neighborhoodbasedCollabFiltering.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Neighborhood Based Collaborative Filtering Recommendation System Example:\n",
    "## Here's an Example implementation in python3  of a Neighborhood Based Collaborative Filtering system. \n",
    "## The dataset can be downloaded here: https://www.kaggle.com/CooperUnion/anime-recommendations-database \n",
    "\n",
    "## We are predicting the best recommendations for around 7 million users. Get ready for some big data. \n",
    "\n",
    "### If you have any questions, feel free to email me at trackait@oberlin.edu"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "import scipy as sp\n",
    "import sklearn.preprocessing as pp\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Importing above, we need to clean the data set. Our first goal is to set up a User-Item matrix, so we can use the algorithm described in the article above. Here's the pseudocode for what we are looking at implementing:\n",
    "1. Set up a user/item matrix, where the rows are the users and the columns are the items.  This is a sparse matrix, with many missing entries.\n",
    "2. Define the mean user ratings for every user.\n",
    "3. Mean Center user Ratings\n",
    "4. Compute the similarity value between the rows(users)\n",
    "5. Store the top-k similar users to users u who have specified ratings for item i.\n",
    "6. With the similarity function, the neighborhoods you predict the missing ratings by: adding the mean value to  (the sum of the similarity of the user times the mean centered value divided by the sum of the absolute value of the similarity value of every user in neighborhood. \n",
    "7. Predict ratings\n",
    "\n",
    "So lets load the dataset and get to work:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "anime = pd.read_csv('./datasets/anime.csv')\n",
    "rating = pd.read_csv('./datasets/rating.csv')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "anime shape:  (12294, 7)\n",
      "raings shape:  (7813737, 3)\n",
      "thats a lot of items: Lets reduce this as we have a lot of users: roughly, 7.8 million.\n"
     ]
    }
   ],
   "source": [
    "print('anime shape: ',anime.shape) \n",
    "print('raings shape: ',rating.shape) \n",
    "\n",
    "print('thats a lot of items: Lets reduce this as we have a lot of users: roughly, 7.8 million.')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3787, 7)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "anime_id      0\n",
       "name          0\n",
       "rating      116\n",
       "dtype: int64"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(anime[anime['type']==\"TV\"].shape) \n",
    "animeTV = anime[anime['type']==\"TV\"]\n",
    "animeTV  = animeTV.drop(['genre','type','episodes','members'], axis=1)\n",
    "(animeTV.notnull()==False).sum() # no nans now"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With one dataframe cleaned and ready to be used, lets look at the ratings matrix."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "unique users:  73515\n",
      "unique anime:  11200\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(6785, 5)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print('unique users: ', len(rating.user_id.unique()))\n",
    "print('unique anime: ', len(rating.anime_id.unique()))# this is a problem, as we only want TV anime. We'll have to sort this out, with merge.\n",
    "#lets sample 100k users to easy computation\n",
    "rating = rating.sample(10000)\n",
    "ratingDF = pd.merge(animeTV,rating, on=['anime_id','anime_id'])\n",
    "ratingDF.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "matching anime got sorted out:  1420\n",
      "NAN size: user:  0 titles:  0\n",
      "now we have to assemble the ratings matrix, with user ratings as rows and movies as columns\n",
      "to remove memory error, lets just sample 100k rows\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>anime_id</th>\n",
       "      <th>1</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>15</th>\n",
       "      <th>16</th>\n",
       "      <th>17</th>\n",
       "      <th>18</th>\n",
       "      <th>19</th>\n",
       "      <th>20</th>\n",
       "      <th>...</th>\n",
       "      <th>32681</th>\n",
       "      <th>32696</th>\n",
       "      <th>32729</th>\n",
       "      <th>32828</th>\n",
       "      <th>32935</th>\n",
       "      <th>32947</th>\n",
       "      <th>32998</th>\n",
       "      <th>33028</th>\n",
       "      <th>33222</th>\n",
       "      <th>33421</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>user_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>8.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>120</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>153</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>160</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>163</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>170</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>173</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>202</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>221</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>226</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>228</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>230</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>232</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>235</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>246</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>248</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>261</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>321</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73221</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73226</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73229</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73234</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73242</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73248</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73255</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73262</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73263</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73272</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73273</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73284</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73286</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73301</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73324</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73329</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73356</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73359</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73371</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73373</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73376</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73389</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73400</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73422</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73424</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73457</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73462</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73467</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73478</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73495</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6058 rows × 1420 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "anime_id  1      6      7      8      15     16     17     18     19     \\\n",
