Unsupervised Learning Project-Document Clustering and Topic Modeling.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Document Clustering and Topic Modeling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this project, we use unsupervised learning models to cluster unlabeled documents into different groups, visualize the results and identify their latent topics/structures."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Contents"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<ul>\n",
    "<li>[Part 1: Load Data](#Part-1:-Load-Data)\n",
    "<li>[Part 2: Tokenizing and Stemming](#Part-2:-Tokenizing-and-Stemming)\n",
    "<li>[Part 3: TF-IDF](#Part-3:-TF-IDF)\n",
    "<li>[Part 4: K-means clustering](#Part-4:-K-means-clustering)\n",
    "<li>[Part 5: Topic Modeling - Latent Dirichlet Allocation](#Part-5:-Topic-Modeling---Latent-Dirichlet-Allocation)\n",
    "</ul>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Part 1: Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import nltk\n",
    "import re\n",
    "import os\n",
    "\n",
    "from sklearn import decomposition\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Read data from files. In summary, we have 100 titles and 100 synoposes (combined from imdb and wiki)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#import three lists: titles and wikipedia synopses\n",
    "titles = open('../data/title_list.txt').read().split('\\n')\n",
    "titles = titles[:100] #ensures that only the first 100 are read in\n",
    "\n",
    "#The wiki synopses and imdb synopses of each movie is seperated by the keywords \"BREAKS HERE\". \n",
    "#Each synoposes may consist of multiple paragraphs.\n",
    "synopses_wiki = open('../data/synopses_list_wiki.txt').read().split('\\n BREAKS HERE')\n",
    "synopses_wiki = synopses_wiki[:100]\n",
    "\n",
    "synopses_imdb = open('../data/synopses_list_imdb.txt').read().split('\\n BREAKS HERE')\n",
    "synopses_imdb = synopses_imdb[:100]\n",
    "\n",
    "#Combine imdb and wiki to get full synoposes for the top 100 movies. \n",
    "synopses = []\n",
    "for i in range(len(synopses_wiki)):\n",
    "    item = synopses_wiki[i] + synopses_imdb[i]\n",
    "    synopses.append(item)\n",
    "    \n",
    "#Because these synopses have already been ordered in popularity order, \n",
    "#we just need to generate a list of ordered numbers for future usage.\n",
    "ranks = range(len(titles))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Part 2: Tokenizing and Stemming"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Load stopwords and stemmer function from NLTK library.\n",
    "Stop words are words like \"a\", \"the\", or \"in\" which don't convey significant meaning.\n",
    "Stemming is the process of breaking a word down into its root."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "We use 153 stop-words from nltk library.\n",
      "['i', 'me', 'my', 'myself', 'we', 'our', 'ours', 'ourselves', 'you', 'your']\n"
     ]
    }
   ],
   "source": [
    "# Use nltk's English stopwords.\n",
    "stopwords = nltk.corpus.stopwords.words('english')\n",
    "\n",
    "print (\"We use \" + str(len(stopwords)) + \" stop-words from nltk library.\")\n",
    "print (stopwords[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nltk.stem.snowball import SnowballStemmer\n",
    "stemmer = SnowballStemmer(\"english\")\n",
    "\n",
    "def tokenization_and_stemming(text):\n",
    "    tokens = [word.lower() for sent in nltk.sent_tokenize(text) for word in nltk.word_tokenize(sent) if word not in stopwords]\n",
    "#     tokens=[]\n",
    "#     for sent in nltk.sent_tokenize(text):\n",
    "#         for word in nltk.word_tokenize(sent):\n",
    "#             if word not in stopwords:\n",
    "#                 tokens.append(word);   \n",
    "    filtered_tokens = []\n",
    "    # filter out any tokens not containing letters (e.g., numeric tokens, raw punctuation)\n",
    "    for token in tokens:\n",
    "        if re.search('[a-zA-Z]', token):\n",
    "            filtered_tokens.append(token)\n",
    "    stems = [stemmer.stem(t) for t in filtered_tokens]\n",
    "    return stems\n",
    "\n",
    "def tokenization(text):\n",
    "    tokens = [word.lower() for sent in nltk.sent_tokenize(text) for word in nltk.word_tokenize(sent) if word not in stopwords]\n",
    "    filtered_tokens = []\n",
    "    for token in tokens:\n",
    "        if re.search('[a-zA-Z]', token):\n",
    "            filtered_tokens.append(token)\n",
