5+C0uuWcOZN65jyTVrWLV+S6tFMgwjI6SaByMiPwKOBfYXkX6cyK4vAytE5BzgGeBUAFXdJCIrgEeAYeB8VS24h/o4TkTaROAu9wfgu8APRKQPZ+Zyunusl0Tki8Bv3e2uVtVgsIHRINt2DXD5Tx5i71CRvRQBuOwnD7Fk3v70TOlusXSGYbSaVBWMqv5VxEdLI7b/EvClkPZe4IiQ9r24Cirks+8B34strFEz/dv30JnLjSgXgM5cjv7te0zBGIaRKSe/McaYNX0iQ8ViSdtQscis6RNbJJFhGFnCFIxRNz1TuvnKB97IhM4c+3R3MKEzx1c+8EabvRiGAVgtMqNBTl40kyXz9qd/+x5mTZ9oysUwjBFMwRgN0zOl2xSLYRhlmInMMAzDSAVTMIZhGEYqmIIxjBZjlRCMdsV8MIbRQm5fv4XLf/IQnbkcQ8UiX/nAGzl5UWhdVsMYc9gMxjBahL8Sws6BYfYOFbnsJw/ZTMZoG0zBGEaL8Coh+PEqIRhGO2AKxjBahFVCMNodUzCG0SKsEoLR7piT3zBaiFVCMNoZUzCG0WKsEoLRrpiJzDAMw0gFUzCGYRhGKpiCMQwjdaxawfjEfDCGYaSKVSsYv9gMxjCM1LBqBeMbUzCGYaSGVSsY35iCMQwjNaxawfjGFIxhGKlh1QrGN+bkNwwjVaxawfjFFIxhtCHbdg1kqkO3agXjE1MwhtFmWFiwkRXMB2MYbYSFBRtZwhSMYbQRFhZsZAlTMIbRRlhYsJElTMEYbc14q4FlYcFGljAnv9G2jFdnt4UFG1nBFIzRlvid3XtxTEaX/eQhlszbf1x0uBYWbGSBlpnIROTTIrJJRDaKyI9EZIKI7Ccid4vIE+7/033bf0ZE+kTkDyLyXl/7kSLysPvZdSIibnu3iPzYbV8nInNbcJlGizBnt2G0npYoGBGZCVwALFbVI4A8cDpwBbBaVQ8DVrt/IyLz3c8XAMcD3xaRvHu47wDnAYe5P8e77ecA21V1HvAN4JomXJqREczZbRitp5VO/g5gooh0AJOArcApwE3u5zcB73d/PwW4VVUHVPUpoA94q4gcDExV1ftVVYGbA/t4x1oJLPVmN0b7Y85uw2g9LfHBqOoWEfkq8AywB/gPVf0PETlIVZ91t3lWRA50d5kJPOA7RL/bNuT+Hmz39tnsHmtYRHYAPcCLfllE5DycGRBz5sxJ7iKNlmPObsNoLa0ykU3HmWEcAswAJovImZV2CWnTCu2V9iltUL1BVRer6uIDDjigsuDGmKNnSjcLZ+9rysUwWkCrTGTvBp5S1RdUdQj4KfB24DnX7IX7//Pu9v3AbN/+s3BMav3u78H2kn1cM9w04KVUrsYwDMMoo1UK5hngaBGZ5PpFlgKPAquAs91tzgZud39fBZzuRoYdguPM/41rTtspIke7xzkrsI93rGXAGtdPYxiGYTSBVvlg1onISuB3wDDwe+AGYAqwQkTOwVFCp7rbbxKRFcAj7vbnq2rBPdzHge8DE4G73B+A7wI/EJE+nJnL6U24NMMwDMNFbFA/yuLFi7W3t7fVYhhG4mRtfRijvRCRB1V1cbDdMvkNo80ZryVzjNZjxS4No42x9WGMVmIKxjDaGCuZY7QSUzCG0cZYyRyjlZiCMYw2xkrmGK3EnPyG0eZYyRyjVZiCMYxxgK0PY7QCM5EZhmEYqWAKxjAMw0gFUzBGW7Nt1wAbNr9seR+G0QLMB2O0LZbBbhitxWYwRltiGeyG0XpMwRhtiWWwG0brMQVjtCWWwW4YrccUjNE2+B36lsFuGK3HnPxGWxDl0LcMdsNoHaZgjDGP36G/F8csdtlPHmLJvP0tg90wWoiZyIwxjzn0DSObmIIxxjzm0DeMbGIKxhjzmEPfMLKJ+WCMTLNt10AsJ32UQz/u/oZhJI8pGCOz1FrqJejQt1Ix9WOK2UgCUzBGJqkWGZb2/uMZU8xGUpgPxsgkcSLDKlVKtsiy+rAabkaS2AzGyCTVIsOqjbItsqw+PMXszfpgVDHbzM+oFZvBGJmkUmRYnFG2RZbVhylmI0lsBmNklqjIsLijbCsVUzueYr4sMDu0e2fUgykYI9OElXqpZZRtpWJqxxSzkRRmIssottRvNGb+Sp+eKd0snL1vYvfUnufxic1gMoiFiVbHRtljB3uexy82g8kYFiYan6RH2Uby2PM8vjEFkzEsf8NoJxp9ns20NrZpmYIRkX1FZKWIPCYij4rI20RkPxG5W0SecP+f7tv+MyLSJyJ/EJH3+tqPFJGH3c+uExFx27tF5Mdu+zoRmduCy6wZCxM12olGnufb129hyTVrOPPGdSy5Zg2r1m9JS0wjJVo5g/km8O+q+npgIfAocAWwWlUPA1a7fyMi84HTgQXA8cC3RSTvHuc7wHnAYe7P8W77OcB2VZ0HfAO4phkX1SjmwDbaiXqfZzOttQdVnfwi8hXgfwF7gH/HUQafUtVb6j2piEwFjgH+GkBVB4FBETkFONbd7CbgXuBy4BTgVlUdAJ4SkT7grSLyNDBVVe93j3sz8H7gLnefL7jHWgksFxFRVa1X7mZhDmyjnajnebaKAu1BnBnMX6jqK8BJQD9wOHBpg+c9FHgB+BcR+b2I3Cgik4GDVPVZAPf/A93tZwKbffv3u20z3d+D7SX7qOowsAPoCQoiIueJSK+I9L7wwgsNXlZyjHUHttnODT+1Ps9mKm4P4iiYTvf/E4AfqepLCZy3A3gz8B1VfROwG9ccFoGEtGmF9kr7lDao3qCqi1V18QEHHFBZaiMWZjs3BdsoZipuD+LkwfxcRB7DMZH9nYgcAOxt8Lz9QL+qrnP/XomjYJ4TkYNV9VkRORh43rf9bN/+s4CtbvuskHb/Pv0i0gFMA5JQjkYFxnKZ/KTWQLG8j2QwU/HYp+oMRlWvAN4GLFbVIeBVHP9G3ajqn4DNIvI6t2kp8AiwCjjbbTsbuN39fRVwuhsZdgiOM/83rhltp4gc7UaPnRXYxzvWMmDNWPC/jHXSCrNOe0aQ1KzLnNPJMtZNxeOdOE7+ScD5wBycaK0ZwOuAOxo89yeBH4pIF/Ak8BEchbdCRM4BngFOBVDVTSKyAkcJDQPnq2rBPc7Hge8DE3Gc+3e57d8FfuAGBLyEE4VmpEwatvO0ZwRJzrrMOW0Yo8Qxkf0L8CDwdvfvfuA2GlQwqroeWBzy0dKI7b8EfCmkvRc4IqR9L66CMppH0tV4m2FyS1IpmHPaMEaJo2D+TFU/KCJ/BaCqe7xkRsMII0nbeTNmBEkqBSt3bxijxFEwgyIyETcCS0T+DDCDslGRpMrkN2NGkLRSMOe0YTjEUTBfwEmwnC0iPwSW4PhLDCN1mjUjSFopxFGwSUWtGUZWkTiBVSLSAxyNk1vygKq+mLZgrWDx4sXa29vbajGMENqtM7ZQ5rFLuz2LSSAiD6pqmU89ThTZalVdCtwZ0mYYTaGdVqYcy7lC4x0bGNRGZB6MiEwQkf2A/UVkulvpeD+3KvGMpkloGG2GLckwNrEcp9qpNIP5W+BTOMrkd772V4DrU5TJMNoaC2Uem1iOU+1EzmBU9Zuqeghwiaoe4vtZqKrLmyijYbQVVmdrbGIDg9qJE0X2TyJyAU55fXBK6P+TWzbGMIw6yFIoszmt42E5TrUTR8F8G6ei8rfdvz+Ms8jXuWkJZRjjgSwELpjTujayNDAYC8RRMG9R1YW+v9eIyIa0BDIMozlYNFt9ZGFg0AjNnLHGUTAFEfkzVf0jgIgcChSq7GMYRsYxp/X4o9kz1jgK5lLgHhF5EifR8rVYJr9hjHnMaT2+aMWMNc56MKtx1l+5wP15narek4o0hmE0DYtmG1+0Iv8qTib/BuBWYIVnJjMMoz0wp/X4oRUz1qozGOBkHJ/LChH5rYhcIiJzUpPIMIymYqtGjg9aMWONVexyZGORw4DPAR9S1XxqUrUIK3ZpGEa7k0YUWd3FLt2d5wKnAR/Emc1clohUhmEYLWK8Jpg2M8w6jg9mHU6i5QrgVFV9MnWpDMOoynjtIJPAEkybQ5wZzNmq+ljUhyJytqrelKBMhmFUwTrI+rEE0+YRJ0w5Urm4XJiQLIZhxMDKxjeGLZfQPOJEkVVDEjiGYRgxsQ6yMSzBtHkkoWDih6EZhtEwY7GD3LZrgA2bX87ELMsSTJtHrCiyKtgMxjCayFgrG59Ff5ElmDaHJBTMfQkcwzCMGhgrHWSWHepjvSryWCBOmPK+wFnAXP/2qnqB+/8nUpLNMIwKjIUO0io2j2/izGB+ATwAPAwUq2xrGLFpdR5Hq88/HhiL/iIjOeIomAmqelHqkhg1M5Y7yLh2+bSuMYt+gXZkrPmLjGSpWotMRD4N7ALuAEZCQFT1pXRFaz5jqRbZWO4gt+0aYMk1a9g7NDqyndCZ477LjyvpeNK6xrjnbwVjedBQiXa9LsOhkVpkg8C1wGcZDUlW4NDkxDNqIcuO0zjEscuneY1Z9QtkedDQqIIYC/4iI3niKJiLgHmq+mLawhjxyGoHGbcTimOXT/Mas+gXyPKgIcuKz8g2cRItNwGvpi2IEZ8sdpC3r9/CkmvWcOaN61hyzRpWrd8SuW2cRLc0rzHNRLt6Ewqzmp1vZWmMRogzgykA60XkHkp9MBc0cmIRyQO9wBZVPUlE9gN+jBMO/TRwmqpud7f9DHCOK8sFqvp/3fYjge8DE3Gi3S5UVRWRbuBm4EhgG/BBVX26EXmzRNYcp/WMvqvlcaR9jWnkkTQy0s/ioAGyO1s2xgZxFMzP3J+kuRB4FJjq/n0FsFpVvywiV7h/Xy4i84HTgQXADOA/ReRwVS0A3wHOwwmj/gVwPHAXjjLarqrzROR04BqctWzahiwl2tXbCVWzy6d9jUn6BRo1cbVi0BDHpJlVxWeMDaoqmDRK8YvILOBE4Es4Ph6AU4Bj3d9vAu4FLnfbb1XVAeApEekD3ioiTwNTVfV+95g3A+/HUTCnAF9wj7USWC4iorUs3zkGyIrjNG1zVhausRqNjPS9jn7JvP257/LjmjJoiDvbClN8nztx/ojpbix8N0briJPJ/xQhBS1VtZEosn/AWRVzH1/bQar6rHvsZ0XkQLd9Js4MxaPfbRtyfw+2e/tsdo81LCI7gB6gLFBBRM7DmQUxZ86cBi5p/JI1k10rqFfJtsKBXutsyz+T3LhlB1+88xFz+BuxiGMi88c2TwBOBfar94QichLwvKo+KCLHxtklpE0rtFfap7xR9QbgBnDyYGLIY4SQJZNdGlQzJ9WjZFsVOVbPbMtr/+AN92cy0s3IJnFMZNsCTf8gImuBz9d5ziXAySJyAo7CmioitwDPicjB7uzlYOB5d/t+YLZv/1nAVrd9Vki7f59+EekApgFtlxiaNcaKOatW4s4yalWyrXKg1zvbMoe/UStVw5RF5M2+n8Ui8jFKTVs1oaqfUdVZqjoXx3m/RlXPBFYBZ7ubnQ3c7v6+CjhdRLpF5BDgMOA3rjltp4gcLSKCU5DTv493rGXuOWx2YpQQJ6S41jDdnindLJy9b6wOt1UOdG+21d0hTOrM090hZbOtsHsza/pE9gwNlxxrz9BwWzr8s7R+zVgmjonsa4yal4ZxQohPTUGWLwMrROQc4BnvHKq6SURWAI+45z/fjSAD+DijYcp3uT8A3wV+4AYEvISjyAxjhLizkkZH7ZVMa630XTkvtDjGZC21KFe6N85YbnSs5vzdXlhiaXLEqUU2AfgApeX6VVWvTle05jOWapEZ9dP33E5O+NZaBoer1yJrpG5Zqwt6RlHpmoDIz/q37+HMG9exc2B0FrNPdwe3nHsUC2fvm7rcaeG//xB9/WYGjKaRWmQ/A14GfgfsTVYso13JanHD29dv4dLbNjBYKB1YRc1K6p1l1OLAb7bvqtKszPs97LN2zIkJDgLOP3ae+ZkSJI6CmaWqx6cuidE2ZNXE4HX6QeUClTvKeiLksuwQr6Yooj5rt3D0sEHA8nueIBiEOtaVaCuJo2B+LSJ/rqoPpy6NMebJctHGsE4foKujei2yWmcZWR7tV1MUlT5rp3D0sOehK5/nvGMO5fp7+9pCibaaOArmHcBfuwmXA7huQVV9Y6qSGWOSsTZy78oLv/jkO5h3UN2BkaFkfbRfSVHEqROXletohKhBwBlHzeGMo+a0hRJtNXEUzF+mLoXRNtQzcm+Wvyaq009auXhkfbRfSVG0ixKpRLVBQLtffzOoGkU2nrAosmRYtX5L2Usb5YNpVamUrHb6rWYs3ptGZR6L11yNZl9TVBSZKRgfpmCSI84DnuWli7NMWp1HVoMzKmEDlHJacU8aCVM2jJqJY2JJyl+T1Auf9Y4D0us8shycEUUrZM66Es7a92gKxmgZSURa1fPChymSrHccEK/zqFdJZjk4A8KvKw2ZK92/rHXeYWTtezQFY7Rs5N5opFU9L3yYIlkyb//MdxxQvfOIurY4322Ww6qjlH/SMlcbZGSt8w4ja9+jKZhxTqtH7o1EWtX6wkcppBs+fGTmOw6o3HmEXdtFK9aTz+Xoylf/brMaVl1tEJGUzHEGK1nrvMPI2vdoCmYck5Upf70hsbW+8GEKKYfwyp7hqsdpRb2w4PkqdR4bNr9cdm3DRRguFhkYjvfdZjGsutogIgmZt+0a4J7HnicfKNwZHGRkrfOOIkvfoymYccxYmPJXotYXPkwhvTpU4OLbNvDBt8xiRW9/6HGaPcurdL6oziPs2oLE+W6zlv8SZxDRiMzeve7ICbsHCyWfhQ1WstR5VyIr36MpmHHMWJjyV6OWF95TSJeu3MDA8Gh4/sBwkRW9/dzxiXewe7BQcpxmz/Kizjf/4KklslUryjlYKFBUGPLVXWv1rKwe0pw1+O91GKctnhV6nqx03mMBUzAZotkv/FiZ8lejlhf+5EUz2XdSJx+75Xe86huxduZy7B4slJWdb/YsL+x8xaLyl9/8JRM6O2paUfO+vhcjv9tW+95qIa1ZQ1RtOo8Vvf1cuPTwMfc+ZAlTMBmhVS981qf8aSjdBTOmUQwkGEfN3Jo9yws7n1f9echdh6XSDMqvbKO+21pmZVmZ5aQxa6hmVhxL5uKsUnXJZCN9al2WN2l6psRf5reZ3L5+C0uuWcOZN65jyTVrWLV+SyLH9WZuEzpz7NPdwYTO6GrKPVO6Oe3IWSVtUaaTJGXr7oheKTIvwj2PPR/r+Qj7br2Rux//ejAead3/rOB/DiZ358s+H2vm4ixiM5gMMNad7WmQtu8j7sxt264BVjzYX9KWtukkyoznsXuwwBd+vonP/uxhPvGuwzjjqDk1yRJnVpaVCMO08T8HG7fs4It3PjKmzcVZwxRMBsiis73VppFmKN245Wyi2tO8L2FmPD+7BhzF87W7H2f5PX1cuyy+STWO7208DXq852Dh7H05/ojXZMIk2C6YgskAWXO2Z8EBHHeUnXZnMLkrXxZltHeoyOSucpNKkpRHhRX5wJtnsmrD1hHl4jEwXKx5dlFtBhd2/wcK6V93Kwg+R6ZYksMUTEbIirM9adNIvUqgmtJtlhLcPVigOy8M+MJ9u/PlORO1EPeeBJ8JgJ/+PtwPUs/solJn6r//4ChVUeWk5WszHXFWK7ev38JlKx8inxMKRa1pJmhUxxRMhsjC6ClJ00ijSiCJKKhGmTV9IpIT8CkYBHbsGWTbroGKvpswJVLrPQk+E2F5PJCOSfXkRTOZf/BUTrjuVwCOki1oU3wxzZidbts1wCW3bSjJFbr4tg1t52dqJRZFZpSQlD8oqci4RqKgkiAYcdaRg6LC+T/8fWRkVVT0VRL35ORFM/n1FUu5+D2H091RPQquUXYPFujuKDWLVbrX23YNsGHzyw1FQDYrem3T1h0lygWcxNRNW3ekcr7xiM1gjBKS8gel6SRudlCEN5PatPUVPnpzLwPDRYYK4TkplWZXSd2TnindfHLpYU1ZN76We52E2bK50WtRoeDRIeJGbZiCMcpIwh+UtBIImkyaHRTRM6WbaRM76crnRopHQrmCqKREkpwdescLVh5Imp4p3Zy2eBY33//MSFtYHlBSiiHq/m3a+grTJnYmnHA7lY6cUxTUoyPntBvJYArGCKVRf1AcJRDXzh4cGX/uxPkcMXNaaO2wNImjICptk4RibHaE37ZdA6zorZ4HlNTsLOz+7Rka5qM398ZadqAWeqZ08/XTFnHpyg3kJUdBi1y7bKH5XxLEFIyRGpVmQnE7yrCR8Wd/tpGJnTmKCtcue2Pqo3iPOAqi2jaNzA5bkfwYV3EkNTuLKto5MBx/2YFayEr0ZrtiCsZIlbCZUC0dZVRBwj1ubkqzo37idEjVtokzO2zWEsHViKs4kjRb+u/fjj2DnP/D34/4vKA1CbdGfZiCMZpOLR1ltYKEXtTPMYcfmJq8QbwOyYuYqleJRNHIEsFJh/fWojjCFGsjeVDe/lmrcmHExxSM0XRqMaf4OzhR2DMcpmyaH/WTli+k77mdXHrbBgYLWvMSwWnJVCkfKWzVzSTlyVqVC6M2RCvUOxpvLF68WHt7e1stxrhg1fotJXb2akUbt+0aYNPWV/jIv/ymJOexIwf/fuExqTn7wzrRbbsGWHLNmpISMhM6c9x3+XENnf/29Vu4dOVDDAaU6D7dHVz/oTePRFEBTZOpkqyVlEfS8rS6Np5RGRF5UFUXB9tbkmgpIrNF5B4ReVRENonIhW77fiJyt4g84f4/3bfPZ0SkT0T+ICLv9bUfKSIPu59dJ+IsrC0i3SLyY7d9nYjMbfqFthFJJND5OXnRTO67/Dg+esyhgHDDL5+smFTXM6WbYw4/gG98cBHdHcKkzjzdHcIZR83hpOVrU0nKi0r4SyPR0/NLBZULwKuDw5x7029H5Liv78VYyac5kVSSBuMkjCZ9j8ISbo3s06pM/mHgYlV9A3A0cL6IzAeuAFar6mHAavdv3M9OBxYAxwPfFhEvvfg7wHnAYe7P8W77OcB2VZ0HfAO4phkX1o5UykyvR+l4+23fPci37+1jYHi0o7pk5UP0Pbczcl8vk/1H5x3NnZ98Jyt6+1NZR6dSJ5pGomdYh+xRUGfRsUrXGCbTq4MFPnpzb+KZ8HGURxYrhAdp9Plt1npNY5mW+GBU9VngWff3nSLyKDATOAU41t3sJuBe4HK3/VZVHQCeEpE+4K0i8jQwVVXvBxCRm4H3A3e5+3zBPdZKYLmIiJpNsCaiIr527h0uWzsjjn3db1oZKDgFFP0MDhc54bpfceX7FnDEzGkVHegbNr+cWlTVpq2vkAv4drwZwTGHHxjpF6jXlFMtmMFP2DWGJUQCDAwnXzssjvLIuu+kXv9QHNOgmfJGabmT3zVdvQlYBxzkKh9U9VkR8UKDZgIP+Hbrd9uG3N+D7d4+m91jDYvIDqAHeDFw/vNwZkDMmTMnsetqF8IivvI54aqfb4p0REcRpqzCGCwon/3ZRqZ05xkuauRLPLkrn8oo2auwOzAcPiO4dtnCUMd3I07tYIc8MFwgl5OypQKirjEsIdKjmtKtp1M8/9h5LL+nryT5MbhvVnNM6s0nqrZfFpa5yBotVTAiMgX4CfApVX3FdZ+EbhrSphXaK+1T2qB6A3ADOE7+ajK3E3E6ltDRakHpzDvOeY84M4cwZTWhM8dwURkulN96b92Ty37yEPMPnsruwULZqoOnLZ7Fit7+xEbJXicSVC4ewRlBkhWe/R3y5K48Jy1fW7ZNd4eEXmNUvhBUVrq1dor+7UE575hDKwZnZDHHpN58okr7AeNiBdBaaZmCEZFOHOXyQ1X9qdv8nIgc7M5eDgaed9v7gdm+3WcBW932WSHt/n36RaQDmAa8lMrFjEHidixhpo7PnTSfL97xSMl2cWYOUWagW889ijNuXMdgiJIB0KJywrfW0pWXEaXjvcQrevtHSsZM7sqze7AwYhuvZ+RcqaP2COuMkixk6W0fNHmdeMRruPr9R4QeL+redndEV1quVSmGbX/9vX2ccdTYmvnX6x+aNX0ie4dL1wHaO+xEL7Zq5dOs0xIF40Z6fRd4VFW/7vtoFXA28GX3/9t97f8qIl8HZuA483+jqgUR2SkiR+OY2M4CvhU41v3AMmCN+V8cau1Ywkwd+3R31Gxf95TVpb4Fnr7ygTey+JAevnrqQi77yUPkpXwxL2exL2VwuPyYnbkcuwcLPL1t94jC3DM0jIgwoSMfa1Tun8nF8YWEdUZpFPcMmrxW/+F5ro7YPqzESrXQ71qVYj1K1G/ODIaSt8pf0Yh/KNiFeH+3auXTrNOqGcwS4MPAwyKy3m37exzFskJEzgGeAU4FUNVNIrICeAQnAu18VfV6oY8D3wcm4jj373Lbvwv8wA0IeAknCq0taPTFrKejCJo66rWvq/evCn6Lpf94G7fu4It3PFLVFwFOJz65Kx/i29HIkvp+wmZywc4njhkuzMkeVnU4LvV8R0ETW7VVNyspxbBnrFYl6t1bLSoDBWVCpxN59pUPvBGFuv0VSSimep7f/u17mNjZwc6B0ZHOxM6OkdlL3JVPx1MgQKuiyNYSnX69NGKfLwFfCmnvBY4Iad+Lq6DaiSQciUkWJqynWKOzGuOof8Xvz/DyHY5f8JqKvojJXXkK6syAdg8WKpq1ojrmqJncfZcfx32XH1fSCVy49PCKnULcqsPBfaKOWe931DOlm7V9L8Y2f5525CxufqBUKUbtHzUDrWZO8/B+v3TlQ4AyMFxbkAg0v5q0n2rfSXDlU8lJ2fc13gIBbEXLjOOPuY/Kzeh7bmdNcfleR+Gt0ljvioiebHHPX0vynado5h20T5msX/rvR/CvHz2a+y4/LrJGl5+ojrmSPMHEvmqJfrUmFlZbtbHe76iWVTO37RpgxYOlSvHHv+3nspUbIvcfnYGO/hX3fnjkc0Jeak/CTGqVVKh/1czzj50XupJonO8rSfnHCi0PUzZGCY5og6Od84+dVzZS9xzg3TWuldFoCKknGzgj0+68IDmpeP56R+VxqhP7zVphPpi4jvF6/Sa1HCuuD6xeM05c01pUCLpjviyPEAQqzkCr3Q+PQtHxqfmJc9+TCqQIu/+X3LaB+QdPZd5B+4TuEyd6rtr31Ypq2K3GFExGCFtU64t3PlLyEiy/p4/giznqAK89NLKSiauS+SbM/DFQUChUTuprxLlazRwXfLmhehRZksmAtZiPwjoafxJnLdcdpBZFF7Ztpc6/lg7Sf2/DfDBAzfc9qQFB2HUMFpyB2leXlQ+Qaomeq/R9jYXqBkljCiYDhD3AV/18E10dpWaErnyO8445lOvv7Yt0gCcxIqpmJ64UyluLIzppJ2fw5Y5z7CTliQpgCBJV1uXcm3r56qkLG7LJ16I0o7aF6M6/lg4yLOjAf48r3feoSs3BSLnzj51X8z2Kml0NDhdDB0hJLeOc5IBmrGDVlH20qpryhs0vc+aN60qiU5wsdS0pfuhVowVKHOBJVtCNUwU3bJsoGcdDpAzUXj14lVs5OZjQ2ZkXHvjM0obvWaXw4Kht/dv0PbeT9ZtfZpHrB/P4/M8eLgkKOOttc7j6lD9vSNYgccqx/HDdM1wfqCRQi2JetX4Ll7jLIvjZp7uDW849qmSV1LDvtiMH+VyurvNH3dtmk2Q0W6aqKRulhJoqVLnyffNDnYaVHOCNjojiOKu96CM/eWHk/Gv7XhxxoL79y6v51uon2tqRCbU5+bftGuC1PZP5wvvml33mLaDWKD1Tunl62+5YlaaDAQy3r9/CScvXctXPH+Gk5WtZtX4L23YN8MvHX+DHIZFySX63cR3hwSKptTrLT140k19c8M4yK0Gl1Tq996y7QxCRus4fdW+bXTzz9vVbePuX1/BX//wAb/9yslXI/ZiJLANETZ1PXjRzJFy33uV5ayXuqonB6KN8Pscdn3gH0yd3jYz2PJPC1+5+nOX39HFtiH3bz1jOD4hrX/ePzoNZ4aM0voBakvW2Lr5tAzmBvOTKZ1wJO6nj+Hlq8WFVYt5B+/DVZeVVKrxBQZQDv95lnCvd26589aTgpN6PbbsGuOS2DQz5Zm9pLT1uCiYjRCmKOE7eWh3B1Y5VzU4c9oJ3552M+t2D4f6ZgQj7tsdYzw+Ic9/iFvucMW1Cw/IkWW9rtCMqV4h7hoYrOqlr7RTjKOpKSxN4hUjjUpLgu2U0wTfsGfTes0rLOFe63kr3dmC4clJwku/Hpq07SpSLJ0caS4+bgskQSSqKRqg2K6rWCUSFp9aa8DjWCgXWE6YaZEJnLpHs77DvaKBQvXRJLcsGAFQoUFtXpxjXEe5Vc/bPqGpdmiBYIuiDN9wf6xmMkrFagmucexv2jiT/fkR9Z8kvPW4Kxhgh2InVG97rhOtucPMlRoka7bZTfoB/lLth88tVS62EkUT2t/87AidXSVQ5afnaivuXR2oVKRSLRBSXZkJHvuFBQ/C5q6So/feiqEpnDvyxJpUGMbXmmFV6BsPC4oOm4eD1joayl78bHmFm1aTfjwUzptKRo+Q77cg57UljCsYAau/EKnUCJy+ayfyDp3L8N3/FcNFXOiNitJtUfkCjNuqw/eMcs1rnFSy14g+zXfqGA1n96PMlNnj/efqe28mlK52llKM6rigZve/hhOt+BcTLVfL283+39/W9GFmItFqVhGqdYqV7FUdpBanm9/JXAy/NMXuC4Ai+2jPolzHu4ncnL5rJvpM6+dgtv+PVwL3syo8uxVCtAGsj+TM9U7r5+mmLuHTlBvKSo6BFrl22MJXBnCkYo+4peKVZzu7BAhM78yWh11GjXS8qLVgTq5YHPqwTOWJG+IqYcfaPW5AxTuflv5de5+2F2d77hxcBCc0Mv339Fi4NCaX1d1zO4milHYVfxt2DBbo78gzW6JD2f7dRhUgbrZJQ63PXv30PWiy9Fx05x8HfHVG1ITzH7BE6c6XKpCufL8kxqzVHpRYlsGDGNIqB9JCujhy/+OQ7mHfQPrEKsDYaLdqsxeBMwYwTanU+NmqiqrV0SjAqLVgo0i+/J7N3LWGdyGf/bWNJQcxq0WvB/eMUZIzbeYXdy2+teaLE0XrdmidKMsO9Y4etkeN3KF+8Yr1r6nBGwxetWF8iY9KFTf2FSButkrBp6w5yUv1eeUzuypdUKwbHzLPyb4+isyM/ck0bNr/M0HCBp7e9yoTOXPmznReGhsvvyRlHzeGMo+bU1enG9R2FbTtYKPKJd81j+uSumgqwNkozfL6mYMYB1cxfaZSwqOWFq6bgvCWM8zlhYMipXtCZz1EoKtcueyOv7Zkc6jz3zDnVZmO11uSqFC4b1Xn572WcKJ6ogIAu3wJiv3z8hTLfyHARNm19hWMOPwCIH91WS8dVqWPyH6uaL+WyEF/E3uECO/YMsW3XQNk5dg8WnPvru3edeaGzI8/C2fuOPOfDhVKfUUDfUygqV75vQcnKqP57Um+nW8usoHQm+wQ3/PJJrr+3r6IvqFKx1axiCmYMUkuHEMcMUYsyqIW4L1yYghssFNixZ5Dep7Zx0Yr1FHwfFwrKkLtc88W3beCuC95Z0XlebTYWVZOrWEe4bLXOC+CVPSErpwXaw47dlZcRMwrA4396JfQ4j/9pVMFA5e8hifBX73n0L2ftX/DMnxXvbT9aNLOUoYLy8VseDJ15Tu7KhyrmyV350Pp4HkV17l13YAG644+oPhOrlVpnBU7CqI6EKdfjC8oypmDGGLV2CHHNX2nZZOO8cP7oGieZr0BR4aM39ZaZRIIMFZStO/aMKMh8Ttg9EM8RHTy/p2D3DhcoFIt0uGVA/JWi/dcSpZjDOi//oGDqxPDXzt8edWx/aZHndoZnfoe1x3Wae4MPiFfqx3se/QEA1RJsq4VrR808dw8WIhf1isq/8vj7E97Am+ZMD80x63tuJ/c89nzV0i1JJwKH3YdGfUFZwxTMGKIeZ3wt5q9m2GSjcLoMQXHWbCpUUSx+tmzfw/wZ07jjE+9g92ChZCQd1yTkKdhNW3fw0Zt7GSjAsHvfVIQ7P/GO0M6nWoLstl0DXLf6iZK6WZ87cX6oqWfBjGmxju1x9CH78c+/eqpMpvfOPyjWfYsafPxw3TN8O9DBhQ1iKs0aPMISbOOGawcHQrOmT6y4qFelYx4xY2rZTAri11ZLIxE46t1sxBeUNawW2RiilnpXHt5IOMl6ZUkzajIpsmeo8jK/YXzm3zZy2j/+mpOWr+W/tu3mQ0e/lvsuP45bzj1qZFEyj0oLTfVM6WbaxC668qXJiF6Vgig853fwnjr1nlbz9bsfL6lb9cU7H+ELJy+guyPHpK483R05vnZqeJhopWP/3b/+ruwFfue8HhYf0lPtlgHRpsnr7+mLtSjWpq2vkIuRnBdWy857Jrvzzv6d+fLjBAdClZ5l77NgbTFwIs06O8oTTPue21miXABuvv8Z+p7bWdKW1kJh1a5nLPpcgtgMZgxRrzM+LfNXUiaDOBnuHjlASgexQHiORy0moaSjrir5GTpzOY6YMY1fX1EeFRQ372Y0esxBgBvPOpKl818TW8ZR0+ToGjafeNc8bvjlkyUZ8lH5K5eFVIMOI+z+BcO18zkBiqgqEzs7Ik1DXm5PWDVif96PP/quI58L/f7Wb345VN71m18uOW6aicDNChcOoxm1/0zBjCEaccYnbf5qxGQQfLBjldDICxcc5ziNN23dEZqoBpVf/DgdRVIBD2E5Gx5ehxv8TsLuaVjns2nrK2XRYwp1zf4UUC1SKORQiuw3uSt2/kpQuXR3CJ8/aQEv7R5keaCUPlBW2QBGqyKPHiPH9R96MwtmTA29555i81bcCfp35h20D189dWGs729RiMksrD2NKEs/rTBNN6v2nymYMUYrRzwejdRGinqwR5z0Idni/iQ0CE9U86j04sftKMLuca2jvbCcDSjN1vYTdk8//eP15AQm+Eb0TicQfu2fXrGeolKToh+dCTnnvHLVRq465YiKyZRhyjMvcNX7FnD6Ua8FKPEheMs3BL/zcCd3jmkTO0NndAAXrdjgrrrpXXN5FeC478i8g/bhrLfN4eb7S30wQV9bWlGWSVHrs9nM2n+mYMYgrXTGQ/0mg0oPdrVscf9L3zOlm9MWzyrpGDpyjimk0otfraOIqsVWz2hv92CBCZ25Mge4+pSD/3xh99S1+jE0UFppd8GMaWVBAgBDhdo6irCZ0HARZk+fVDGpL0x5FhSu+LeN/G7zdj501FxmTZ/Iwtn7Riax7jupixnTJlRU+MH7fvpbZpcoF3DCwu//4zZOWjijpD3uO3L1KX/OWUfPrboAWBYGdmHU82w2s/afKRijZuo1GcR9sI9f8JqK2eLbdg2wIrDwVU6EOyIivfyELeO7bddAWSXci95zOPtN6mJuz6TIDjLKjOPdozA8JbBz7zBX37FppMTL509aELvS7sLZ+/K1UxdyyW0PMVgohm4Tr6OIitTTih301h17I4+4oncLdz70p5E8lrAk2IHhIh/7wYMUUU5bPIsVvf0lpXb6t+9h++7Bsvv+gwf+K/ScLzbobJ930D4tXVmyXuqdiaRt8vNjCsaomXpNBpUe7FpGYmGKqrsjXzHSKyi/X6H4KwZ7x/zfv3hsZPtgYJK/g4yS07tHl7iFKv0Ui8rnb9/oBio4MnumqStXbWI4IkTb3wlEObRr6SjCZkLBcOkw88sre4YqHtefx3LHJ94Rqjhfdf1Ft/5mM/96rlPqZeOWHVz9c0fpDhWLFIM+rAh9eERIFeBKZqN6nNtZXK+o3plIM01+pmCMuqjHZBD1YAM1jcQaHYHFXfjLIyxQyusgK8l58qKZzJg2gWX/9EBJe1h9seEiTJvQya3nHlW2PYT7bsIc2lErMobRM6Wbr526sCSK7Nplb6xoGlTgspUbKh7XozPnhHd733lOpCwwY7CgnPHd33Dl++Zz5aqNJXXVgoTp3byUhyBXUgb1KIqsrlfUyHsQNZM3E5kRSdKjtmr71+MLCnuwN23dETkSg/KM8kZHYLWERXvkc84sKdhBalErjhg7O/JlvpiuvIQqmYtWrOerpy4sczwvObSHq05ZUDXRs9qKjEG27RrgtT2TufOTToJqMFy63DS4AZDItUyCeOv/LJy9b2kSa2D/weEiV616JHLNmUp0dpSGIFerThBXUfQ9t3PEL7N7sNA0n0UtNPoeBGfyaczMTMG0CdVGbZVKujdy7HooN1E5pWH8DBWLbNyygw/ecH/oeRtxulYKi47q/HMIx71uf+54+LmS9gG3FlalcwURETpyWtahDrp5PPddfhxnHT2X7933NCsf3Mzv+1/mxG+tHQnJDVP2O/YMcvUdmypWf/YT9p36M91DC4BKrqZFD/3r//RM6eaYww/k2mULuSRkCYJ8jqiJSyiTOvMjJspghFulwUocRRHM7j9t8ayy52VguMDQcKEs9LoZuSV+GnkPmjEzMwWTAM1+qMLOX2nUVq2ke73Hrvdaw47ZkXNyIEbKqVRZVwXqj6bzJxiGJQqesuhgbl//bEnbUFHLlAtEL3EcPFeYWfDi2zaURYJ5Hd6s6RO57cHNDBWUQV9hz517h0vK4HhO8pyUzyyiRtn1JpwWtEiMCi+j58/nytZ5H/EdfWttiW+qqJStsig4bcFKNN0dOf7xw0cyY9qEMtNONbNRNZNSWHb/it5+/v6E1/P1ux9Hi8pAQSmqsuyfHmBCp+Ogi7t+UBr434Na+qJmRJOZgmmQLDj/Kj0oO/YMVS3pXu+xvc9rVaxhx5zY2cH1H3oz0yZ2Mmv6xJrXComi0mqPOREuXrGBAV8kVndHnr9ZcijvecNBfOrHG0pW5Iyi3koKYR2t1+FFlfS/6uebGCyMzlL8prQgA8OFkdlVMCQ6jGoJp5870fGT+MmJExjQkc+VFRl9dbDAR2/uLZsxT5/cxSffNS80GfPSlQ9RVGWooHR35lCFE+cfyH8+9vzIthe9+3D+feOfWPngM3R3lOYIVVLo/dv3jAxcvOCO84+dVyJzVHb/fpO6uOMT7+CEb60FRmefnukzzvpBaeNf1sLzp1Xqi5oRTWYKpgGy4vyr9KBsfml3xF7x7OhRx65kvqr3mF7Yb9RaIbU+/NXNhg+VKBf/OSZ35Ssql0ldeYpuKC6UZ6mH5dQEn4l5B+3DV5dF2dDDbVG5XEidnACdOWGoqORywknL15aFAv/tMYeW5efsHSqWmfqCirF/+x4mdnaUrFI6uauD//M//pwXdw3wv+58tOyeDQw7Zr/5B08tK0QKWraSp6d0QUdkXP2H5/nhOW/l6W2v8ssnXuB/3zUa4eet1Bm2amhUoufnTpzPtt2DLF/zON+5948sv+eJESU4t2dS6D2d2zPJreacK4sKhHjrB6XJtl0DXBKYEV98W3kSqp9mRJOZgmmAZiYsVSLqQVnb9yKXrXyobPuwyr21HPtzJ87ni3dWNl/VI6+XMR9Ww6u7o7YinZXWsgciS5145+jfvqesNLyf//GmGbzrdQey6dmdXLpyTcmoMa6pxHOye1Wg/QpqwYypZSYjoEwxBOnKC7iFVLxtvVmOdx++ubqvbD+v7H0QTzFu2zXAjj1DZXk3e4cLXHzbevKSi1TIWlRO+NZauvLCroHSkv7X39tXspLn7sFC6AJhZ9y4jo6c8GrE9YeV/PE/T/5n9eo7HmG4UKSgjJgfPbNxZ0e+7L57xTIr+e6cBNDGBkSNEGcRuzDSTiA1BdMAaU4xa/XrBB8UgCXXrCnrRLvywlcDlXurnStsJNuoYo16sMOOPakzzz9++MhYJj2IXss+J8KmrTuYNrGr/Bxdef7xzDePvIxhpeH93LJuM7es21zWftGK9Yg4CZWVlG9Y4IXfyd4zpZsz3jqnzB/gMaEzh6rywbfMLpmdnH+sU6wyqAiq4S97H8Qv63ChSGdemNCRH8kfcnRGtB/KUdLKYMg6a8HnZnJXvkyJeh1nWPDFyDYR713Y8xT2tXpm4wUzptKRz40s1QCjxTL9AyPPF+P3wQC+ZZALnH/sPLbvHmySfzYq+qJ6VEa9vsw4tLWCEZHjgW8CeeBGVf1yksdPa4pZr1/H/6Bs2Pxy1U60lnP5j7199yADw7Ut6lVNXo8wpV1EWeBLpqsWjh21lr3nEwjLmi+qlszqRhIlQxRVJcJCbYOdaGkNsNIRNDASwr3iwf7yg3kogLD4tftx4dLDSwYW199bPkOpRFeF2WGYrBSV6z/0JkA4/4e/KzGZ1cpQ0THNeSbGsEXFqsqfJ1L+sOcp2vTpVDAIliE6bfGskdmQf8bphdn7n0N/hejr1jzB1+5+vEQJpeWfDZvxduQoeW9aQdsqGBHJA9cD7wH6gd+KyCpVfSTJ8yQ9xUzKrxPaUQc60XrO5SkkzxcQtdpjvVRT2tUUYrUcl4Fh5Yt3PlLi7I0aGChOqK0TulwkT02RtCMElW9UDbB//tWTfP/XT9OZc1b1zAVtRT72Do9+X/ddflzJ7Cd4/xa/djpr+7aFHie4DHOQMFkLrnJbMGNq1fI2UUzqzFEETjtyFictX1tifq00cwwiwC8uOCZS/igT7xd+vim0gkFYGaIVvf287qCpZYvYRSmLb615ouTY3owsTf9sz5Ruvn7aopFVYb1ZcSvzdKCNFQzwVqBPVZ8EEJFbgVOARBUMJDvFTMqvE2d2Veu5wlYwrLTaY71EKe16w2uDeOuxVCroGFaSXnJCWcJOFTpDR9fhx/ju2idLTGtxOtmw7yuYzHrS8rVl+03uyo/UC6v03UWVhXllz1DJM5aDSP9IkK68cPUpR7Bo9r6ctHxtyfcZpvxHQrGRkQoKHlO6O6qWCAp7nvaZ0BFawSBs5p8X4ao7Hgn15wWfmzBfiEfa/tksFuRsZwUzE/AbyfuBo4Ibich5wHkAc+bMCX7cdJL061R74Go9V2gNsCqrPdZLmNKuZT2XS0Oi0Dyi1mOpdq7uzhzFwfKk0Ep847RFnLSwdKQbVgOsIydOGG5h9F525wUVcSKXCkX+ZomTfDkQEtYcxLu2sA5zcneeq963gHe9/sCqndDUiZ0V271n7J7HnuezP9sYawGyXE541+sPjPw+w5T/hUsPZ9PWV9xKANWvP0jwu456N0LfiUKRro5ciQ8pWllEzzqb4fRP059SD+28ZHLYN13WNajqDaq6WFUXH3BAPCdymngdZFJLHPdMiV56tdZzNbMKayPnP3nRTH59xVIufs/hdHeMLss7oTMX+36GJhoWlS+ecgTdHbmRJX47c0JXHj7+3w4luOpvRw7e9mf7lx27Z4pTA8y/ZPJVpyygEFjjRnKO+eqWc4/i11ccx+V/+QauXVbbsxF1HXGUC4za9oPX5bft90zp5l2vPxApu37HvOj5ILrd3z2ZK32fwee2Z0o3xxx+QM3XX4mwdyPsnbjyfQvK/DZRz33Y/YLaoyDbBdGIhZvGOiLyNuALqvpe9+/PAKjq/4naZ/Hixdrb29skCSvTzOoAtZxr1fotZWa3ZiaW1np+79rCHLL1nivqmKvWbymzgceRzb9/nGur9dlo9DuLe11h5wnWnQvKXI9szXg36v1uoPR+DRcLfPK4w0tyfdoREXlQVReXtbexgukAHgeWAluA3wJnqOqmqH2ypGCyTBZK42RR+SYhW1rX1iy56jlPq5+nuNQi51i5pqQYdwoGQEROAP4BJ0z5e6r6pUrbm4IxDMOonSgF085OflT1F8AvWi2HYRjGeKSdnfyGYRhGCzEFYxiGYaSCKRjDMAwjFUzBGIZhGKnQ1lFktSIiLwD/ldDh9gdeTOhYSZFFmcDkqpUsypVFmcDkqpV65XqtqpZlqpuCSQkR6Q0L22slWZQJTK5ayaJcWZQJTK5aSVouM5EZhmEYqWAKxjAMw0gFUzDpcUOrBQghizKByVUrWZQrizKByVUricplPhjDMAwjFWwGYxiGYaSCKRjDMAwjFUzBJICI7Ccid4vIE+7/0yO2+7SIbBKRjSLyIxGZkBG59hWRlSLymIg86q6l03K53G3zIvJ7EbkjTZniyiUis0XkHvc+bRKRC1OS5XgR+YOI9InIFSGfi4hc537+kIi8OQ056pDrQ648D4nIr0VkYRbk8m33FhEpiMiyLMgkIseKyHr3Wfp/acsURy4RmSYiPxeRDa5cH6n7ZKpqPw3+AF8BrnB/vwK4JmSbmcBTwET37xXAX7daLvezm4Bz3d+7gH2zIJf7+UXAvwJ3ZOR7PBh4s/v7PjhrDs1PWI488EfgUPf72BA8B3ACcBfOyq1HA+uacH/iyPV2YLr7+19mRS7fdmtwKqwva7VMwL7AI8Ac9+8Ds3CvgL/3nn3gAOAloKue89kMJhlOwemkcf9/f8R2HcBEdzG0ScDWVsslIlOBY4DvAqjqoKq+3Gq5XNlmAScCN6Ysj0dVuVT1WVX9nfv7TuBRnMFDkrwV6FPVJ1V1ELjVlS0o683q8ACwr4gcnLAcNculqr9W1e3unw8As1KWKZZcLp8EfgI8nxGZzgB+qqrPAKhqVuRSYB8REWAKjoIZrudkpmCS4SBVfRacDgg4MLiBqm4Bvgo8AzwL7FDV/2i1XDgjmReAf3FNUTeKyOQMyAXOYnGXAcWIz1slFwAiMhd4E7AuYTlmApt9f/dTrsTibJM0tZ7zHJxZVtpUlUtEZgL/HfjHJsgTSybgcGC6iNwrIg+KyFkZkWs58AacAfDDwIWqWtc72NYLjiWJiPwn8JqQjz4bc//pOCOFQ4CXgdtE5ExVvaWVcuE8A28GPqmq60Tkmzjmoc+1Ui4ROQl4XlUfFJFjG5ElSbl8x5mCMxr+lKq+koRs/sOHtAXzCeJskzSxzyki78JRMO9IVSL3dCFtQbn+AbhcVQvOwDx14sjUARyJs6z7ROB+EXlAVR9vsVzvBdYDxwF/BtwtIr+q5zk3BRMTVX131Gci8pyIHKyqz7pmirCp7ruBp1T1BXefn+LYqxtSMAnI1Q/0q6o3Cl+Jo2AaIgG5lgAni7Ps9QRgqojcoqpntlguRKQTR7n8UFV/2og8EfQDs31/z6LcnBpnm1bIhYi8Eces+Zequi1lmeLKtRi41VUu+wMniMiwqv6shTL1Ay+q6m5gt4j8EliI49dLizhyfQT4sjpOmD4ReQp4PfCbWk9mJrJkWAWc7f5+NnB7yDbPAEeLyCTXtrkUx37fUrlU9U/AZhF5ndu0FMfx2Gq5PqOqs1R1LnA6sKZR5ZKEXO53913gUVX9ekpy/BY4TEQOEZEunOtfFSLrWW402dE4JtdnU5IntlwiMgf4KfDhlEfiNcmlqoeo6lz3eVoJ/F2KyiWWTDjP1ztFpENEJgFHkX6fEEeuZ3D6AUTkIOB1wJN1nS3tqIXx8AP0AKuBJ9z/93PbZwC/8G13FfAYsBH4AdCdEbkWAb3AQ8DPcKOAWi2Xb/tjaU4UWVW5cEw+6t6r9e7PCSnIcgLOSPaPwGfdto8BH3N/F+B69/OHgcVp35+Yct0IbPfdm94syBXY9vukHEUWVybgUpwB3UYcc2vL75X7vP+H+1xtBM6s91xWKsYwDMNIBTORGYZhGKlgCsYwDMNIBVMwhmEYRiqYgjEMwzBSwRSMYRiGkQqmYAzDMIxUMAVjGGMUt2hq5N+G0WrsgTSMDOAWOryE0STO/wl8D6dc+gvAR1T1GRH5Pk512zcBvxORHv/fwMXNl94wwjEFYxgtRkQW4BTbXKKqL4rIfjjLBdysqjeJyN8A1zG6fMDhwLvVKdz4ff/fzZfeMKIxE5lhtJ7jgJWq+iKAqr4EvA1noTVwygr5qxLfFlAmwb8NIxOYgjGM1iNUL7Xv/3x34LPg34aRCUzBGEbrWQ2c5vpTcE1kv8apdAvwIWBti2QzjLoxH4xhtBhV3SQiXwL+n4gUgN8DFwDfE5FLcZ38rZTRMOrBqikbhmEYqWAmMsMwDCMVTMEYhmEYqWAKxjAMw0gFUzCGYRhGKpiCMQzDMFLBFIxhGIaRCqZgDMMwjFT4/3Y7DzYEAUvBAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "tfidf_crew_cast_corr.plot.scatter(x='corr', y='num_votes')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "No popularity bias is present."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ranking with Collaborative Filtering Embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "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>id_right</th>\n",
       "      <th>similarity</th>\n",
       "      <th>id_left</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>253</td>\n",
       "      <td>0.343631</td>\n",
       "      <td>2</td>\n",
       "      <td>0.237054</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>110</td>\n",
       "      <td>0.289399</td>\n",
       "      <td>2</td>\n",
       "      <td>0.242349</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2011</td>\n",
       "      <td>0.279259</td>\n",
       "      <td>2</td>\n",
       "      <td>0.248665</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>165</td>\n",
       "      <td>0.244945</td>\n",
       "      <td>2</td>\n",
       "      <td>0.234542</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1097</td>\n",
       "      <td>0.237925</td>\n",
       "      <td>2</td>\n",
       "      <td>0.233132</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   id_right  similarity  id_left    target\n",
       "0       253    0.343631        2  0.237054\n",
       "1       110    0.289399        2  0.242349\n",
       "2      2011    0.279259        2  0.248665\n",
       "3       165    0.244945        2  0.234542\n",
       "4      1097    0.237925        2  0.233132"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cb_ranking = pd.read_csv('output/movie_ranking_cb.csv')\n",
    "cb_ranking.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "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>corr</th>\n",
       "      <th>num_votes</th>\n",
       "      <th>num_likes</th>\n",
       "      <th>avg_rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.393939</td>\n",
       "      <td>27138</td>\n",
       "      <td>12345.0</td>\n",
       "      <td>3.246499</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>0.115152</td>\n",
       "      <td>25699</td>\n",
       "      <td>14913.0</td>\n",
       "      <td>3.568368</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>-0.151515</td>\n",
       "      <td>35900</td>\n",
       "      <td>21916.0</td>\n",
       "      <td>3.613858</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>-0.236364</td>\n",
       "      <td>62129</td>\n",
       "      <td>55297.0</td>\n",
       "      <td>4.292038</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>0.090909</td>\n",
       "      <td>23348</td>\n",
       "      <td>14857.0</td>\n",
       "      <td>3.746595</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        corr  num_votes  num_likes  avg_rating\n",
       "2  -0.393939      27138    12345.0    3.246499\n",
       "21  0.115152      25699    14913.0    3.568368\n",
       "34 -0.151515      35900    21916.0    3.613858\n",
       "50 -0.236364      62129    55297.0    4.292038\n",
       "62  0.090909      23348    14857.0    3.746595"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cb_corr = spearman_corr(cb_ranking, 'similarity', 'target', 'id_left')\n",
    "cb_corr = cb_corr.join(ratings, how='inner')\n",