       "user_id                                                                   \n",
       "4           NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "5           NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "7           NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "17          NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "30          NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "33          NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "35          NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "36          NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "80          NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "102         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "120         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "123         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "130         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "153         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "160         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "163         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "167         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "170         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "173         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "202         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "221         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "226         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "228         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "230         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "232         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "235         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "246         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "248         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "261         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "321         NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "...         ...    ...    ...    ...    ...    ...    ...    ...    ...   \n",
       "73221       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73226       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73229       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73234       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73242       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73248       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73255       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73262       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73263       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73272       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73273       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73284       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73286       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73301       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73324       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73329       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73356       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73359       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73371       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73373       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73376       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73389       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73400       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73422       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73424       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73457       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73462       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73467       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73478       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73495       NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "\n",
       "anime_id  20     ...  32681  32696  32729  32828  32935  32947  32998  33028  \\\n",
       "user_id          ...                                                           \n",
       "4           NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "5           NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "7           NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "17          NaN  ...    8.0    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "30          NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "33          NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "35          NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "36          NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "80          NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "102         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "120         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "123         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "130         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "153         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "160         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "163         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "167         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "170         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "173         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "202         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "221         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "226         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "228         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "230         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "232         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "235         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "246         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "248         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "261         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "321         NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "...         ...  ...    ...    ...    ...    ...    ...    ...    ...    ...   \n",