    "    return filtered_tokens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['look', 'father', \"'s\", 'arm']"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tokenization_and_stemming(\"she looked at her father's arm.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Use our defined functions to analyze (i.e. tokenize, stem) our synoposes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "docs_stemmed = []\n",
    "docs_tokenized = []\n",
    "for i in synopses:\n",
    "    tokenized_and_stemmed_results = tokenization_and_stemming(i)\n",
    "    docs_stemmed.extend(tokenized_and_stemmed_results)\n",
    "    \n",
    "    tokenized_results = tokenization(i)\n",
    "    docs_tokenized.extend(tokenized_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a mapping from stemmed words to original tokenized words for result interpretation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "angeles\n"
     ]
    }
   ],
   "source": [
    "vocab_frame_dict = {docs_stemmed[x]:docs_tokenized[x] for x in range(len(docs_stemmed))}\n",
    "print (vocab_frame_dict['angel'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Part 3: TF-IDF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In total, there are 100 synoposes and 538 terms.\n"
     ]
    }
   ],
   "source": [
    "#define vectorizer parameters\n",
    "tfidf_model = TfidfVectorizer(max_df=0.8, max_features=2000,\n",
    "                                 min_df=0.2, stop_words='english',\n",
    "                                 use_idf=True, tokenizer=tokenization_and_stemming, ngram_range=(1,1))\n",
    "\n",
    "tfidf_matrix = tfidf_model.fit_transform(synopses) #fit the vectorizer to synopses\n",
    "\n",
    "print (\"In total, there are \" + str(tfidf_matrix.shape[0]) + \\\n",
    "      \" synoposes and \" + str(tfidf_matrix.shape[1]) + \" terms.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'analyzer': 'word',\n",
       " 'binary': False,\n",
       " 'decode_error': 'strict',\n",
       " 'dtype': numpy.int64,\n",
       " 'encoding': 'utf-8',\n",
       " 'input': 'content',\n",
       " 'lowercase': True,\n",
       " 'max_df': 0.8,\n",
       " 'max_features': 2000,\n",
       " 'min_df': 0.2,\n",
       " 'ngram_range': (1, 1),\n",
       " 'norm': 'l2',\n",
       " 'preprocessor': None,\n",
       " 'smooth_idf': True,\n",
       " 'stop_words': 'english',\n",
       " 'strip_accents': None,\n",
       " 'sublinear_tf': False,\n",
       " 'token_pattern': '(?u)\\\\b\\\\w\\\\w+\\\\b',\n",
       " 'tokenizer': <function __main__.tokenization_and_stemming(text)>,\n",
       " 'use_idf': True,\n",
       " 'vocabulary': None}"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_model.get_params()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Save the terms identified by TF-IDF."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "tf_selected_words = tfidf_model.get_feature_names()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<100x538 sparse matrix of type '<class 'numpy.float64'>'\n",
       "\twith 19003 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "#tf_selected_words"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# (Optional) Calculate Document Similarity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[1.         0.1996283  0.23342595 ... 0.3581728  0.3277855  0.17714484]\n",
      " [0.1996283  1.         0.26626879 ... 0.26910321 0.22856581 0.10870124]\n",
      " [0.23342595 0.26626879 1.         ... 0.27738654 0.24669308 0.12805055]\n",
      " ...\n",
      " [0.3581728  0.26910321 0.27738654 ... 1.         0.55040973 0.08151334]\n",
      " [0.3277855  0.22856581 0.24669308 ... 0.55040973 1.         0.13005234]\n",
      " [0.17714484 0.10870124 0.12805055 ... 0.08151334 0.13005234 1.        ]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "cos_matrix = cosine_similarity(tfidf_matrix)\n",
    "print (cos_matrix)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Part 4: K-means clustering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "\n",
    "num_clusters = 5\n",
    "km = KMeans(n_clusters=num_clusters)\n",
    "km.fit(tfidf_matrix)\n",
    "clusters = km.labels_.tolist()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.1. Analyze K-means Result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create DataFrame films from all of the input files.\n",
    "films = { 'title': titles, 'rank': ranks, 'synopsis': synopses, 'cluster': clusters}\n",