    "cb_corr.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:ylabel='Frequency'>"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAD4CAYAAADrRI2NAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/d3fzzAAAACXBIWXMAAAsTAAALEwEAmpwYAAAQwklEQVR4nO3df4xldX3G8fcjYhQlCt0FV8o6agiWNgg4UCu0Qq1GIIi0sZUYpda6mkoiqW1c0SiJaYKNYtu0tYIQkfqjWn5IFVuBWolV0YHys6uidlVgC6slLlojBT/9454119nZnTOzc+7d2e/7ldzMueeec8/Dmcuz55753nNTVUiS2vGoaQeQJE2WxS9JjbH4JakxFr8kNcbil6TGPHraAfpYs2ZNzczMTDuGJK0qN9100/eqau38+aui+GdmZpibm5t2DElaVZJ8e6H5nuqRpMZY/JLUGItfkhpj8UtSYyx+SWqMxS9JjbH4JakxFr8kNcbil6TGrIpP7kqr3czGT/VedvP5pw6YRPKIX5KaY/FLUmMsfklqjMUvSY0ZrPiTHJrks0k2JbkzyRu6+ecluSfJLd3tlKEySJJ2NOSonoeBN1bVzUn2B25Kcm332Huq6l0DbluStBODFX9VbQG2dNMPJtkEHDLU9iRJ/UxkHH+SGeBo4EbgeODsJK8E5hi9K3hggXU2ABsA1q9fP4mY0s/0HXfvmHutRoP/cTfJE4DLgXOqahvwXuAZwFGM3hG8e6H1qurCqpqtqtm1a3f4ykhJ0jINWvxJ9mVU+h+qqisAquq+qnqkqn4KXAQcN2QGSdLPG3JUT4CLgU1VdcHY/HVji50B3DFUBknSjoY8x3888Arg9iS3dPPOBc5MchRQwGbgtQNmkCTNM+Sons8DWeCha4bapiRpcX5yV5IaY/FLUmO8Hr+aspTr4kt7K4/4JakxFr8kNcbil6TGWPyS1BiLX5IaY/FLUmMsfklqjOP4tUfzuvjSyvOIX5IaY/FLUmMsfklqjMUvSY2x+CWpMRa/JDXG4pekxjiOX9oNXt9fq5FH/JLUGItfkhpj8UtSYyx+SWqMxS9JjbH4JakxFr8kNcbil6TGWPyS1BiLX5IaY/FLUmMGK/4khyb5bJJNSe5M8oZu/oFJrk1yV/fzgKEySJJ2NOQR/8PAG6vql4DnAK9PcgSwEbi+qg4Dru/uS5ImZLDir6otVXVzN/0gsAk4BDgduLRb7FLgJUNlkCTtaCLn+JPMAEcDNwIHV9UWGP3jABw0iQySpJHBiz/JE4DLgXOqatsS1tuQZC7J3NatW4cLKEmNGbT4k+zLqPQ/VFVXdLPvS7Kue3wdcP9C61bVhVU1W1Wza9euHTKmJDVlyFE9AS4GNlXVBWMPXQ2c1U2fBXxiqAySpB0N+dWLxwOvAG5Pcks371zgfOBjSV4NfAd46YAZJEnzDFb8VfV5IDt5+PlDbVeStGt+cleSGmPxS1JjhjzHL+3UzMZP7dHPp53ru683n3/qwEm0XB7xS1JjLH5JaozFL0mNsfglqTEWvyQ1xuKXpMZY/JLUGMfxS6uU4+m1XB7xS1JjLH5JaozFL0mNsfglqTEWvyQ1xuKXpMZY/JLUGItfkhpj8UtSYyx+SWqMxS9JjelV/El+ZeggkqTJ6HvE/3dJvpzkj5I8achAkqRh9Sr+qjoBeDlwKDCX5MNJXjBoMknSIHqf46+qu4C3Am8Cngf8VZKvJvntocJJklZer+vxJzkSeBVwKnAtcFpV3ZzkKcAXgSuGiyhpEvpe31+rX98vYvlr4CLg3Kr68faZVXVvkrcOkkySNIi+xX8K8OOqegQgyaOAx1bV/1bVZYOlkyStuL7n+K8DHjd2f79uniRplelb/I+tqh9uv9NN7zdMJEnSkPoW/4+SHLP9TpJnAz/exfIkuSTJ/UnuGJt3XpJ7ktzS3U5ZXmxJ0nL1Pcd/DvDxJPd299cBv7fIOh9g9EfhD86b/56qelffgJKkldWr+KvqK0meCRwOBPhqVf3fIuvckGRm9yNKklZS3yN+gGOBmW6do5NQVfOP5vs4O8krgTngjVX1wEILJdkAbABYv379MjajldJ3fPfm808dOImkldD3Im2XAe8CTmD0D8CxwOwytvde4BnAUcAW4N07W7CqLqyq2aqaXbt27TI2JUlaSN8j/lngiKqq3dlYVd23fTrJRcAnd+f5JElL13dUzx3Ak3d3Y0nWjd09o3teSdIE9T3iXwP8Z5IvAz/ZPrOqXryzFZJ8BDgRWJPkbuDtwIlJjgIK2Ay8dlmpJUnL1rf4z1vqE1fVmQvMvnipzyNJWll9h3N+LslTgcOq6rok+wH7DBtNkjSEvqN6XgP8I/C+btYhwFUDZZIkDajvqZ7XA8cBN8LoS1mSHDRYKq1KXs99ZbgfNbS+o3p+UlUPbb+T5NGM/kArSVpl+hb/55KcCzyu+67djwP/NFwsSdJQ+hb/RmArcDujIZjXMPr+XUnSKtN3VM9PGX314kXDxpEkDa3vl63/Fwuc06+qp694IknSoJZyrZ7tHgu8FDhw5eNIkobW6xx/VX1/7HZPVf0F8JvDRpMkDaHvqZ5jxu4+itE7gP0HSSRpRe1NnwvwuyFWRt9TPePXzX+Y0QXWfnfF00iSBtd3VM9JQweRJE1G31M9f7yrx6vqgpWJI0ka2lJG9RwLXN3dPw24AfjuEKEkScNZyhexHFNVDwIkOQ/4eFX94VDBJEnD6HvJhvXAQ2P3HwJmVjyNJGlwfY/4LwO+nORKRp/gPQP44GCpJEmD6Tuq58+SfBr49W7Wq6rqP4aLpd2xN43b1uq1lNeh4+4nq++pHoD9gG1V9ZfA3UmeNlAmSdKA+n714tuBNwFv7mbtC/z9UKEkScPpe8R/BvBi4EcAVXUvXrJBklalvsX/UFUV3aWZkzx+uEiSpCH1Lf6PJXkf8KQkrwGuwy9lkaRVadFRPUkC/APwTGAbcDjwtqq6duBskqQBLFr8VVVJrqqqZwOWvSStcn1P9XwpybGDJpEkTUTfT+6eBLwuyWZGI3vC6M3AkUMFkyQNY5fFn2R9VX0HOHlCeSRJA1vsiP8qRlfl/HaSy6vqdyaQSZI0oMXO8Wds+ulDBpEkTcZixV87mV5UkkuS3J/kjrF5Bya5Nsld3c8DlvKckqTdt1jxPyvJtiQPAkd209uSPJhk2yLrfgB40bx5G4Hrq+ow4PruviRpgnZ5jr+q9lnuE1fVDUlm5s0+HTixm74U+DdGF3+TJE3IUi7LvBIOrqotAN3Pg3a2YJINSeaSzG3dunViASVpbzfp4u+tqi6sqtmqml27du2040jSXmPSxX9fknUA3c/7J7x9SWrepIv/auCsbvos4BMT3r4kNW+w4k/yEeCLwOFJ7k7yauB84AVJ7gJe0N2XJE1Q32v1LFlVnbmTh54/1DYlSYvbY/+4K0kahsUvSY0Z7FSPJPU1s/FT047QFI/4JakxFr8kNcbil6TGWPyS1BiLX5IaY/FLUmMsfklqjOP4JTVrpT8/sPn8U1f0+YbiEb8kNcbil6TGWPyS1BiLX5IaY/FLUmMsfklqjMUvSY1xHP+A+o4RXi1jf6XVwuv775pH/JLUGItfkhpj8UtSYyx+SWqMxS9JjbH4JakxFr8kNcZx/KuIY5MlrQSP+CWpMRa/JDXG4pekxlj8ktSYqfxxN8lm4EHgEeDhqpqdRg5JatE0R/WcVFXfm+L2JalJnuqRpMZMq/gL+EySm5JsWGiBJBuSzCWZ27p164TjSdLea1rFf3xVHQOcDLw+yW/MX6CqLqyq2aqaXbt27eQTStJeairFX1X3dj/vB64EjptGDklq0cSLP8njk+y/fRp4IXDHpHNIUqumMarnYODKJNu3/+Gq+ucp5JCkJk28+KvqW8CzJr1dSdKIwzklqTEWvyQ1JlU17QyLmp2drbm5uWnHALwmvqTJ2nz+qcteN8lNC10SxyN+SWqMxS9JjbH4JakxFr8kNcbil6TGWPyS1BiLX5IaM81v4NqjOD5fUis84pekxlj8ktQYi1+SGmPxS1JjLH5JaozFL0mNsfglqTF7/Th+x+dL0s/ziF+SGmPxS1JjLH5JaozFL0mNsfglqTEWvyQ1xuKXpMZY/JLUGItfkhpj8UtSYyx+SWqMxS9JjZlK8Sd5UZKvJflGko3TyCBJrZp48SfZB/gb4GTgCODMJEdMOocktWoaR/zHAd+oqm9V1UPAR4HTp5BDkpo0jevxHwJ8d+z+3cCvzl8oyQZgQ3f3h0m+NoFsi1kDfG/aIZZptWY392SZe7IWzZ137tbzP3WhmdMo/iwwr3aYUXUhcOHwcfpLMldVs9POsRyrNbu5J8vckzWt3NM41XM3cOjY/V8E7p1CDklq0jSK/yvAYUmeluQxwMuAq6eQQ5KaNPFTPVX1cJKzgX8B9gEuqao7J51jmfaoU09LtFqzm3uyzD1ZU8mdqh1Or0uS9mJ+cleSGmPxS1JjLP55khyY5Nokd3U/D1hgmcOT3DJ225bknO6x85LcM/bYKXtK7m65zUlu77LNLXX9aWVPcmiSzybZlOTOJG8Ye2xi+3yxy41k5K+6x29LckzfdYfWI/vLu8y3JflCkmeNPbbg62YPyX1ikh+M/f7f1nfdKef+07HMdyR5JMmB3WPD7u+q8jZ2A/4c2NhNbwTeucjy+wD/DTy1u38e8Cd7am5gM7Bmd/+7J50dWAcc003vD3wdOGKS+7z7XX8TeDrwGODW7RnGljkF+DSjz6s8B7ix77p7QPbnAgd00ydvz76r180ekvtE4JPLWXeauectfxrwr5Pa3x7x7+h04NJu+lLgJYss/3zgm1X17SFD9bDU3Cu9/u5YdNtVtaWqbu6mHwQ2MfoU+CT1udzI6cAHa+RLwJOSrOu57lSzV9UXquqB7u6XGH3GZtp2Z79Nc58vddtnAh+ZSDI81bOQg6tqC4zKBjhokeVfxo6/sLO7t8uXTPCUSd/cBXwmyU3dZTGWuv4QlrTtJDPA0cCNY7Mnsc8XutzI/H98drZMn3WHtNTtv5rRO5ftdva6GVrf3L+W5NYkn07yy0tcdwi9t51kP+BFwOVjswfd39O4ZMPUJbkOePICD71lic/zGODFwJvHZr8XeAejX9w7gHcDf7C8pDtsbyVyH19V9yY5CLg2yVer6oaVyLcrK7jPn8Dof5BzqmpbN3uwfT5/8wvMmz8eemfL9LpUyYB6bz/JSYyK/4Sx2VN53dAv982MTrX+sPv7zlXAYT3XHcpStn0a8O9V9T9j8wbd300Wf1X91s4eS3JfknVVtaV7i37/Lp7qZODmqrpv7Ll/Np3kIuCTK5G5e+7dzl1V93Y/709yJaO3pDcAS/nvnkr2JPsyKv0PVdUVY8892D6fp8/lRna2zGN6rDukXpdKSXIk8H7g5Kr6/vb5u3jdDG3R3GMHAFTVNUn+NsmaPusOaCnb3uGswdD721M9O7oaOKubPgv4xC6W3eG8XFdc250B3LGi6XZu0dxJHp9k/+3TwAvH8i3lv3ul9cke4GJgU1VdMO+xSe3zPpcbuRp4ZTe65znAD7rTV9O+VMmi20+yHrgCeEVVfX1s/q5eN3tC7id3rw+SHMeo177fZ91p5u7yPhF4HmOv+Yns70n8hXs13YBfAK4H7up+HtjNfwpwzdhy+zF6cT1x3vqXAbcDt3W/6HV7Sm5GIwxu7W53Am9ZbP09KPsJjN4q3wbc0t1OmfQ+ZzRq5+uMRmy8pZv3OuB13XQYfdHQN7tMs7tad8Kv7cWyvx94YGz/zi32utlDcp/d5bqV0R+ln7sn7PPFcnf3fx/46Lz1Bt/fXrJBkhrjqR5JaozFL0mNsfglqTEWvyQ1xuKXpMZY/JLUGItfkhrz//kus+6qm1IoAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cb_corr['corr'].plot.hist(bins=30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We see more negative values than positive ones. It's a surprise, since collaborative filtering learns user interactions. We are expecting a result more correlated with the ranking by user similarity. One reason could be the inclusion of negative interactions (dislikes) in the collaborative filtering factorization."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    382.000000\n",
       "mean      -0.007711\n",
       "std        0.354699\n",
       "min       -0.854545\n",
       "25%       -0.260606\n",
       "50%       -0.018182\n",
       "75%        0.269697\n",
       "max        0.793939\n",
       "Name: corr, dtype: float64"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cb_corr['corr'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='corr', ylabel='num_votes'>"
      ]
     },
     "execution_count": 60,
     "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": [
    "cb_corr.plot.scatter(x='corr', y='num_votes')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "No popularity bias was found, indicating that the bias term in factorization algorithm was fitted properly."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Comparing Ranking Performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAucAAAGuCAYAAAAzuKjxAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/d3fzzAAAACXBIWXMAAA9hAAAPYQGoP6dpAACwyUlEQVR4nOzdeXgc1Zkv/u85tfQqqVtqrZblfcXYYFYDBrMzkLBN4pAACcmdASaTYZLJDYtZghMIZCEhcyEJ+SU3CcSEkASSwB0IGDCLF7yBjcG7ZUu2FkvW2lJ313LO7w9ZAnnV0t1V1f1+nkcPbqlV9aoPVfXWqfecw6SUEoQQQgghhBDHcacDIIQQQgghhPSh5JwQQgghhBCXoOScEEIIIYQQl6DknBBCCCGEEJeg5JwQQgghhBCXoOScEEIIIYQQl6DknBBCCCGEEJeg5JwQQgghhBCXUJ0OwEtaW1udDsGVGGMIBAJIJBKgNa1yB7VrbqJ2zV3UtkcXi8WcDoGQIaOeczJqnHMEg0FwTv875RJq19xE7Zq7qG0JyQ10BBNCCCGEEOISlJwTQgghhBDiEpScE0IIIYQQ4hKUnBNCCCGEEOISlJwTQgghhBDiEpScE0IIIYQQ4hKUnBNCCCGEEOISlJwTQgghhBDiEpScE0IIIYQQ4hKUnBNCCCGEEOISqtMBOOXFF1/E66+/jt27d2PevHn41re+5XRIhBBCCCEkz+Vtcl5cXIyFCxfi/fffR3d3t9PhEEIIIYQQkr/J+VlnnQUA2LVrFyXnhBBCCCHEFajmnBBCCCGEEJfI257zoWhsbERjY+PAa5/Ph6qqKgcjcidFUQb9l+QGatfcRO2au6htCckNlJwfwxNPPIHFixcPvF60aBEefPBBByNyt8LCQqdDIBlA7ZqbqF1zF7UtId5Gyfkx3HLLLbjyyisHXvt8PrS3tzsYkTspioLCwkJ0dXXBtm2nwyFpQu2am6hdcxe17dFFo1GnQyBkyPI2ObdtG7ZtQwgBIQQMwwDnHKr68UdSWVmJysrKgdetra10wjuG/s+U5BZq19xE7Zq7qG0J8ba8Tc7/+Mc/4plnnhl4vXz5clxwwQX4+te/7lxQhBBCCCEkr+Vtcv6FL3wBX/jCF5wOgxBCCCGEkAE0lSIhhBBCCCEuQck5IYQQQgghLkHJOSGEEEIIIS5ByTkhhBBCCCEuQck5IYQQQgghLkHJOSGEEEIIIS5ByTkhhBBCCCEuQck5IYQQQgghLkHJOSGEEEIIIS6RtyuEEkIG+yi5FG123cBrxhj0lA7DMCCl/Pj7YJjmOx8xdbwDURJCCCG5jZJzQggMmcD7yb8iwIqgs8DB7zIoksO2BYCPk/O4OAABgfnq/3IkVkIIISSXUXJOCEGDuQkKNFSpM8FYX7UbYww+nw+pVGpQz3lARNBgboItTShMcypkQgghJCdRzTkhBHuMdQjx4oHE/FiCLAIJgUZrcxYiI4QQQvIL9ZwTkucMmUCjtRnl6rQhvZ8zBSFejN3GWlRrszMcHRkNW0q8FE+i9xNPPjjj8CUspFIpCCkGvb9cUXBeyJftMAkhhHwCJeeE5Lm9xgZwKAiyoiH/ToiXYJ/5AUyZgsYomXOrnYaFl3tSiCkMDGzg+4ppw7btQe9NSYkuYWBuQEMBp4eqhBDiFErOCclzu4xVQy5p6RdiUbQA2GtuwAT99MwFR0bl/aSJYs4wTf94bMDRxhJIKbE+ZeKDpImzgnTDRQghTqHkfBh0XYfPRxetQzHW1yMXCoUGXeyJ+8WtNuzv2IHx/rnQlEMGdx5sV1XTgCO0a6EsxR57DWYXXJiNUMkwCSnxfksXKnQNmjY4OQcATdMOO15LbIn3TYFLCwqyGitJDzoXE5IbKDkfBsMwYBiG02G4jqIo0HUdPT09hz0qJ+62KfEafCwExfbDFOagnzHGoPh8sEzziBf6EGLYm9qI5s46BHk0WyGTIdppWOi2BaZrEqb5cdv295ybR2jXKAM+SqTQ1NmJEJW2eA6di4+OOtaIl9DZl5A8JaXETnMlwjw20OM2HD4Who+FsTO1KgPRkdFalzAQ4QzaMNq2kDNoDNiQNI//ZkIIIRlByTkheWq/vQM9og0FvGxEv88YQ5jHsNNYAXnIrB/EWUJKrEuaKFaGd4pnjKGYc6xO0BNCQghxCiXnhOSpHal3EGLFUJk+4m0U8FIkZCearW1pjIyM1jbDQkJKxIaZnANATOHYadrosumGixBCnEDJOSF5KCXiqDffR6Eysl7zfirTEeIl2G68nabISDqsSRiIKgzqCMqVCjhDgDGso9IWQghxBCXnhOShXca7UKAjyEY/kLOQl2Gv+QESoisNkZHRMqTEeykTsREO6GSMoVhhWJVIpTkyQgghQ0HJOSF5RkqBbcZbKFBKRzQQ9FABVgSNBbAztSIN0ZHR2pQ0ISQQHUFJS79SRcE+S6DZohk/CCEk2yg5JyTPNFvb0CvaUMjL07I9xhgKeCm2G28fthw8yb5VCQMlCocyihuvIGco5IwGhhJCiAMoOSckz2xNvYkwj41qIOihCnkZkjKOBmtT2rZJhq9bCGwxrBENBD1UicLxbsKAoMVsCCEkqyg5JySP9Ih27LM2pa3XvJ/CNBTwGLamlqV1u2R41iVM+BhQxEdfrhRTODqFxC6TSlsIISSbKDknJI9sT70NHwvBzwrTvu1CXoFmaxu67Oa0b5sMzcpECiUKT8tYAp0xRDnHql4aGEoIIdlEyTkhecKWJnYY76CQl6cleTuUn4cRYEXYlnor7dsmx9dk2dhnCZQqStq2WaowrE+ZMKi0hRBCskZ1OgDiLaInDrtuz6Dv2QpHTzgMIx6HOGThElZQALW6JpshkqOoM9+DLU0U8NKM7aOQl2OnsRJzAp+GxvwZ2w853KpeA4WcIZiGkpZ+xQrHLtPGxqSJUwPpG6NARq9tpwUrOfimiXMbveEedMcNCPHxuVj1MxRPoss9IV5BRysZlsRf/wzzvbWAOvh/nTgYJA7pXbMsAEDht78HXpD+MgoyPFuSr6OAl4Kz9PWsHirMS3DA3oNaYzWm+s7N2H7IYEJKvJs0UJqGgaCfxBlDscKxMmFQcu4iPfttrPpJHIwD7NAmZ3HgE086pOj7OvfeAoTLM3fsE0LSh5JzMmTSMmF+uBHqlGngJbGB7zPG4PP5kEqlID95UZAS5ob1MDdthG/eOU6ETA5qtXajXezFOG1uRvfDGEcBL8PW1BuYos/PSPkMOdxWw0KPkJihp79SsUzh2GRY6LAFImlO/snItO2woAaAyrmDb5iOdi5ues9E+06LknNCPMJTyXk8Hsfjjz+O9evXIxAIYOHChbj88ssPe9+yZcvws5/9bOC1lBKpVAp33nknzjrrLHzwwQe455574PP5Bt7zmc98BgsXLszK3+FV1tbNgG2DRYa2qiRjDDwShbF+DSXnDtuaWoYQL4HGAhnfV5FSgT3mWjRbW1GhTc/4/kjf3ObFCoOWgZuhAs4QYAxrkwYuClGpkhu07bCgh4d+o6SFGQ5stzD2LN/x30wIcZynkvMnnngCtm3jN7/5DRobG3Hfffehuroas2fPHvS+BQsWYMGCBQOv161bhx/+8Ic45ZRTBr5XVFSEJ598Mluh5wRj/VrwaDHYMAac8ZIYrI82QXR1gRdSaYsTEqILdeZ6VKjZSZRVpiPMY9iSeoOS8yxICokNSROT9cyczhljKFE4VvYauDDoo6chLnBguwV/dOjJuS/M0LbDymBEhJB08swzymQyieXLl+OGG25AMBjEpEmTcMEFF2Dp0qXH/d1XX30V55xzzqCecjI8MpWC+eEH4MUlw/o9Fi4A8wdgbliXocjI8exILYfGAgiySNb2Wcgr0GB9iLh9IGv7zFfvJw1wBkTTOBD0UGUKR7MtsNeiOc+dluoSSLZL+AqH3t6+IoZku0Syg1bwJcQLPJOc79u3DwBQU/PxzB8TJ07Enj17jvYrAIDu7m6sXr0aF1100WHf/+IXv4j/9b/+Fx5//HF0d3enP+gcYn64EWAYcklLP8YYWLQYqTWrMhQZORYhbWwz3kIhL8tqj6efFcDPCmhaxSxYkTBQwjl4BtvXzxkinGFVr5GxfZChadthgWuAFhx6e6sBBkUH2ndR7zkhXuCZspZkMolAYHC9bCgUQiKROObvLVu2DBUVFZg+/ePH69XV1fjpT3+K6upqtLW14Wc/+xkeffRR3HvvvYN+t7GxEY2NjQOvfT4fqqqq0vDXeI+55l3w4hLwI5S09Cd9R0v+lNIymBvWA60tUMorMhonGaw+9R5MmUChMvy5zY/Xrsf73UKlAjvN5TiZXwmV0UwfmXDAsrHLtHGSXxtyO7GDPeyMM2AYHakxVcGapInPRjkUKm1xTPsuG75CDs4P71vrv0HjjA1qWsYYfIUc7TttVJ9Gg0IJcTvPJOd+v/+wRLynp+ewhP1QS5cuPazXPBqNIhrt6wGOxWK4+eabceuttyKVSg0qfXniiSewePHigdeLFi3Cgw8+ONo/xXOs9jYc2LEVwZPmQj1GaZCuHyUB8/kgiiLAxvcQXfiFzARJjujV+rcQ0SoQ9IdGvI2jtutxxOQYtMXr0Kx8iFlFF4x4/+ToXt/fhkJFQbHfP+ybKF0bXruO0SVqO7uxR/XhlMKR//9ERqdjRzeCUe2YZZraEY7ZYARo3yEGrn2EEPfyTHI+ZswYAEB9fT3Gjh0LAKitrcW4ceOO+ju7du1CXV0dzj///GNum3MOKeWgqacA4JZbbsGVV1458Nrn86G9vX2kf4JnJV57BSwQhOXzw04dvpQ3Ywy6rsMwjMM+w4H3FBej8+1l4BdeCnaEHh+Sfm1WPZpS2zFOm4vUEdrteIbSrsdTwEuxru0FVNkn0UDCNJNS4vUDHSjmDIYx9HITxhl0TYdhGpBieO1awjle2d+KiTaVtzjBTAh07DVQPls74jHNGYOm6zANA+KQY1YJCbTuMNG89wD0UP6dg+mmhHiJZ5Jzv9+Ps88+G0uWLMFtt92G5uZmvPbaa7j99tuP+jtLly7FKaeccthBuXHjRpSXl6OsrAwdHR345S9/iZNOOgl+/+BpwiorK1FZWTnwurW1FbadXwOipJRIrnwHvKRk4PWx3nvU5LwkBrlnN1KbP4Q2fWZGYiWDfdT7GoIsCo0FRpxcA8du1+Mp5BXYY65Dk7ENZerkEcdADrfLsNBmC0zQtOG1z8F6BymG364xhWNT0kSXaSJEN9lZ17rNBFMALXTkc3F/KYs4wjGrBQGuAq3bDZSfqGUhWkLISHnq7HrLLbcAAG666SYsXrwY119/PebMmQMAWLhwIT788MOB95qmiTfffPOwkhagr0f9rrvuwsKFC/GNb3wDhYWF+K//+q/s/BEeY++phWg7AB4rG9V2mKqBF5fAeHd5miIjx5IScewx16FIKXc0Do35EGYl2Jp6w9E4ctGqhIEoZ/Bl8YlEhDNoDFiXNLO2T/KxA9tN+IrYwLiB4WCcwVfIcGAbtR0hbueZnnMACIfDuPPOO4/4s2effXbQa03TsGTJkiO+9+qrr8bVV1+d7vByUmrVcvBoFCwN01DyWCnMDz+A6ImDh8JpiI4czQ5jBVSmI8iKnQ4FhUoF9pob0SvaEeT0aDkdTCmxLmlgnJrdwX0fz3mewrlBmpo221o3W/AVjLxPTS/gaN1CM7YQ4nae6jkn2SWTCZjvrx91r3k/VhQBfD4Ya99Ny/bIkQlpY1vqTRRkefrEowmwQugshG2pt50OJWdsTJoQEihWsn8KL1MU1FsCTTTneVYZcYHuBgFf0ciPaV8RQ7xRwIjTfOeEuBkl5+SojPfWAQof9tzmR8MYg1JSCmPlO6OqgSbHttfciJSMo5A7W9LSjzGGQl6OHcY7sCQNJEyHFQkDJYozUxoGOUMhzXmedQe29c1vrodH3uZ6uG++c+o9J8TdKDknR5Va8RZ4SSyts6vwsjKIA62wa3embZtksC2p1xHmpVCYe6rWCngpbGlhj0ErxY5Why2wzbBQ6kCveb+YwvFu8vAZQUjm7P/QhD/CR/U0jDEGXxHH/o+o7pwQN6PknByRtbcOorEBSll6e1+Z7utbMXTlO2ndLunTZtWj1a5FRKk8/puziDMFBbwUm1Ov0VOTUVqdMBBkDAUjGBSYLqUKR4+Q2GxQD2w2SCGxf5M1qpKWfv4IQ8uH1rCn0SSEZA8l5+SIjBXvgEWiYP5jL/I0EkppGcyN70H09qR92/luc+o1hFgxdBZ0OpTDRJRKdItmNFvbnA7Fs6SUWJ4wEFNG14M6WurBgaErqLQlKzr22DB7JALR0V+y/VEOs1eiYzeNGSDErSg5J4eRySSM99ZAKU3PQNBDsUgU0HQYa1dnZPv5qld0oM5cjyKX9Zr3U5kPYR7D5tRSp0PxrF2mjTZboFR1/tRdqnB8kDIRFzS4MNOaNhjwRxgUffQ3ZIrG4C9iaNpAN1aEuJXzZ3jiOsZ7awHOwaKZmYaPMQYlFoOx/C0qcUijrall0FkQAVbkdChHVcSr0GhtRqfd6HQonrS8N4VizrM6t/nRRDiDjwFrEpTkZZKUEo1rTfjT0Gvez1/M0bDWpPMvIS5FyTkZREqJ1PK3wEtK0zoQ9FC8rByirRX2LhoYmg6mTGB76m0U8UpXTJ94NH4eRpBF8FGSes+HKyEk1idNV/SaA3032TFFwdu9BiV5GdRRayPZKRGMpa/dgyUcqS6J9l1U2kKIG7njLE9cw66vg2hK/0DQQ308MPStjO4nX2xPLQcDQwGPOR3KcRUpVdhtrkGv6HA6FE9ZmzSgMCDq4EDQQ5UrHC22QK1JSV6m7F1lIBBNT0lLP0VnCEQZ9q5KpW2bhJD0oeScDJJa+TZYtBjM78/4vpTScpgfbIDoiWd8X7nMliY2p15DoVIBxtx/SAdZBDoLYUvydadD8QwpJd7qTaFU4eAuejLi4wxRzvFOLyV5mWClJBrWGmntNe8XLFXQsM6ElaKnHoS4jfuv5CRr+lYEXZexgaCHYpEIoOu0Yugo1RrvwpIpFPEKp0MZEsYYIrwS2423kRJ0YzYUe0wbjZZAuaI4HcphylWO9UkTPTQwNO0a1vbV8wdK0n+pDhQzMAY0rKExA4S4DSXnZICx/uBA0DStCHo8AyuGrnibalZHSEgbm5L/QJFSAc7cl7gdTZjHoDAdW1JvOB2KJ7x1cCCo30UlLf2inEFjwLs0MDStpJSofT2FYBkHy0C7M84QKuWofT1F519CXIaSczIgteLtjA8EPRQvK4NoO0Arho5QrbEaKRlHEXfn9IlH09d7XoWtqTdgiF6nw3G1uBBYlzRR7pKBoIdijKFUUbCsl1YMTafWzRZ6WwTCFZm76Q5XKOhtFWj5iBaTIsRN3Hm2J1ln7a0/OBA0OyUt/QYGhq5antX95gIhbXyQ/B8U8gooTHU6nGEr4KXgTMXm1GtOh+Jqy3tT8DHmqoGghypXOTpsQSuGptGOl5MIlnKovsy1u+JjCJZy7Hg5mbF9EEKGj5JzAgAwVmVuRdDjUUrLYG54DyJBPajDsctYiZSMI6JUOR3KiPT1no/BltTrSIpup8NxJVtKvNlroMzhFUGPR2cMMYXj9R4aGJoObTsstNfaKBiT+VK1gjEKOnbbOLDNzPi+CCFDQ8k5gTQMGOvXgMdKHdk/i0QBVYW5fq0j+/ciSxrYmPx/iChVnuw171fAS6EyHz5MvuJ0KK70ftJEr5CuLWn5pEqVY6thocmiaRVHQ0qJLX9LIFTKoQUyf0OmBfpqz7f+PUm154S4hHev6g7QdR0+n8/pMNKuZ9VyMAC+8nIwPvyeGoa+C4imapAY4cm9rAzWmpWIXXbFyH4/z7zX9f9gw0TMV5O5gaAHe2pVTQMyeNEu5eOwPfUWTin5NArUkoztx2uklHi9vRnluoqArqdtu/098JqmpTUZiwIotiTeMgT+VzSStu3mm4b3E+jcY6P6tAA0bXg3ZSNt2+IJCvauSaB7p4YxJ2f/6SkhZDBKzofBMAwYRu7NSNC9bClYtASWLQB7+NOhMcbgU3wwrVEsB10cg7lhPTq2boZSVT2ybeSJlIhjfdffEFWqYVsCNjIzhR1jDIrPB8vM7DLfPlkEHwtjeevTODt0U8b24zU7DAt1KRNzfRpMM30lB4wx+Hw+mBlo1zKFYXlXDy7zKShU3N/b7zbCllj3+26Eyjmg2jCHubjTiNtWBULlHOuXtCE8oQBccW8J1UjlYscayV109sxz9oFW2LW7wLM8EPRQLBAAK4ogtXqlo3F4wQfJl6AwFYU8s6u4ZgtjDMVKDfaYa9Fm1Tsdjmu8Ek8iprhz+sSjKeYMfsbwJi1KNCJ73k4h2SFQODb706IWjlWQ6hTY8xa1HSFOo+Q8zxlrVoGFC8BDYadDAS+JwVy3BtKiGR+OpstuxnbjbRQrNa4eIDhcAV6IMC/F2sSfqO4VQINp4yPDQpUHas0/iTGGSpVjWW8KSUHtOBypboFtLyRROFaBomX/2FY0hsJqBdteTCLVRQtKEeIkb535SVpJIWCsWQVe4o46X15SAmmkYG35yOlQXGtd4s8IsCIEWXYWisqmEqUGB+zdqDffdzoUx/2jJ4koZwhncc2BdClVOBiAt6n3fFg2P5eAojOEK5xr81AFh6IzfPSXhGMxEEIoOc9rdu0uyK5O8BJnZmk5FFNU8OISpNZQacuRNJgfosnagpg6Pqd6zftpzI8Ir8K6xJ9hydwb2zFULZaN9UkTY1TvrPj6SZwxVKoKlvakYNBTkCFp3WqiYa2JyATF0WObMYbIBAWN6020bqGpFQlxCiXneSy1dhV4tBgsjTNBjBaPlcLa/CFET9zpUFzFlibWJJ5FIa+EzoJOh5MxUaUalkzhozyeWvHleBJFnKHIwwMqyxUOCxLLe/P3JmuobEPigyW9CFdw+Aqcb3NfAUdBJcfGJb2wDbq5IsQJzp8JiCOkacDcsB6s2B0lLf1YYRGg6TA3vOd0KK6yOfkaUiKOYmWs06FkFGcKipVx+Cj1KrrtFqfDybpWy8ZqD/ea91MYQ6Wi4B89Seo9P45tLyZgJiSKatzT5oU1CqykxNa/U3kLIU6g5DxPmR9uAoQELy52OpRBGGPgxcUwqLRlQNw+gE2pl1Gi1Hh6waGhCvMS+Fkh1iT+mHeDQ1/q7zX30AwtR1OhcphSYjnVnh9Ve62F2jcMRCao4Kp72pwrDNGJKna/aaB9Fw3QJyTbKDnPU8bad/tKWkaw6FCm8VgZ7Po62AdanQ7FcVJKrEn8ET4WRpi7Y2xApjHGEFPHo9nalleDQ/d/otc8F8YUKAdrz1+Op5CimVsOYxsSG57sRaiMIxB136XYH+EIlXNseJLKWwjJNvedEUjGiZ44rG2bwWMxp0M5Ih4KgYXDMNavcToUx+01N6LJ2oJSdUJOJGxDpbMgIrwKaxPPwpRJp8PJihcP9ppHPFxrfqgKhcOGxFvUe36YbS8mYHQLRMa7r4OkX2ScAqNHUHkLIVmWO1cBMmTmxvcBTe+r73YpHi2BufbdvCtr+CRTJrEm8UdEeFVODwI9mqhSDQEbGxIvOh1Kxu0z+2ZoGevxWvNDKYyhSlXwSk8SCeo9H9C2s6+cJTrJXeUsh+Lqx+UtbTuovIWQbKHkPA8NlLS4uCeWx2IQB1phN+x1OhTHbEz8PwhYiCrVTofiCM4UlCjjsc14M+dXDv17PIESznNyyftyhYOB4bWe/HgCcjxWSmLD73oQLufwR9zf3v4IR7iC4/3f9cBK0Q0WIdng/jMDSSvR0Q57T61rS1r6MZ8frCgCY/1ap0NxRLu1F9uMZShRxoOz3OpNHY4Qj6KAx/Bu7xIImZurFtYaFj5MWajWcvN0zBnDGJXjtZ4Uuu3cbMPh2PK3BMwEUOTicpZDFY1TYKeALX+l8hZCsiE3rwbkqIz31oEFQ2ChsNOhHBePFsNcvwZS5NcFXUqBdxNPI8RjCPHcWwl0uEqU8egSzdhhvON0KGknpcTz3QmUKtyTq4EOVanC4WMML8fzu/f8wDYTdW8biE5SwBX3Prk8FFcYIhMV1L1joHUrLU5ESKbl7tWAHJGx7l0wl5e09OMlMch4N+zaXU6HklU7jRXosBsQU8Y5HYorqExHVBmL9xN/Q0J0OR1OWm02LNSads7Vmh+KMYaxmoK3EwYOWLbT4TjCSklseKpvsSF/kfcuvf6ivvKWjU/1UnkLIRnmqUmT4/E4Hn/8caxfvx6BQAALFy7E5ZdffsT3XnnllfD5fANJ6MyZM3H//fcP/PzFF1/En//8ZyQSCZxyyin42te+hmAwtwfd2fubIJoaoc052elQhoRpGlgkCuO9NVAnTXY6nKxIijjeS/wVxcpYqMzndDiuUcQrEBctWJ94DmeHbnI6nLQQB3vNyxWOQA7Ma348Uc5QwBlejCfxpUjI6XCybuvfE7BTQMk0796IFY1TsH+jhS1/S2DWwty+XhLiJE8l50888QRs28ZvfvMbNDY24r777kN1dTVmz559xPf/5Cc/QXX14YPp3nvvPTzzzDP4zne+g4qKCvzkJz/BE088gW984xuZ/hMcZaxfC1ZQCBbwzkmVF5fAfH895DULwRTvXtSG6v3E38CZiiJe6XQorsIYQ0yZgD3mWkyx5qNMneR0SKO2PmlivyUw1685HUpWMMYwVlWwJmniYtNGlZb7x3O/9loLe94yUDpT9VQ5y6G4whCZoKDubQNjTtURneipFIIQz/DMs7VkMonly5fjhhtuQDAYxKRJk3DBBRdg6dKlw97W66+/jgsvvBATJ05EMBjE9ddfj3feeQepVO7OxSulhLFuNXjUXSuCHg8vLoY0DVhbNzsdSsYdsPZgl7kSMWW8J8qOss3PC1DIy7Gm9xnPDw61pcTf40lUqhx6HrV1kcJRwjn+Fs+fgYXCktj4+77FhrwwO8vx+CMcoTKOjb/vhbCovIWQTPDMmWLfvn0AgJqamoHvTZw4EXv27Dnq79xzzz248cYb8Z3vfAd1dXUD39+zZw8mTJgw8HrcuHEQQqChoSEDkbuDXb8Hsr3N9bO0HIopKni0GMa61U6HklFSSqxNPIsCXooAd+/8804rVmoQF63YaaxwOpRRWZkw0G0LjMnxWvMjGatxfJiysMvIj3mzd72WRLJdoGhc7rR1ZLyCZKfArqX5PcCXkEzxzDOpZDKJQCAw6HuhUAiJxJF7YL73ve9h2rRpME0Tzz33HO677z787Gc/QzAYRDKZRCj0cc0jYwzBYPCwbTU2NqKxsXHgtc/nQ1VVVRr/quxJvLcWPBIF9/nTvu3+Xt5M9fbykhjMDz8At0ywDMTvBrtT69Bm78U4/WTX9Jpnul1HQmM+RJUx2JD8Oyb6T4fOA8f/JZcxpMT/xJMYoynQHJihhR2sb2ecAQ48gAgrCsoUgb/Hk/hmaW7fiPa22djxcgqR8SpUPfNtzQ8eq5yxjDatojFExqnY8XIKY88MIFiSOzcehLiBZ5Jzv99/WPLc09NzWMLeb9asWQAATdNwww034I033sDmzZtxyimnwO/3o7e3d9D7e3t7D9vWE088gcWLFw+8XrRoER588MF0/DlZJW0b7evXQq8eC92XuUGGuq5nZLuyvALx2l1Qd25H4dnnZmQfTrKkiff2PI+YPhZhv/uSlUy160iV6RPQ3dOCnfItnBW9zulwhu3FlnYYAMaHglAcvPHRNefadZKqYlVXD/ZqPpwY9s4YmOHa8H8b4StQUDw2kNWbXC0Lx6xeLZFo7cW2vxo4/397s9OKELfyTHI+ZswYAEB9fT3Gjh0LAKitrcW4cUObbu6TJ8Zx48ahtrYW5513HoC+MhfO+WG94rfccguuvPLKgdc+nw/t7e2j+jucYGzeBJHohSgsykhdPWMMuq7DMAxImZkaRBaNom3Za7BnnpiR7TtpS2IZklY3yvQprhr3kI12Hakor8Z7Hf8P49mZnioDSgqJ5/a3o0rlsAwDThR2MM6gazoM04AUzrSrAqBM5fj9vmbcUVrkqqcz6dK6zUTd2h5UzNFgGEZW9skZg6brMA0DIgvHbGENR/26XmxftR+xae4e2ByN0poRxDs8k5z7/X6cffbZWLJkCW677TY0Nzfjtddew+23337Ye+vq6mCaJsaPHw/LsvCXv/wFhmFg2rRpAIALLrgAP/7xj3HeeeehvLwcS5YswTnnnAPfIb3KlZWVqKz8eNaM1tZW2Lb35uhNvrsCvLgEUJSMJllSyoxtn8dKYX20CWZ7G3ihd5Kx47GkgY2JF1GkVIEjs+0zUpls15EKsRJ0sAZs7HkJpwY/63Q4Q/ZGPAkBiXKFO/eZHqx3kMLZdq1WFLyXMrEpkcJMn7sTu+GSQmLTs3GEyji0EMva59xfyiKydMxqIYZQOccHf4xj/l0FAyVThJDR8cyAUKCvJxsAbrrpJixevBjXX3895syZAwBYuHAhPvzwQwBAR0cHfvSjH+G6667DV77yFWzduhWLFy9GONy3KubJJ5+Mz33uc1i8eDFuuukmKIoysO1cIxMJmJs+AC/x1kDQQ7FwAVggAGP9GqdDSasdqXdgSwtFvMLpUDyFMYaIMgY7jHfQKzqcDmdIUkJiaU8KVaoyUBucz/ycoUzheKE76bqbv9FqXG8i3iRQVJP7tdhFYxX0tgg0rKOVQwlJFyZz7ayYQa2trU6HMGypd1cg8be/QDtpLliGBp8xxuDz+ZBKpTJ6kbX31kGmDBTcfk9OPAa3pYm/dt2HMC9BVBnjdDiHyVa7jpSUEnutDzBBOw1zg9c6Hc5xLY0n8XJPEif7NEeTcze1a1JIrE+Z+Go0hBk50nsubIk3F3dBCzIUjcvuw2mn2rZzjwWjR2LB/YWuncc95rGZykh+81TPORk+Y9Vy8OKSjCXm2cRjZRAtzbD31h3/zR6w21gLSyap13yEGGOI8EpsN96GId09b7Yp+3rNK6nXfJD+3vOX4rkzJd++dw2kuiUKxuR+r3m/gjEKjLjE3lXZqa0nJNd5P2MjR2W3NMOu3wNeVuZ0KGnB/H6wSBTG6pVOhzJqUkpsTr2GAl4GzvLnIp5uYV4CDhU7U+6e9/zdhAHjYK05GaxKVbDLtFGbA/OeC1ti+0tJhCsUcDV/bsK4yhCuVLDjpSSE7b6nbIR4DV0pcpixeiVYQSF4MHT8N3sEj8VgrF8DaXq7h6bF3oUu0YwipfL4byZHxRhHgVKGbak3IV26aqiQEq/0pFCuKI5OnehWQc4QUxhe6fF+73njOhNGt0RBVf5dWgsqOYweiYa1VHtOyGjl3xkkT0jbhrF6ledWBD0eXlwCCAHzgw1OhzIqO1LvIMyKobHMzTufLwp5GXplO5qt7U6HckSbUhY6bYFKlU63R1OpKPggZaHF8t5sWP2klNj5ShKhcp5Xveb9uMoQKuPY+Y/cG+BLSLbR1SJHWZs/hEwmwEtKnQ4lrRhXwItjSK182+lQRsyUSdSZ7yGs5FbbOEVlOoKsBLuMVU6HckRLe5IoVTg06jU/qkKFo4gzLOt1zzz/w9W6xULPfoGCyvwtUyuoUtDbKtC62fslSoQ4iZLzHJVa+TZ4cQxM9cxU9kPGy8pg1+6C3bLf6VBGZK+5EQwcIUaLYqRLgRJDvfk+LOmucqcG08Yu00YF9ZofV7nCsTJhIOXQwkijVftaCsEYh+LL35swRWcIxjh2vebdmyxC3ICuGDlIdLTD2rYlZwaCHoqHwmCFhUitWu50KCOyx1iHEI+CMTr80iXIopCQaDQ/cjqUQd7qTSHCGUI5MFtSppUoHAzAmqS7brCGome/jdYtFsKV1M7hSo4DWy3Em71bokSI0+hMkoNS764AC4bAwgVOh5IxvKQU5uqVkJa3Hp9aMoUmaytCvMTpUHIKZxxBHkW96Z6xCCkhsTppoIxmaBkSzhhiCsdbHixt2fNWCr4iBj1Mba2HOXxFDHve8l47EuIWdCbJMdK2++Y2j5XmxEI9R8NjpZCmAXPTRqdDGZYmaxsAiQArcjqUnBNiETRYm1wza8t7KROQfT3CZGjKFQUNlsBe0zs33bYhUb/KQKiM2rlfqIxj7yoDtuHNEiVCnEZnkxxjbfkIsrcHvDQ3S1r6MUUBL4khteItp0MZlkZzMwI8Ak4lLWkX5FEYMoE2u97pUAAAK3pTKFE4LTo0DAHOEOEMKxPeKW1pfM+EtIFACR3T/YIxDkigcb132pEQN6GzSY5JrXi7b0XQHBwIeiheVg571w7YrS1OhzJkjdZmBFih02HkJIVp8LMCNFvbnA4FbbbALtNGKQ0EHbaYwrE2YcL2yHR8dW/3DQR167L1TmCcIVDCsedtSs4JGQm6cuSQvoGgm8FLy50OJSt4KAxWUAjjXW8MDE2KbsRFCyXnGeRjYVck5+sSBkKMIUy95sNWonAkpMRWD6wYGm+y0bHbRqicLqWHCldwdO6xEW+igaGEDBedUXJIavWqvoGgBbk7EPRQPFYKY/VKSNv9F4BWuxYcKnSWOyu2uk2AF6LVqnV8EZQ1SQPFCs/pcR+ZojKGYoVhrQdKW+pXHhwIGqJL6aG0YN/A0PoVNDCUkOGiM0qOkELAeHc5eCyWVwkBL4lBJpMwP9rkdCjH1Wrthp8V5FX7ZJuPFcBEEt3CuTnwWywbDZZAjAaCjlgx59iQMmG5uLRF2BJ7V5l99dXkiIIxjr2rTAjLve1IiBvRWSVHWDu3Q3Z1gsdyeyDooZiqgheXwPDAnOetVi10FnQ6jJymQocKPw7YdY7F8H7SRJgxBDndhI1UVOEwJbDNxaUt+zeZsJKSkvNjCMY4rJTE/k2m06EQ4il0VskRxuqVYMUlYJrmdChZx0vLYG3bDNHV5XQoRyWlRLtdDz8POx1KTmOMwcdDaLOcS87fS5qI0ODAUVEZQ5RzbEi6N6mrX24gWMLBVWrro+Fq34qhdcvdX6JEiJvk/pQeaaTrOnw+n9NhHEYkE+j44H34pk6H6kByztB3cdJUDRLZf3wpi0tgB4LgH32Agosvy/r+hyJuHYDZmURIK4LGPXIDdbD8RtU0wMXlBYfyyxC6WAMKHBh70WnZqLM6MDfkh6YoWd//UPSXVWma5nht/rGUSOADw8K/hsOuKwVLdNho2dyByjl+aJp72tmNbVtUydGwIQnVCiIQdc9nRYibUXI+DIZhwDDc1wNgrH0X4BwiXADTzH5PE2MMPsUH0zIduyCwaBRd77wJnHm2I/s/nn3mNnCogKXCZO7tDfwkxhgUnw+W6Vy7joQqAmi16tDV1ZX1pO7d3hR8DPDbNkzhjsWQDsUYg8/ng+nydi2CxDZbYHN7B8Zq7rpU7Xw9CS3AoARtmKZ72tmNbcuDElqQYevrbZh0id+xONzYsUbI0VBZSw4w1q0GjxaD8fxtTiVWCtHUAHt/k9OhHFGHvQ8+FnJdD2Au8vEgTJlAUma/zGlTykSE0ywt6aAzhkLOsCnlrrpzKSXq3jEQLKV2HgrGDpa2vGO45oaBELfL32wuR4ieOKwd28BLYk6H4igWCIKFC2C8t87pUI6ow26AxqjnJhs0BMDA0WE3ZHW/tpTYYliI0CwtaVPIGT5wWd15+04biTaBUCm181CFyjgS7QJtO9w/5S0hbkBnF48zN20EdB2sgBa24dFimO+vdzqMI+qw99FMLVnCGIOPhdFpZ/cpym7ThiGBCM3SkjZRzlFv2eh1UYlQ3YoUAsUcio/aeagUnSFQzFC/nOY8J2QoKDn3OHPje+BFUXq8CoAXF0O0NMNubXE6lEGEFOgWLZScZ5HKfOjMcs/55pSJQs6g0rGYNgWcQQVcs1qomZBoWm9Sr/kIhMoUNL5vwux1z40WIW5FZxgPk6lUX0lLcbHTobgCCwTBQiGYH33gdCiDxEUrBGxKzrNIZwG0i+wm5x+mLBRSr3laMcZQqDBscUndecNaA0wB/FFq5+HyRxkUFWhY664yJULciJJzD7N2bgMAsMIihyNxD1ZY1Ffq4yJdogkKNCjQnQ4lb+gsiC67KWsD0BJCYq9lI5LHg7IzpYhzfJRyR0JX93YKwRgHo5uwYWOMIRDj2PM2lbakw7Jly8AYw9q1azO2jx/96EdDeir/qU99CgsWLBh4/dvf/haMMbS2tmYstlxHVxIPM7dsBi+K5PUsLYfiRVHYu2shU+65AHTajdBZkEqPskhnAVhIISE7s7K/7YYFBUCYkra0K+Ic7UKizXa2HKKz3kJ3g0ConObqHqlwuYJ4o0BnnTuehJDMuOKKK7By5UpEIhGnQ/Esyuo8zNr6EQ0EPQQrLAAgYe3e6XQoA2imluzrn7Gl027Myv62GSaKOAenG7C0CzDAz4BtDvee168w4I8waAFq45FS/Qz+KEP9CvetF0LSp7S0FGeeeSZU1V3rE3gJJeceJbq6IA60UknLIRhXwAqLYO3Y7nQoA2imluzrm7ElhE6RnRlbNqcs6jXPEMYYCjh3dFCobUjsW903tzkZnWApx741BmwjP+Y8X7lyJS644AKEQiEUFRXhC1/4Avbv3w8A2L17NxhjePLJJ/Gv//qviEQiKC0txY9+9CMAwDPPPINp06ahsLAQ1157LTo6Og7b/v79+3HttdciFAqhsrIS3/ve9w57z+bNm3HVVVehqKgIoVAIV1xxBXbuHNyB1dXVhS9+8YsoKChAaWkpbr/9dljW4cfc5s2bcd5558Hv92PSpEl48sknD3vPoWUt/X/n73//e3zta19DNBpFZWUl/vf//t+H7eP555/HtGnT4Pf7cfrpp2Pt2rUIh8O4//77h/R55wo603iUVbsD0DSwICV9h2KhMKwd25wOA0D/TC2t0Cg5zzqV+dFpZb7nPC4Emm2BIoWS80wp5AxbDcuxRWyaNpiQAgiU0CVztIIlHFIATe+7YxxBJq1cuRILFixAUVER/vjHP+KXv/wl1qxZgyuvvHLQ++655x4UFhbiT3/6ExYuXIhvfetbuPPOO/HTn/4UP/jBD/D444/j9ddfx+23337YPm6++WZMmjQJzz33HG644Qbcfffd+MUvfjHw8127duGss85CW1sbfvvb3+Lpp59GS0sLLrzwQqQ+Uf75la98Bc8//zwefvhh/O53v8OHH36Ixx57bNC+kskkLrnkEjQ3N+Opp57Cww8/jAcffBDr1w9tCuO7774bnHM8++yzuOWWW/DII4/gV7/61cDP33vvPXz2s5/FzJkz8dxzz+HLX/4yrrvuOkdWPncaPXPwKKt2F3hBIdUxHwEvKIS1bQukZYKpmqOx9M3UYsFHyXnW9c3Ysi/j+9lpWFABhOlYzJgizrHTtHHAFoip2a/5rl9hIFjCwekGbNQY71sxtH5FCmNOz+1B8nfeeSdOPfVUPPfccwPX6lmzZuHEE0/E//zP/2DmzJkAgLPOOguPPPIIAOCCCy7AX/7yFzz22GPYs2cPSkpKAAAbNmzAr3/9a/zyl78ctI8LLrgAP/zhDwEAl156KZqamvDggw/i5ptvBuccixcvRjQaxauvvgq/3z+wvwkTJuDXv/41vvrVr2Lz5s147rnn8Ktf/Qpf+cpXAACXXHIJJk2aNGhfv/3tb9HQ0IAtW7ZgypQpAIDZs2djxowZmDp16nE/jzPOOAP//d//DQC4+OKLsXTpUvz5z3/GrbfeCgB46KGHMGHCBPzlL38BPziWLhAI4Mtf/vJwPvacQN0AHmXX7gQLhpwOw5VYOAwIG/a+vU6HQjO1OChbM7ZsNywUclrKPZP8B+vOdzhQ2pJoF2jbbiFYRpfLdAmWcrTt6FtpNVf19vZi+fLl+OxnPwvbtmFZFizLwrRp01BZWYk1a9YMvPeiiy4a+LeiKJg4cSJOOumkgcQcAKZOnYqOjg7E4/FB+7nmmmsGvb722muxd+9e7N3bd/175ZVXcNVVV0FV1YEYotEo5syZMxDD6tWrIaUctC1VVXHVVVcN2va7776LWbNmDSTmADBt2jTMmjVrSJ/JJZdcMuj1zJkzB+IEgDVr1uBTn/rUQGIO4LAY8gWdbTxIWhbspgawcIHTobgSU1WwcBh2/R6nQ6GZWhyks2BWZmzZZlgooHrzjOqvO9/uQHLesMaAFmLQw9TG6aKH+z7PfWtyd2Boe3s7bNvGN77xDWiaNuiroaEB9fX1A+89dFYTXdeP+D2gr7Tkk8rKyo74urGxr6SvtbUVjz766GExrFixYiCGxsZGaJqGaDQ6aFvl5eWDXjc2Nh62vyO972iO9Dd98u9pbGxEaWnpoPdEo1FomrNPwJ1AZS0eZDc3ArYNFqae86NhgSCsuj1weo6UDrsBKs3U4ggNfjBwdNlNCPJIRvaREBKNlsAsnU6lmVbAmSPJ+d5VfSUtdIOdPowxBIo59q4yMOkSX05+tpFIBIwxLFq0CFdfffVhP4/FYmnZT//g0kNfV1ZWAgCKi4txxRVX4Ktf/ephv1tQUDDwXtM00d7ePihBb25uHvT+ysrKI9aXNzc3ozgNiyFWVlaipWXwCt/t7e1Uc+528Xgcjz/+ONavX49AIICFCxfi8ssvP+x9W7ZswR/+8Afs2LEDQN9jl3/5l39BVVUVAOCDDz7APffcA5/v46TpM5/5DBYuXJidP2SU7L31fathOlxP7WYsGHJFz3nHwZ5zkn2fnLGlAtMzso/dpgUGmt88Gwo5wy5TossWKFSy89C3u8FGz36BylPoXJtuwVKOxnUmuhsECsfk3tzxoVAI8+bNw+bNm/HAAw8c8T27d+8e9X6ef/75QeUozz33HKqqqlBdXQ2gr2Rm06ZNOPnkk6EoR/6cTzvtNDDG8Pzzzw/UnFuWhb/97W+D3nf66afjySefxPbt2wdKW7Zu3YpNmzbh3HPPHfXfctppp+HFF1/EI488MlDa8te//nXU2/UiTyXnTzzxBGzbxm9+8xs0NjbivvvuQ3V1NWbPnj3ofT09Pbjoootw++23Q9d1LFmyBA888AB+9rOfDbynqKjoiFMAeYG9rx6gWVqOiYVCsOt2Q5ommEOPxKSUiIsWlKmTHdk/AVTmQ5fdfPw3jtBOw0IhZzS/eRYEGYMGYKdp4WQlO2M4GtYZ8BUyqH5q33RT/Qy+IobG9QYKxwScDicjfvjDH+KCCy7A5z73OVx33XWIRqPYu3cvXn31VXz5y1/G+PHjR72P119/Hd/61rdw8cUX49VXX8Xvf/97PP744wPJ7eLFi3Haaafh0ksvxc0334zy8nI0NTXhzTffxPz58/H5z38eM2fOxNVXX42vf/3rSCaTGD9+PB5//HHYtj1oXzfddBMeeOABfPrTn8YDDzwAKSXuvfdeVFRUjPrvAIC77roLp512Gv75n/8ZN998M3bv3o1HHnkEuq4PqkPPB575a5PJJJYvX44bbrgBwWAQkyZNwgUXXIClS5ce9t5TTjkF8+fPRygUgqZpuPrqq7F37150dXU5EHn62fV14AFKzo+FBUOAlLCbGhyLISE7YMOExnLzwuMFGvwZXYhoh2EhRL3mWcEYQyHn2