       "73221       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73226       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73229       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73234       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73242       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73248       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73255       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73262       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73263       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73272       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73273       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73284       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73286       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73301       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73324       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73329       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73356       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73359       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73371       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73373       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73376       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73389       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73400       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73422       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73424       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73457       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73462       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73467       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73478       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "73495       NaN  ...    NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "\n",
       "anime_id  33222  33421  \n",
       "user_id                 \n",
       "4           NaN    NaN  \n",
       "5           NaN    NaN  \n",
       "7           NaN    NaN  \n",
       "17          NaN    NaN  \n",
       "30          NaN    NaN  \n",
       "33          NaN    NaN  \n",
       "35          NaN    NaN  \n",
       "36          NaN    NaN  \n",
       "80          NaN    NaN  \n",
       "102         NaN    NaN  \n",
       "120         NaN    NaN  \n",
       "123         NaN    NaN  \n",
       "130         NaN    NaN  \n",
       "153         NaN    NaN  \n",
       "160         NaN    NaN  \n",
       "163         NaN    NaN  \n",
       "167         NaN    NaN  \n",
       "170         NaN    NaN  \n",
       "173         NaN    NaN  \n",
       "202         NaN    NaN  \n",
       "221         NaN    NaN  \n",
       "226         NaN    NaN  \n",
       "228         NaN    NaN  \n",
       "230         NaN    NaN  \n",
       "232         NaN    NaN  \n",
       "235         NaN    NaN  \n",
       "246         NaN    NaN  \n",
       "248         NaN    NaN  \n",
       "261         NaN    NaN  \n",
       "321         NaN    NaN  \n",
       "...         ...    ...  \n",
       "73221       NaN    NaN  \n",
       "73226       NaN    NaN  \n",
       "73229       NaN    NaN  \n",
       "73234       NaN    NaN  \n",
       "73242       NaN    NaN  \n",
       "73248       NaN    NaN  \n",
       "73255       NaN    NaN  \n",
       "73262       NaN    NaN  \n",
       "73263       NaN    NaN  \n",
       "73272       NaN    NaN  \n",
       "73273       NaN    NaN  \n",
       "73284       NaN    NaN  \n",
       "73286       NaN    NaN  \n",
       "73301       NaN    NaN  \n",
       "73324       NaN    NaN  \n",
       "73329       NaN    NaN  \n",
       "73356       NaN    NaN  \n",
       "73359       NaN    NaN  \n",
       "73371       NaN    NaN  \n",
       "73373       NaN    NaN  \n",
       "73376       NaN    NaN  \n",
       "73389       NaN    NaN  \n",
       "73400       NaN    NaN  \n",
       "73422       NaN    NaN  \n",
       "73424       NaN    NaN  \n",
       "73457       NaN    NaN  \n",
       "73462       NaN    NaN  \n",
       "73467       NaN    NaN  \n",
       "73478       NaN    NaN  \n",
       "73495       NaN    NaN  \n",
       "\n",
       "[6058 rows x 1420 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print('matching anime got sorted out: ',len(ratingDF['name'].unique()))\n",
    "\n",
    "print('NAN size: user: ',(ratingDF.user_id.notnull()==False).sum(), 'titles: ',(ratingDF.anime_id.notnull()==False).sum()\n",
    "     )\n",
    "print('now we have to assemble the ratings matrix, with user ratings as rows and movies as columns')\n",
    "print('to remove memory error, lets just sample 100k rows')\n",
    "ratingDF.rating_y.replace({-1:np.nan},regex=True,inplace=True)\n",
    "ratingMatrix = ratingDF.pivot_table(values='rating_y',index='user_id',columns='anime_id')\n",
    "ratingMatrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We now have our ratings matrix. Its quite sparse. From this, we can get two really useful pieces of information besides from what we already have-- the mean rating for each user, and the L2 score for each user. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "#ratingMatrix['mean']\n",
    "#means = ratingMatrix.mean(axis=1)# we cant calculate L2 if mean is a column, so we save it to a temporary series\n",
    "\n",
    "#ratingMatrix['L2'] = ratingMatrix.apply(lambda x : np.sqrt(np.sum(np.power(x,2))),axis=1)\n",
    "#ratingMatrix['mean']=mean\n",
    "#ratingMatrix['L2']\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#sanity check\n",
    "#lets get rid of the nans now. \n",
    "ratingMatrix.fillna(0,inplace=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "# now we normalize it with l2\n",
    "    \n",
    "meanCenteredRatingMatrix = pp.normalize(sp.sparse.coo_matrix(ratingMatrix.to_numpy()).tocsr())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "#lets see how affected our matrix is:\n",
    "#meanCenteredRatingMatrix = meanCenteredRatingMatrix.T\n",
    "#meanCenteredRatingMatrix = meanCenteredRatingMatrix.loc[:, (meanCenteredRatingMatrix != 0).any(axis=0)]\n",
    "#Optional: save as csv so you dont have to rerun the ENTIRE kernel.\n",
    "#meanCenteredRatingMatrix.to_csv('animeMatrix.csv')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With our user Matrix mean centered, now lets find the cosine similiarity for each item. To do this, we have to convert to a scipy sparse matrix format, so we can take advantage of its sparse matrix computation speed ups (and plotting shortcuts on matplotlib)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.lines.Line2D at 0x7f51b3472588>"
      ]
     },
     "execution_count": 11,
     "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": [
    "#meanCenteredRatingMatrix = pd.read_csv('animeMatrix.csv')\n",
    "plt.spy(meanCenteredRatingMatrix,precision=0.01, marker='o', markersize=1,aspect='auto',markeredgecolor='black',rasterized=True)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "That sparse matrix is not as bad as it usually has to be! Sampling and Mean centering does wonders. Lets compute the cosine similarity."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "itemSimilarityMatrix = cosine_similarity(meanCenteredRatingMatrix.T,dense_output=True)\n",
    "userSimilarityMatrix = cosine_similarity(meanCenteredRatingMatrix,dense_output=True)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The Easiest thing to do now is to get the similarity matricies into pandas dataframes so we can create a prediction function. Now we have item-item correlations, and user-user correlations. We can recommend items to a specific user, and an anime playing website can look at the similiarity of users. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "#print(userSimilarityMatrix.shape,ratingMatrix.columns.shape)\n",
    "\n",
    "\n",