    "frame = pd.DataFrame(films, index = [clusters] , columns = ['rank', 'title', 'cluster'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rank</th>\n",
       "      <th>title</th>\n",
       "      <th>cluster</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>The Godfather</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>The Shawshank Redemption</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2</td>\n",
       "      <td>Schindler's List</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3</td>\n",
       "      <td>Raging Bull</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4</td>\n",
       "      <td>Casablanca</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>One Flew Over the Cuckoo's Nest</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6</td>\n",
       "      <td>Gone with the Wind</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7</td>\n",
       "      <td>Citizen Kane</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>8</td>\n",
       "      <td>The Wizard of Oz</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>9</td>\n",
       "      <td>Titanic</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   rank                            title  cluster\n",
       "0     0                    The Godfather        0\n",
       "2     1         The Shawshank Redemption        2\n",
       "4     2                 Schindler's List        4\n",
       "0     3                      Raging Bull        0\n",
       "2     4                       Casablanca        2\n",
       "2     5  One Flew Over the Cuckoo's Nest        2\n",
       "1     6               Gone with the Wind        1\n",
       "2     7                     Citizen Kane        2\n",
       "2     8                 The Wizard of Oz        2\n",
       "2     9                          Titanic        2"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "frame.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of films included in each cluster:\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>cluster</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>62</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   cluster\n",
       "2       62\n",
       "1       13\n",
       "4       12\n",
       "0        7\n",
       "3        6"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print (\"Number of films included in each cluster:\")\n",
    "frame['cluster'].value_counts().to_frame()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average rank (1 to 100) per cluster:\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rank</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cluster</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>29.571429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>64.230769</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>50.854839</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>52.833333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>36.500000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              rank\n",
       "cluster           \n",
       "0        29.571429\n",
       "1        64.230769\n",
       "2        50.854839\n",
       "3        52.833333\n",
       "4        36.500000"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped = frame['rank'].groupby(frame['cluster'])\n",
    "print (\"Average rank (1 to 100) per cluster:\")\n",
    "grouped.mean().to_frame()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<Document clustering result by K-means>\n",
      "Cluster 0 words:son,army,killing,father,family,battle,\n",
      "Cluster 0 titles (7 movies): \n",
      "The Godfather, Raging Bull, Amadeus, The Lord of the Rings: The Return of the King, Gladiator, Braveheart, Dances with Wolves\n",
      "\n",
      "Cluster 1 words:love,relationship,war,marries,home,children,\n",
      "Cluster 1 titles (13 movies): \n",
      "Gone with the Wind, Forrest Gump, The Best Years of Our Lives, City Lights, Midnight Cowboy, Mr. Smith Goes to Washington, Annie Hall, Out of Africa, Terms of Endearment, Giant, Network, Wuthering Heights, Yankee Doodle Dandy\n",
      "\n",
      "Cluster 2 words:killing,police,car,friend,n't,father,\n",
      "Cluster 2 titles (62 movies): \n",