JXF0pbG9Sb8Uc9cJj3HH+nrPc9VZ511Ft555x3E43F8+ctfxuWXX47vfOc7CAaDmDw5PZ02TzzxBLZu3YprrrkGTz31FL773e8OKmGZPHkyVq9ejZKSEnz1q1/FpZdeijvvvBM9PT2DOjb/7//9v7jyyitx++2344tf/CKmT5+Or33ta4P2FQgE8Morr6CsrAzXX3897rjjDtxxxx2YO3duWv6Wk08+Gc8++yw++ugjXHPNNfj1r3+N3/72t7BtG0VF+bWmC5NOTRw7TDt37sS3vvUtPPfccwPfe+ONN/DXv/4VP/3pT4/5u++88w7+v//v/8Pvfvc7AH1lLffddx8KCgqgaRrmzp07MPn+sfRPqO8kKQQ67/7fUCdOAo+OvsYrHRhj8Pl8SKVSjs1DfCTmxvfgv+Jq+M44y5H9N5vb8HrP/8FEbZ4nayrd2q7D0WXvR7doxjVFhy/MMVpCSnyzuROTdBUlWSqzSAcvt2u9aUMCuCOW+cHwPS023lzcjYq5mmdWBfVa25oJiab1Js69twDh8syWtqSrxptk19KlS3HxxRdj2bJlOO+885wOJ2s8U9aSTCYRCAzugQyFQkgkEsf8vaamJjzxxBO45ZZbBr5XXV2Nn/70p6iurkZbWxt+9rOf4dFHH8W999476HcbGxsHRjwDgM/nG6hbd4rddgAwDfBQ2DUJX38cbomnHwsEIRr2HrXOLtN6zAPQWdCzj+Pc2q7DofMAEnYXwAUUlt7ypkbTggmgUPHWYEF2sKefcQZ4bCa7QoVjc8qE4Bxahj/z1o8OztIS9M7x219exRnzRNPqwb5ZW1o32yiqoulmCfDVr34VF154IUpKSvDhhx/iu9/9Lk4++WTMnz/f6dCyyjPJud/vPywR7+npOSxh/6TW1lbcd999+MxnPoNzzjln4PvRaHRgRHIsFsPNN9+MW2+9FalUatAg0SeeeAKLFy8eeL1o0SI8+OCD6fqTRqR7+xYwnw++ggLXJQT9Uz25RkEh0LDvsOmhssU60AM9FRz0/5QXua5dh4GLQsAElLCFqH74FGCjsa6tE2GFo+Dgwh5eo2vea9cSKWGnTHT6g5gSzOzn3ralF6FizZPHr+ahYzYYlWjfKhH9jDPnaeIuHR0d+I//+A+0traiqKgIl112GX70ox95tpNrpDyTnI8ZMwYAUF9fj7FjxwIAamtrMW7cuCO+/8CBA7j77rtxySWXHHcSe845pJSHPQa85ZZbBi2z6/P50N7ePpo/Y9R6tmwGC4ZgGO6ZH5YxBl3XYRiGqx6l2n4/rB3b0XbgAJgDB3Zrzz4wwQctkewlbm3X4ZBSgoGjsX03oKe39v+jjjgCgOfal3EGXdNhmAak8F67hjnDxgNtiKUyN5ZD2BLNWxIonqx6qn05Y9B0HaZhQHjkmNUKBJq3GDjQ0gauZq7DyalOGjI8Tz/9tNMhuIJnknO/34+zzz4bS5YswW233Ybm5ma89tpruP322w9774EDB7Bo0SIsWLAAn/nMZw77+caNG1FeXo6ysjJ0dHTgl7/8JU466aSBpW37VVZWDswVCvT1xB86ejnbzN27gEDAlcnSkW5wnMSCIcAyYTY2QKmoPP4vpFncPgCV6a76TEbCbe06XBrzI261wVbSe+zuTJkIcea9z+ZgvYMU3mzXIGOoTZmwA5nrHe7YbUGYgK/QW+3bX8oiPHTM6gUMwgTa9xiIjPdMSnJM3d3dEL09kIcsGDRakXHj07o94l6eOhJuueUWPPbYY7jpppsQDAZx/fXXY86cOQCAhQsX4tvf/jZOOOEEvPLKK2hsbMTzzz+P559/fuD3H3/8cZSWlmLXrl149NFH0d3djVAohLlz5+JLX/qSU3/WkEkhYO+rhzpugtOheALTdcAfgFW325HkPCm7EGD5NcLcjRToSMj0ztRkSIn9tsAJiqdOoTkhzBl2m5ntJOnYbUMPs4z25JI+XGXQCxjaa62cSc5Fbw9av30XZPLYY+KGK/LbZ9K6PeJenjoSwuEw7rzzziP+7Nlnnx349+c//3l8/vOfP+p2rr766iOu2OV2Yn8zkEqBhcNOh+IZPBSCtbsWvtPnZX3fKdmDMKcZApymMBUpGU/rNvcdnDWEFh/KvjBn2GXaSAkJX4Y+/449FrQgtW22aAGGjt3OPpVOJ5lMQiYT0E88CcyXnic8MuWeUlaSeZ5KzvOdtXtn38qguvcGKDmFhQtg79qe9f1KKWHJJDgdYo5j4DBEb1q3WW/ZCDEGxWWDsvNBkDEwAHstG5P0zBxfXfU2JedZpIUYuvbmTnLej/l0cF96Bi57YfYdkj75NfzV46yd26nXfJhYYSFEawtEvDur+xWwICHB6RBzHAeHifTWfu41LQSp19wRCmMIMYa9GSptkUKip0VQcp5FWpCht1V4coAyIZlAmYNHSClhbdsKVlDodCiewoIhQNNg7diW1f3asm/VO0aHmOMY+EB7pEudaSNIveaOCXCGfVZmVgpNdkhIG1A9svBQLlD9DNIGEu3UP5wuu3fvBmMMyTQPSiXZQZmDR4jGBsieOHgk4nQonsIYAyssgrnlo6zuVx58COm2uejzExtoj3QQUqLZEtRz7qAgY6g3M5PIJdoEwADFO1OFe57iA8CARBv1nBMCUHLuGebmTWDhAqo3HwFeFIG1+UNIkb1emf4ec4/MZpbjZFqfYLTZAiZAPecOCnKGZsvOyHSByQ4B1ffxSqok8xhjUP19nz0hhJJzzzA/2ABWRNPyjQSPRCF74rDr67K3T9Y3UC2dPbZkZCQkFJa+gYONloAKwEe5m2OCjMEA0J6BGmUjLqHo1LjZpmgMRpx6M0aioaEB1113HSoqKlBcXDxoNrrf//73GDt2LMrKyrBo0SKILHZSkZGjqSQ8QHS0w95bB/XEOU6H4klM18GKIjA2vgc1S4s4KNDQV06RezMQeI2ADRXpW02y2bYR4oxKlhzkY309S02WjWIlvX1MZq8EU9K6STIEXAXMXkoch8u2bXz605/GGWecgW3btsHv92PlypUDP3/ppZfw0UcfYf/+/bj44osxZcoUfPnLX3YwYjIU1HPuAcb768GCwb7BjWREeLQY5ntrslbawhiDxvywkZlBa2ToJAR8PJi27TWZNnRKzB3FGEOIM+y30n88W0kJplD7ZhvjgEVjF4dtzZo1qK2txaOPPorCwkLouo7zzjtv4Of3338/CgoKMGnSJPznf/4nlixZ4mC0ZKgoOfcAY+0qsGgJ9dSNAi8pgezuhl27M2v79LMwbEkLRzjNlib8rCBt22u2Bfx0LDrOx4AWK/1PpmxTgprXAYzBNqisZbjq6upQU1MDXT/yCOaampqBf48bNw779u3LVmhkFCg5dzlr316IpkYopaVOh+JpTNPBo8VIrV55/DenSZBFYVFy7jhLphDk0bRtr9UW8FPy5jgfY9hvp7/nXJgSoPbNOsYAYVFyPlw1NTWoq6uDaR55uti6urpB/x4zZky2QiOjQMm5yxmr3gErioD501czm694rBTmhvWQiURW9leglMFEKiv7IkcmpA0TSYR5LC3bs6REt5DUc+4CPsbQmonk3Ab1nDuBAZKG6AzbaaedhnHjxuG//uu/0N3dDdM08eabbw78fPHixeju7sauXbvw05/+FJ///OcdjJYMFSXnLiZTKRjrVoOXljkdSk5g0WJAUWGsezcr+yvk5bAkJedOMmXfjVghT88x1HEwGfTRNHuO8zGGzgwk56DOW2cwmnp2JBRFwQsvvICGhgZMmDABFRUV+O///u+Bn1922WWYOXMmzjjjDHz2s5+lwaAeQbO1uJix9l2AMfDiEqdDyQmMMSilpUi99Qb0s84F45m9N40olUjJOKQUYIzug52Qkr3wswLoaRoQ2iEkGABan8Z5OgMMAAkhEaCbJZLHqqur8Ze//OWw7/evA/Av//Iv2Q6JjBJlDC4lhUDqzdeglJZlPInMJ7ysAqKjHdbWzRnfV1SphoSAIXszvi9yZCkZR1QZm7btdQkBndHKr27QP2NOF83bnBuo15yQAZT1uZS1+UOIjnbw8gqnQ8kpTNfBY6VIvvFqxvfl42EEWRRJ2Z3xfZEjM2QvYuqEtG0vLiRNo+gS2sH/xtO8EBE95HIOHVqE9KHTkEslX38FPFYGptED9HTjFVWwd+2Atbc+4/sqUydTcu4QIW0kZTfKlElp22avkFQL6BKMMWjoa5O0o15cZ1ByTggAqjkfFl3X4fP5Mr6f1M7tsPfUInDK6eCadvxfcBg7eEbVVA3SC1e1oiLIkhjsN5cieuttGd3VODYb+zo2QVVV75VCHIxX1TRPjtSK291gJsP4yIlQeXpuckXKhspNaB44Lo+m//9DTdMGalK9SkuZgM+HgoL0LdCm6QY4l55sYy+3LecCmqahoCB9axIQ4lWUnA+DYRgwjMzPWx3/23NgJTHYmgb7KHOXugljDD7FB9MyPXNBYBWVSLy3Dh27dkApLc/YfiJiPEyZQI/RldZVKrOBMQbF54NleqddP6nbakWJMgGJnhSQpikt46kUIOVR5xT2AsYYfD4fTI+26ydxCXQlEuhm6as7N00TQghPtrGX21YIAcsy0d2dmSeN2ehYIyRdKDl3GbthH6ytH0E9cY7ToeQ0XlAIVhRB8rVXELruxoztJ8SjKOBl6JVt8MFbybnXJWQXxuunp3WbtrfynZzHGJD2qbGpjR3jsfuJ45IpA+m6bZQpWtAun1By7jLJpS+DR4vBQ2GnQ8l5SmUVzPVrIC69AjxanLH9VKuzscdci6hSnbF9kMFMmURKxjFGm5XW7eZY7uB5DOlP6IQNqn12Qg4tQhQZNx6R3z7jdBjEw2hAqIvYLc0wN74HXkXL62YDK4qAhcJILlua0f1U67PRKzthSer5yJa4OIAQL0EhT2/JEp0w3UUCSPcU59KmGVucwBgg6NEUIQDoWuMqyddeASuKgBcUOh1KXmCMQamsgrFqBUR3V8b2E1PGw8fC6BFtGdsHGaxXtKNGm5v2QbgaY94Y9JwnbPnxlIpp26YpaUo/BzAO2NR/QQiAEZS1GIaBF154ASeddBImTUrfFGX5TnS0w1y/BurU6U6HkldYtBjM70Pq7TcQuPyqzOyDcYzT5mKvuRFFCs1bn2mmTCEhOzFOm5v2bfsYkImZ+8jI2JDwp7nr3E5JMCWtmyRDwDiDncqNg6u7uxsp0QNTJtO63crI+LRuj7jXsJNzXddx/fXX4+WXX6bkPI2Sb74GFgqBFUWcDiWvMMbAK6qQWv4W/BdcAuYPZGQ/4/RTsM14C5Y0oDKauz6T4qIFIV6SkRr/EOew0r5VMhJSShgSCKa5m9tMSCgadZ1nG1cAK5kbyXlK9OBPrXfBlIm0bve2CNWx54sRDQidPn066uszv4BLvhCJXhirlkMZN8F7c2HnAF4Sg72vHqlVK+BfcGFG9hFTJiDIihAXrYgoVRnZB+nTI9owWT8nI8dSIWcwcm1KCY8y0VdzXqiktzrTSkioAToPZxtXgWRnbhxbpkzClAmM1U9KW2cMjVnKLyM6qz300EN44IEHsG7dunTHk5eMlcsBRQUviTkdSl5inEMpq0Dqrdcg7cxMF8AYw3j9dMTFgYxsn/RJiV4kZTfG66dmZPtRhSMpAUEJuuNSB9sgytObnJu9EopKyXm2cY3B7M2t40plOjTuT8sXPXHNLyPqOb/99tvR2tqK008/HbFYDGVlZYN6qRhj2LBhQ9qCzGXStpF65w0oZeXUa+4gXlYOe99emJs2QJ+T/lplAJign46PUq/AkL3QGc15ngndYj+ivBqFSmYWlio92EublECQDldHJYVEiDH40lhzLiwJKwlw7y0O6nlcA+xU34BcKisi+W5Eyfmpp2amVyofmR9tguzpAZ86w+lQ8hpTVfBYDKm3l2UsOS9SKhDhY9Btt6BEHZeRfeQzKSV6RBtmBz6VsX2EOYOfAQkpEaTJsB2VkBLlanp7zVPdfT23ik5tm239CbnRLREops9/NLZv347rrrsO27Ztw+233459+/ahvLwcixcvPuL7Kyoq8Mwzz2DBggUAgPvvvx+PPfYYkskkOjs7oSg0Qjrbhp2cW5aFr33ta6iurkZ5eeaWPc8Xxoq3wEtiYCqtB+U0Xl4Ba+P7sJuboJRnZlaVib4z8UHif1Asa+hJSZolZCcspDBOOyVj+2CMoVJV0CMEStJc60yGp1dKzNLSe95MdvSt56hQBUHW9X/myQ6BQDEdW8Mxfvx4/OIXv8Bll10GAPj+97+PM888c0Slx/X19XjooYdQW1uLqioaH+WUYR8BnHPMmzcPH3zwQSbiySuiox3W9q3gpXST4wY8GAIrLEJq9cqM7WO8diosJJGUmZtXPV/FRSsq1Rnw88yurjteU9CTrjW5yYj1CImxepqT83YBxdc3rR/JLsYZFD+QaKeDa7Rqa2tx4oknjuh3d+/ejWg0Som5w0aUnE+cOBEdHR0ZCCe/GOtW902fGM5sMkGGjpeUwFz7LqTIzAXCzwtQoU5Ht2jJyPbzlZA24uIAJupnZHxfEzQVcSEgaVCoY1JCIimBCVp6H7cn2gRUPyXmTlF9DIk2Ss6H48Ybb0RdXR2uueYahMNhzJs3D8uWLcPXv/51hMNhrFmzBjfddBPuvPPOgd/5yU9+gjFjxqCsrAyPPPLIwPdffvllXHLJJdi/fz/C4TCuueYaJ/4kghHO1rJo0SI88MADaGxsTHc8ecVYvwY8WkLlDS7Ci2OQvT2wdu3I2D4m6GegR7RBSroIpUuPaAcDR5U2K+P7mqirMNBXVkGc0SkEgoyhLM2lRb0HBA1GdJCiMyQO0HlxOJ566inU1NTg+eefRzwex8qVKzF//nw8+uijiMfjOO200wa9/9VXX8UDDzyAF154AXV1ddi+fTtaW1sBAJdddhleeukllJWVIR6P4/nnn3fiTyIY4YDQP/3pT2hubsbEiRMxe/bsI87W8re//S1tQWZCPB7H448/jvXr1yMQCGDhwoW4/PLLs7Z/u7UForkJ2pyTs7ZPcnxM08AiUZgb34c2eWpG9lGtnYh3AfTIdoRZSUb2kW/i4gDGaumbU/hYIgpHTOHoFBIhKo11RKeQmKqrae/Y6G3pK2shzlB0oKeFkvNM+sMf/oCbbroJc+f2TXzw8MMP45e//KXDUZFDjSg5j8fjmD59+qDXXvPEE0/Atm385je/QWNjI+677z5UV1dj9uzZWdm/+dGmvpKWAE2p5za8KALzw42Q13w2I081VObDGO1EtNv1CHNKzkfLlhZ6ZRvG6wuzts+ZuopNKQtVNI4766SU6BICF/nSn0X3tgr4Cqnn3Cmqn6G3lZLzTGpoaMCcOXMGXkciERQWFjoYETmSEV1a3njjjXTHkVXJZBLLly/Ho48+imAwiEmTJuGCCy7A0qVLs5acW1s+BCugA8KNeCQKe/cuiNb9UDI0WHecdgr2mhshpA3OaJqq0egVbVCZDxXqtKzt8wS/huUJA7aUUKgsLat6ZF+9+Uxfeicjl1Ii2SEQLKM7LqeoPoZkhw0pJA3KHYbhdCJVVVWhrq5u4HVnZye6umiCArcZ9UNZKSW6u7s9NThq3759AICampqB702cOBF79uzJyv6lbcOq3QVeWJSV/ZHhYX4/EAjA2rk9Y/uo1GaAgaFXdmRsH/kiLtpRrc7O6k3OVF0FB9AhvHPeyxXttkSVyhFJc7250S0hrL4EkThD8TNI++P55snQlJeXY+fOnUN67+c+9zn87ne/w/vvv49kMom77roLPM2r7JLRG3EXwZtvvonFixdjxYoVME0Tmqbh7LPPxv3334/58+enM8a0SyaTCAQCg74XCoWQSCQGfa+xsXHQoFefz5eW6YWspkbANMALi3JiMGj/35ALf0s/Hi6AvbsWytnnZWT7CgKo0k5Ap92IAiWWkX2MlhfaVUgbCdmO8f5TsrpQhgJgpl9Dk2UjpnrryUd/jyTjDPBgBUGHEDg35E97e6cOznGu+pmr/58/Fn4wbs6YF5sWmr/vv0YHECr21nHlpLvuugu33XYb7r77bvzHf/zHMd976aWX4q677sIVV1wB0zRxxx13IBZz5zUon40oOX/11Vdx+eWXY+rUqbjrrrtQUVGBxsZG/PnPf8aFF16I//mf/8FFF12U7ljTxu/3H5aI9/T0HJawP/HEE4NW1Fq0aBEefPDBUe+/64P3wYMh+HNsCkVdz6GVOwqLIPbVIxqNZmwX05WzsKzl/0LXdVcnA25u1y6zFYxxzCg9EyrPbpzzmYpf7tsPTdcHkiIv0TX3tuvRJGyBLpHCeeUxRP3prTnv2hqH6u9GIOhP63adoLn4mD0e1W+CGwFEowVOh+IZV111Fa666qqj/vy3v/3toNff/OY38c1vfnPQ634LFixAU1NT2mMkwzOi5Pyee+7B5Zdfjr/+9a+Dkopvf/vbuPrqq3HPPfe4OjkfM2YMgL6VsMaOHQugb9L+ceMGL6l+yy234Morrxx47fP50N7ePur9x3duB/wBpFKpUW/LDRhj0HUdhmF4qrzpWITPB7N2J9paW8Ey1CMbEeNhyiS6kgfg5+67EHmhXTvN/ShXp6K7swdAT1b3PVEI2FJifyKJqIdWC2WcQdd0GKYB6bGynAbTQqnCEertQXuiN63bbt2bgKLD0+dlzhg0XYdpGBAuPWaPR/EBrfu6UdRupXW7mexoISTdRpScf/DBB1i8ePFhvX2MMfzbv/0brr322rQElyl+vx9nn302lixZgttuuw3Nzc147bXXcPvttw96X2VlJSorKwdet7a2wrbtUe/fbmoCfLprE56RklLmzt/k8wNCwGxtgRIrzcguNAQR4WPQI9rhY+59iuLmdk3ILkxWz0nLcTlcOoAZPhUtlo2IlwavHax3kMK97Xo0bbbAWQEdIgOLhCU6bHCNee4z+aT+T0W4+Jg9Hq4yJNptR45pQtxiRN094XB4YFDlofbu3YuwB8o1brnlFgDATTfdhMWLF+P6668fNL1QJom2A2A+7z86zWm6DjAG0d6W0d1UajOQFN0Z3UeuMmUKhuzJ6iwthzrVr6PNFp7tpfSShJDoEhKnBDJTspHqFuBU5uw4rgBGtxcr5glJnxH1nF955ZW48847UV1djUsvvXTg+6+88gruvvvuY9Y+uUU4HB60nG02yZ5u8EjEkX2ToWGMAT4/ZDyziXO5OgVbUm9ASgHGvFMa4QZJ0QmdhVDIMzPd5VCc6NMgAbQLiRLFQ73nHtRqC5QrHJUZGoBrdEtwmkXRcVyj2VoIGdGp6Ic//CE++OAD/NM//RMKCwtRXl6O5uZmdHd347TTTsMPf/jDdMeZM6SUkMkkoNJVwO2YqkL2preu9VAxZQIkbKRkD/zMfXXnbpaQ3ShVJjo6mNbHGU70adhr2SjxUN2510gp0SYEzg9mbvlOMyHBVLrBchpXGMxEbiTnljTSNiOSJY30bIh4wogyxGg0ipUrV+LFF1/EO++8g/b2dhQXF+Occ87BFVdcQXNmHottA0IAWZz2jYyQokAamT0h6jyIMC9FUnbDD0rOh8OQvShVJzodBk4NaPigw6QFiTKoV0rEhcQp/vQuPPRJdkpCoTnOHccUwEp6OznXmB8aC6DeeN/pUIhHjSg5f+2111BXV4cvf/nLg2YzAfqm7Bk3bhzOP//8tASYc/oHMtFF3P0YA0TmByWVKOPQZtcd/41kgJQCKRlHsVJz/Ddn2EyfBs76SltiVNqSEa22QLXKMzqnvG0CKg0FchzjgDCdjmJ0YkUV+Lei3zgdBvGwEXVx33PPPWhubj7iz1paWnDPPfeMKqic1n/t9nbHQH7I0iC/YqUapkxmZV+5wpAJSAhElDFOhwKNMczxaThAs0tkhJQS7bbEaRkaCPrxjvDx+Zk4h8GzM80Qki4jSs4//PBDnHrqqUf82dy5c/Hhhx+OKqicphx8WCFpNLrrSQlomXuM3q9IqYQhe+iCNAyG7IXOgvBzd8wMNdevo92WsKkN065HSvRIiZP93l1YhwyDBOguieS7EZW1MMbQ2dl5xJ+1t7fT/KTHwDjvS/joM3I/28rKlJeFvBwCNiwY0JC5AW+5xEQSBbzM6TAGTPepUKi0JSMO2AJjVQXFGR5wy5SsPSwjxyLh+Sktmzo6cde+ViTS/D/UM7Mmp3V7xL1GlJyfccYZePzxx3HttdcOmilBSomf/exnOOOMM9IWYC5i/gCk6fGiujwgTRMsGMr4foK8GAwcpkxAY5ScD4Upk4gpE5wOY4DG+mZtqTNtxGjWlrTqsCUuDGW+11zRGST1mThOiL7pFL0sKSUSUuKkgA49TePLDLpzzCsjSs4XL16M888/H7Nnz8ZNN92EyspKNDQ04Mknn8S2bduwbNmyNIeZW1hBAUDJuatJIQDDAC/I/AwqnHEEWBEs6d1lw7PNlibCSszpMAY5ya9hQ9KEkBKcBnynRa+QiEuJkzI4S0s/LQCYvZQAOU3aElogN44fnTH40zV7XQZWxSXuNaLkfN68eQPL3d9xxx0QQoBzPvD9M888M91x5hQeLYY40Op0GORYDk6hyCPRrOwuyCOwQMn5UFlIIcSy0zZDNVPvW5CoU0hEqbQlLdpsgVKFozSDs7T008IcyS7qOneaMAG9hJ4+kfw24pVwzj77bCxfvhyJRALt7e2IRCIIBoPpjC1nKbFSiMYGp8MgxyBTSYArYIVFWdlfiBej3d6blX15nZQSpkwiwLPTNkPl4wxTdBVttkCUSlvSolMInJalgaD+QoauOuo5d5qwJHxFdHNL8tuoryCBQABVVVWUmA8DLy3rWyWUuJZMJMCjxX0DeLPAzwshQL12QyFgQ0LAzwudDuUws/0auoSkmXfSwJQSnUJiVhZKWgDAH+GwqdrQcbbZ1xaE5DM6AhyglFdCJnohs7DADRkZmegFr6zM2v78rAC2tLK2Py+z0ZdB+Zn7VlQ9QVfRKyVyZPVxR3XYAj4GjNeyM3VHoJh7fmXKXGCnJAJRSk1GY/v27TjllFNQUFCA7373u7j11lvx7W9/+6jvr6ioGDRW8P7770csFkM4HKbZ9xwy4rIWMnJKRRUgJWRvL1jYfQkGAWQyCbVqbNb252dhCFC33VDYsu9z8rHMz6QzXCWqgpjC0S4Egl6fD85hHUJiuq5BydLg2kAJh53qK6vgKpVVOEHYElYSCJZScj4c48ePxy9+8QtcdtllAIDvf//7OPPMM7Fu3bphb6u+vh4PPfQQamtrUVVVle5QyRDREeAAFgiARaKQPT1Oh0KOQEoJGY9Dqa7O2j51FhpIOsmxCVhQoYMzdya/s3wqumyaWWE0pJToEgIn+LLXfxQq6/v/iXrPnWMdfOTU3xZkZGpra3HiiSeO6Hd3796NaDSa1sTcpNnpho2Sc4eo48ZD9sSdDoMcgUz0ArYFZey4rO3Tx0OwYFKt8hDY0oLG3DvGZYZPQ6eQENSWI9YrJZISmO7L3oTXWoBBL2A0naKDzF4JvYDlzFSK2XDjjTeirq4O11xzDcLhMObNm4dly5bh61//OsLhMNasWYObbroJd95558Dv/OQnP8GYMWNQVlaGRx55ZOD7L7/8Mi655BLs378f4XAY11xzzTH3vXXrVlx++eWIxWKIxWL42te+BgBYtmwZKioq8OMf/xhVVVW48sorAQBPPfUUTjjhBEQiEcyfP39gNfmnnnoKF1988cB2zzzzTMyfP3/g9YUXXoglS5aM/sPyEErOHaKOnwjZSz3nbiS7u8GKIuAF2RtwqLMgAEmDQodAwDr4ebnTZE2FBNAlKMkbqQ4hEVN41me9KRyjwOyhdnOK2StRUEW95sPx1FNPoaamBs8//zzi8ThWrlyJ+fPn49FHH0U8Hsdpp5026P2vvvoqHnjgAbzwwguoq6vD9u3b0draN7XzZZddhpdeegllZWWIx+N4/vnnj7rfeDyOiy66CGeffTbq6+tRX1+P6667buDnra2t2LNnD3bt2oXnnnsOL7zwAu6991784Q9/wIEDB3DDDTfg05/+NAzDwPnnn48VK1bAMAzE43HU1tZi+/bt6O3thWEYWLlyJRYsWJCRz8+tqOZ8GHRdh8+XnhUcfTNPROJvf4EKgGneXg6Noa+XQ1M1SHj/wiZ6exCYNh0FWViAqB+3y4BugKuA5pbl8Q7W+qqa5q51zaVAkBdmtX2GowDAhK4EuoVEqQuP7f5VnTVNc+2Tmh7TxknhYNbbuHSKjT0reqG5sN2Gwgtteyx20kbZSQHXHtu54A9/+ANuuukmzJ07FwDw8MMP45e//OWwt/Piiy+iqKgId99998D3zjnnnIF/Synx0EMPwe/3AwB+/vOf44477sDs2bMBALfccgt+8IMfYNWqVTj33HNRUVGB1atXo6enB/PmzYOUEitWrIDP58OYMWMwZsyY0fzZnkPJ+TAYhgHj4OI0oyUjUUDXYbQdAC8uScs2ncIYg0/xwbS8X5YhpYTV1QmtZgK6u7uztl/74OeWMhNg3B2HJWMMis8Hy3RXu5p2CgEUZ7V9hmu6yvFObwqm6b7H84wx+Hw+mC5r135CSrRbNiYymfU2DlTYSHbZMAxjINH1Ere37bFIKZHssuGvsDPS7unqWPO6hoYGzJkzZ+B1JBJBYeHwnxLX1dVh8uTJR/15SUnJoCm2d+/ejW9961u46667Br5nGAb27dsHAFiwYAHeeOMN9Pb2YsGCBZBS4o033oDP58u7XnOAylocwxQF6vhJEF2dTodCPimVAhIJqJOnZnW3CtOgQIMNmk7xeGzYrpxG8ZOm6iq6hITlsQTJDeJCwgYwWc9+eUNknAphAlYi67vOe1aib3XQyHh3dE54yXBuJKuqqlBXVzfwurOzE11dXcPeZ01NDXbu3HnUn/ND1gipqanBY489ho6OjoGv3t5efP7znwfQl5wvW7YMy5Ytw4IFCw57nW8oOXeQOm06pIt7//KR6OwAKywCL4llfd80Y8vQCGnBz8NOh3FM4zUFKoBOqjsftk4hMUblCGZpAbBP8kc4fIUMqW6abSfbUt0CegGjOc5HoLy8/JiJ8id97nOfw+9+9zu8//77SCaTuOuuuw5LpIfiU5/6FNra2vDwww8jkUggkUjgnXfeOer7/+3f/g0PP/wwNmzYACkl4vE4XnjhhYGnJOeffz5WrlyJXbt2Yfbs2ZgzZw527NiRl/XmACXnjtKmzoDsiUOmUk6HQg4SXZ1Qp81w5JG2nxfARnrKpnKZDdP1PecKY5ikq+ikKRWHrVsIzNCdq/kunqzC6KKbqmwzuiWKJ1Gv+Ujcdddd+MEPfoBIJIJ77733mO+99NJLcdddd+GKK65ATU0NJk2ahFhs+J1R4XAYr776Kl5//XVUVVWhpqYGf/rTn476/quuugrf/va38aUvfQmRSARTpkzBU089NfDz6upqVFVVYd68eeCcg3OOefPmobq6Ou/qzQGASa8Vpjmof0Rzukgp0bV4EXhZGZSyirRuO5v66xxTqZTn6hw/SQoB8721CF73RehzTs76/t+MP4G4aEWpOjHr+z4St7ZrrbEaZ4e+jGptttOhHNPSniRe60lhdhanAxwKt7Yr0Df2YnXSxL9FQ5jh0Oe25+0Utr2YRMVJ7mq3oXBz2x5P0/smplzux/jzMlMbPpIEdKR2t3fgzn2tOD3ogz9NT4CSQuC/JmZvYTziLOo5dxBjDOqMWRCdVHfuBjLeDdg2tKnTHdl/iBdTWctxCGnDhokgizgdynFN01V0CwnDY0mSk7qFBAMwSXeuB7V4sgqzR8JKUbtli52SMHskSqZQzzkhACXnjtNPOBGyowNS0ONvp4n2NijjxoMFAo7sP8SLqazlOEzZVwIW4sUOR3J8Y1QFAcaotGUYOoTAOE2B7uBMKeEKDi3EkOqkdsuWZKeAFmQIV1BK4iZLlixBOBw+7CsSiTgdWs6jI8Fh6pRpAABJveeOklJCdnZCO/Ekx2II8xgMmfDc4+hsMpGACh90FnI6lOPijGGqrtCg0GHoFhIzHS4DYoyhZKqKFNWdZ02qS6JkqgrGvTd9ZS67/vrrEY/HD/vq6OhwOrScR8+QHMZ8PqhTpkK0tYFHo06Hk78SCcjeHmgnOFfHXKiUwYYJGyZU6I7F4WamTCDMY56Zg3qGT8M2o++GyysxO8WUEl1CYpqDJS39SmeoaN1iUrtlgZQSqS6BCefn3jzkhpRAmp6KU3lcfnH+LEigzZmLxPN/oguBg0TbAfDyCigOTKHYL8xLwcBhyF6ojJLzIzFkAjFlvNNhDNl0XcUzEkhKIECH9jF1CgkfA2o055dvL5mmwkr0zb2tBY//fjJyVrLvc45Nz510xM8YAozh/QSVKZKRyZ2jwcO0E05E4k9PQ3Z3gRUWOR1OXhId7dDPOtfRGBSmooCXIiV7EETE0VjcypQJRJVqp8MYspiqIMIZOoRAgDufdLpZhy0wRVehuKCDIliiIFDMkOwQ0ILUbpmU7BDwRxmCsdz5nCsiRfhNhK7lZOSo5twFeDAEdeJkiAPpnaqRDI1MJCDj3dBPyv70iYcqUSbAkD1Oh+FKQtpIyjiK1RqnQxmWE3waOmnA9zFJKV1Rb/5JpTM1qjvPglSnQOlM97Q7IW5AyblLaHNPg2hvp8GADhAHWsHLyl0x13ypOh4pSs6PKCXjYACKFW/N9Tvdp6LLlnRsH0NSAr1SYoYL6s37lc7UkOoUkDSgN2OkkEh1SpTOdE+7E+IGdES4hDZrDhJ//gNkVydYUcTpcPKGlBKivQ2+c893OhQAQKk6CYbshSVTUFnuDZAajYTsQoSP8dznMk1XYaFvJpJCxfmSDTfqEAIRzlCquqe0oXiKCimAVLeEv4jaLROMbglhAyVTc6vn/EBTJ/5xVyusRHpv7L74zOS0bo+4FyXnLsGDQajTZkC0toJTcp41srcXsicO7aRTnA4FAFDIK+BjIfSKThQqZU6H4ypJ0Y1xmjvaaTiCnKNaVdAhBAoVelh5JJ1C4AQXlbQAgBZgKKpR+mqii6jdMiHZIVBUo0DLsdHSVlLCSkhUnKRD1dPzt1kGPcHJJ55Jzn//+9/jpZdeghAC8+fPx8033wxVPTz8jo4O/OpXv8KmTZuQSCQwZswY3HjjjTj55I/ria+88kr4fL6BmVFmzpyJ+++/P1t/ylHpp56B3j88CSkmgNHgsawQrfuhjB3n6Cwtn8QYQ4U6A232HhSCkvN+QtpIyE5UajOcDmVEZvlUrKCZG45ISIlOW2KGz32Xo7JZGva8lXI6jJyV6paoOSd3Z6ZSdQbVn64bOxq3kk/cdzY8gldeeQVvvfUWfvzjH8Pv9+O73/0unn32WXzhC1847L3JZBKTJk3Cl7/8ZUSjUaxatQoPPfQQHnvsMZSVfZzs/OQnP0F1tbtmfdBmzgI4h2hrgxIrdTqcnCelhGhrQ+CKq5wOZZAx2izsNTfQ1Jqf0Cs7wKGgVJ3odCgjMt2n4eWeFCwpoVKbDhIXEjaAqS6qN+8Xm65i24tJ2KaEolG7pZNt9tWb59IUioSkiyee1S1duhRXX301ysvLUVRUhIULF2Lp0qVHfG9FRQWuueYalJSUgHOOs846C6WlpdixY0eWox4+punQTjqFZm3JEtnZAZgmtJPmOh3KIFXqTAj09RSTPj2iDVXaTCjMXaUPQzVeU6AxoIMGFx6mQwhUqwqC3H2Xo6IaBaofSHVQr2W6pTolFF/fZ0wIGcx9Z8MjqKurw/jx4wdeT5gwAa2trejpOf6sFgcOHEBjYyNqagZPv3bPPffgxhtvxHe+8x3U1dWlO+QR850+D7KjHdKgR6mZJlpboJ0wCzzorqXgdR5EuToFcXHA6VBcQUqBXtGOGs1dN1HDoTCGqbqKDpuSvEN1CYlZLixpAQDGGUqmaUh20k1VuiU7BGLTVHAaJE3IYRw/I9q2fcyfK4qCZDKJUOjjBKr/34lEYtD3D2UYBn7wgx/gkksuGVTC8r3vfQ/Tpk2DaZp47rnncN999+FnP/sZgsHBS8E1NjaisbFx4LXP50NVVdWw/r7h4hMngxeXQLS0QK32xpRx/aUXXirBkKYJ0XYAoWs/B0VxX8/NRN8ZWN3zDICJYMyZe2i3tGuP6AAAjPXPgcLc11ZDdYLfhxe7eh3/PBlnH//X4XsFS0p0CYmZAZ8rj0MAKJ+l48C2XgDOHwvHww/GxxlzummPSUqJVJfExAvd2+5etn37dlx33XXYtm0bbr/9duzbtw/l5eVYvHjxEd9fUVGBZ555BgsWLAAA3H///XjssceQTCbR2dlJbeQAx5Pze++9F5s2bTrizyKRCJ588kn4/X709vYOfL//34FA4KjbNU0TDz/8MCKRCP71X/910M9mzZoFANA0DTfccAPeeOMNbN68GaecMngmiCeeeGLQ/8yLFi3Cgw8+OLw/cATYhZeg/f/9HfrESa6/GHySrntnYE+qZT+UgkKUnXkWmAsfp88W5+Pd2j/AUOIo1Jwdf+B0u+7vbcOk8OkoK3Z+HvrROCNo4I+dPRCqhoALZm3RNeeP1y7DhMYYTi4vdW0tvnamiY1L9oDbGvSQN5IUzeXnYqPHhpUwMPnMGAqi3ixVc5Px48fjF7/4BS677DIAwPe//32ceeaZWLdu3bC3VV9fj4ceegi1tbUZ74x0s927d2PChAlIJBLw+/1Z37/jyfn3vve9476npqYGtbW1mDGjb6aG2tpaxGKxo/aam6aJ73//++Cc41vf+tZx7/qOlgDfcsstuPLKKwde+3w+tLe3Hzfe0RKz5kD8+RkkW/Z7YlpFxhh0XYdhGJ5YaEVKCbOpAb6zzkVHp3vrusdqJ6E1VQufKHRk/25oV0ua6LJacBq/LivHXib5pUQhZ9ifTKLCwfm8GWfQNR2GaTi+wE6LYWGSrqK7o8PROI5JBYIxjs7mJArHOH7JPCbOGDRdh2kYEC4+F3fvtxEo5rC0OLJ1WEej0ezsyAVqa2vx2c9+dkS/u3v3bkSj0bQm5qZpQtPoJmw4nO++GYILL7wQf//737F//350dXXhj3/8Iy666KIjvteyLPzgBz+AaZq44447Dptusa6uDjt37oRt20ilUnj66adhGAamTZt22LYqKysxd+7cga/y8nLYtp3xLxkMQZ05C/b+ZkgpPfEFwPEYhvolursg43Hop52ZlfYc6dck/SzExQGYIpm37dplNyHAilDKJjneHqP9EkJgmq6i0xbOHgMHE3IpnD8W40Jguq443jbH+yqdqSLV6fzndbyv/oRcuCCWY30lOwVKZ6pZbcNcdeONN6Kurg7XXHMNwuEw5s2bh2XLluHrX/86wuEw1qxZg5tuugl33nnnwO/85Cc/wZgxY1BWVoZHHnlk4Psvv/wyLrnkEuzfvx/hcBjXXHPNMfe9detWXH755YjFYojFYvja174GAFi2bBkqKirw4x//GFVVVQOdnE899RROOOEERCIRzJ8/Hx9++OHA9y+++OKB7Z555pmYP3/+wOsLL7wQS5YsOWYsnZ2duPXWW1FdXY2ioiLMnz8fiUQCAPBf//VfqKmpQUFBAebOnYs333xz4PdWr16N0047DYWFhSgrK8O///u/AwDOPfdcAEAsFkM4HMbLL798zP2nm7u7AQ665JJL0NLSgm984xuwbRvnnnsuFi5cOPDz+++/HzNnzsTChQuxZcsWvPvuu9B1Hddff/3Ae7761a9iwYIF6OjowM9//nO0trZC13VMnjwZixcvRjgcduJPOyrfWfPR86ufQxoGmMsfUXqN2N8MddoM8Gix06EcU5kyGWEeQ5doRrFSc/xfyDFSSnSLFszQL3Ss7j7dpvk0bEyZkJKmyTSkRLeQmObCKRQPFZuhoW5535OG/pp9MjJSSKQ6BUpnUk9qOjz11FN4++23B5W1LFiwANdddx1uvfXWw97/6quv4oEHHsCrr76KmTNn4utf/zpaW/tmiLvsssvw0ksv4brrrkNTU9Mx9xuPx3HRRRfh1ltvxV/+8hcAGFRG09raij179mDXrl2QUuKFF17Avffei7///e844YQT8Ktf/Qqf/vSnsWXLFpx//vm49dZbYRgGDMNAbW0tGGPo7e2FqqpYuXIlnnzyyWPG86UvfQmMMbz33nsoLi7GqlWrwA+WrJ5yyilYtGgRIpEIfvrTn+Kzn/0sdu/ejWAwiP/8z//EbbfdhhtvvBE9PT3YuHEjAOCtt94amHwkL8tahoIxhhtuuAE33HDDEX/+yQWEZs2ahb///e9H3dbs2bPx85//PN0hpp06eRp4UQSipRnKGG8MDPUCaZoQB1oR+PSxewTcgDGGab7zsCHxIqK8OmcS1KHqle0wZQqTfGc5HUraTNVVpCSQkEAwz3O8TlvAz4AqB0t8hqpkqgpIINUl4Y/kecONUqpLQoqDnynJuj/84Q+46aabMHdu3+xXDz/8MH75y18OezsvvvgiioqKcPfddw9875xzzhn4t5QSDz300EBi+/Of/xx33HEHZs+eDaCvbPgHP/gBVq1ahXPPPRcVFRVYvXo1enp6MG/ePEgpsWLFCvh8PowZMwZjxow5aixNTU3429/+hv3796O0tG+M1tlnnz3w80921H7zm9/EAw88MDDOUNM07NixAy0tLSgtLcW8efOG/VlkQn5d7T2EcQ797PNgt7RASvfWDnqNaNkPVlAAdfoJTocyJBP1MyAh8nJaxU67CeO0U+DnBU6HkjbFCkeEM3QKN8+lkR1dQmKyrg7MMOJmqo8hOlFBsp3abbSSHQKR8QpUv/vbPRc1NDQMmlo6EomgsHD445rq6uowefLko/68pKRk0Ax4u3fvxre+9S1EIpGBr8bGRuzbtw9AX2//G2+8gWXLlmHBggUDr994442BWWSOFUtRUdFAYn6oRx55BDNmzEBRUREikQg6OzsHnhb8+te/xvbt2zFz5kyccsopeP7554f5SWQGJecupp9+JmCZkO1tToeSE6SUEC374Tv7PFfO0HIkGgtgkj4PnaIpr27SUqIXvbIdM3wXOB1K2k3TVXTTYkSIC4lpundKG8pmaUh1UbuNVqpLouxE77S7FwynRK6qqmrQ2i6dnZ3o6uoa9j5ramqwc+fOo/6cH3KNrampwWOPPYaOjo6Br97eXnz+858H0JecL1u2bFBy/snXx4vlkwn3J7399tv43ve+hz/+8Y9ob29HR0cHioqKBq6nU6ZMwdNPP43m5mbce++9+NznPof29nbHyw69kaHkKR4MQTv5NNj7m50OJSf0L+6kn+GtMonpvvORlN1IyuGfQL2qUzSgTJmMqFp9/Dd7zBSfhm4h8upm61CGlIhLiSm6+0ta+pXN0mDEJaxk/rbbaFlJCaNbomwWJefpVF5efsxE+ZM+97nP4Xe/+x3ef/99JJNJ3HXXXYcl0kPxqU99Cm1tbXj44YeRSCSQSCTwzjvvHPX9//Zv/4aHH34YGzZs6BsMHo/jhRdeQHd3NwDg/PPPx8qVK7Fr1y7Mnj0bc+bMwY4dO7By5crjJucVFRX49Kc/jVtvvRWtra2wbRsrVqxAKpVCd3c3VFVFLBaDZVl48MEHB92M/P73v0dLSws45ygsLISUEqqqorS0FJzzIX+u6UbJucv5558H2dEO8Yl53snI2M1N0E4+FTzkrsG/xxNWYhirnYQO0Xj8N+cAS6bQLVow03/x8d/sQVM0BUkJ5HOO1yUkfB6pN+8XKufwRxkSbVTaMlKJdgF/hCFcQalHOt111134wQ9+gEgkgnvvvfeY77300ktx11134YorrkBNTQ0mTZqEWCw27H2Gw2G8+uqreP3111FVVYWamhr86U9/Our7r7rqKnz729/Gl770JUQiEUyZMgVPPfXUwM+rq6tRVVWFefPmgXMOzjnmzZuH6urqY9ab9/vd736HoqIinHjiiSgpKcGiRYsghMCll16KT33qU5g+fTrGjRsHTdMwduzH4/j+8Y9/4IQTTkA4HMa///u/4+mnn0ZBQQGCwSDuuecenH/++YhEIvjHP/4x7M9oNJjM5+6bYTrSI5Ns6H7sx5C9vVAnTnJk/8fDGIPP50MqlXJtb6Ds7YW58T2Ev3En1DHe641ts+rxcvz7GKueBB8/+qq46eRUu7ZatRAQuKLgbscfLWaClBJ3t3ShVOGOzHfuhuN1l2EhqnB8tdhbN8of/aUXTe+brp1pxA1teywtH5momKNh5meCx39zmo0kAR2p5t0d+Medrag+3QfVn54bESspcMF/0eQQ+YJuXz3At+BCiAMtkKbpdCieZTc1Qhk/0ZOJOQAUq2NRrk5Fh2hwOpSMsqWFLrEfs/yX5mRiDvQlUFPzvO48LiWm+rw3W0fFSTqSHRK2mb9tN1K2KZHskKg4iaYGJuR4KDn3AG3miWAFhRD7jz3vKDmyvukTW+BfcOSFq7xilu8ydIsWmDLpdCgZ0yka4WcFqNHmOh1KRvUl5/lZHmEenN98igfmNz9UdIICPcyQOJCfbTcaiQMCeqhv1hviDUuWLEE4HD7sKxKJZD2Wurq6I8YSDoexcuXKrMeTaZScewDjHP7zLoS9fz9knl7QR0M0N4EXFkGdOcvpUEalTJ2CYqUa7fY+p0PJCCFtdNlNOMF/CTjL7Qv4ZF1FQgLJPOw97xISOoBqD9Wb92OcoXKuRsn5CCTaBCrnarSIk4dcf/31iMfjh311dHRkPZaampojxhKPx10zN3k6UXLuEfrp8wBIiAPO1L17lRQCdkszfAsu9Mz0iUfDGMMs/z+hW+yHJQ2nw0m7LtEMhemYqJ/pdCgZV6pwFOTpfOedtsBEXYXi0bKlMaf3lbbQrC1DZ6Ukku0SY07Pr5IWy5CwkiI9Xwb9/5ZPvPdcMU8xnw++s+bDWL0KPFaas/W46SZa9wOMQz81NxK+MeosFPBSdNgNiKnjnQ4nbaQU6LQbcYL/MijMnYPt0qm/7rzetFHudDBZFpcSZ3iw3rxfUY2CUBlHz36Bohrv9f47oWe/QLCUo2hcfnxeqp9BDTA0vZ97nSgkO7x7hsxDvnMWIPXm65AdHWDRqNPhuJ6UEqKpCb5zzgPz+ZwOJy0Y45jlvwyrepcgKquhsNw4hLtFCwBgiu+c47wzd0zXVXyYMiGlzJubbeNgvfk0D9ab92OMoeYcHdtfSqJwLM+bthspKSV6W2xMvsyfN59VSUURvvCbIqfDIB7m7ef8eYYXRaDNPQ12U27P2JEusr0NMpWE7+zznA4lrWq0ufCzAnTmyLznUkp0iEZM8y2AxvxOh5M1030akhLodeGUd5nSaQsEGMMYD9abf9KYM3TYJpBsy5+2G6lku4RtANVn5FdJCyGjQcm5x/gvuBiyswPi4Kpa5MiklLCbGqGfdiZ4QYHT4aQVZwpO8F+CLrsJQtpOhzNqPfIALJnCVN8Cp0PJqqjCUapwdNj5k+B1CIkZPhXc4z2oeohjzKka4s3eP/4yLd5ko+pUDXqY0g1ChoqOFo9RyiqgnnAi7EbqPT8W2dUF2d0F3/m5ucrkBP0McKahS+x3OpRRkVKiw27EZP1s+Lm3FqRJhxN9at4MCpVSolMIzPLlxpiCCRf4kWyXMOL50X4jYcQFku0SEy7InydihKQDJece5L/wMsi2VojeXqdDcS27qQHanLlQSrK3Klw2qUzHdP18dIkmV64EOFQJ2YmU7MYM/4VOh+KIWT4NHULC8nAbDlW3kDAkMMPDg0E/qaBKQWyGiu4GSs6PprtBIDZdReEYb5cxEZJtlJx7kFozDuqkqRANuTnf9WiJeDdkexv8F17qdCgZNcU3H7Y0ERcHnA5lxDpFI8ZqJyPEi50OxRGTdBU+BrTbuZ/gtQmBcZqCAo9PafpJU/7Jj94WAbM392+uhsvslehtEZj8T9RrTshw5UYXRpboug6fS2b90K7+Z7T8+GEo48eDB4KOxsLQVz+qqRoknL9IJZsa4Z99MiJTpzkdSkYVoADTxXmo7V2LYq0y/Ts4WBesahqQgZ7dlOhBj9GGS0v/EwV6bo0LGI65vSZ2JFOo0rJT7tE/Y4amaVl76iKlREfKwpXFBSjIoTEgBXOAnTNNxBtslM1wfsCjE217NO0NKZRO92HcSTSzGCHDRcn5MBiGAcNwybyllWOg1IxHqm4P1ImTHQ2FMQaf4oNpmY5fEERPHHZrCwI3fAXdeTBodiI7Gx+KpehKHUCAF6Z124wxKD4fLDMz7dpq1SOmTIA/VYLuVO631dHMURneNW0kFANqFgZKMsbg8/lgZqhdjyQuBHqEwEwmcu64nHy5hhU/SiFUAccHPTrRtkdixAXizRZm3+h3TXu7pWONkKHIneeLech/6RUQLfshk0mnQ3ENu2Ef1GkzoVaPdTqUrChQSlGlzkKXaHI6lGHpK8dpwQzfRU6H4rjpugqdMbTlcGlLiyUwQVMQVXLvkhMZr6J8torOOtvxzgk3kFKis85G2YkqIuOp/4+Qkci9M2UeUadMgzJ2HOyGvU6H4gqipwfyQCv8l13hdChZNd13PuKiFZZMOR3KkHWJZvhYGGO0E50OxXEKYzjVr6E1R5NzISUOCIEzA86XfWTK9GsCSHX2LVGf75LtEqlOielXB5wOhRDPouTcwxhj8F/2qYO95wmnw3GcvW9vX6/52HFOh5JV5epUhHkpOu1mp0MZEiklukULpvkWgDM6BQHAvKCOdiGRFLmX3LULCVsCc/25m5yHShWMO09H5x4bMgfbcKik6Os1H3eujnA5zdBCyEjRldHj1CnToIwbD3tffveei3gcsq0V/n/6tNOhZB1jDNN856FbtEBK9/e+9sp2WDKFifo8p0NxjbGqgiqVo8nOvUVt9lsCc/0aAtzbCw8dz5TLAwBDXk+t2N0gAAlMuZxmaCFkNCg59zjGGAL/dBVEawtkIn/nPRf76qGecGLe1Jofarx+OgQs9Mh2p0M5rm57P8ZqJ+XlokNHwxjDeUEfWm0BkUN1ywkh0SYEzg3m/mA8LcAw49oAuvbasJK504ZDZaUkuvfZmHGtH1qQUgtCRoOOoBygTpoMddIUWHvrnQ7FEaK7C6K9DYHL8q/XvJ/OAhinnYJu290rhloyhbhswxTffKdDcZ1T/ToYgJYcqj1vsmxUqxzj9fwYGDjmdA1F4xR07LacDiXrOndbKKxWMOb03C1fIiRbKDnPEf7Lr4I80AoRjzsdSlZJKWHvrYd20ilQKqucDsdRk31no0e2wXTxwNAusR8hXoxSZZLTobiOjzOcE/ChyRI5MeuHJSX22wIXhvKnxIExhhM/H0SyXaL3QO7cZB1P4oBAok3ixOuDYDlevkRINlByniPUmnFQTzgR9r46p0PJKtnRAdndlZe15oeKKRNQwEvRLdzZey6lRFwcwBT9nIHFUshgC0I+9EqJjhwYVNhkCYQ4w1x/dhZXcouCSgUTL/ahc7cFYXm/HY9HWBIduy1MvMiHgkoaBEpIOlBynkMCV1wF2dEB0dnhdChZIaWEva8e+hlnQymJOR2O4xhjmKSfhbg44Mqe16TshiF7MV4/zelQXCuicJzm19BgeXtgqC0lmmwbF4f8UPLwRmzyZX7oYY7OPd5ux6HorLOhhTgmX5Y/T0gIyTRKznOIUlYB7dQzYO+td2Vylm7iQAtkKgn/xf/kdCiuMV4/DYbsQUq6r7ypW7SgXJ2KII84HYqrXRr2o1NIdHq49rzZFlDAcFYwP+uPFY1h9g1BxJsFkp3ebcfjSXUJxJsEZt8QhKLn300YIZlCyXmOCVx6BWRvD2TbAadDySgpBMS+vfCddyF4YXqXrfeyII+gTJnsutIWKQV6xAFM1M9wOhTXK1MVnOzXsM+jvedCSjRaNi4O+aDnYa95v+LJKsaepaOjNjfnPpdCon2Xjep5Okqm5MeAX0KyhZLzHMMjUfjOPR/2vnpIkbs9NqK5CWAM/gW0/PuhJuhnoEe0u+rpSa/sgIREtTbH6VA84YqwHx0e7T1vsgU4WF5Mn3g8068OgHGgs96bN1rH0lVvg3FgxjW0Eigh6UbJeQ7yXXAJpJAQ+72xYuRwScuC3bAP/kuuAPNTneOhxmpzYMNCQnY4HcqAuGjFGG0WNEYJ21CUqwpO9WvYa9muusk6HltKNFg2Lg354KNZO6AFGE78QhDd+wSMuPdutI7G6BHo2idw4ueD0ALUzoSkGyXnOYgHgvBffBnshn2QVu7Nt2s37AUPh6HPO8fpUFxJ50FUqtMRF+4obRLSRo9ox3jtVKdD8ZQrwn50CW/N3NJoCeiMYX6IbsL6lZ+ooeIkDe27vHWjdTRSSnTsslE+W0P57PyaiYeQbPFModjvf/97vPTSSxBCYP78+bj55puhqkcOf9GiRdi6dSsU5eNpnZ599tmBf2/atAm/+MUv0NTUhPHjx+O2225DTU1Nxv+GbPKdfS5Sby+D3bgP6thxToeTNjKVgmhqRPD6m8AUmrbraMbpp2J17x8gpQBjzt6D98oOMDBUajMdjcNrYqqCswI6NqRMRDhz/fST1sFe838uDOR1rfmRnPDZAN78ThfiDQIFY7x93oo3CliGxKzPUTkLIZniiZ7zV155BW+99RZ+/OMf4xe/+AV27do1KNk+kn/5l3/Bs88+O/DVr6urCw8++CA+85nP4A9/+APOOOMMPPDAA7Dt3KoJZKqGwBVXQTQ1QhruXZRmuOy9dVCqxkCbfbLTobjaGG0WBGwkZJfToaBHtKFKmwWV5efMHaPxT2E/EkKi1QO153stG4UKx7wAtfOhfIUcM/45gM56G1bSu73nVkqiq97GzGsD8BV6In0gxJM8cXQtXboUV199NcrLy1FUVISFCxdi6dKlI9rWypUrUVVVhQULFkDTNFx77bVIJBLYtGlTmqN2njZnLnh5Bey9e50OJS1ETw9Ey34Ervxn1/ciOk1nAZSrUxwvbZFSoFe0o0ajm6mRKFI4Lgj5sNcSEC4uiUhJiSZL4KqC/JzXfCiqz9QRGaego9byZHmLlBIdtRYKaxRUn0k3YIRkkieS87q6OowfP37g9YQJE9Da2oqenp6j/s6SJUtw/fXX45vf/CZWr149aFsTJkwYeK0oCsaNG4c9e/ZkJHYnMc4RuPKfIVqaIXt7nQ5n1Oy9dVBnnAB14mSnQ/GEGm0uEqLD0UQgITshIFBFJS0jdnHIDwmJZhf3nu81bVSqHCf5qAb5aBjrGxya7JBItHkvOU+2SSTbJU78fBCMBvsSklGO15wfr5xEURQkk0mEQqGB7/X/O5FIDPp+vy996UsYO3YsNE3DmjVr8MMf/hAPPvggpk6dikQigXA4POj9oVAIiUTisO00NjaisbFx4LXP50NVVdWw/j6nKdNmwJg6Hfa+emhTp2dkH/292JnszRadHZAd7Qjd8rVBYwnI0dX452B14mkY6IGfFQz799PRrj2iHRXaVPjVw49TMjRhBbi8IIgXu3tRriqj7pnuT6wYZ0Aa8v1eIdBsC9xWUgjtKOOASJ+iMQomXezH7rdSCEQ5uJLecyY/+P8GZywdTTtA2BIde2xMvMiPSDX1mhOSaY6fSe+9996jlpREIhE8+eST8Pv96P1Ez2//vwOBIw9ImTZt2sC/zzrrLLz77rtYuXIlpk6dikAgMGhb/ds70raeeOIJLF68eOD1okWL8OCDDw79j3OJ4A03of7eO6AkElAjkYztR9czc9KWUqJ3314UnHMeSqdTD+xQRRFFWe9EJGUninyxEW9npO0qpUTC7MTcyOWIFkVHvH8CXFVUhNe37UGTZJjoT89MKLqWnuN1e08vZoYCmFdZlpbt5brTPl+Efav3oKcRiE3KzKw2WprPxQd2JqH5OE7/QhU0vyceuBPiaY4n59/73veO+56amhrU1tZixowZAIDa2lrEYrEj9pofCed84NF+TU0NXn311YGfCSGwe/dufOYznzns92655RZceeWVA699Ph/a29uHtE9XCRdCn3saEju3Q5txQtp7uBlj0HUdhmFkpITCPtAKOx6HcsGl3vz8HVTFZ2FLchkirHrYvzvadk2JOAzRi2JzMrVbGnw67MfTHT0oYxLaKI5hxhl0TYdhGqNeubLbFmg2LHylKERtPAwzPxPA+l/HEYgBqj9952POGDRdh2kYaRujYCUl2usMzP1KGPFEJ3D4Q2ZPiEapg4B4h+PJ+VBceOGFeP7553HqqafC7/fjj3/8Iy666MgrQ8bjcWzduhWzZs2CqqpYu3Yt3nnnnYEe8Hnz5uG3v/0t3nzzTZx11ln4+9//jkAggFmzZh22rcrKSlRWVg68bm1t9eysLr5Lr4Dx/e9AtLeBR4szsg8pZdqTcykl7L310M85Dygo8Ozn75RKdSbel3+HKZJQR7gA0EjbNS7aUMQr4QO1Wzqc6lPxssJRb1qYoI3i1H2w3kGK0R+vdaaFk3waxiiM2ngYymYriIxX0LHbRMm09NXp95eyiDSeizv2WCgap6BsjkJtTEiWeCI5v+SSS9DS0oJvfOMbsG0b5557LhYuXDjw8/vvvx8zZ87EwoULYds2nn76aezduxeMMVRWVuIb3/gGZs7sK4coLCzEokWL8Itf/AL/5//8H4wfPx733HNPztcxKyUx6GeeA3Pje2CRqGdmOxEt+yEtE/4LL3E6FE+K8DHwswL0iHYUKRX/f3t3Hh5Vfe8P/H3WmezbTDYhJGTDkLBDEUSQTS2KwgWsFrH2Klq0ffTWreAGKFq3ep+nUqmgFrkuXIRq69N73cDeWm9B630KsvlTfCwSQkImmWQy6znf3x+R0ZAEssxyMnm/nifPk7PMOZ+Z7zPnfOZ7Pud7Yrpvr+lGiT4ppvtMZLIk4fI0OzY1taFQFbDF+TvcZJhoMgVuSeNTentLkiRULU7CB79sRWqBadlhCf0tJtrqTYy9LnXAnDOIEoEkBuKYTnHS0NAQ7xD6xXS74V53H9SSUsg5fa9BPp0kSbDZbPD7/RHtORemieDe/4Nt2kwkzb0kYtsdbP7XswUnQv8P+Vrl2Vf+jv60qyFCOBLcjdmptyJXLe3Va6l7Qgj88mQLQgIo1fvWtxKJ76sQAp8GQhihq7gmkzf79tX/veDByc9CcI5UI5L8RvJYLIRA/f4QsktVjL1u4LexwxG5cx5RtFnz5zpFhZyeDn3qdBjHjg6IcXbN+jpAAPYLZsQ7lAGtUBsJr2iOaZt7RRNU6HAoxTHb52AgSRIuS0vCCcOEr5/14v3RZAq0mALfT2WveX9UXGZHoLV9iEKr8bkEAm6BysvYxkSxxuR8kLFfOBsiEIB50tpXAYRpwqg9BtuFsyHZ+Zjo/shXK2EgBJ9oidk+28wm5KkVkKXELheLhypdxTmqgq9D8an/FULg65CByUk6clS2b38k5ygomqrDfdSwVIeJEALuowaGTtGR7GAbE8Uak/NBRk5Ngz71Api1xyx1MjidWX8CgAT71OnxDmXA0+VkZClD4BVNMdmfEAI+0cIHD0WJJEm4NM2OE4YJfxy+w83f9JpflBKdYQAHm9K5dgTbrPVgIl+jQLBNoOxi9poTxQOT80HIPn02hN8H0RjfR7t3R5gmzOPHYJs+E5KdJ4dIKFSr4DVj03Megh8B0YY8tXc17tRzVbqKfFXG18HY955/HTIwwa6x1zxC7JkyiqbqaPnaGr3nQgi0HGvvNbdnMkUgigd+8wYhOS0N+vemwjhea4mTwenMkw0QhgHb+ew1j5R8tRI+4YYpop/MtZlNsEvpSJOdUd/XYCVJEi5Jbe89D8bwO9xito/QMpe15hE1fI4dQY+Aryn+x2N/s0CgVaB0DtuYKF6YnA9S9hmzIDytEO7meIfSgRACZl0t9CkXQE5Kjnc4CcOhlkCCBK9wR31fXuFGvjqCQ69F2WibhgxZwvFQJB/UfmbHQgaqbSry2WseUUlZMgonaGitjf844i3HDBSM15CUzfSAKF747Ruk5KxsaKPHwTheG+9QOhDNTRBtbbBPmxHvUBKKImnIUYrhNaP7Y0wIAZ/pRr5WEdX9UPu457NS7KgzjIg9DfJMfKbASUNgTgp7VKNh+Gx7+wgpntj92DpdwGPC5xIYPpttTBRPTM4HMduMWRCuRoi2tniHEmbUHYc2dgLkjMx4h5Jw8rUR8IvWqO4jCC9C8CNPZXIeC5OTdABAgxH9hK7WMFCoyhiusdc8GtIKFWSXq2itjV9y3lprIrtMQfo5bGOieGJyPoip5wyFUjwcRp01es9FWxuEqxG2C2bGO5SElKuURr3u3Gu6kSRlIkXOjto+6Fs2WcKUJB11UU7ODSFQHzJxYYqd5UpRVDLThrYGE0Yw9rXnZkigrcFEyUz2mhPFG5PzQc42fSbMhnqIUCjeocCoOw5lWAnUc4bEO5SE5FBLAMhRHe/cJ9zIU8ujtn3qbFqyDW5ToNWMXoLeYJhQJGC8XYvaPgjIHalCT5XQdiL2veeeEyb0FAm51X178iwRRQ6T80FOq6qBlJT8zbji8SOMEMyT9bCx1jxqFElDtjIUXjN6N4X6zFbkMjmPqVxVQZmmoC6KN4bWGyYmJ+nQ2GseVZIsoeh8HZ56M6YjaQkh0FZvYuj5OiSZbUwUb0zOBzlJUaCfd35773kch1U0Gxog6TZo1aPjFsNgkKuWR63uPCj8CMKLXLU0Ktun7p2fbMNJw4QRhe9wmynQbApMTeZDh2Jh6Hk2BNsEAi2xOx4HWtqHTxx6HtuYyAqYnBNs35sC0eaBaI3d491PZ56shzbpPEgqL6lGU67aXncuROR7WX2mG7qUjDQ5N+LbpjMbbdcgS0BjFGrPTxgGhqoyCjh8YkzYM2XkVKrwxLC0pa3eRE6FiqQspgREVsBvIkHOzIJaXhm30hazzQPhdsP2vfPisv/BxKkMhwkDfuGJ+LZ9wg2nMpw3DMaBJkkYZ9fQYES2t1UIgUbDxOQk9qjG0tDJOryNJswIt2dXhCnQdtLEkPP0qO+LiHqGyTkBAPRJU2A2noQwYv8QDLP+BJShw6A482K+78FGl9t7tqNxU6hfeJCrlkV8u9Qz30uywWVG9omhblPAL4BxSbwRNJbyRmmQJMDnin5y7m1s30f+aLYxkVUwOScAgDayBlAUmK7GmO5XCAGz8SR09prHTK5aBp8Z2eTcFCH4RCsc6vCIbpd6brimIFWWIjrmeYNholxXkSbzVBFLii4hb4wG78nod5Z4TxrIH6NB0XnFi8gqWODbC7quw2ZL3Mu7oQnfg+/gfmgFhb16nYT2g7qmahDoXU+P4WpEMBRC9pQLIKek9Oq11DdD5ZH4yvUJVFU9cwnKN8tUTQPO0hvbarRADsoYllkFRWIPXLxM9hn4u8eLIq37NjjV5pqmnfEmcCEEmnxBzMtJQ1paasRjpTMru0DF/3zcAEVSIas9S5x72ranmCEBryuACcsykJaW1K94iShymJz3QiAQQCAQiHcY0VMzBsZf/weBtjZIZzi5n06SJNgUG4KhYK9HfAnV1UGtOBce0wRa4ndD6mCSZhQiKLzwBluhSd0/cESSJCg2G0LBs7erx3AhUxmCtlYfAF+EI6aeGqkA74QMtAUC3Q57KEkSbDYbgmdp12bDhF8IVMBAC7+bMZc8VEDRJbjr/EjJ7dnNuD1t21M8JwwoGpA8LIiWlvg/6yKaErljjRIPr1VSmDq8DFJyCszGkzHZnzBNmE2N0MdPjMn+qF2q7IAupUS0tMVvepCrsN483oZrCpIkCa4IlLY0GiZKNQUpLGmJC1mRkD9GDdeER4O30UT+GA2ywpIWIivhUZfCJFmGNmZ8zOrORXMTIAS0c6tjsj9qJ0kSHEoJfCIyDyMSQsAr3HCy3jzuZElCjU2Fq59PCxVCoMkUGGPnCB7xVDBOh89lwgxFPkE3QwI+l0DBOLYxkdUwOacO9DHjIJqbIILBqO/LbDwJdcRISLzcGHO5alnEhlMMiDaYCDE5t4hRdg1NhoDZj1FbvAJoEwLVNlY+xlNOhQpFb+/hjjSvy4SsATmVbGMiq2FyTh0ow0piUtrSXtLigj5mXFT3Q11zqiXwiVaYov91pj7hRoqUDbucFoHIqL8qdQ0m2odB7CuXacKhyHDwwUNxJSsS8kZr8Lmi8NCwRhN5o1nSQmRFTM6pA0mWoY0aA7PJFdX9CHczYJrQRoyM6n6oa9lKESTI8InWfm/LJ1o4vrmF2GUJJZqCpn6UtrgNk73mFpE/RofXJSL6QCLTEPC6BArGsKSFyIqYnFMn2qix7aUtoejdvW82NkItq4Rk7360EIoeRdKQrQyB1+x/3bnfbIWTybmlVNu0PvecG9/Um1fZOCSmFThGqJBkwNcUueTc3yQgye3bJiLrYXJOnajDywBNj1rvuRACZrML2uixUdk+9UyuWg5/P3vOQ8KPALysN7eYETYVLaZAoA915y2mgASgTGfiZgWKJiF3pBbRunNvowlnlcoHDxFZFJNz6kRSFGgjayBcUUrOW1uAQABaFUdpiadctRQ+4YYQfT/pe003dCkJ6XJeBCOj/jpHbR9SsbkPQyo2mSaGaQr0Mz2gimIqf0x73bnox30Epwgh4Gsykc+SFiLLYnJOXdJqRsNsdkH0c0i2rpiNjVCKiiGn8gbCeHIqpTBh9GvUFp9ww6mUnvlJoxRzsiShXFfQ3IdkrtUUOJclLZbiHKlBGIDf3f/k3N8sYIaA3JG8MkJkVUzOqUtaxbmAEO03bkaYcDdBqxkT8e1S7+hyMtLlPHj7Md65T3iQq5ZHMCqKlEpdQ2svk/OQEHCbAhUsabEULUlCdrkK78n+d5Z4G01kl6nQknn6J7IqfjupS5LNBrWsAmZjZB9IJLxtEB4PtOpREd0u9U2eWgF/H58Uaogg/BypxbLKdRUeIeDvRd15iymgACjSOISi1RSM1eBrMiH6MX59uKRlLK+MEFkZk3PqllYzBqa7qV8ng9OZrkbIDicUhzNi26S+y1XL4RXuPrWxV7ihQEeWMiQKkVF/5asykiQJ7l7UnbtNE0WaAo1lSpaTN0pDyAcEWvt+PA56BEJeIK+GyTmRlTE5p25pVdWAzwfhicyTJAHAbG5mSYuF5KplMBBEQLT1+rVe0w2HWgJZYi+rFcmShOG60qshFVtZ0mJZtnQZGUVKv0Zt8Z40kT5UgT2Tp34iK+M3lLolp2dAPmcITFdkSltEMAjhboY2siYi26P+S5LTkSo74BW9v7fAL1qRr1ZGISqKlFOlLT1hflNvXsrk3LLyx2rw92O8c1+TQAFLWogsb8Achbds2YI//elPME0T06ZNw/Lly6GqncOvr6/HzTff3GGe3+/HvHnzsHz5cgDA/PnzYbPZwiNMVFVV4YEHHoj6exiI9Jox8H/wPoCifm/LdDVCSkqGUlTc721R5OSrlTgePAT0ogPcECE+GXQAKNFUtJo+hISAepZSFY8QMAEUawPmtDDo5I3ScOh1H4JeAS2pd6VHIa9AoFUgbxSTcyKrGxBH4bfeegt//vOf8eSTT8Jut2Pt2rXYunUrrr766k7rOp1ObN26NTzd2tqKa6+9Fueff36H9X71q19hyBDWyp6NNrIGvv/6I4TfD8lm69e2zOYmqFXVkGResLGSXLUcRwJ7IITo8ZCIPuGGAhU5yrAoR0f9UaQpkNFerpKpnLlt3YZAniIjSWa9uVWl5ilIdsjwNprQzuldOZm30URSjozUfJahEVndgEjO33nnHVxxxRXIy2t/0MmSJUvwzDPPdJmcn+7999+H0+lEVVVVtMNMSHJ+IaSMTJiuRij5BX3ejjBNiKYm6NWjIxgdRUKeWg4DAXwd2gtF+vbBJBIkyIYM0zQh0PFSepvZhFy1lPXmFqdJEgpVBS2mQOZZmsojBMpZ0mJ5+WM0HPsoAPQyOfc1mSgYz15zooFgQByJv/rqKxQXF4enS0pK0NDQAI/Hg5SUlDO+9t1338WsWbM6zb/nnntgGAbKy8vxox/9CEVF/S/bSESSJEGrHoXg/n39S86/GS9dLWeNstUkyRmYkLQYbWZTh/mSJMFut8Pn83U5mssQjcNhDgSluoK9vtBZ12s1TZRwCEXLyxul4Yt3/TCCAorWs6scRlDA1yyQN4pPBSUaCOKenBuGccbliqLA5/N1SMJP/e/1es+YnH/55Zf44osvsGrVqg7z161bh8rKSgSDQWzfvh333Xcf1q9fj+Tk5A7r1dbWora2Njxts9lQWFjY4/eWKGw1YxD48C+AaUJSOp+8T5VCnKkkwnS5oJWVQz3tMyZrODd5Zqd5iqIgPT0dbrf7rN9Tsq4Sm44PvQEA7d9R6ZuyFUmWgG8G/ggIAa8Ahtt1KF18x8k6ckplaEkSfC6B1LyOJYLyN8dgWZLw3TFd/E0mtCQJjjI93P5EZF1xT87vvfde7Nu3r8tlmZmZ2Lx5M+x2O9ravh3q7dT/SUlJZ9z2O++8g3HjxiEnJ6fD/OrqagCApmlYunQpdu7ciQMHDmD8+PEd1tuwYQNWr14dnl65ciUeeuihnr+5BCHGT0SrqkHxtEJz5na7nq533SsjhECwpRmZF30fGVlZ0QqToiQ9PT3eIVA/jE4O4HlXK4Smw658m8zp2rffV3cgCF2ScK7TEU7wyLqGTgii/rAPtqKu7wPSTjsWu9xeDB2Xguyc7FiER0T9FPfkfN26dWddp6ioCEeOHMG5554LADhy5AgcDscZe81DoRDef/993HTTTWfdfnc9vjfeeCPmz58fnrbZbHC5XGfdXiJSK0fAX3ccZnpGp2WSJEHXdQQCgS7LH0yPB2ZbG4IlpYP28xuI2HOeGHQhYJeARp8PDlWBJEvQNR2BYADimzHQXYEQhqgKmpua4hss9UhWpcCXHwbh80odesJlSYKm6wgGAjC/ORYLU8BzMoiKSwfv+QsAstgxRANI3JPznpg1axZ27NiBCRMmwG6349VXX8Xs2bPP+Jrdu3dDCIFJkyZ1mP/VV18hGAyiuLgYoVAIr732GgKBACorO9dCFxQUoKDg2zrrhoaGQZukqFU1CB4+CNM0u/0xI4ToOjl3nYSclw+kpQ/az28gMwyD7TbADVEVtJomcoQcLmUR5rffV48pMNqusJ0HiOwKBcIAfG4T9oxvr4acKmUxv3Ms9rlNmCEgu0Jm+xINEAMiOZ87dy7q6+tx2223wTAMXHDBBViyZEl4+QMPPICqqqoO8959911Mnz4dmtbx7vSmpib85je/QUNDA3RdR1lZGVavXo3U1NSYvZ+BSBtRBa/fD+FphZSa1qvXmm439Anfi1JkRHQ2wzQFn/iC3S5vEyaKOL75gKElScgsVuBzdUzOu+JzmcgsVqAlcwhbooFCEl11dVKXGhoa4h1CXLmffASSLEEZ0nFkG0mSYLPZ4Pf7O/Wci2AQwb/vQerNt0EtHh7LcKmfFEVBVlYWXC4Xe9wGuI+8AbzsbsMEu97p+xoQAnt8QaxypKFA5c2gA8Xnb/lwZKe/w0OFujoWn9gbxLDpNpRdZI9XqJbgcDjiHQJRj/GnNPWYVlUN0+3u1WvM5iZAt/GpoERxNERT4BfA4UAIXwRCONzmwxeBEI4EQjgUCEECkKvwdDCQOM5VEWgRMALd968ZQQG/W8B5Lq+KEA0k/MZSj2nnjoT/nf+CCAYhaT17mIVoboJWeS6fCkoUR7mKjClJOlq+uWdE11QEgu11yTkAilQFCkdpGVDSz1GgJknwNZtIcXZ9xcPfZEK1A+lDeEWEaCBhck49pgwdBtjsMJuboDicZ11fCAHT3QzbzLkxiI6IuiNLEq7OaH/GAMuVEoMkS8ipUNB63ERKN4djX7NAToXKsc2JBhh2Z1KPSbIMraISormpR+uLtjbA74dWMSK6gRERDUKOSg2B1u7LWgKtAjmVPbvKSUTWweScekWtrILZ0tLlkImnE81NkJ25kDMyox8YEdEgk12uIujpuu7cCAgEPQLZZbxATjTQMDmnXlHLKwGfF/D5zrqu2eKGWlkVg6iIiAaf1DwZqh3wuzsn5363gGID0gp4micaaPiTmnpFyXFAysiE6W6CkpTU7XrCNCHcbmjlFTGMjoho8JBkCZnFKpq+DMHwK4AEqGr7E7JbjxvIHMZ6c6KBiD+pqdfU8koId8sZ1xGeVsA0oAwvi1FURESDz5DzdCQ7ZJghATPYPnyiGQSSsmUMnaLHOzwi6gP2nFOvaWUVCO79PwghIHUz/JpwuyHn5UNOSo5xdEREg0fheB2F49uTcI7EQ5QY2HNOvaaUlAJ+P+Dvvu5ceFqhlrGkhYiIiKg3mJxTr8nZOZBS02C2dF3aIoRovxm0pDTGkRERERENbEzOqdckSYJSXALR4u56BZ8XCAahFg+PbWBEREREAxxrzqlP1OLh8O3fByM9A5IkIahpMILB9l7zxpOQUtM4vjkRERFRLzE5pz7RKs+F//33YB4/BkCCkCSYQgAQgCmgjZ8U7xCJiIiIBhwm59QnSsE5yLh/Xfv/HCGAiIiIKCJYc05EREREZBHsOe8FXddhs9niHYblnBrrPCUlBUJ0fow0DUxs18TEdk1cbFuixMDkvBcCgQACgUC8w7AcRVGg6zo8Hg/LWhII2zUxsV0TF9u2e+xYo4GEZS1ERERERBbB5JyIiIiIyCKYnBMRERERWQSTcyIiIiIii2ByTkRERERkEUzOiYiIiIgsgsk5EREREZFFMDknIiIiIrIIJudERERERBbB5JyIiIiIyCKYnBMRERERWQSTcyIiIiIii2ByTkRERERkEUzOiYiIiIgsQo13APHwj3/8A6+++io+//xz6LqOzZs3xzskIiIiIqLB2XNut9sxe/Zs/PjHP453KEREREREYYOy57yiogIVFRXYu3dvvEMhIiIiIgoblD3nRERERERWlHA954ZhnHG5oig93lZtbS1qa2vD0zabDYWFhX2OLVGd+kx789mS9bFdExPbNXGxbYkSQ8Il5/feey/27dvX5bLMzMxe3fy5YcMGrF69Ojy9cuVKPPTQQ/2OMVGlp6fHOwSKArZrYmK7Ji62LdHAlnDJ+bp16yK2rRtvvBHz588PT9tsNrhcrohtP1EoioL09HS43e6zXrmggYPtmpjYromLbdu9rKyseIdA1GMJl5z3hGmaCIVCCIVCAIBAIABJkqBpWof1CgoKUFBQEJ5uaGjgAe8MDMPg55OA2K6Jie2auNi2RAPboEzOP/30U6xatSo8vWjRIuTm5mLjxo1xjIqIiIiIBrtBmZzX1NTgjTfeiHcYREREREQdcChFIiIiIiKLYHJORERERGQRTM6JiIiIiCyCyTkRERERkUUwOSciIiIisggm50REREREFsHknIiIiIjIIpicExERERFZBJNzIiIiIiKLYHJORERERGQRTM6JiIiIiCyCyTkRERERkUVIQggR7yBoYKutrcWGDRtw4403oqCgIN7hUISwXRMT2zVxsW2JEgN7zqnfamtrsXr1atTW1sY7FIogtmtiYrsmLrYtUWJgck5EREREZBFMzomIiIiILILJOfVbQUEB7r//ftY4Jhi2a2JiuyYuti1RYuANoUREREREFsGecyIiIiIii2ByThFRV1eH+fPnIxAIxDsU6qVjx47htttuw5VXXolXXnkF69evx0svvdTt+suWLcPevXvD0y+99BJ++MMfYsmSJTAMIxYhExERJSw13gEQUWxdf/31+MlPfoLx48cDAF577TVUVlbiV7/6Va+3VV9fj23btuHZZ59FTk5OpEMdtI4dO4bHHnsMx44dw4IFC9DY2IjMzExcffXVXa6/bNky3HHHHaipqQHQ/oPpzTffRDAYxMsvvwxFUWIZfsJiuxBRLDA5Jxrk6urqMHXq1D699sSJE0hNTWVi3k/8wWRNbJfIq6urww033IBt27ZB1/V4h0NkSUzOqddOnjyJ5557Dnv37kUoFMLIkSNx/fXXAwB27dqFV155BcFgEHPmzMHSpUshy6yesoonn3wS9fX1ePjhhyHLMoYNG4bPPvsMBw4cwPPPP4+HHnoIb775JrKysnDttdcCAF5//XXs2LEDhmFg4cKF4W19/PHHePjhhxEMBrFkyRKMGTMGK1eujNdbSyhW+8EUCoWgqjxdsF2IKBb4raZeMQwDDz74ICoqKvDMM89A0zQcOnQovPzjjz/Gr3/9azQ3N+O+++5DYWEhZs+eHceI6bv+7d/+Dfv37+/QG7hy5UpMmzYNl1xySaf1P/nkE2zduhVr1qzB0KFDsXHjRrjdbgDA+PHjcf/99+Oxxx7D5s2bY/o+Ekk8fzAdPXoUmzZtwuHDhwEA06ZNw0033YS9e/fisccew8KFC/H73/8excXFeOCBB7Bz505s27YNjY2NGDZsGFasWIGioiLs3LkT7733HtauXQsAuP3226GqKh555BEAwD333IPZs2djxowZUfoUI4/t0s7j8eB3v/sd9uzZA6/Xi+LiYqxevRo2mw2bNm3CBx98AI/Hg4KCAlx//fWorq4GABw+fBjPPPMMvv76a+i6jqlTp+Kmm27CL37xCwDA0qVLAQB33XVX+FhERO2YnFOvfPbZZ6irq8Ojjz4KTdMAANXV1airqwMAXHXVVUhOTkZycjIuu+wyvP/++0zOB7A///nPmDlzJkpLSwG019D+93//d5yjSizx+sHk9Xpx33334eKLL8bdd98NAPj888/Dy91uN+rr6/Hb3/4WQgjs3r0b//Ef/4FVq1ahqKgIb7/9NtauXYv169ejpqYG69evRzAYRCgUQl1dHSRJgt/vhyzLOHjwIG677bZIfWQxwXZp99RTT0GSJDz11FNITU3FoUOHwldDS0tLsXjxYqSkpOCNN97AL3/5S2zcuBE2mw3PPvssLrvsMlx44YXw+Xz48ssvAQAPP/wwbrjhBmzZsoVlLUTdYL0B9UpDQwOcTmc4MT+d0+kM/5+bm4uTJ0/GKjSKgsbGxg5tmpqaiuTk5DhGRN/9waTrOpYtW4a+PK5iz549SE5OxpIlS2Cz2WCz2VBVVdVhnWXLlkHXddhsNvzpT3/CwoULUVJSAkVRcPHFF0OSJBw6dAgOhwNZWVnhnuURI0agoqICBw4cwOHDh5GTk5PwddaJ2C4ulwt/+9vfcPPNNyMjIwOKoqCqqip8/J8xYwbS09OhKAoWLFiAUCiEf/7znwAARVFQW1uL5uZm2O12jBgxotefBdFgxZ5z6hWHw4H6+vpuax3r6+uRkpIS/j/RT8iJLjs7G/X19eFpj8eDtra2OEZEjY2NKCkpCU/39QdTfX39GZ8kmZaWBpvNFp4+ceIEXnjhBbz44ovhecFgMPwDvLq6Gv/4xz/g9/vDpQ179+6Fpmnh0UoSWSK2y6njeUZGRpfLd+zYgbfffhuNjY2QJAltbW3hqwU/+9nP8NJLL+Hmm2+G0+nEkiVLcN555/XsQyAa5JicU6+Ul5fD6XRi06ZNuOaaa6DrOg4ePBjuXX3llVfws5/9DG63G3/4wx+waNGiOEdMp8vMzMTx48d7tO60adPwxBNP4MILL8SQIUOwefNmSJIU5QjpTCL1g8npdGLXrl3dLj+9nR0OBxYuXIhZs2Z1uX5NTQ3effdd+P1+/OQnP4EQAhs2bICmaZgzZ06v4xtoErFdnE4nPB4P3G430tPTOyz79NNPsW3bNjz44IMYNmwYZFnGVVddFb5aUFhYiNtvvx2maWL37t149NFHefwg6iGWtVCvKIqCe++9F42Njbjhhhtw7bXX4g9/+EN4+bhx43DLLbfg9ttvx9SpU1lvbkGLFi3C9u3bcdVVV2HLli1nXHfcuHFYtGgR1qxZg3/9139Ffn5+p5M09V9vfzC99957+OKLLxAIBPqc8EycOBEtLS3Ytm0b/H4//H4/9u/f3+36l1xyCbZt24YjR45ACAGv14vdu3eHE9CamhocPHgQx48fR3FxMUpKSlBbW4tDhw4N2J7zwd4uWVlZmDRpEtavXw+32w3DMHDgwAEEg0F4vV7Isoz09HQYhoGtW7fC6/WGX7tz5040NzdDlmUkJSVBCAFZlpGRkQFZlnv8uRINRuw5p15zOBzhO+6/64033gAAzJ07N9YhUS9MnjwZkydP7nb5rbfe2mF6wYIFWLBgQYfpU2pqajhSSwQsWrQIzz77LLZs2YJ58+adcd3v/mA6NSpIX34wJSUlYc2aNdi4cSNee+01KIqC6dOnd6pvPmXy5Mnw+/146qmnUFdXF66FPlUq4XA4kJ2dHe5FBYARI0bgq6++GrDlbWyX9uPBc889h5/+9KcIBAIoKSnB/fffj7Fjx2LixIlYsWIF7HY75s+fD4fDEX7dJ598gueeew5+vx8OhwM///nPw2U+S5YswapVqxAKhXDHHXdg3Lhxvf6ciBKZJPpyxwoREREREUUcy1qIiIiIiCyCZS1ERAlq165dWL9+faf5iqLg5ZdfjkNEBFirXerr63HzzTd3uWzNmjUcApEoDljWQkRERERkESxrISIiIiKyCCbnREREREQWweSciIiIiMgimJwTEREREVkEk3MiIiIiIotgck5ElrZr1y5IkoSPPvooavt4/PHHe/So9UsvvRQzZswIT7/wwguQJAkNDQ1Ri42IiAYXJudERH00b948fPjhh8jMzIx3KERElCD4ECIioj5yOp1wOp3xDoOIiBIIe86JKCI+/PBDzJw5EykpKcjIyMDVV1+NEydOAAC+/PJLSJKEzZs344YbbkBmZiacTicef/xxAMArr7yCyspKpKenY+HChWhqauq0/RMnTmDhwoVISUlBQUEB1q1b12mdAwcO4PLLL0dGRgZSUlIwb948fP755x3WcbvdWLZsGdLS0uB0OnHnnXciFAp1ua3p06fDbrejtLQUmzdv7rTO6WUtp97nli1bcMsttyArKwsFBQW4/fbbO+1jx44dqKyshN1ux6RJk/DRRx8hNTUVDzzwQI8+byIiSkxMzomo3z788EPMmDEDGRkZePXVV/Hb3/4We/bswfz58zusd8899yA9PR3/+Z//iSVLluCOO+7A3XffjX//93/Ho48+iqeffhrvvfce7rzzzk77WL58OUpLS7F9+3YsXboUq1atwjPPPBNe/sUXX2DKlClobGzECy+8gJdeegn19fWYNWsW/H5/eL0f//jH2LFjBx555BH87ne/w6effopf//rXHfbl8/kwd+5c1NXV4cUXX8QjjzyChx56CH//+9979HmsWrUKsixj69atuPHGG/HEE09g48aN4eWffPIJFi9ejKqqKmzfvh3XXXcdfvCDHyAYDPZo+0RElMAEEVE/XXDBBWLKlCnCNM3wvH379glJksSbb74pjhw5IgCIK6+8Mrw8FAqJvLw8kZKSIhoaGsLzf/7zn4vMzMzw9M6dOwUAcc0113TY59KlS8WQIUOEYRhCCCGWLVsmSkpKhNfrDa9z4sQJkZKSIp5++mkhhBD79+8XkiSJTZs2hdcJBoOiqKhIfPdw+Jvf/EbIsiwOHz4cnnfw4EEhSZKYPn16eN7zzz8vAIj6+nohhAi/z8WLF3eIderUqWLWrFnh6cWLF4uysrJw7N/d1v3339/VR0xERIMEe86JqF/a2trwwQcfYPHixTAMA6FQCKFQCJWVlSgoKMCePXvC686ePTv8v6IoGD58OMaMGYOcnJzw/IqKCjQ1NaG1tbXDfhYsWNBheuHChTh69CiOHj0KAHjrrbdw+eWXQ1XVcAxZWVkYPXp0OIbdu3dDCNFhW6qq4vLLL++w7b/97W+orq5GeXl5eF5lZSWqq6t79JnMnTu3w3RVVVU4TgDYs2cPLr30Usjyt4fg02MgIqLBick5EfWLy+WCYRi47bbboGlah79jx47hn//8Z3jd00c10XW9y3lAe2nJd+Xm5nY5XVtbCwBoaGjAU0891SmGv/71r+EYamtroWkasrKyOmwrLy+vw3RtbW2n/XW1Xne6ek/ffT+1tbWdbiTNysqCpmk92j4RESUujtZCRP2SmZkJSZKwcuVKXHHFFZ2WOxyOiOzn1M2lp08XFBQAALKzszFv3jysWLGi02vT0tLC6waDQbhcrg4Jel1dXYf1CwoKuqwvr6urQ3Z2dv/eyDfbr6+v7zDP5XKx5pyIiNhzTkT9k5KSgvPOOw8HDhzAhAkTOv0VFxdHZD87duzoML19+3YUFhZiyJAhANpLZvbt24exY8d2iqGyshIAMHHiREiS1GFboVAIr7/+eodtT5o0Cfv27cNnn30Wnnfo0CHs27cvIu9l4sSJ+OMf/wjTNMPzfv/730dk20RENLCx55yI+u2xxx7DzJkzceWVV+IHP/gBsrKycPToUbz99tu47rrrIpKgv/fee7jjjjswZ84cvP3229iyZQuefvrpcN326tWrMXHiRFx00UVYvnw58vLycPz4cbz//vuYNm0arrrqKlRVVeGKK67ArbfeCp/Ph+LiYjz99NMwDKPDvn70ox/hwQcfxGWXXYYHH3wQQgjce++9yM/P7/f7AIBf/OIXmDhxIv7lX/4Fy5cvx5dffoknnngCuq53qEMnIqLBh2cBIuq3KVOm4C9/+QtaW1tx3XXX4fvf/z7WrFmD5ORklJWVRWQfGzZswKFDh7BgwQK8+OKLWLt2bYcSlrKyMuzevRs5OTlYsWIFLrroItx9993weDwYNWpUeL3nnnsO8+fPx5133olly5ZhxIgRuOWWWzrsKykpCW+99RZyc3Pxwx/+EHfddRfuuusujBs3LiLvZezYsdi6dSv279+PBQsWYNOmTXjhhRdgGAYyMjIisg8iIhqYJCGEiHcQRESD3TvvvIM5c+Zg165dmD59erzDISKiOGFyTkQUBytWrMCsWbOQk5ODTz/9FGvXrkVhYSE++ugjlrYQEQ1irDknIoqDpqYm/PSnP0VDQwMyMjJw8cUX4/HHH2diTkQ0yLHnnIiIiIjIIthFQ0RERERkEUzOiYiIiIgsgsk5EREREZFFMDknIiIiIrIIJudERERERBbB5JyIiIiIyCKYnBMRERERWQSTcyIiIiIii/j/AJXDYgMfrVoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tfidf_corr['embedding'] = 'tfidf'\n",