    "itemSimilarityDF= pd.DataFrame(itemSimilarityMatrix,index=ratingMatrix.columns, columns=ratingMatrix.columns) \n",
    "userSimilarityDF=pd.DataFrame(userSimilarityMatrix,index=ratingMatrix.index,columns=ratingMatrix.index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1., 0., 0., ..., 0., 0., 0.],\n",
       "       [0., 1., 0., ..., 0., 0., 0.],\n",
       "       [0., 0., 1., ..., 0., 0., 0.],\n",
       "       ...,\n",
       "       [0., 0., 0., ..., 1., 0., 0.],\n",
       "       [0., 0., 0., ..., 0., 1., 0.],\n",
       "       [0., 0., 0., ..., 0., 0., 1.]])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def animeID_to_name(animeID): # quick function to get from anime to name\n",
    "    return (animeTV[animeTV['anime_id']==animeID].name).values[0]\n",
    "\n",
    "animeID_to_name(itemSimilarityDF.index[9])\n",
    "itemSimilarityMatrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From here, you can get the top users and top movies. But first, lets look at our prediction error. For this, we have to make a prediction function.\n",
    "to make a prediction, we are going to create a peer group for every user that contains the set of similiar users to the user we are predicting who have also rated the item the user has predicted. With this peer group, we can develop \"neighborhood based collaborative filtering\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([226, 228, 230])"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "user_id=userSimilarityDF.index[20:25]        \n",
    "usersRatedItems=ratingMatrix.loc[user_id,:]\n",
    "#usersRatedItems = usersRatedItems.sort_values(by=usersRatedItems.columns.values,ascending=False).index[1:]\n",
    "usersRatedItems = usersRatedItems[usersRatedItems.apply(lambda x: x > 0)]\n",
    "usersRatedItems.dropna(axis=1,inplace=True,how='all')\n",
    "usersRatedItems.dropna(axis=0,inplace=True,how='all')\n",
    "meanUser = usersRatedItems.mean(axis=1,skipna=True)\n",
    "usersRatedItems.fillna(0,inplace=True)\n",
    "pd.Series(np.subtract(usersRatedItems.T.sum() , meanUser)).index.values\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "#def recitems(user_id, topk): #find the top k best prediction for what the user wants given the user's similiarly related users:\n",
    "\n",
    "\n",
    "#predict()   \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def findPeerUsers(user_id,anime_id,topk=50,debug=False):# remember user ids are the columns.\n",
    "    if debug==True:\n",
    "        user_id=userSimilarityDF.index[20]\n",
    "    usersRatedItems=ratingMatrix.loc[user_id,:]\n",
    "    usersRatedItems = usersRatedItems[usersRatedItems>0].index\n",
    "    if debug==True:\n",
    "        anime_id=usersRatedItems[0]\n",
    "            # we find the closest 50 users\n",
    "    closestUsers =  userSimilarityDF.sort_values(by=user_id,ascending=False).index[1:]#we cut out 0 since its always going to be the user\n",
    "    peerGroup=pd.DataFrame(columns=usersRatedItems)\n",
    "    for peer in closestUsers:\n",
    "        \n",
    "        peersRatings = ratingMatrix.loc[peer,usersRatedItems]\n",
    "        peersItems = peersRatings[peersRatings>0].index\n",
    "        if len(peersItems) > 0:\n",
    "            x=pd.DataFrame(peersRatings,peersItems).T\n",
    "            peerGroup = pd.concat([peerGroup,x],axis=0)\n",
    "    peerGroup.fillna(0,inplace=True)\n",
    "    peerGroup.index.name='peer_id'\n",
    "    return peerGroup\n",
    "\n",
    "def findReviewersOfItem(anime_id,debug=False):#returns a vector of all of the users who have rated an anime:\n",
    "    if debug==True:\n",
    "        anime_id=itemSimilarityDF.index[20]\n",
    "    itemReviews = ratingMatrix.loc[:,anime_id]\n",
    "    itemReviews = itemReviews[itemReviews>0]\n",
    "    return itemReviews\n",
    "\n",
    "def peerSim(user_id,peerGroup,debug=False):\n",
    "    if debug==True:\n",
    "        user_id=userSimilarityDF.index[20]\n",
    "   # peerID = peerGroup.index\n",
    "    sim=userSimilarityDF.loc[user_id,peerGroup]\n",
    "    return sim\n",
    "def meanCenteredRatingPerUser(user_id,debug=False):# returns vector of actual ratings- mean ratings for each item that the user rated.\n",
    "    if debug==True:\n",
    "     user_id=userSimilarityDF.index[20:25]        \n",
    "    usersRatedItems=ratingMatrix.loc[user_id,:]\n",
    "    #usersRatedItems = usersRatedItems.sort_values(by=usersRatedItems.columns.values,ascending=False).index[1:]\n",
    "    usersRatedItems = usersRatedItems[usersRatedItems.apply(lambda x: x > 0)]\n",
    "    usersRatedItems.dropna(axis=1,inplace=True,how='all')\n",
    "    usersRatedItems.dropna(axis=0,inplace=True,how='all')\n",
    "    meanUser = usersRatedItems.mean(axis=1,skipna=True)\n",
    "    usersRatedItems.fillna(0,inplace=True)\n",
    "    return np.subtract(usersRatedItems.T.sum() , meanUser)\n",
    "        \n",
    "def predict(user_id,anime_id,debug=False):\n",
    "    if debug==True:\n",
    "        user_id=userSimilarityDF.index[20]\n",
    "    peerUsers = findReviewersOfItem(anime_id).index.values\n",
    "    \n",
    "    meanPeerRatings = meanCenteredRatingPerUser(peerUsers)\n",
    "    meanPeerRatings = meanPeerRatings[meanPeerRatings>0]\n",
    "    sim = peerSim(user_id,pd.Series(meanPeerRatings).index.values)\n",
    "    denom= np.sum(np.abs(sim))\n",
    "    if denom < 0.0001:\n",
    "        denom+=1\n",
    "    userMean =ratingMatrix.loc[user_id,:].mean()\n",
    "    #print(denom)\n",
    "    return np.add(userMean,np.sum(np.dot(sim,meanPeerRatings)) /denom)\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "actual:\n",
      "  0.0\n",
      "0.004929577464788733\n"
     ]
    }
   ],
   "source": [
    "x=ratingMatrix.index[33]\n",
    "items=ratingMatrix.loc[x,:]\n",
    "actual = items[np.isnan(items) == False]\n",
    "d = predict(x,actual.index[0])\n",
    "\n",
    "print('actual:\\n ',actual.values[0])\n",
    "print(d)\n",
    "#list(filter(np.isnan(items),items))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Excellent! Now we can look at how to calculate error lost to judge our accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate(user_id):\n",
    "    usersRatedItems=ratingMatrix.iloc[user_id,:]\n",
    "    watchedAnime = usersRatedItems[usersRatedItems>0].index\n",
    "    #print(watchedAnime)\n",
    "    error = []\n",
    "    if watchedAnime.shape[0]==0: \n",
    "         error.append(np.nan)\n",
    "    for i in watchedAnime:\n",
    "        actual = ratingMatrix.loc[user_id,i]\n",
    "        if np.isnan(actual):\n",
    "            #print(er)\n",
    "            continue\n",
    "            \n",
    "        elif not np.isnan(actual):\n",
    "            predicted = predict(user_id,i)\n",
    "            if np.isnan(predicted):\n",
    "                continue\n",
    "            else:\n",
    "                er = np.power(np.subtract(actual,predicted),2) \n",
    "                error.append(er)\n",
    "    \n",
    "            #print(er)\n",
    "    return np.mean(error)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "nan"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "z=ratingMatrix.iloc[4,:]\n",
    "z = z[z>0]\n",
    "evaluate(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[1.2398333663955564e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 4.959333465582226e-05,\n",