      "The Shawshank Redemption, Casablanca, One Flew Over the Cuckoo's Nest, Citizen Kane, The Wizard of Oz, Titanic, The Godfather: Part II, Psycho, Sunset Blvd., Vertigo, On the Waterfront, West Side Story, Star Wars, E.T. the Extra-Terrestrial, 2001: A Space Odyssey, The Silence of the Lambs, Chinatown, Singin' in the Rain, Some Like It Hot, 12 Angry Men, Gandhi, Unforgiven, Raiders of the Lost Ark, Rocky, A Streetcar Named Desire, To Kill a Mockingbird, My Fair Lady, Ben-Hur, Doctor Zhivago, Jaws, The Good, the Bad and the Ugly, Butch Cassidy and the Sundance Kid, The Treasure of the Sierra Madre, The Apartment, High Noon, The Pianist, Goodfellas, The Exorcist, The Deer Hunter, The French Connection, It Happened One Night, Rain Man, Good Will Hunting, Fargo, The Grapes of Wrath, Shane, The Green Mile, Close Encounters of the Third Kind, Nashville, The Graduate, American Graffiti, Pulp Fiction, The African Queen, Stagecoach, The Maltese Falcon, A Clockwork Orange, Taxi Driver, Double Indemnity, Rebel Without a Cause, Rear Window, The Third Man, North by Northwest\n",
      "\n",
      "Cluster 3 words:george,marries,family,woman,friend,john,\n",
      "Cluster 3 titles (6 movies): \n",
      "It's a Wonderful Life, The Philadelphia Story, An American in Paris, The King's Speech, A Place in the Sun, Tootsie\n",
      "\n",
      "Cluster 4 words:soldiers,commanded,captain,general,orders,war,\n",
      "Cluster 4 titles (12 movies): \n",
      "Schindler's List, Lawrence of Arabia, The Sound of Music, The Bridge on the River Kwai, Dr. Strangelove or: How I Learned to Stop Worrying and Love the Bomb, Apocalypse Now, From Here to Eternity, Saving Private Ryan, Patton, Platoon, All Quiet on the Western Front, Mutiny on the Bounty\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print (\"<Document clustering result by K-means>\")\n",
    "\n",
    "#km.cluster_centers_ denotes the importances of each items in centroid.\n",
    "#We need to sort it in decreasing-order and get the top k items.\n",
    "order_centroids = km.cluster_centers_.argsort()[:, ::-1] \n",
    "\n",
    "Cluster_keywords_summary = {}\n",
    "for i in range(num_clusters):\n",
    "    print (\"Cluster \" + str(i) + \" words:\", end='')\n",
    "    Cluster_keywords_summary[i] = []\n",
    "    for ind in order_centroids[i, :6]: #replace 6 with n words per cluster\n",
    "        Cluster_keywords_summary[i].append(vocab_frame_dict[tf_selected_words[ind]])\n",
    "        print (vocab_frame_dict[tf_selected_words[ind]] + \",\", end='')\n",
    "    print ()\n",
    "    #Here ix means index, which is the clusterID of each item.\n",
    "    #Without tolist, the values result from dataframe is <type 'numpy.ndarray'>\n",
    "    cluster_movies = frame.ix[i]['title'].values.tolist()\n",
    "    print (\"Cluster \" + str(i) + \" titles (\" + str(len(cluster_movies)) + \" movies): \")\n",
    "    print (\", \".join(cluster_movies))\n",
    "    print ()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.2. Plot K-means Result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca = decomposition.PCA(n_components=2)\n",
    "tfidf_matrix_np=tfidf_matrix.toarray()\n",
    "pca.fit(tfidf_matrix_np)\n",
    "X = pca.transform(tfidf_matrix_np)\n",
    "\n",
    "xs, ys = X[:, 0], X[:, 1]\n",
    "\n",
    "#set up colors per clusters using a dict\n",
    "cluster_colors = {0: '#1b9e77', 1: '#d95f02', 2: '#7570b3', 3: '#e7298a', 4: '#66a61e'}\n",
    "#set up cluster names using a dict\n",
    "cluster_names = {}\n",
    "for i in range(num_clusters):\n",
    "    cluster_names[i] = \", \".join(Cluster_keywords_summary[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x648 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline \n",
    "\n",
    "#create data frame with PCA cluster results\n",
    "df = pd.DataFrame(dict(x=xs, y=ys, label=clusters, title=titles)) \n",
    "groups = df.groupby(clusters)\n",
    "\n",
    "# set up plot\n",
    "fig, ax = plt.subplots(figsize=(16, 9))\n",
    "#Set color for each cluster/group\n",
    "for name, group in groups:\n",
    "    ax.plot(group.x, group.y, marker='o', linestyle='', ms=12, \n",
    "            label=cluster_names[name], color=cluster_colors[name], \n",
    "            mec='none')\n",
    "\n",
    "ax.legend(numpoints=1,loc=4)  #show legend with only 1 point, position is right bottom.\n",
    "\n",
    "plt.show() #show the plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Part 5: Topic Modeling - Latent Dirichlet Allocation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.decomposition import LatentDirichletAllocation\n",
    "lda = LatentDirichletAllocation(n_components=5, learning_method = 'online')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "tfidf_matrix_lda = (tfidf_matrix * 100)\n",