    "tfidf_crew_corr['embedding'] = 'tfidf_crew'\n",
    "tfidf_crew_cast_corr['embedding'] = 'tfidf_crew_cast'\n",
    "cb_corr['embedding'] = 'cb'\n",
    "\n",
    "cols = ['embedding', 'corr']\n",
    "dfs = [tfidf_corr[cols], tfidf_crew_corr[cols], tfidf_crew_cast_corr[cols], cb_corr[cols]]\n",
    "corr = pd.concat(dfs, ignore_index=True).dropna()\n",
    "corr['embedding'] = corr['embedding'].astype('category')\n",
    "\n",
    "g = ggplot(corr, aes(x='embedding', y='corr', color='embedding', fill='embedding')) +\\\n",
    "    geom_violin(alpha=0.6) +\\\n",
    "    scale_y_continuous(breaks=[-1.0, -0.75, -0.5, -0.25, 0, 0.25, 0.5, 0.75, 1.0], limits=[-1, 1])\n",
    "g = g.draw()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can observe that text embedding performed a little bit better, but there is a large overlapping between the distributions. So, there is no strong evidence of difference between methods."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='cb_corr', ylabel='tfidf_corr'>"
      ]
     },
     "execution_count": 62,
     "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": [
    "tfidf_cb_corr = tfidf_corr[['corr']].rename(columns={'corr':'tfidf_corr'}).join(\n",
    "    cb_corr[['corr']].rename(columns={'corr':'cb_corr'}), how='inner'\n",
    ")\n",
    "tfidf_cb_corr.plot.scatter(x='cb_corr', y='tfidf_corr')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In addition, we can observe that there is no strong correlation between text embedding and collaborative filtering embedding. It may be a indicative that both methods are complementary. So, combining than can provide a even better performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='tfidf_corr', ylabel='tfidf_crew_corr'>"
      ]
     },
     "execution_count": 63,
     "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": [
    "content_corr = tfidf_corr[['corr']].rename(columns={'corr':'tfidf_corr'}).join(\n",
    "    tfidf_crew_corr[['corr']].rename(columns={'corr':'tfidf_crew_corr'}), how='inner'\n",
    ")\n",
    "content_corr.plot.scatter(x='tfidf_corr', y='tfidf_crew_corr')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If we compare content based methods, we observe a stronger correlation. Let's check it numerically:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.2214131380567743, 1.2560464682861095e-05)"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from scipy.stats import pearsonr\n",
    "\n",
    "pearsonr(tfidf_cb_corr['tfidf_corr'], tfidf_cb_corr['cb_corr'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.7216452817938481, 1.1361322483897958e-62)"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pearsonr(content_corr['tfidf_corr'], content_corr['tfidf_crew_corr'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Precision and Recall at K\n",
    "\n",
    "Another way to check ranking performance is to transform the problem into a classical binary classification problem, so that we can take advantage of better known metrics. Precision and recall are the most common ones.\n",
    "\n",
    "First of all we must label some items as 'relevant'. Than, we look for relavant items in the top K positions of the ranking. K in [5,10] are common choices. Here we will choose k=10.\n",
    "\n",
    "Precision is given by the number of relevant items in the top K divided by K. In other words, it measures the proportion of the top K items that are relevant. Recall is given by the number of relevant items in the top K divided by the total number of relevant items. In other words, it measures the proportion of relevant items that are in the top K items.\n",
    "\n",
    "### Defining Relevant Items"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "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>id_left</th>\n",
       "      <th>id_right</th>\n",
       "      <th>similarity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>364</td>\n",
       "      <td>0.335812</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>480</td>\n",
       "      <td>0.317939</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>367</td>\n",
       "      <td>0.314605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2</td>\n",
       "      <td>588</td>\n",
       "      <td>0.298155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2</td>\n",
       "      <td>589</td>\n",
       "      <td>0.263034</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   id_left  id_right  similarity\n",
       "0        2       364    0.335812\n",
       "1        2       480    0.317939\n",
       "2        2       367    0.314605\n",
       "3        2       588    0.298155\n",
       "4        2       589    0.263034"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "movie_sim = pd.read_csv('output/movie_similarity.csv', usecols=['id_left','id_right','similarity'])\n",
    "movie_sim.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:ylabel='Frequency'>"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "movie_sim['similarity'].plot.hist(bins=30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can define a threshold for the user correlation similarity to label items as relevant or not."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:ylabel='Frequency'>"
      ]
     },
     "execution_count": 27,
     "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": [
    "movie_sim.groupby('id_left').agg(\n",
    "    similarity=('similarity', lambda x: x.quantile(0.5))\n",
    ").plot.hist(bins=30)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:ylabel='Frequency'>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "movie_sim.groupby('id_left').agg(\n",
    "    similarity=('similarity', lambda x: x.quantile(0.75))\n",
    ").plot.hist(bins=30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we will user 0.3 as the threshold, so that movies with user correlation greater than or equal 0.3 will be considered relevant similars to the target movie."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "movie_sim['relevant'] = np.where(movie_sim['similarity'] >= 0.3, 1.0, 0.0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:ylabel='Frequency'>"
      ]
     },
     "execution_count": 30,
     "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": [
    "movie_sim.groupby('id_left').agg(\n",
    "    num_relevants=('relevant', 'sum')\n",
    ").plot.hist(bins=30)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(12480, 5)\n",
      "(5280, 2)\n"
     ]
    }
   ],
   "source": [
    "movie_sim['num_relevants'] = movie_sim.groupby('id_left')['relevant'].transform('sum')\n",
    "print(movie_sim.shape)\n",
    "\n",
    "movie_sim = movie_sim[movie_sim['num_relevants'] >= 3].drop('similarity', axis=1)\n",
    "ids = movie_sim['id_left'].unique()\n",
    "movie_sim.set_index(['id_left','id_right'], inplace=True)\n",
    "print(movie_sim.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To make a better evaluation we will consider only those target movies with 3 or more relevant similars.\n",
    "\n",
    "### Text Embedding Performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0    5141\n",
       "1.0      79\n",
       "Name: relevant, dtype: int64"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_sim = pd.read_csv('output/movie_similarity_tfidf.csv')\n",
    "tfidf_sim = tfidf_sim[tfidf_sim['id_left'].isin(ids)]\n",
    "tfidf_sim.set_index(['id_left','id_right'], inplace=True)\n",
    "tfidf_sim = tfidf_sim.join(movie_sim, how='left')\n",
    "tfidf_sim.fillna(0.0, inplace=True)\n",
    "tfidf_sim.reset_index(drop=False, inplace=True)\n",
    "\n",
    "tfidf_sim['relevant'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    174.000000\n",
       "mean       0.016667\n",
       "std        0.057985\n",
       "min        0.000000\n",
       "25%        0.000000\n",
       "50%        0.000000\n",
       "75%        0.000000\n",
       "max        0.400000\n",
       "Name: precision, dtype: float64"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_pr = precision_recall(tfidf_sim, 'similarity', 'relevant', 'id_left')\n",
    "tfidf_pr['precision'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    174.000000\n",
       "mean       0.061658\n",
       "std        0.191476\n",
       "min        0.000000\n",
       "25%        0.000000\n",
       "50%        0.000000\n",
       "75%        0.000000\n",
       "max        1.000000\n",
       "Name: recall, dtype: float64"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_pr['recall'].describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The results are very bad! In general, the relevant movies are not appearing in the top 10 given by the embedding ranking.\n",
    "\n",
    "### Collaborative Filtering Performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0    5197\n",
       "1.0      23\n",
       "Name: relevant, dtype: int64"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cb_sim = pd.read_csv('output/movie_similarity_cb.csv')\n",
    "cb_sim = cb_sim[cb_sim['id_left'].isin(ids)]\n",
    "cb_sim.set_index(['id_left','id_right'], inplace=True)\n",
    "cb_sim = cb_sim.join(movie_sim, how='left')\n",
    "cb_sim.fillna(0.0, inplace=True)\n",
    "cb_sim.reset_index(drop=False, inplace=True)\n",
    "\n",
    "cb_sim['relevant'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    174.000000\n",
       "mean       0.007471\n",
       "std        0.030439\n",
       "min        0.000000\n",
       "25%        0.000000\n",
       "50%        0.000000\n",
       "75%        0.000000\n",
       "max        0.200000\n",
       "Name: precision, dtype: float64"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cb_pr = precision_recall(cb_sim, 'similarity', 'relevant', 'id_left')\n",
    "cb_pr['precision'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    174.000000\n",
       "mean       0.048851\n",
       "std        0.199571\n",
       "min        0.000000\n",
       "25%        0.000000\n",
       "50%        0.000000\n",
       "75%        0.000000\n",
       "max        1.000000\n",
       "Name: recall, dtype: float64"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cb_pr['recall'].describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The same bad results for the Collaborative Filtering.\n",
    "\n",
    "But it is not a problem, since the user correlation similarity is just a reference for us. In the previous tests we saw interesting recommendations given by embedding similarity. We had to choose a method to compare with and we tried a reference that provides good recommendations, but is sparse (you need users in common to link a pair of movies) and very expensive to compute. Maybe a method based on the most popular movies (always recommend the most popular movies) could provide a better reference.\n",
    "\n",
    "Anyway, in the recommendation dashboard we will see good results for the embedding methods."
   ]
  }
 ],
 "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": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}