       " 4.017060107121603e-05,\n",
       " 4.959333465582226e-07,\n",
       " nan,\n",
       " 4.959333465582226e-05,\n",
       " nan,\n",
       " 4.017060107121603e-05,\n",
       " 2.430073398135291e-05,\n",
       " nan,\n",
       " nan,\n",
       " 4.017060107121603e-05,\n",
       " 2.430073398135291e-05,\n",
       " 1.7853600476096008e-05,\n",
       " 1.7853600476096008e-05,\n",
       " 0.00014332473715532632,\n",
       " 3.1739734179726246e-05,\n",
       " 4.959333465582226e-07,\n",
       " 2.430073398135291e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 4.959333465582226e-07,\n",
       " 2.430073398135291e-05,\n",
       " 2.430073398135291e-05,\n",
       " nan,\n",
       " 3.1739734179726246e-05,\n",
       " 2.430073398135291e-05,\n",
       " nan,\n",
       " 4.959333465582226e-07,\n",
       " 0.00012695893671890498,\n",
       " nan,\n",
       " 4.959333465582226e-05,\n",
       " 4.017060107121603e-05,\n",
       " 4.959333465582226e-07,\n",
       " 3.1739734179726246e-05,\n",
       " 2.430073398135291e-05,\n",
       " 1.7853600476096008e-05,\n",
       " 2.430073398135291e-05,\n",
       " 1.2398333663955564e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 4.017060107121603e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 2.430073398135291e-05,\n",
       " 2.430073398135291e-05,\n",
       " 1.7853600476096008e-05,\n",
       " 1.2398333663955564e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 7.934933544931562e-06,\n",
       " 4.017060107121603e-05,\n",
       " 4.959333465582226e-05,\n",
       " 3.1739734179726246e-05,\n",
       " nan,\n",
       " 3.1739734179726246e-05,\n",
       " 2.430073398135291e-05,\n",
       " 1.2398333663955564e-05,\n",
       " 2.430073398135291e-05,\n",
       " 1.7853600476096008e-05,\n",
       " 4.959333465582226e-07,\n",
       " 4.959333465582226e-05,\n",
       " 4.959333465582226e-07,\n",
       " 2.430073398135291e-05,\n",
       " 3.1739734179726246e-05,\n",
       " nan,\n",
       " 1.7853600476096008e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 4.959333465582226e-07,\n",
       " 3.1739734179726246e-05,\n",
       " nan,\n",
       " 4.017060107121603e-05,\n",
       " 2.430073398135291e-05,\n",
       " 3.1739734179726246e-05,\n",
       " nan,\n",
       " 4.959333465582226e-07,\n",
       " 1.7853600476096008e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 1.7853600476096008e-05,\n",
       " 2.430073398135291e-05,\n",
       " 4.959333465582226e-07,\n",
       " 4.017060107121603e-05,\n",
       " 3.1739734179726246e-05,\n",
       " 2.430073398135291e-05,\n",
       " 4.959333465582226e-07,\n",
       " 1.7853600476096008e-05,\n",
       " 2.430073398135291e-05,\n",
       " 4.017060107121603e-05,\n",
       " 3.1739734179726246e-05,\n",
       " nan,\n",
       " 4.017060107121603e-05,\n",
       " 4.017060107121603e-05,\n",
       " 0.00017903193810751833,\n",
       " 2.430073398135291e-05,\n",
       " 4.959333465582226e-05,\n",
       " 2.430073398135291e-05,\n",
       " 2.430073398135291e-05,\n",
       " 4.017060107121603e-05,\n",
       " 3.1739734179726246e-05]"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x=ratingMatrix.index[1:100]\n",
    "error = []\n",
    "for i in x:\n",
    "    error.append(evaluate(i))\n",
    "error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f519aa6a3c8>]"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZcAAAD8CAYAAAC7IukgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJztvXt4XOV56Pt7Z0ajuyVbkm35gm1sY2MwEDAkJGkSLuGaE6cppLDbU5om5Zy9Yaen7T4p7O6m56TleTbt2aWnLWnLSUjSNA1JSNp4JzSEAmlSU4xtDJiLDcIXLNnGkizJus/tPX+stUZLo7lrJM1o3t/z6NHMt9b61lqzLu/3Xj9RVQzDMAyjlAQW+gAMwzCMxYcJF8MwDKPkmHAxDMMwSo4JF8MwDKPkmHAxDMMwSo4JF8MwDKPkmHAxDMMwSo4JF8MwDKPkmHAxDMMwSk5ooQ9goWhvb9f169cv9GEYhmFUDPv37+9T1Y581q1a4bJ+/Xr27du30IdhGIZRMYjI8XzXNbOYYRiGUXJMuBiGYRglx4SLYRiGUXJMuBiGYRglx4SLYRiGUXJMuBiGYRglx4SLYRiGUXJMuBiGYZQxiYTynb0niMUTC30oBWHCxTAMo4w5cGKQz3/vFZ4/cnahD6Ug8hIuInKTiBwWkS4RuS/N8loR+ba7fI+IrPctu99tPywiN+bqU0TuddtURNp97f+niLzk/r0qInERWeYuOyYiB91llnZvGMaiYTIad/7H4gt8JIWRU7iISBB4GLgZ2AbcKSLbUlb7DDCgqpuAh4AH3W23AXcAFwE3AV8SkWCOPncD1wPTygyo6p+q6mWqehlwP/CvquoX5de4y3fkf/qGYRjlTTShAMTc/5VCPprLVUCXqh5R1QjwGLAzZZ2dwNfdz48D14mIuO2Pqeqkqh4Futz+MvapqgdU9ViOY7oT+FYex24YhlHReL6WWHzxCZfVwAnf9263Le06qhoDhoC2LNvm02daRKQBRwv6nq9ZgZ+IyH4RuTuffgzDMCqBaNzTXCrLoV+JVZH/F2B3iknsg6raIyLLgadE5JCq/ix1Q1fw3A1w3nnnzc/RGoZhzAJPqCxGzaUHWOv7vsZtS7uOiISAFqA/y7b59JmJO0gxialqj/v/DPCPOGa3GajqI6q6Q1V3dHTkNSWBYRjGghKrUM0lH+GyF9gsIhtEJIzzct+Vss4u4C73823AM6qqbvsdbjTZBmAz8EKefc5ARFqADwM/8LU1ikiz9xm4AXg1j/MyDMMoeyKuzyVaYZpLTrOYqsZE5F7gSSAIPKqqr4nIF4F9qroL+ArwDRHpAs7iCAvc9b4DvA7EgHtUNQ5OyHFqn27754DPAyuBV0TkCVX9rHs4vwj8RFVHfYe4AvhHJ36AEPAPqvrj4n8SwzCM8sHTXOIVFi0mjoJRfezYsUNtJkrDMMqdv/v3Y3zhB6/x3269kM/+wvkLeiwisj/fdA/L0DcMwyhjohWquZhwMQzDKGOSeS4mXAzDMIxS4QmVqBWuNAzDMEpFJOYIFTOLGYZhGCXDy2+ptFBkEy6GYRhlTDKJ0sxihmEYRqmYqi1mmothGIZRIpK1xRZh+RfDMAxjgUhqLuZzMQzDMEpF1PJcDMMwjFIzNVmYmcUMwzCMEuFNcxw1zcUwDMMoFZ7GEjefi2EYhlEqFvNkYYZhGMYCETGHvmEYhlFqYhaKbBiGYZSaqdpiZhYzDMMwSoRNFmYYhmGUnKTmshiFi4jcJCKHRaRLRO5Ls7xWRL7tLt8jIut9y+532w+LyI25+hSRe902FZF2X/tHRGRIRF5y/76Q7/EZhmFUKpVaFTmUawURCQIPAx8FuoG9IrJLVV/3rfYZYEBVN4nIHcCDwC+LyDbgDuAiYBXwLyJygbtNpj53Az8EfprmcH6uqh8r4vgMwzAqEi9abDGaxa4CulT1iKpGgMeAnSnr7AS+7n5+HLhORMRtf0xVJ1X1KNDl9pexT1U9oKrHCjiHfI7PMAyjIvE0l8Xo0F8NnPB973bb0q6jqjFgCGjLsm0+fabjahF5WUT+WUQuKuD4ABCRu0Vkn4js6+3tzWN3hmEYC0vM8lzmnBeBdap6KfCXwD8V2oGqPqKqO1R1R0dHR8kP0DAMo9R4jvzFmOfSA6z1fV/jtqVdR0RCQAvQn2XbfPqchqqeU9UR9/MTQI3r8C+4L8MwjEphSnNZfGaxvcBmEdkgImEcB/2ulHV2AXe5n28DnlFVddvvcKPJNgCbgRfy7HMaIrLS9eMgIle5x95fTF+GYRiVQqVm6OeMFlPVmIjcCzwJBIFHVfU1EfkisE9VdwFfAb4hIl3AWZwXPO563wFeB2LAPaoaByfkOLVPt/1zwOeBlcArIvKEqn4WR2j9RxGJAePAHa4AS3t8Jfl1DMMwFphKrS0mzvu5+tixY4fu27dvoQ/DMAwjKxv/6xPEE0pjOMhrX7xpQY9FRPar6o581q0kh75hGEZVoarJ/JZK01xMuFQ5Q+NRrv+zf+WNU+cW+lAMw0gh6vOzmHAxKoqTg+N0nRnh8OnhhT4UwzBS8CLE6muCxBNKJbkxTLhUOV7WbyRWWWGOhlENRGOOMKkPB4HK0l5MuFQ5ntodqbDSEoZRDURdzaUu5LyqKykc2YRLlWOai2GUL54wqXM1l2gFJVKacKlyPOFSaUXxDKMa8J7L+hpHuMRNczEqBW9kZJqLYZQfno/FEy6muRgVQ8Q0F8MoW7y6YkmHvmkuRqXgCZVJEy6GUXZ4g786zyxm0WJGpZD0ucQq56Y1jGoh6dD3zGIVNAg04VLlRCt0ljvDqAamkiidV7VpLkbFYKHIhlG+RGdoLiZcjAohGjOHvmGUK55ZzIsWq6QJw0y4VDneSMgc+oZRfnihx1b+xag4vJs3amYxwyg7vOfSM4tZKLJRMXhRYlZbzDDKj9QkylgFPacmXKocK/9iGOVLavmXRWcWE5GbROSwiHSJyH1plteKyLfd5XtEZL1v2f1u+2ERuTFXnyJyr9umItLua/8VEXlFRA6KyHMicqlv2TG3/SURsbmLC2DKLFY5N61hVAueGazWDUVeVA59EQkCDwM3A9uAO0VkW8pqnwEGVHUT8BDwoLvtNuAO4CLgJuBLIhLM0edu4HrgeMo+jgIfVtXtwB8Bj6Qsv0ZVL8t3fmfDwRMqleLQ/8IPXuU3vrZ3oQ/DMOYF/2RhsPhCka8CulT1iKpGgMeAnSnr7AS+7n5+HLhORMRtf0xVJ1X1KNDl9pexT1U9oKrHUg9CVZ9T1QH36/PAmgLO08jAVIZ+ZQiXk4MTnB6aWOjDMIx5IeIKk4ZwCFh8SZSrgRO+791uW9p1VDUGDAFtWbbNp89sfAb4Z993BX4iIvtF5O4C+ql6kkmUFaK5ROMJakLmKjSqg1iytphzz1eSbzS00AdQKCJyDY5w+aCv+YOq2iMiy4GnROSQqv4szbZ3A3cDnHfeefNyvOVOpZV/icYT1ARkoQ/DMOaF1Npiiy0UuQdY6/u+xm1Lu46IhIAWoD/Ltvn0OQMRuQT4MrBTVfu9dlXtcf+fAf4Rx+w2A1V9RFV3qOqOjo6OXLurCiqt/EssrtQETXMxqoPUJMrFZhbbC2wWkQ0iEsZx0O9KWWcXcJf7+TbgGVVVt/0ON5psA7AZeCHPPqchIucB3wf+V1V909feKCLN3mfgBuDVPM7LoPJCkSNmFjOqiBlVkSsoWiynWUxVYyJyL/AkEAQeVdXXROSLwD5V3QV8BfiGiHQBZ3GEBe563wFeB2LAPaoaByfkOLVPt/1zwOeBlcArIvKEqn4W+AKOH+dLTqwAMTcybAXwj25bCPgHVf1xCX6bqiA5n0uFaC5mFjOqiVg8gQjUhiqvKnJePhdVfQJ4IqXtC77PE8DtGbZ9AHggnz7d9r8A/iJN+2eBz6ZpPwJcmtpu5EdF+lzMLGZUCZG4UhMIUBPwHPqVI