    "tfidf_matrix_lda = tfidf_matrix_lda.astype(int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LatentDirichletAllocation(batch_size=128, doc_topic_prior=None,\n",
       "             evaluate_every=-1, learning_decay=0.7,\n",
       "             learning_method='online', learning_offset=10.0,\n",
       "             max_doc_update_iter=100, max_iter=10, mean_change_tol=0.001,\n",
       "             n_components=5, n_jobs=1, n_topics=None, perp_tol=0.1,\n",
       "             random_state=None, topic_word_prior=None,\n",
       "             total_samples=1000000.0, verbose=0)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lda.fit(tfidf_matrix_lda)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5, 538)\n"
     ]
    }
   ],
   "source": [
    "topic_word = lda.components_\n",
    "print(topic_word.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_top_words = 7\n",
    "topic_keywords_list = []\n",
    "for i, topic_dist in enumerate(topic_word):\n",
    "    #Here we select top(n_top_words-1)\n",
    "    lda_topic_words = np.array(tf_selected_words)[np.argsort(topic_dist)][:-n_top_words:-1] \n",
    "    for j in range(len(lda_topic_words)):\n",
    "        lda_topic_words[j] = vocab_frame_dict[lda_topic_words[j]]\n",
    "    topic_keywords_list.append(lda_topic_words.tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(100, 5)\n"
     ]
    }
   ],
   "source": [
    "doc_topic = lda.transform(tfidf_matrix_lda)\n",
    "print (doc_topic.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<Document clustering result by LDA>\n",
      "Cluster 4 words: family, george, father, john, love, friend\n",
      "Cluster 4 titles (31 movies): \n",
      "The Godfather, The Shawshank Redemption, Raging Bull, One Flew Over the Cuckoo's Nest, Gone with the Wind, The Godfather: Part II, On the Waterfront, Forrest Gump, West Side Story, It's a Wonderful Life, The Philadelphia Story, An American in Paris, The Best Years of Our Lives, My Fair Lady, Doctor Zhivago, Braveheart, The Pianist, Goodfellas, City Lights, The King's Speech, It Happened One Night, A Place in the Sun, Midnight Cowboy, Annie Hall, Out of Africa, Good Will Hunting, Terms of Endearment, Tootsie, Giant, The Grapes of Wrath, The Green Mile\n",
      "\n",
      "Cluster 1 words: army, soldiers, killing, men, war, commanded\n",
      "Cluster 1 titles (21 movies): \n",
      "Schindler's List, Lawrence of Arabia, Star Wars, 2001: A Space Odyssey, The Bridge on the River Kwai, Dr. Strangelove or: How I Learned to Stop Worrying and Love the Bomb, Apocalypse Now, The Lord of the Rings: The Return of the King, Gladiator, Saving Private Ryan, Raiders of the Lost Ark, Ben-Hur, Patton, Butch Cassidy and the Sundance Kid, The Treasure of the Sierra Madre, Platoon, Dances with Wolves, The Deer Hunter, All Quiet on the Western Front, Shane, Stagecoach\n",
      "\n",
      "Cluster 3 words: singing, police, filmed, apartment, love, musical\n",
      "Cluster 3 titles (13 movies): \n",
      "Casablanca, Citizen Kane, Sunset Blvd., The Silence of the Lambs, Chinatown, Singin' in the Rain, Amadeus, The Apartment, High Noon, Mr. Smith Goes to Washington, Close Encounters of the Third Kind, Nashville, The Third Man\n",
      "\n",
      "Cluster 0 words: car, police, house, killing, n't, father\n",
      "Cluster 0 titles (28 movies): \n",
      "The Wizard of Oz, Psycho, Vertigo, E.T. the Extra-Terrestrial, Some Like It Hot, 12 Angry Men, Gandhi, Unforgiven, Rocky, A Streetcar Named Desire, To Kill a Mockingbird, The Exorcist, The French Connection, Rain Man, Fargo, Network, The Graduate, American Graffiti, Pulp Fiction, The Maltese Falcon, A Clockwork Orange, Taxi Driver, Wuthering Heights, Double Indemnity, Rebel Without a Cause, Rear Window, North by Northwest, Yankee Doodle Dandy\n",
      "\n",
      "Cluster 2 words: water, captain, mr., bar, ship, attempting\n",
      "Cluster 2 titles (7 movies): \n",
      "Titanic, The Sound of Music, From Here to Eternity, Jaws, The Good, the Bad and the Ugly, The African Queen, Mutiny on the Bounty\n",
      "\n"
     ]
    }
   ],
   "source": [
    "topic_doc_dict = {}\n",
    "print (\"<Document clustering result by LDA>\")\n",
    "for i in range(len(doc_topic)):\n",
    "    topicID = doc_topic[i].argmax()\n",
    "    if topicID not in topic_doc_dict:\n",
    "        topic_doc_dict[topicID] = [titles[i]]\n",
    "    else:\n",
    "        topic_doc_dict[topicID].append(titles[i])\n",