1zsKa1ypsxilXHTxuJqZjGjaojFE4SCQjAoye+Vgj2lVY4nXOIJrQiVO2JmMaOKiCWUUEAIuff8oiv/Yixe/BpLJZjGzCxmVBPe/e7d84stFNlYxPgFSiU49R2zmGkuRnUQiyuhoOAp64uqtpixuPELl0rQXCLxBKGA3bZGdRBNOJqLiFATFDOLGZWD3yxWCYmU0XiCsDn0jSoh6ksaDgUC5tA3KoeoG0fvfS53nAx9M4sZ1UEsnkg680MBqZioTjDhUvVE4wka3Yqr5S5cEgl1o2fstjWqg2hcCXmaS1AqIqLTw57SKicaVxrcukXl7tD3Sl+YWcyoFmKJRFJTDwUD5tA3KodoPEFjrae5lPeoyAvDNLOYUS3E4jrNLGahyEbFEI0nkppLuTv0PbOd5bkY1ULEl9cVsmgxo5KIxrVifC7ehGYhEy5GlRDzCZeaQKDsn1E/9pRWMQm35EtDbWVoLp5JIGxmMaNKiCWcJEqAYMAc+kaF4DnIPc2l3Kc6NrOYUW1E41PRkaFgoOz9on7sKa1ivBu10dVcyl3ljppZzKgyHLOYo7k4Gfrl/Yz6sae0iom6ZrAGT3Mpc7NY1MxiRpXhmMWc17SZxYyKwdMEKk1zMbOYUS1EYj7NxRz6RqUQdUdBlaO5mFnMqC5iiURyFsqg5bkYlYJnFmv08lzK/MaNWhKlUWV4Jfdhkea5iMhNInJYRLpE5L40y2tF5Nvu8j0ist637H63/bCI3JirTxG5121TEWn3tYuI/IW77BURudy37C4Recv9u6vwn6E6mTKLVZbmEjbNxagS/JPj1Sy28i8iEgQeBm4GtgF3isi2lNU+Awyo6ibgIeBBd9ttwB3ARcBNwJdEJJijz93A9cDxlH3cDGx2/+4G/trdxzLgD4H3AlcBfygiS/P9AaoZTxNoqJAkSjOLGdWGN80xLE6z2FVAl6oeUdUI8BiwM2WdncDX3c+PA9eJiLjtj6nqpKoeBbrc/jL2qaoHVPVYmuPYCfydOjwPtIpIJ3Aj8JSqnlXVAeApHEFm5MB7WdfVBAgGpOw1l0jMzGJGdRHzVUVejJOFrQZO+L53u21p11HVGDAEtGXZNp8+8z2OYvoymB59VROUstdcPJOAmcWMakBVicQTydB7myysjBGRu0Vkn4js6+3tXejDWXAiPuESDgYqJkPfzGJGNeDltCTncwksPs2lB1jr+77GbUu7joiEgBagP8u2+fSZ73Hk3ZeqPqKqO1R1R0dHR47dLX78JezDoUDZm8WiZhYzqohYUrj4osUWmc9lL7BZRDaISBjHQb8rZZ1dgBeldRvwjKqq236HG022AccZ/0KefaayC/g1N2rsfcCQqp4CngRuEJGlriP/BrfNyMF0s1j5J2hFzSxmVBHJ59NXW6ySosVCuVZQ1ZiI3Ivzwg4Cj6rqayLyRWCfqu4CvgJ8Q0S6gLM4wgJ3ve8ArwMx4B5VjYMTcpzap9v+OeDzwErgFRF5QlU/CzwB3IITFDAGfNrdx1kR+SMcgQXwRVU9O9sfphrwC5fK0FzMLGZUD56WktRcKswsllO4AKjqEzgvd3/bF3yfJ4DbM2z7APBAPn267X8B/EWadgXuybCPR4FHs56EMYOIzyxWUwEVVy2J0qgmUn2MjkO/vJ9RPzYErGJiKQ79yXLXXBJWW8yoHrzyTGFfVeRyN137sae0ikmaxUIBakIV4HNJOvTttjUWP97gLxSwqshGheE3i9VWgkM/niAgzkNmGIudaKrPJRggllAcD0H5Y8Klion5olFqQuWfoR9NJExrMaqGWIoZ2CsDUylOfXtSq5hpZrFK0FxiasLFqBqS0WKBqTwXoGJMY/akVjH+6KuKcOj7pnw1jMVOJGVyPC/fpdwHgR4mXKoYzwzmmMXKX3OJmVnMqCKmKmhMOfT97eWOPalVTCyRIBgQAgHHoV/utcUiZhYzqohktJgvFBnM52JUANG4Ts3PHQwkQ33LFTOLGdWEl+dS44sWAyqmBIwJlyomEpsyM4VD5a+5mFnMqCbS5bk47eU9CPSwJ7WK8b+sHc2lvIWLmcWMaiI1z8XMYkbF4IT2ujduSMpeczGzmFFNROOpeS6uWazMn1MPEy5VjPOydm4Bz6Ffztm/ZhYzqglLojQqFmcK1SmzmGp5J2hZEqVRTURnJFF6mkv5PqN+7EmtYmJxTdpzwyHnVihn01gknkger2EsdlLzXLx7P2rRYka54zeLef/LORw5lkjYLJRG1eCZxfyThUF5Wxf82JNaxUTi00ORASbj8YU8pKyYWcyoJvwVNGDKoV/ulTQ87EmtYqI+n4v3v5xno4yaWcyoIjzH/VTJfctzMSqEdD6Xcs518QcgGMZiJzYjFHkRmsVE5CYROSwiXSJyX5rltSLybXf5HhFZ71t2v9t+WERuzNWniGxw++hy+wy77Q+JyEvu35siMujbJu5btqu4n6L6SOdzKWeHfixuZjGjevBXLXf+LzKzmIgEgYeBm4FtwJ0isi1ltc8AA6q6CXgIeNDddhtwB3ARcBPwJREJ5ujzQeAht68Bt29U9bdV9TJVvQz4S+D7vv2Pe8tU9eMF/wpVSsT3svZu4HKeMMzMYkY14RWWFXHu+eAi1FyuArpU9YiqRoDHgJ0p6+wEvu5+fhy4TpxfZCfwmKpOqupRoMvtL22f7jbXun3g9vmJNMd0J/CtfE/SSE/Ml/FeKaHIprkY1UIsrklTGEwNAKOLSLisBk74vne7bWnXUdUYMAS0Zdk2U3sbMOj2kXZfIrIO2AA842uuE5F9IvK8iKQTRt62d7vr7evt7c18xlWC3yyWdOiXseYSi2tSCBrGYid1MGXlX+aeO4DHVdUfM7tOVXcA/wH4cxHZmG5DVX1EVXeo6o6Ojo75ONayJuozi1WC5hKNJ6aN5AxjMeMPuAFfVeRFpLn0AGt939e4bWnXEZEQ0AL0Z9k2U3s/0Or2kWlfd5BiElPVHvf/EeCnwHvyOK+qJxJPEA5VhrNQVYklzKFvVA+ptfRqFmH5l73AZjeKK4zzck+NyNoF3OV+vg14Rp0KiLuAO9xosg3AZuCFTH262zzr9oHb5w+8nYjIVmAp8O++tqUiUut+bgc+ALye7w9QzcTiiaSqndRcytQs5kXOmFnMqBaicaXGp6kn81wqpPxLKNcKqhoTkXuBJ4Eg8KiqviYiXwT2qeou4CvAN0SkCziLIyxw1/sOzss+BtzjmbPS9enu8veAx0Tkj4EDbt8ed+AECPhF94XA34pIAkdY/ndVNeGSB9Fp0WKeWaw8R0XR5MRJZhYzqoNYPJEsVgm+qshl+oymklO4AKjqE8ATKW1f8H2eAG7PsO0DwAP59Om2H8GJJkvX1/+Vpu05YHvWEzDSEoknqHHNYuXu0E+d28IwFjvRxHSfi01zbFQM08q/lLlDP5lQZmYxo0qIxhLJumIwpbmUc4kmP/akVinxhKI6Fd6YjKEvW+HiHFfYkiiNKiGWqrkswiRKYxGSNDOFUpIoy9wsFgrYLWtUB9GUPJdkKHKZDgBTsSe1SokkNYHKqC1mZjGj2nBq6U1pLiJCTVAWVYa+sQiJpUyh6gmZctdczCxmVAuxRGKGph4MiJnFjPJmyizm3AKBgBAKSNn7XMwsZlQLkZQMfXAmDivXZzQVe1KrlOQsdykZwOUaiWJmMaPaiKUp1BoMmuZilDnRFJ8LOE79cjeL1ZhZzKgSUqsig6O5l+sAMBUTLlVK6hSq4Ggu5evQtyRKo7qIJhIzNPWaoFi0mFHepDOL1Zax5hJLzspnt6xRHcRSaouBOfSNCiCdWawmWL4O/YiZxYwqI7W2GLh+URMuRjnjmcVqKs7nYresUR1EUvJcwEkdMLOYUdZ4BSpTfS7lqrmYcDGqjUx5LotpsjBjERJJ87IOhwJlXHLf07TMLGZUB6kzUYLzvJrmYpQ1ycm3UvJcIrF4pk0WFNNcjGrDX7XcIxQ0zcUoc2IphSvBETTlGkMfTRPdZhiLmdSqyOD5XMrzGU3FntQqJZKmnEo5O/SnAhDMLGYsfhIJJZ7QGT6XUCBgk4UZ5U16s1glhCLbLWssfqKJ9KH3oaCUrXUhlbyeVBG5SUQOi0iXiNyXZnmtiHzbXb5HRNb7lt3vth8WkRtz9SkiG9w+utw+w277r4tIr4i85P591rfNXSLylvt3V3E/RXWROp8LQDgULFvNJRqzJEqjekhWLU/1uSymJEoRCQIPAzcD24A7RWRbymqfAQZUdRPwEPCgu+024A7gIuAm4EsiEszR54PAQ25fA27fHt9W1cvcvy+7+1gG/CHwXuAq4A9FZGmBv0PVEUtjFqsJStmWf4klEgRkasIkw1jMpE6J4REq43SBVPIZBl4FdKnqEVWNAI8BO1PW2Ql83f38OHCdiIjb/piqTqrqUaDL7S9tn+4217p94Pb5iRzHdyPwlKqeVdUB4CkcQWZkIZLGLFbO5V8iabKVDWOxMmUWW8SaC7AaOOH73u22pV1HVWPAENCWZdtM7W3AoNtHun39koi8IiKPi8jaAo5vXpmIxhkajy7kIeQknVmsrJMoYzojLNPPRDTO/uNn5/GIFo5EQukbmZy3/cXiCYbGyvt+XmxkqqUXCgamhSIPjkXKNu+lkoaC/xNYr6qX4GgnX8+x/gxE5G4R2Sci+3p7e0t+gB5/9MPX+eSXdqNaviOMWBoHeTnP5xJLJLJGij38bBef+tvnq+Il+Ojuo3z4T55lLBLLvXIJ+Mbzx7n2f/y0rO/nxUZycrwZk4VNBd1EYgk+9CfP8s0978z78eVDPsKlB1jr+77GbUu7joiEgBagP8u2mdr7gVa3j2n7UtV+VfWGa18Grijg+HD7eERVd6jqjo6OjiynPDte6R7i7d5R3nx3ZM72MVsiaWy65RyKHE0zcZKfa7YuJ55QfvrmmXk8qoXhiYOnGI3EOdI7Oi/7e+fsGP2jEcYi5ZlguxjJNH+Rvyryu+cmODcR4/C7w/N+fPmQj3DZC2x2o7jCOA76XSnr7AK8KK3bgGfUGebsAu5wo8k2AJuBFzL16W7zrNsHbp8/ABCRTt/+Pg684X5+ErhBRJa6jvwb3LYFQVV5u9cRKk8fenehDiMnzstacNxcDt58LuU4Qo3ENKtwuXRNK22NYZ45tLiFS//IJAdODAIk77O5ZngiNu2/Mfck51tKzXPxWRdODU0AcNr9X27kFC6u/+NenBf2G8B3VPU1EfmiiHzcXe0rQJuIdAG/A9znbvsa8B3gdeDHwD2qGs/Up9vX7wG/4/bV5vYN8DkReU1EXgY+B/y6u4+zwB/hCKy9wBfdtgXh1NBEcoT3zBvl+6KLxmZqArXuxETlaBrLZRYLBoSPbFnOTw/3lq0NuhT89HAvnuzvOjNfwiU67b8x92TSXGqCkkyiPDU0DsDJwfH5Pbg8CeVeBVT1CeCJlLYv+D5PALdn2PYB4IF8+nTbj+BEk6W23w/cn2EfjwKPZj2JecIbTb7v/GW8cPQsZ0cjLGsML/BRzSSWmKkJeDdyNJ4gXGZz1ecyiwFcd+FyvvdiNy++M8hVG5bN05HNL88cOsOKJbU0hEPzprmMTLqay6RpLvPFVCjyzKrI8VTN5VyFai5GYbztjibv/tD5JBR+erg8tZdIfKYm4EVjlaPfJRLTnKHIv7C5nVBAePqN8jVHzoZILMG/vtnLtVuXs7GjibfPzI/Pxcxi84+nncyc5jiQDFP2zGGDY1HGy9AfZsKlxLzdO8qSuhAfuWA5Hc21PF2mPoB0ZrGapFms/IRLLJEgnKOuWHNdDe89f1nZ/uazZe+xs4xMxrh26wo2Lm/kaN/ovOQ8jLhCZcSEy7wR8SpSpCZR+gpX+s1hnomsnDDhUmK6zoywcXkTgYBw7Zbl/Oxwb1m+rNOZmbzv5Ziln49ZDODarSvoOjPC8f75GdXPJ0+/cYbaUIAPbmpnY0cTkXiCE2fH5ny/55Kai/lc5gtPc0lX/iWWUFSV0+cmqK8JAuXp1DfhUmLe7h1hY0cTANdeuJzhyRh7j5Vfcl80TTlvz6FfjmaxaGzm8abj+guXAyy6qDFV5elD7/L+jW3Uh4NsWu7cY/Phd5ly6JvmMl9M1RabWf4FIJ5QTg1NcOnaFgBOmnBZ3JybiHJmeDL54H9wUzvhUICnyzBqLBqbORGRpxmUY7RYNJGf5rKurZGNHY1l+ZvPhrd7RzneP8a1F64AYGP7/AiXSCzBpDvYMIf+/JGMFpsRiuwIm/FonL6RSS5b65RRPG1mscWNl9TmaS6NtSGuPr+t6FH06GSMX/3yHr63vzvvbb788yN85mt7c66XzsxUzg79dLPyZeK6C1ew52h/MsppLnju7T6+8INX5y0n6Fn3Hrp2q6OZtTTU0N5UO+dOff9vuBBmsR+81MPOh3cvSD2trjMj/JfvvsxEdPbO8r9//jjfeP541nV++MpJbv+b57j9b57jwR8fAtJoLq4PpmdwHFXY0N7A0oaaZORYOr63v5sv//zILM+gcEy4lBAv72BjR2Oy7dI1LUU7XhvCQQ72DLH/nYG8tzk3EePZw2dyPhBOKHJKDH2ojH0ueZrFAC5Z00I0rnQPzJ0/4q9/+jZ/9+/Hee3kuTnbh5+3zgyzYkktq1vrk20bOxrnXHPxC5SFMItF48rLJwY5tgA+tJ+92cvj+7v56eHZl4p65GdH+NMfH8o6cAuKUBMMUBMMsGJJHbdsX8mG9sZp63ihySfOOppKZ0s9nS31WYXLo7uP8qdPHmZ0njXPvPJcjPx4u3eEmqCwdllDsq21wclxOTceZWmB+S4iwpaVzRw6lf8LbOvKZhLqCLqLV7dkXC8Sm1lluNw1l3znclnq/uaDc1Rn7OxohOfe7gfgRwdPZf2dS0XfSIT2ptppbRuXN/HEwVOo6rRKC6XEL1AWIlps68pmAA6dGk5aBOaLQbf47I8OnuKmi1cW3c/oZIx33MCL3V19XONqn6ncvL2Tm7d3pl3m4Q0IvUCOzpY6OlvqMvpcYvEEb50ZIRJL8PShM3z80lXFnkbBmOZSQt4+M8K6tsZpL8HWhhpg6kYtlK0rm3nz3ZG8zS9b3IfxjRwCKZ2ZKRyaSqIsN6KJ/M1iLfXubz5HwuXJ104TTyhrl9VtPkkfAAAgAElEQVQnX+5zTf/IJG0pwmVTRxODY1HOjkbmbL9+4TI8Of9msU3LmwgIHD49Pxqin6Ex53d9+o13Z2Uae9NX++tHB0/N6piCnubiauWdrfV0ttZlDEU+1j+WHCw+8crs9l0oJlxKyNu9I2xKGV1NjaKLewFsWdnMyGSM7oH8HHbr2xqpDQU4fDp7MbtofKZZLBx0whrLUnMpwCzmaYhD43Pz0n3i4CnWtzVwz0c2cbx/bF5MY30jEdpTNN+NyYixuTMZeWax9qbwgmgudTVB1rc3cijH/TwXeAPCsUh8VqYx79h3rFvKT147PavnK5TUXMZprg3RVBuis6U+YyLlIVco71i3lGcPn5lX05gJlxIRjSc43j/GxuXTbaQts9ZclgDkFBYewYBwwYrmnJVS0+a5lLPmUoBZrHUONRfPJHbL9k5uuGglwYDwxCxHo7lQVfpHJ2lvTjGLub69ufS7eA79zpb6BQtFvnDlkgWp/Ds4FuWiVUtY2lAzq2t8+PQwDeEg/9uHN3JuIsbut/uK7stz6HcPjLGypQ6AlUuc/+nKwBw+PUwwIPzW9ZuZjCXmNUTfhEuJON4/RiyhM+zC3ouu2HlGPDPXoQLMAltWNucc6S3WJEpwAiFqglK0QM/GT1yT2C3bO1nWGOb9G9vm3DQ2GokzEU3QlqK5rGqpp64mMKcFLD2Bsqq1LplMOd9sWdnM8f6xeXdID45HaWuq5aaLV87KNHbo9Dm2rGzmQxe001wbmpV5yvOTnjg7Rqcb3NHZ6giXU2kKWB46PcyG9kbev7GdjuZafjSPpjETLiXCGz3OEC6zNIs11YZYs7S+ILPA1pXN9A5P0p9ltsL0ZrFydujPPN5MiAgt9eE50Vx+dPAU69oauGiVo1Hesr2TY3NsGvOuY6rPJRAQzm9vmjfNZWQBfC4wNcB6c561l6GxCK31NdyyvZPRIk1jqsrh08NsXdlMbSjI9dtW8JPX3y3aOuCVgxmNxOl0NZbOFkfIpIsYO3x6mC0rmwkGhJsvXjmvpjETLiXCe8DP75huFltS5wTkzWYUvXVlc95mMZh6GLNtE0uX51LGJfcL0VzACaQotc/FbxLzorNunAfTWN+Icx5tTTOjDTctn1vhcm4iSjgUYFljmIloYkFMplvzuJ/ngsHxKK0NNVx9flvRprEzw5MMjEXZssI5h1u3dzI0HmV3V3GmsaCv1lgus9iIG6W21bfv+TSNmXCZBfuPn+Vru4/ytd1HeeaNM6xcUkdzXc20dULBAM11oVmNoreuXMKRvlEmY/mp5VOmtMwPYyQ+s8rwlOaSfT+vnzyXPO+v7T7KviLL2wxPRPna7qO80589H0VV004RkI3W+pq0v/lTrxdv3vBMYrf6wkVLYRp74ejZab9n6kvU01w6UjQXcDTl7oHxGec0NB7lyz8/MuuChsMTMZbUhWh2B0mZ/C7ReIK/fPqtnHOLvNI9mDb/6NlDZ/inA2knkGXt0gYawsFp9/N4JM7Dz3ZxpoBy86+dHOJYX37BD4mEMjQepbW+hlAwULRpzDvmLa7v9Bdc01ix5in/M7DKNYfVh4MsbaiZ8dt7mp73PtixfhkdzbVz7iP0sDyXWfC5b71Ej++C3pohRt0ZRRcvXLasbCaeULrOjHDRqtw5FR1NtbQ1hrOO9JxQ5Olmpjq3CN5YlgdIVflP39zPMZ9AWN1az+77rs15XKmMReJ88Yev0zcS4b/cuCXLsboVYvM0i4Hzm/cMTn/xdJ0Z4Tf/bh+/d9NW/uNHNhZ8vP/W1UdnS13SJOZx6/ZO7vv+QV7pHuLSta0F9/ufvvkifT4T5tXnt/Gtu9+X/J5Nc9m4vBFVONY/mgz+AMfH98c/eoPJWIJ7rtlU8DF5jEzEaK6rSQ6aRiZiaecnevbQGf7HU2+ybdUSVvkSPf2oKp/+6l4+vKWDP/vUZdOWfeuFdzjeP8Yn3rN6xnYBN0jF73f88Wun+NMnD3P5eUtZ7o7cc/Gfvvkia5c28PeffW/Odc9NRFGFFtesfev2VXzrhRM8/cYZbr0key6KHy+E2tO+PNPYc2/3F5Wf5I+YXNlSP+1zavFK7/m/sNO5LzzT2P98+WTBloBiMM2lSMYjcXoGx7nnmo0c+IOPcuAPPspf3vmetOu21ocZKNLnAoWbBZLJl1ls1OlurvpwkIZwkP6RzMe67/gAx/rH+ONPXMyBP/gon/7Aes4MTxQ1al+xpI4PXdDB917szlrBYGpWvvxv15b6cDJPwcN7+L67/0RRx3u0b5TNK5pnvBBuuaST2lCAxwso0+MxNB6lb2SS377+Ag78wUe5ZftKugenj+w9zSXdS/3arct5+Qs3TBMsAOe1NfDeDcv47r7iztVjeCJKkxvyCs5LNx3f3d9Ne1MtH76gI2Nf3QPj9I9G0lbw7RuZpL05c5KxZxr2zuXx/d2sXVbPe/OcFG5gNMLx/jFe7h7M6/fwtN6lbrTn1Rvb6Gyp4/H9J/Lan8eh005lBX8C9f23bOXp3/1wUYmvfrPYqpa6aZ9TEykPnx6mMRycVtXht67bzO77rp1zwQImXIrGK0exdeUSljaGWdoYJhBIf7O0NqQ30eTLhvZGwsHcuSt+tqxs5s3TwyQyvLRjcZ0xERFAe1PttFF0Ko/v66YxHOSTl69maWOY1a31RONatGZ22xVrODU0wXNZwjNjSc2lMJ9Lqp/LO68jvaO8+M5gQcepqhzrG2VDW8OMZUvqarjp4pX84KWegs0mnplma2czSxvDrGtr5PTQxDRh2z8aYUldiNpQcMb2DeFQMtw9lduuWMOx/jH2H8+/fFAqwxMxmutCSd9hunptfSOTPHvoDJ+8fHXWCd0O9gwl15/Zx8wKBH62rGxmYCxK7/Ak3QNjPPd2P7ddvjbjM5dp38MTMY7nMMPClI/US4IOBoRPXr6af32zl3cLMMUdOjWcNIl5LG+uS1oJCsX/DKz0CZeVLXUzile+ceocF6xsnvYbtTU5s5jOByZcisR7KaTW/klHS/3szGKhYIBNy5sKjhgbj8aTZSf8qKozE2WaB7O9KZxRuIxFYvzwlZPcsr0zeYN2uLkX2QRSNq6/cAUt9TV8d1/mUX8knn4+8WwsbahhLBKf5qfyjjEcLFzL6B2ZZDQSz3i9b79iLecmYjz1emGzYHqDFK9fT1j7f8/ekcmsL95MONcpmPW3zcXIZIym2lDSLJbO5/JPB3qIJZTbrliTta9Xuj3hMlMz7stxjsnKE6eH+f6LPajCJy+faULLhCdcUj9nwovubKmf0jhuu2ItCYXvv5jeN5RKLJ6gq3ckaXkoBV6ei/+agFMGZmAsmhzcqCqH3x0u6b4LJS/hIiI3ichhEekSkfvSLK8VkW+7y/eIyHrfsvvd9sMicmOuPkVkg9tHl9tn2G3/HRF5XUReEZGnRWSdb5u4iLzk/u0q7qcojKPuS2F9HsLF0VxmF7m0dWVzQbkunpkknUCKJTJrAu1NtfQNpz/WH796mtFInNt3rJ22PkBvhm1yUVcT5OOXruLJ105nFMBFmcUavCz9qT77RiLUBIWPXdLJD18+WdDUsMf6HCGd6XpfvbGNVS11BQuto32jiMB5bj06z4Th9+U5pV8Kq0sHTlXuW7Z38qODpxiLFBd+Ouz6XJqSDv3p10hVeXx/N5euaeGCFdlfZK+6L/WBsQgxX9TZWCTGWCSeVbgk7+dT53h8fzfv39g2rYZfLg52D7G6tZ5wKJCXcBlK0VzAGQDsWLeUx/M0qx7rHyUSS5RYuDjPQGfLdD9TajjymeFJBseiM8yl80nOp1VEgsDDwM3ANuBOEdmWstpngAFV3QQ8BDzobrsNuAO4CLgJ+JKIBHP0+SDwkNvXgNs3wAFgh6peAjwO/Ilv/+Oqepn79/GCfoEiOdY3SntTbdIWnY2lDWGGxqMZTVT5sGVlM++em8xbSF2wohmR9H6a5Ms6nVmsObNZ7Lv7ulnX1sCV65dOrd80O80F4PYda5iMJTJG0BRlFkuTvNo3MklbYy2371jL8GSMn7x+Ou/+cmmqwYDwS1es4edv9RY0K+CxvlE3GdIxk3jO8J4Bv3CJ0NZYuOYCcPsVaxiZjPHjV/M/Vz/DE1GafdFiqWax106e49DpYW7zDTjSoaq80j1IOBRAlWn10LzBTHsWAbqsMczy5loe23uCd86O5dSSUjnYM8R7zmvlws4lvNKd2yTqmbG9+8jj9h1reLt3lAMncvcxFSlWQuHiau8rZwgXN5HSNY3Nxb4LJZ+n9SqgS1WPqGoEeAzYmbLOTuDr7ufHgevE8VbtBB5T1UlVPQp0uf2l7dPd5lq3D9w+PwGgqs+qqmfjeR4o7O4qMcf6xtjQnt/IqaW+hoTObrKlrZ2ZNZF01IeDrFvWkFbbiWZ5Wbc31XI2ZWQJTkbwvx/p57bL10xzRHovhNkIl+2rW7hgRRPfzeAsjWQRhplIVzDUcxq/d8My1iytL8hcdLR/lFBApjlHU7ntijWO2eRAIf2OTRNYXnipP6y0fzSS1dmdjas2LOO8ZQ1FBRuoKiOTsWnCJdUs9vj+bsKhAB+/JHu13XfOjnFuIsb7zm8DHFOfh/c5tbxNKltWNnO0b5Sm2lBBVYr7RybpGRznkjUtXLK6hVd7zuUc6HnCpSVFuNyyvZO6mvzMql7pFW/ywFLgmYZTNRdP2JxyIyS9SurlbhZbDfif+m63Le06qhoDhoC2LNtmam8DBt0+Mu0LHG3mn33f60Rkn4g8LyKfyOOcZs2RvlHWt+U2icFUln6xJWCguESyLRmSL6NZfBgdTWFnZJmiIX3vxW5E4JMpI8alDWGCAZkhXBIJJzP51Z6h5F8mM5SIcPsVaznwzmDaUiZTs/IVEIpcP7PsvmfXDwSE265Yw+63+6aZnzzOTURnOOaP9Y1y3rKGrA7rdW2NXLV+GY/v606e81vvDmc0oagqR3tHWO8bpDhhv6GkcInFEwyMFa+5iDjn+tzb/cky7X7OTaQveAhOFnhCodkNJggHA9OixSZjcf7ppR5u2LYiY1CBh2eKunaLE03m97v0Zcnj8eM9A7f6fH754O17++pWtq9uYWQyNm1+mMlYfIZJdnA8QnNtaMb1bq6r4ZaLO/mfL5/MGbzxximn9Eq6QIxiCSbNYtMHOd73g+59t+/4ACuW1CbfPQtBxTn0ReRXgR3An/qa16nqDuA/AH8uImmTGETkblcI7evtLb7K6fCEEz6aj78FfIUUZ5Exvry5ltaGmoL8Lhd2LuFY/yhnhqebabL5MJJmrhQfyo9fPc37NrTNGLkHAsKyxvCM9b/3Yjc3/vnP+Nhf/lvy70hf5kzyne9ZRTAgaUeExUaLwfSyO33DUxFJv3T5GlTh23una0vxhPKJh3dz//cPTms/2jea1/X+1JVrOdI3mjznjz70M/45g0lqYCzKuYnYjEHK6tb6ZI7O2bEIqtlNRrnwHN8/TGN2/NUv7+E/f+vFtNt5VZCbap3fsrkuNK0y8p4jZxkci+blWD/YPUQ4GOD9m9oB6BueGoz0JcvbZD/H7WucHKJPXVmgScwNJLh49RK2r3HyxPx+l9//x1f55Jd2T9tmaCyaUWB+8vI1DE/E+Le3Mkc4RuMJXjoxmMwxKRXNdSECMtM8Wx8Osry5lq89d4yP/eW/8dTr73JxHjlxc0k+4r8H8BtU17ht6dbpFpEQ0AL059g2XXs/0CoiIVd7mbYvEbke+H3gw6qavDtVtcf9f0REfgq8B3g79URU9RHgEYAdO3YU7QDxQhnPz1e4NMy+Sq+I8GefupS1S/N3Yu68bDV//i9v8c3n3+G3P3pBsj3by7o9TfSXqnKsf5QPui+GGdukCV/uOjNCOBjgr/7DVO5PNgfs8uY6/urO93DFuqUzlhVjFvNeDN6INFlZ2BUua5c1cM2WDv5hz3HuuWZjcnT5L2+8y5HeUc6NR5NJbomEc/7v35j+/P384ntW094UTtZne+CJN/j/fn6EW9Ik2B7N4MdZ3Vqf1Fz6kwmUxWkuAGuWNrBpeRMvHO2fljw6MBpJRnB1nRmZYb7xnPeeSaypLjTNLHbcHf3nk9j7SvcQF3Y2J805/aM+4eIOTHJpZ7du7+T89saCJ2c72DPE+e2NNNfVsHl5kNpQgFe6h9h52WpOD03wTwd6iKsSiSWSJZC80i/p2OYm0R5Powl6/POrp+kbmeQTl5V2cq72plp+9LlfSBs88Q+/+d7kVOsAlxWR0FtK8nla9wKb3SiuMI6DPjUiaxdwl/v5NuAZdWwBu4A73GiyDcBm4IVMfbrbPOv2gdvnDwBE5D3A3wIfV9VkcRwRWSoite7nduADwOuF/AiF4r0U8tZcZll23+ParSvYnCMix8+G9kau3bqcb+45Pi0kN1tobzoHvRPimGD10vT+hnThyz2D46xqreOGi1Ym/5bUZTed3Ly9M222dTFmsebaEMGAJAX60HiUaFynaQC/8cEN9I1E+OHLUyP6r+4+CjhmG68KwbvDE0xEE2zoyH29gwHhI1uWJ8/50+9fz4F3BjmQZqrqYxnuo1Wt9Zwcmi5ciglF9nPl+qXsOz4wzdewz5f/8rXnjs7YxquC7AmX5rrQtGixnsEJaoKS05yVSCiv9gyxfU0LTbUhakOBGWaxlvqa5Is9E8GAFDXr50F33+CE9W9btSSpufz988eJJRRVpuWvDI5FkqbVVJY21FBfE8xa6uaru4+yvq2Ba7akn3VyNlzYuWRaMqXHpuXN0563fCsXzBU5hYurQdwLPAm8AXxHVV8TkS+KiBeZ9RWgTUS6gN8B7nO3fQ34Ds7L/sfAPaoaz9Sn29fvAb/j9tXm9g2OGawJ+G5KyPGFwD4ReRlHMP13VZ1T4ZJ8KeTpc/Fi5VMzxueDT39g/YwXaHaz2EwHvfcQZSrr0dFUOyN34eTgeEZhVChJTasAzcWpjFyTrIyQtOv7nMYf3NTO5uVNPLr7KKrKayeHeP7IWW53/Up73ZppSQ0jz+vt57Yda2muDfHV3cdmLDvWP0owIDO00VWtzuRPo5OxvE1Gubhy/TKGJ2K8eWbKB7fv2FnCwQA7L1vF9/b3zPAJepFhSeFSWzMtWuzk4DidLfU5ExmPnx1jeDLG9tUtiIgb7j7dLDYbs182eocnOTU0wXafULpkdQuv9QwxFonxzT3Hk4Lb738bHMusuYgIq1rrMgqXA+8McOCdQX79/evzTvJcjOT1tKrqE6p6gapuVNUH3LYvqOou9/OEqt6uqptU9SpVPeLb9gF3uy2q+s/Z+nTbj7h9bHL7nHTbr1fVFakhx6r6nKpuV9VL3f+eMJozjvaPsnJJHfXh/Bx1XsTJwBxNu5uN1BcoOLM6Qnrhkm5k6T10mSKl2prC9I5MTnNcnxycYFVLaYRLMXkuMD1Lv3d4pgYgInz6Axt47eQ59h4b4Gu7j1FfE+T3b72Q1oYa9h51hMtUjkv+JkmPptoQn7pyLU8cPDUjRPlo3yhrltbPGLH7I8Y84dJepEPf48r1TpkU75wAXjh2lu1rWvjfP7yR8Wicx/a+M22bKbOYc/+mmsVOutppLrzQ3+2rHTNNe3PttGixXAmUs8HLrblkzZSJ6OLVLYxG4jz01JsMjEX53Rsck7FfWGQzi4GrXWYQLl/dfYzm2lDO8OzFTsU59MuBY32jBb1owqEAjeHgnM3pno3UFyg489FDerNYupFlLs2lvamWSCyRDLWOxhO8OzyRcf1C8cx4oQJHga31NcnRePIlnfIS+8X3rKa1oYY/e+owP3jpJL90xWpaG8LsWLcsaTY61j9KOBQoWljedfV64qp84/lj09qP9aePOPQnUvaPOomfS+pnV7JjzdJ6ViypTd4D45E4r/YMceX6ZVzYuYSrz2/j7/79+LQQ9OG0ZrFU4ZL7NznYPURtKMDmFY5Pp6MpnGIWi+QMQy6WV7qHEGFasVFP0Hzl346ydWUzv+gWy/Tu80RCs5rFYHrQhZ/TQxM8cfAUn7pybV45cIsZEy5FcLRvNK+yL35aG8KzihabDd4L9Ms/P8LIZIxz7mg+kyaQOrLsGRinriaQLOI3Y/1khJmzzemhCVQzazqF4mkuuWzyqfh/8ynhMv2FUR8OcudV5/H8kbNE4gl+/f0bAMdHcbRvlN7hSY72jbJuWUPRJo7z2hr46IUr+Ic970wrz+HkSs28j7wX9snBCSc7v7G2qCKHfkSEHeuXJadHeLl7kGhckwmxn/7AenoGx/mJr3zNVLSY85JcUleT1GZi8QSnz03kdY0P9gyxbdWS5P2WGgDSNzyZ029TLAd7BtnY0USj70W/saOR+pogCXX8bnU1QdqbwkkNfSQSI6Hk1Fz6RiZnhCN/4/ljxFW56+r1c3I+lYQJlwIZGosyMBbN29/i0dpQM6s8l9ngvUB/8vq7XPyHT/LrX90LkLF4XurI8uSQM0LN9IKbijBztsml6RRKMaHIMH1Ol76RSYIBYWmauP9fu3odwYDw4Qs6khFTO1wz0r5jZ11NtXB/i59Pf2ADA2PRZKh178gkI5Mx1qcphLliSR3BgLhmscis/S0eV61fxsmhCboHxpLmMS8677oLV3Desga+9cKUaWx4IooINLo5JU21IUYmY6gqp89NkMhjAJF05vt8Hu1NtZwdjZBIKBPROMOTsTnzubzac27avsFx6l+8egnLGsN8/FInmmuVTxMZypBA6ce7t/2zP6oqj71wgusvXMF5aa5rtVHdelsRFFJTzE+6Kr3zyX/8yEaWN9cmX9RNdSEuXZM+8qa9qZaXu6fyAHoGs49QU4MAvEinfOzx+VCsWazFJ9D7hiMsy1C5urOlnq9/+qpp0WDbV7dQGwqw5+hZjvePcc3W2UX9vO/8ZVx+XisPP9vlVCvOUqssGBBWLnEcxk5dsdKM6ne4Wsq+YwPsPT7AlhXNySS7YED4wKZ2fvzqVODH8GSMpnAo+Zs114VIqDMPz0n3RZxrABFNJLj/lgunZYq3N4WJJ5SBsQjj7sh/LnwuQ+NRTp+bSJul/sAvbmciGp8qu9NST5c7o2ey9EuWBES/X8zTPk+fm6B/NMKHskw7UE2YcCmQQqoh+2mtD3NoaO7mWc/FkroaPv2BDXmt29YUTo4sA+4I+sIsL9eOlPBlry5WqTSXos1i9WGGJ2NE44mcTuMPbp6ewxIOBbhsbSs/OniKSDxR8PVORUT43Ru28Ctf3sNjL7yTzDA/vz19aZDVrfV0u5rLxhKVD9m6cgnNtSH2HO3nxeMD7EzJwdi0vImBsWhSoHnl9j38lZHz1U5rQ0F+9X3rprX5Nd25FC5etYd05VdS80RWtdbzs7d6UdWkKTWbWWxNq6OZ+CPMkvvrKF25l0rGzGIFklrFNl9aZjkb5XzS3lSbHFlOxuL0Dk9mfYksawwjMuVz6RmcoL0pXPScFakUbRZzXw7n3Am5CjW9XLl+Gb3uORVqBk3H+ze28b7zl/FXz77N66fOUROUjNrdqtY6egbGpyV+zpZgQLh83VJ2vXSSkclYMoLMw3sJv+W+JJ2ilVMvWH9l5J7B4rVTfy6VNxHaXDj033bPY/Py3Llhq1rrGIs4ZWAyFa30s6KlFpHpEWZvvZtZmFUjJlwK5Fj/9Cq2+eLZ/2czK+B8MfXwT80amE24hIIBljaE6fX5XEqltcCU5hIqYD4XmJ682jcSKdhpfKVvlsPZai4wpb30jUzyDy+8w9ostcq8RMqJaIK2NDNQFsuV65cy6tYSuzJlFsfN7kvRG4GPTMaSAgWmosaGJx3NZWlDTVETT/mFS6ZAi1Lw1plhakOBvPKt1rjrdA+MJ83X2eql1YaCdDTVThMuXb0jtDbUzJn/qNIw4VIgx4qIFAPnRRdLaPLBLmf8D/+UiSv7CNWfpX9ycLxkOS4w5XMJF6i5eA7ZwbGIM+FWgaPjy89rJSBQXxNkxZLSjKyvXL+MD13QQSSWyJqUuaq1Hm8cUiqfC0wFKqxqqZvhR+tsqaMxHEwKlxlmsdqpysizSZL1hHzv8GQyCGSuzGLndzSlzWZPZSpCbzyZ7JwtFNnb5qQvHLnr3RE2dTTNOrJvsWDCpQBU1amGXEQy3VSV3oUJRy6EjuYpB32uBEoPL7xUVUuuuRRrFvMiw7oHxonEEgWPKJvrariwcwnr2xtL+sL4XbfOW7ZBiv/3LuVI+LK1rYSDgRlaCzia1cblTdOEiz9XwzORjUzEnPI+RQ4gltSHCAedRN3e4Umaa0MlM6H6eevMSFIby4VfuAyORWkMB3P6+Pw14MDRXMwkNoU59AsgnlD+87Wb2NZZeH2jFl/xyjUzazOWFe2+keXopKNppU5OlG6bAycGODceYzQSL1mkGDhmMRHyGoH68cxi3suymNHxg790SXLmzlJx6dpW/uZXr+DStZnvI79WUMpRfV1NkEd+7Qo2ZnA6b1rexHNd/cDULJQezX6fy8B4XoU80yEitLma7kQ0Pif+lvFInJ7BcT6VZ5Z8W2OYcCjAyaEJBsaieZWqX9Vax7+88S6qytnRCGdHIyZcfJhwKYBQMMDdH0pbzT8nyZkRK8Cp31JfQ01Q6BuJMDAaYXlzbc45KbzpkfPVdAohEk8UrLXAlLboOVqLeUkXUygxH3JNduWfDKpUeS4eH8lSTHHT8ia+/2IPwxNRhieiLPGZxTz/S8/gOKOR+KyusafpTkTjc+KjeLt3BNX8nesi4mbdjzMZjWfNcfFY3VrPZCxB/2gkGTxgwmUKM4vNE95IaKACzGIiQluj8/B7CZS5aG8OMx6N85ZbGLFURSvBMYsV6m8BZ6QtQjJ/Ya7qV80FzXU1yRf7shI69HPhhdEeOj3MZCwxzSzWFA5Nmzp7NqZPz0fXNxKZ0zDkfM1iQLIYZbaildPXnzKlefeYCZcpTLjME0tLMKfLfNLeHE469PMZoXoviJdPOMmXpTnHJP0AAApaSURBVI4WKzRSDJyJzFrqa5K5ScVOFbxQrGqtZ4k7C+R84U3p8NI7TrFJv0M/EBCawiEOv+sJl+JNn56mO1dFK7vOjBAMCOsKCCFf3VpPjxstVrBwOTNCQzhY0kCWSseEyzyxpILMYuA8/L3Dk8l5WXLhRQAd7BkkHAqUNHw2WqRZDBxzZCyhiMCyBZzytRjOW9bAinmek2Pt0nrCwQAvuvPPNKfMwdNUF+Idd5Ks2Win7c2OZjw4Fp0T4fLWmWHWtTUUlHi7qrWeM8OT9A5PJqfJyMZUgdEJus6MsLGjqapL7KdiPpd5oq4mSH1NsCKixcARLnuOnGUylsjPLOa+IF7tOcfqLHXIiiFapFkMoKUhDP1jLGsIZ8wpKVf+263bGJ6c38FIKBjg/I5GDriaiz/PBRxN5tSQExY+m2kA2ptqk4ESc6FRdhUQKebh3edDeWourb5Jw7rOjHD1+W1FHetipbKetgqntaGmcsxiTbXJ0hz5+lwAxqOljRQDT3MpTlh5gRSV5G/xOK+tIa8phEvNxuVNnHZnZWyeIVyc37OztW5Wo3S/E7/U1yYSS3C8f6xg/4ff/JstO9/DmzTszXeHOTU0UbIyPYsFEy7zSEv9whavLAT/w5+Pz8U//3mp7c6Oz6VIs5g7Aq00f8tC4q+N1VybYhZzHfyzvcb+agmlFi7H+0eJJTSvsi9+/IOofDQXb5s9boVpc+ZPx4TLPLKQZfcLxT8dcD7CJRwKJMM3SxkpBo5ZrFifi5dIWYmay0LhTeoF6TQXV7jMMmDDn9tS6rlcshWszIY//DufPBdwysZEYk4FiULNcIsdEy7zSGv9wk0YVijey7i+Jpj3KM7TdkoZKQaO5hIu0izWUsFmsYXC/1LOZBab7QDCfz1KrVV6hTfP7yisTJMzaZhzXPmYxWBKgwsHAwUXs13s5CVcROQmETksIl0icl+a5bUi8m13+R4RWe9bdr/bflhEbszVp4hscPvocvsMF7uPcqPSfC7ghJvm65z3tillAiWUyCxmwiVvNrQ34rlT0jn0AVbP0q/WWl9DMCA0hINFFb/MRteZEdYsrS+qX++88tVcvIHU+vbMRUirlZy/hogEgYeBm4FtwJ0isi1ltc8AA6q6CXgIeNDddhtwB3ARcBPwJREJ5ujzQeAht68Bt++C91HoDzEftLgThlVGZeTCtRDP1FF6zUWLd+gnhYv5XPKlNhRkXVsj4VBgRo6NV7xyttc4EBDaGsNzFIZcfI0v77wK8blAfmX9q418RO1VQJeqHlHVCPAYsDNlnZ3A193PjwPXiTPc3Qk8pqqTqnoU6HL7S9unu821bh+4fX6iyH2UHa31YSKxBBPRxEIfSk6WNoQJBqQgLcSznXfmqENWKLPLc3F9LnNQv2oxs7GjKSlI/DSVyOcCjjZZaqEfTyhHekeKnrDLO698yr/AlJZukWIzyUdvXA2c8H3vBt6baR1VjYnIENDmtj+fsu1q93O6PtuAQVWNpVm/mH1MQ0TuBu4GOO+88zKe8FyxtbOZj13SSSSeoJ6yVK6SBALC799yYXJq3Hy4fccaVrcWPtdNLq4+v43GNC+6fLhqwzJ+8xc28N40VYCNzPzmL2zgSN/MGmQ3XLSSs6ORrNMF5Mu9124qeOrqXIxFYtyyvZOrirzen9qxlvam2rzv4bXL6vnctZv4pcvTvnKqmqpKolTVR4BHAHbs2DHvtqlrtiznmixFA8uN3/hgftMie1y0qmVO8jI+f9PWordtrA3x+7emWnGNXLz3/DbemyYpcHVrPb97w5aS7OOW7Z0l6cdPc10ND/3yZUVvv2VlM1tW5m/iEhF+p0S/x2IjH1tDD+CvW73GbUu7joiEgBagP8u2mdr7gVa3j9R9FboPwzAMY4HIR7jsBTa7UVxhHOf5rpR1dgF3uZ9vA55Rx2u9C7jDjfTaAGwGXsjUp7vNs24fuH3+oMh9GIZhGAtETrOY69+4F3gSCAKPquprIvJFYJ+q7gK+AnxDRLqAszjCAne97wCvAzHgHlWNA6Tr093l7wGPicgfAwfcvilmH4ZhGMbCIJUQFjsX7NixQ/ft27fQh2EYhlExiMh+Vd2Rz7qW9WMYhmGUHBMuhmEYRskx4WIYhmGUHBMuhmEYRsmpWoe+iPQCx4vcvB3oK+HhVAp23tWFnXd1kc95r1PVjnw6q1rhMhtEZF++EROLCTvv6sLOu7oo9XmbWcwwDMMoOSZcDMMwjJJjwqU4HlnoA1gg7LyrCzvv6qKk520+F8MwDKPkmOZiGIZhlBwTLgUgIjeJyGER6RKR+xb6eOYKEVkrIs+KyOsi8pqI/JbbvkxEnhKRt9z/+c8kVkG4U3EfEJEfut83iMge97p/263kvegQkVYReVxEDonIGyJydTVccxH5bfc+f1VEviUidYvxmovIoyJyRkRe9bWlvb7i8Bfu+b8iIpcXuj8TLnkiIkHgYeBmYBtwp4gs1lmoYsDvquo24H3APe653gc8raqbgafd74uR3wLe8H1/EHhIVTcBA8BnFuSo5p7/F/ixqm4FLsX5DRb1NReR1cDngB2qejFOlfY7WJzX/GvATSltma7vzTjTl2zGmb33rwvdmQmX/LkK6FLVI6oaAR4Ddi7wMc0JqnpKVV90Pw/jvGRW45zv193Vvg58YmGOcO4QkTXArcCX3e8CXAs87q6yWM+7BfgQ7hQXqhpR1UGq4JrjTD1S705C2ACcYhFec1X9Gc50JX4yXd+dwN+pw/M4kzgWNHWoCZf8WQ2c8H3vdtsWNSKyHngPsAdYoaqn3EWngRULdFhzyZ8DnwcS7vc2YFBVY+73xXrdNwC9wFddk+CXRaSRRX7NVbUH+H+Ad3CEyhCwn+q45pD5+s76fWfCxciIiDQB3wP+D1U951/mzgK6qEINReRjwBlV3b/Qx7IAhIDLgb9W1fcAo6SYwBbpNV+KM0rfAKwCGplpOqoKSn19TbjkTw+w1vd9jdu2KBGRGhzB8k1V/b7b/K6nGrv/zyzU8c0RHwA+LiLHcMye1+L4IVpdkwks3uveDXSr6h73++M4wmaxX/PrgaOq2quqUeD7OPdBNVxzyHx9Z/2+M+GSP3uBzW4USRjH6bdrgY9pTnD9DF8B3lDVP/Mt2gXc5X6+C/jBfB/bXKKq96vqGlVdj3N9n1HVXwGeBW5zV1t05w2gqqeBEyKyxW26Dmfq8EV9zXHMYe8TkQb3vvfOe9Ffc5dM13cX8Gtu1Nj7gCGf+SwvLImyAETkFhybfBB4VFUfWOBDmhNE5IPAz4GDTPke/iuO3+U7wHk4FaU/paqpDsJFgYh8BPgvqvoxETkfR5NZBhwAflVVJxfy+OYCEbkMJ5AhDBwBPo0zAF3U11xE/m/gl3GiJA8An8XxLyyqay4i3wI+glP9+F3gD4F/Is31dQXtX+GYCMeAT6tqQfPCm3AxDMMwSo6ZxQzDMIySY8LFMAzDKDkmXAzDMIySY8LFMAzDKDkmXAzDMIySY8LFMAzDKDkmXAzDMIySY8LFMAzDKDn/P+5L+dstouFPAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range((99)),error)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### It looks like all of our error is under 1. perfect! \n",
    "# What if we decided to use this to optimize a regression model?\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.00017903193810751833"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "error=np.array(error)\n",
    "error=error[~np.isnan(error)]\n",
    "error.max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}