    "for i in topic_doc_dict:\n",
    "    print (\"Cluster \" + str(i) + \" words: \" + \", \".join(topic_keywords_list[i]))\n",
    "    print (\"Cluster \" + str(i) + \" titles (\" + str(len(topic_doc_dict[i])) + \" movies): \")\n",
    "    print (', '.join(topic_doc_dict[i]))\n",
    "    print ()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Appendix: K-means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.datasets.samples_generator import make_blobs\n",
    "X, y = make_blobs(n_samples=300, centers=4,\n",
    "                  random_state=0, cluster_std=0.60)\n",
    "plt.scatter(X[:, 0], X[:, 1], s=50);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW4AAAD8CAYAAABXe05zAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzsnXV4HVXawH9nrsfTSL1UqdNSBYoXFmeLs8BiW8qyuC2+sMsHiy2uxYsUFmdxryB16i5JJW0al6szc74/5jbNTebe3KRJ9fyeJw9k5pwz7wzknXfe84qQUqJQKBSKPQdtVwugUCgUiuahFLdCoVDsYSjFrVAoFHsYSnErFArFHoZS3AqFQrGHoRS3QqFQ7GEoxa1QKBR7GEpxKxQKxR6GUtwKhUKxh+Fsi0Vzc3Nl9+7d22JphUKh2CuZM2dOiZQyL5mxbaK4u3fvzuzZs9tiaYVCodgrEUIUJDtWuUoUCoViD0MpboVCodjDUIpboVAo9jCU4lYoFIo9jDbZnFS0DYW1JfyydTkSGJPXl26pubtaJIVCsQtQinsPwJQm9y36iG+K5rOt7cWzK77m2I4HcOeg09GE+nBSKPYl1F/8HsDkdT/z7eYFhEydcPQnZOp8V7SQt9ZO39XiKRSKnYxS3HsAb6ydRtCINDoeNCO8sXbqLpBIoVDsSpTi3s3RTYPycE3c85WRABFT34kSKRSKXY1S3Ls5DqGR6vTGPZ/icOMUjp0okUKh2NUoxb2bI4TgzG6j8WiN95HdmpMzuo1GCLELJFMoFLsKpbj3AMb3HsugrG74HO66Yz6Hm4GZXZnQ55hdKJlCodgVqHDAPQC35uTZkX9hdtkafty8CBMY22EQI9r13GFrWzcNphUv5beSlficHo7vOIR+mZ1bR3CFQtEmKMW9hyCEYGROL0bm9Gq1NcvDNYz/7QVKQ9X4jTAagg8LZ3BcpyHcPvA05YJRKHZTlKtkH+afC96nKFCO3wgDYCIJmhG+LprP10Xzd7F0CoUiHsri3guo0YO8uvpH/rdhDgEjTN+MTvy1z7GMSGCdV4RrmVW2Gl2ajc4FjQhvrZ3G8Z2GtqXYCoWihSjFvYcT0MNc8uuzbPKXE5EGAAsqCrl+ziTuHnwGx3Q8wHZeSagat3ASwbA9vyVY1Sw5SoLVvLVuGr+XryPD5eOETgcytsMgXDbRMAqFYsdQf1VtTNAI88vWFVTrQQZndaVnWvuk580qXU3QiDA0uzt53gzbcZ9smMWWQGWd0t5GyIzwwJJPOLL9QJxa4zjv9t7MRnPq0y01Jyk5DWny0OJP+GjDrJjjc8vW8ta66Vzb9wRKQzV0SWnHgMwuym+uULQCSnG3Id8VLeTeRR+gITCkiQSGZHfjoQMvIMXpiTvv841zeWjJp2gIJKBLg+M6DuG2geMaKeHPN84jaDZOhwcwTJMllRs4IHu/RufSXT6Obj+IH7YsItwg89LrcHFxzyOTusenl3/FJxsat6kLmTrLqzZx9axX8TicSCDfm8njwy+ic0q7pNZWKBT2qM3JNmJF1Sb+tfB9AkaYWiNE0IwQMiP8Xl7APQveiztvbtkaHlz8Sd08vxEibOp8W7SAZ1Z83Wi8nsBqjkiDFVVFSCltz982cByDMrvidbhwCgcezYVbc3JxzyM5NL9fk/dYq4d4r+A3TOzXBzAw8RthAkaYwtoSLp8xEd2ML7NCoWgaZXG3EZPWTmtkyQKETZ1fSlawNVhl6/54edWPthZ00IzwQeEMLu9zDN56iThHtB/A+toSwjYKPGzqPLniSz7ZOJsnR1xMtjst5rzP6eb50ZexpHIDc8vW4nW4ODJ/ALlRuVZXb2ZBRSEpDg9j8vuS1iD1fk7ZGoSABHo7BomkRg8ybesyjmo/MLlJCoWiEUpxtxHLKzfGtUTdmoOC2q22intFdVHcNTUhKApU0CMtv+7Y2d0O5oPCGeiRgO31gkaE1dWbuXHOG7xy8BW26w7I7MKAzC51vwf0MDfPe4P55QUIBJoQGItM/j7gj5zSZTibAxXcteBdllRsSOgnt8NvhFleuSlpxa2bBl8XzefDwpnUGkFG5vTmvO5j6OjLbtZ1FYq9ib1acUspWVldhN8I0zu9QyOLsS3J9WZQ4C+xPadLk3aeNNtzGS4flRG/7bmINMh0pcQca+dJ47WD/8YDiz9mZukq21eFLk1WVm9mdfVmeqV3aFL2exd9wO/lBY2+GB5e8ikdfFn8Y/67lIdrE7pI4uHRnHHvvZHcpsHVs19lSeV6AtGytoW1pfxvw2yeHTU+5mWjUOxL7LU+7nllazl1ykNcNmMi1895nRN+uJ8nln2BYRO33Bacu98hMbVF6tPJlx03uuTsbgfj1VyNjmsIBmV2tVV6nVPa8dTISxmU2TWuPA6hMb+8gLJQ/BKxYGVTTi1eauvmCZkRHlv6OX4j1CKlDZZX5diOg5Ma+9nGOSyu2K60wfLp+40wd81/N67vXqHY29krFXdBzVaunfMaW4KV1iafHiJk6nxQOMN2g68tODy/P8d1HILbJhQvw5VCjR60nXdGt9EMzo4tKOXVXGS6U7jngLMSXrNzSg4C+3C7gBHi4aX/45SfHuScaY8zp3SN7bj1taVxY68lsN5fEqNIG5LtTuXMrqO5od+JeDQXjqg8DjQ8movbB45r5GuPx/uFM+JGzGwNVbG2tjipdRSKvY290lXy+tqpRGwiF4JmhPcKfuMvvY4mNUE4XmsghGBC77F8tel3aJDksqRiPbfOe5unR17aaJ5Tc/DkiEuYVryMTzfMJmCEODy/P6d0HkGaK7Gr5+z9DuanLYttlZ3Eirk2gLW1xVw353WeGXlpTKjglkAFy6s2EU6gmL0ON6E4jRtSHB5uHnAqx3SwLOqROX34b8GvrKstpmdae87e7+AY/3xTxHu5gfUFUROJf16h2JvZKxX3nLI1cV0iTk1jTc0WBmd1a3M5Plw/09alEJYG88vXUVCzlf3S8hqddwiNI9sP4Mj2A5p1vUFZXbmw5+G8vmYqEVPHRBIv6CNkRnhq+Ve8eNDlREydfy78gJ+2LMYlHHGfndfh4oyuo5lc8DOBaH2T+hjS5JDc/et+75XentsGjUtKdlOa/FaykqlblqJpGsd0GMyQrP3YEqy0lUc3DXqmJ5fMpFDsbeyVijstgTVtSLPNre1tzC1ba+srBnAIB8uqNtkq7pZSFQnQN6MT1/Y9nlU1mykJWf7qeCyoKMSUJg8v+R9TtiyxGhFjL69bc3BU+0FM6D2WNTXF/Fa6IqYPpldzcdOAU2ISi4JGmDfXTuOj9bOo1UP0zejI5X2OYVi7njFrB/QwV8x8ibW1xQSMMAIrCalfRiecaBjEKm6v5uKPXUfs1M1mhWJ3Yq9U3Kd3HcUTy760dRlETIO75/+XcV1HcUqX4bhbUEtjSeUGnlr+FfPK16KhcWheX67uewJdG6SJ53jS4lq8Ash0p9icaT5SSp5f+S1vrZuOK+pTN6TJOd0OYSrxFbcmBDWRIF9smhf3BSOwvgA0NKZuWcKHWV3594F/4ouN85hc8DPloRp6p3fkkl5HMqxdj7p5YVNnwoyJrK0prnOtzCtfx7WzX+euQafzh05D6sY+ufxLVtUUEY66tyQQMMIsqdzICZ0OZHrJMvx6CE0IwqbBKV2Gc23fE3fwqSkUey57peI+tcsIvtz0Oyuqixp1RzekyfLqIp5Y9gWfbZzDC6MnNEt5Lygv5KpZL9e9FEwMphQvZXbZGl4/+MoY5X1619FM27rMtkO7Q9MY0cDybCnvFvzC5HU/WxZzPQXcVAf4g3P6UOAvwSk0Gjs+LKyUe7OuiuCTy74k3enjlC7DOaXL8Lhrf71pPutqtzbyh2+roXJUh4G4NCe6afD5xrl1Srvh2Glbl/L5UbeyrGoTNXqQfhmdyXD5Et6XQrG3k1RUiRDieiHEYiHEIiHEZCHEbv2N6tKcPDdqPNf3PYk+aR1sIy2CZoTV1Vv4eP3MZq398NJPG1nyEolfD/H8ym9jjg9r14MTOw3D69ge3ucSDrwOF5f3Poab573Jsd//H+OmPMzrq6fUKfgVVUV8smE2X2+az4qqIirDsXHdAT3MksoNFNSWYEqTl1fbZ1s2FbJ3UufhZLlSbf3V8QiaEZ5d+U2ToXifbpht+8IC63nNLy8AoEYPNXKF1Kci4kcTGgMyuzAqp7dS2goFSVjcQojOwDXAACllQAjxX+Bc4LU2lm2HcGlOxnUdyScbZiPjKLCgGeHD9TM5e79DklqzOhJgdfUW23MmkmlbY90SQghuGXAqR7UfwHsFv1ESqmZwdjd6pbXn0aWfETIjSKAy4uel1d/zzeb5pDg8LKvahC6Nuk05h9AYldOLOwaezruFv/Lfgl9xahqGaZLrTadWDyX/YKJYXxmSLcGKZs8tjoZZJiqUFYoTxreNbV8G6S4vbs1pGwUEVmEqhUIRS7I+AifgE0JEgBRgU9uJ1HrMLl2dMIUcaJbSM5uwMu2sUCEEo3P7MDq3D2C5ao7/4f5GFnLI1FlVvcVKL28QRWFIkxklqzhr+mNIKQmZEULRIRv8ZUnLX5+wqXPf4o8IGmEEIu7LzQ4BTbqXjmw/kDU1xba+81o9xKySVQzM7EqmO4Vzuh3MW+t+bqTsvQ4Xl/Q8Imm5FIp9hSZdJVLKjcAjQCFQBFRKKb9pa8F2lGWVG7lp3psJq+cJiNlQa4pMdwpdU+zrVAtEnXJOxOKKDYQN+41AiYwbimciCRhhW5dISytc1+hBdGk2KwtSQ3BoXj/bGt/1Ob3raFIcbrQ40r1X+Bvn/fwkZaEaxvcey2F5ffFoTlyaA4/mxK05ObXzCP7YZWTC6+imwfebF/LIkv/x4srvWF9bmvS92LHNDVVYa1+uQKHYHRBN+SqFENnAB8A5QAXwHvC+lPLNBuMmABMAunXrNrygoKBNBE6G8nANp0/9T5PWtFM4eGvM1c1KCplZuoob57zRyDr0Ody8ctAV9EoQW2xIk7sXvMc3bdDPUUPgEFqziz41hUNodS8Tt3CQ6vIy6eArae/LanLuRn8Z9y36kDlla2xfDU6hcUrn4dw26DQA1tYU82vJChxC47C8/nRKSVxIqjhYyWUzXqAy7MdvhHEKDYfQuKjnEYzvPbZZ92lKk+dWfMu7Bb/giLqh2vsy+dcB59Bfdb1X7ASEEHOklCOSGZvM5uQxwFop5VYpZQT4EGjkFJZSTpRSjpBSjsjLa73Y5Jbw8fpZSdV8vnPQ6c1S2gCjcnrz2PAL6ZPeEUdUURyY3Z0XR1+eUGkDPLr0M6ZsWdKs6zWH1lbaAEhJl5Qcuqfm8eeeh/POodcmpbTBqslySa8j434R6NKMaUrcIy2f87ofyjn7HdKk0ga47fe3KQ5W1TU71qVJyNSZtHYqM0tXJSXjNp5e8TXvFvxC0IxQq1v10wtqS7hi5ots8pc3ay2Foq1JxsddCBwkhEgBAsBYoHHLk92IeeXr4qZlg1U34/hOQzix84EtWn9ETi/eGnM1AT2MJgQeR+OiUA0pD9fwyYbZceOld5SWFn1qCiEEY9sP4sq+xyUcVxKq5qVV3/Nt0UJ0adAvoxOFtSX4jXCCmBFa/Dw2+EtZUVVk61oKGhHeXjudUTm9k1qrRg/yXsFvthuqYVPn7XXTuWnAKS2SU6FoC5pU3FLKGUKI94G5gA7MAya2tWA7Qq4nPeGG24mdD+SO6Of5juBz2lf/s2NhxXpcmqNJRSUQOIRo1H1dw4qpbol6dmtOnMKBEODXQ81aQ5cm8yvWcdlvL1AarmFQZhf+3PNw+qR3rBtTEqrmgp+foirir5N7Xvm6pNZvaWnWokAFLs0Z9wXdnE3bFVVFODWtbsO3Pno0FV+h2J1IKo5bSnm3lLKflHKQlPLPUsrmx5/tRMZ1GYknTtRDqtPDrQP/iCZ2bmFEl3AQ0JuOl3aicX6PwxiQ2YVsVypezUUHTyZE+082F6/m5JTOw7hv6Ll8e/SdHFyvlkgyaAgWV2xgfkUBG/ylfFO0gL/8+nxMKv2rq3+MUdrJ4tFc/G3/xJZ8PLqktEuY7dm9GaUEUhzuhHHpicIeFYpdwV5Z1vWA7P0Y13VkTOKLU2h4HS7+PfS8mLKlAT3MB4W/cdlvL3DZby/wfsFvSSnYhkgp2RyooDhYCUBpqJrlVZvqkmeCRiQpd0YEgyPyBnB615G4HU6CZoTNocoWu0JM4IIehzMmry9OzcG/hpxjW2o2/nwZ4zs3kQTNCHcv+C+RqOL8tmhBs5S2SzjI8aRz/9BzmxXVU5+OvmwGZ3XDafMC9mgu/tzj8KTX6pvRKW7dE6/m4rQmIlsUip3NXpnyDnBD/5M5PL8/7xX+RnGwkoFZXTl3v0PoUi+crzLs55LfnqUkWF0XZrekcgOvrPmRNw6+ihxvelLX+nHzIh5d9jkVYT8SiSYEumnicViJJUfk929WA4dnV37Noor1cWtRJ4tDaByc2yemq3qGy8cjwy7k7/PeREorUcaruUBAtiuNikgtYLlsdKnbpqJvY1bpag7J69sspe0QGmd0G811/U7c4a+efw/9E3+d+RKbA+UEjDBuzeomf+X+xzHEprN9PIQQ3HPAWdwwZxLhaFVFsJR2n/QOnNRl2A7JqVC0Nk2GA7aEESNGyNmzd+v9SwDuW/Qhn2+cZxvr7RIOHhp2AWPy+iZcY1rxUm7//Z2EmYIu4SDF6aYyEkhKrmb0303IUe0H8s8Dzo758thGRdjPl5vmUlBbSo/UPE7ofCDpTi9LKjewrnYr+d5MXlr1fVxfdYrDzR2DTufYjgdw27y3+WHL4qSTeDyak9O6juL6fichREuj0C2klMwqW83CikLSnF7GdhhMrie5F25DVldv4fU1U5hXvpZUh4fTuo7itG6jWlSITKFoLs0JB9xnFbeUksO/vTth9IlHOHnlkCtiNuIarnHGtEfZ4G866cMlHGhCJLzeNjTEDkWJCOC2gacxruuOfeK/svpHXl39o63Mbs3Ju4deR+eUdqyp2cIlvz7XrJonPoeLBw+8gIOSSFpSKPYFWjuOe6/EkGaTER5hqfP6mikArK7ezKNLP+P23yfz34JfqIkEqdaDbA4kF+NrSJMUhwdnE4/cJRw4dtCFIIFnl3/dpHumOhJg8rqfuW3e2zy69LNGdVhO72pZmw1tYo/m5JDc/etcMD3T2vP8qMvon9EZl3Dg1py092ZyetfRda3LGhIwItzx+zvcu/ADllftERUUFIrdhn3W4gY4bcojbAwkDhvr6M3ilC7DeX3NVHRTx0DidbhwCSdPjLiICTMmJuXj9WhOJvQey68lK5ldZt/vEaCrrx1bgpWEWyGZ5oo+x3JJr6Nsz62o2sRfZ76EbhoEzQgOBE7NyaW9joyZs7p6C7f/PpmiQDlOTSNsGhzdfhB3DDrNNn69IuwnYurketL5tWQFd85/N2ELMg2BS3Ny5f5/4NzuY3b4nhWKPZXmWNz7tPPu8t7HcN/iDxO6L9yak0lrpsb4sINGhCAR7pj/DoOyuvJ7edPp/RI4odOBzC8vjDvGKRxc1+8kVtZs5rXVP+3w5uRLq37gsPz+9E7vEHPclCY3zJkUo1ANJIYZ4dXVPzE6t09dfHWv9Pa8e9h1rK0ppjxcS4+0vITNfrPqNYfokZbf5FeNiVU06+kVX3NYfv+YjVSFQmHPPusqATi+81Au73NM3JRsr8NFmssbV7FXhv2M6zKqrutMPBxCY3yvo8n1ZlAejdqww5Amt82fzKcbZnNk/gA6eJNLLY+HLk0mRV099fm9vCCuFRw2dd4r/LXR8R5p+Qxr1yPpDu1ghewNb9ezyecDVuXFzzbOSXpthWJfZp9W3GDFOL9y0F/xaq6YmGCfw82w7J6QIANTIPA6XDw5/BLbZg3bxlzb9wQu7nUkAMOye8RVZBJJ2NTZFCjnp+Il9EpvzwNDziPFYZ8A4mjiP59EMqdsbaPjW4NVmHE8ZCayVWtz3DfkXAZmdmn0fBuiS4PiYFWrXVeh2JvZ5xU3wMCsbnx0xE1c1PMIBmZ2ZXROH+4ZfCb/Gf5n+md0irtZqEuT7ml5DM/pyfndD20UdufVXIztMIhz6jVqOGu/g3CJpi3QoBlhbtla5pSviVua1sCkf0biynV+o3GSq0AQNO0jQJxo9GtizeaQ5vIycfTlvHTQXzmz20FxQ+t8DjeDs7q22nUVir2ZfdrHXZ8cTzqX9zmWy/scG3P8nO5j+Gzj3EYRGk6hsX9GR3qmWRUBr+57PPtndOTV1T9RFCwnx53Oed3HcEa30TGxyvneTJ4e+RdunfcWNXoQI1rRzo6AEaagtgSncNh2X3drTg7I6sayqo1xgwdNm43TRL0oJZKz9zs47vmWsn9GR65LP5Gfty5nU6A85nlua8xwXMehrX5dhWJvRCnuJtgvNZf7hp7LnfPfRWD5oTWh0TUlh0eGXVA3TgjB8Z2GcnynppXPoKyu/O/IW1hcuYEvNs7l0w1zCEt75Z3p9JHq9BAwGheH0oTgj11G8sH6GXEjW9q5Yn3SJaFq1tTat18DSHV522yDUBOa1Qt0zuus95eiRUtnZbpSeHT4Rc0q2qVQ7MsoxZ0Eh+X356ujbuerovn8XLyMaj1An/QOlIVqmrVZVx8hBIOyupLu8vHpxjm2qZIpDjdHdBjAZX3GcsXMlwgYEQJGuM4l8+CB59M7owODsroyv7yg0RIe4eTMBtZzyIhYDRewd7+0RXhoffK9mbw15hqWVW6k0F9Ce28WB2R12+EMSoViX0Ip7iRZW1vME8u+iLo2IiwoL+CTDXMY3+soLopuPLaE/VJzOTyvH9O2Lo8JOXQKB7meDI5qPxCX5uR/R97CtK3LWFtTTL4ng6M7DKqrWnf34LO45NdnCRiRujXcmpMMVwoV4VpWVW+uCwns4MvCq7lsO7ALBAe2sOhTc+mX2Zl+qrOMQtEi9ukEnGQxpMmJP/6b8nDjUD6P5uKlgy6nb0anFq+vmwbPrPiaD9fPQGL5pQ/L688tA8fFxEUnoioS4OP1M/lpyxLWRpv0RqRRl1hzfKch3D7wNIQQfFg4g8eXfdEoTtwbvZf9d+BeFApFy1AJOK3M7NLVhOI0+I2YOu8XztihxgxOzcG1/U7kiv3/QGmomgxXCqnNrAGd4fJxYc8jWFNTzIqqTXWlWLcl1nxdNJ8hWftxcpfhHNVhIO8XzmBVzea6+WlOLw8OPY/9MzpRowf5etN81tZsoXNKDid0GkqWO7XF96dQKFoXpbiTYGuoOm4st4mkKMl6JU3h1px09DXdazEefj3Ed5sX2qbLB40Ik9ZO5diOB3DJr8/V1Q3fRsTQ+XzTPDwOF9fOfg1DmgTNCB7NxXMrvuH+oX/i0Px+LZZNoVC0HkpxJ0GvtPZxFbdLc9TFUm/0lzFx1XdML14GCA7P789lvccm1fi2NSgP1yYsUFUcrOTbzQspD9c2ikIJSZ1vixbw05YlMbHf23zmt/8+mU+OvLnFm7EKhaL1UAk4SdA/szPdUnJtMxUdQuPMbqMpqNnKBb88xdeb5lOtB6nWA3y5aR4X/PIU62ubLvvaGuR40mzjtrfRwZfFlC1LEpZfjZfsI5F8vnHeDsuoUCh2HKW4k+TxERfTPS0Pn8ONR3OS4vCQ6vTwyIF/pr0vi0eXfY5fD8fU0TaR1Oohnlz+xU6R0etwc3ynobbZiV6Hi4t7Hhm3F+c2InGSgUKmTmFtSWuIqVAodhDlKkmSXE86b4+5hoUVhays3kw7dypj8vvh1pwY0mRm6Spbd4pEMn3rcqSUOyVW+cb+J1NQW8Lyqk0EjTAOoeEQGn/sPILjOg4hzell2tZltla3ROJ1uG3PeTQXPdLy21x+hULRNEpxNwMhBAdk78cBDfoZSikxE4RVmlIikXELUbUmXoebF0ZdxsKKQmaWrsKtuTiy/QC6peYCcEje/gzK7MqCisKYuHGv5uLSXkfx1rrptopbE4KTOqveiwrF7oBS3K2AU3PQJ70DK6qLbM8PzOyyw41xm0O8FwxYaedPjLiYyet+5r3CX6mM+OmWksf43kdzRPsBHJK3P3+b9TK6aeA3wvgcbgTwn2EXkuHy7bR7UCgU8VEJOK3ErNLV3DBnUqOmwR7NxRMjLmbYTspIbA3Cps5PWxZTWFtCB182Y9sPUnVEFIo2RiXg7AJG5vTioQPP5+Eln1IcsupKd/Bm8fcBp+5RShusePI/dByyQ2ssrCjkmeVfs6CiEKfQOLrDQK7Y/zjaezNbSUqFYt9FWdytjJSSraEqBII8b8auFqfNKQqUM2nNVH7euhyX5uCkzsPond6BO35/J+brw4Eg3eXj7THXkLsPPBeForkoi3sXIoQgfx+xKldXb2b8jBcIGZG6hJ5XVv+IIc1G9csNJDV6kFfX/MTNA07dFeIqFHsNKo5b0WLuW/QRfj0Uk4UZNvVGSnsbujT5bvPCnSWeQrHXohS3okVUhGtZXrUpbuedeCQKm1QoFMmhXCV7KIY0WV61Cd006JvRCU+DfpdtTSCa3BOJkyJvh4bg0Ly+bSiVQrFvkJTiFkJkAS8Bg7B6tVwqpfy1LQVTxOf7ooU8sOQTIqaBENaG6OV9juVP3cfsNBnyvZl4Ha5GNb23IRAxmaQC8Dnd/KX30TtJQoVi7yVZV8kTwFdSyn7AEGBp24mkSMTs0tX8c+H7VEb8+I0QtXoIvxHmuRXf8OmGnRfJ4xAal/c5Fq/W2NL3aC6u7ns8XVNycEZT7g/M7sFLo/9Kl5ScnSajQrG30qTFLYTIAA4HLgaQUoaB+OXlFG3Kcyu/tbVyg2aE51Z+yymdh++0/o1ndBtNxNR5YdV3SGnVOkl3+rhz8OkclNuH87sfSo0exKU58DpUAo9C0Vok4yrpCWwFXhVCDAHmANdKKRv38VK0OcurNsU9VxXxUxGp3ak1s8/tPoYzuo1mdfUWXJqTnmn5dS8OIazYbYVC0bok4ypxAsOA56SUBwK1wK0NBwkhJgghZgshZm/durWVxVRsI1FZVinBq+18y9alOemX2Zle6e1Vt3aFYieQjOLeAGyQUs6I/v4+liKPQUo5UUo5Qko5Ii8vrzWhfFdWAAAgAElEQVRlVNTjxM4H4hKORsc1BMPb9VA1RRSKfYAmFbeUcjOwXgixLY5rLLCkTaVSxOWy3seQ782MsbxdmoN0l5dbB47bhZIpFIqdRbJx3FcDbwkh3MAa4JK2E0mRiAyXjzfGXMWHhTP5YtNcdNPk6A4DOWe/Q8jxpO9q8RQKxU5AFZlSKBSK3YDmFJlSKe8KhUKxh6FS3hV7NDI8C1n7MujrwNkTkXopwt200SKlhPBvSP87YJaCexQi5U8IR+zGujSrITIXcIJ7BEJ42uZGdnOksRVkCBydEDuxm5PCHqW4FXssZs1EqHkGCFgHjLXI0HRk+nVoqZfGnSelRFbdBcH/gYzOjfyO9L8K7SYhXIOtMbVPQ81EENuyQ01k+u1oKWfHWTcEwS+R4RkgshG+0xCuPq13w7sAGVmIrLwT9NWABloaMu0mtJTTd7Vo+zRKcSv2SKS+AWqeAkL1jwJBqH4M6T0R4ehgPzk8LVZpWwdBhpHlV0LeFKT/Tah5yVpf1rtG1f8htTyE96hYeYyNyNJzQNaA9AMOpP9NZMpFaBk32t+DWYP0vwGBD0GGwXMYInUCwtmt+Q+kDZD6amTZn6P3E8UMQtU9mEi0lDN2nXD7OOqbR7FHIoOfA/Z1v0FC8Iv4c2vfaqC065+sQobnxVryMQSRNU80nlZ+NZgl9ZScAQTBPwkZ+jl2bGQRZun5yOJhUPMYGAVgFkHgA2TpH5GR3SPaVtY8AzJocyYINQ8jm1EZUtG6KMWt2DMxKwD7yoQQRppVCeYWJ1jYAcZ6y3KOh74y5leprwV9FfYvkgCy9pXtY8PzkaXnQ2SWzVgDZC2y8o4E8u1EQtOJ+3I0A9ZzUuwSlOJW7JEI9zAQqXFOpiLcQ+NPdh1IXC+hDINrW/XieBdPif3dKKrnB7fB2Lh9+ap/Ym/J10NfiTQSvVx2EiJRFq4JcTZqpZTIyBJr49hM8AJUtBiluBV7Jp6jQGsHNEz/d4KWD+7D4k4VqRcDdorWDe6D0Vy9wHOEzdrRMb4zG1yyu6Xw7a8GTmuDUpo1oC+LK9d2wsiSEzAr70WaZUmMbyN8p2L/nADhs55zA2R4JnLrEciy85Dlf0UWH4JZ9QAyTjs7RctQiluxRyKEE9FuMriGAF4Q6YAHXEMR7d5C2NRzqZvr7IbIfg5EpmW1i1RrrnsUIutRa0zG3dEXg7feTB84uiHSroxdz9EJ3COxV3IeROr46L83I9lNVkNgMrLkVMzaV5G1byD1VcnPbwVE6mUg2sWRz4+suj32kL4KWXYZmJstX7+sxvLzT0ZW/6ftBd6HUJmTij0eqRda/lZH12ZFZEgZgfAMy1/uGoRwdo89b1Yh/e9B8EsQTvCejkg5FSG8jdcyK5Bll4CxNmp9OwAJGf+ICR80S05q5CNvGid11r/nMETW44iEbozWw6x9A6r/Deg2Zz2IvK8Qjs7W2Iq/Q/BT7P3iXkT+rwgtjntLoTInFfsWwtkN4RnT7DA6IVwIz6EI38mNlDaA0DLQ0v6Clvs+Ws47aKln2ypta2wWtJsMrpHEWNb+d60Xy7Zx6XcRa8Ung44V9hiC0HRk9UPNnJ8YGVmOWTYBc/NAzM2DMMuvROprrJPhX7FX2oDQIDRj++/hmcTdzBTO6AauojVQiluhAKQ0oxtq8zBNP2btJMziwzE398csPgyz9rWm/bSVN0L4N6xQwBAQBn0xsvTsuk064TkIkf0iOPtjWdEaOHoByVrQQfC/h7QN02s+MrIEWXYOhKdgRemEIfQdsvRMpL467gakNVnEbspqCRp4SD3xeUWzUAk4in0OaWy1Yqe19ghnV2RoKrLy1mgMthaN8RbUWZrmFiupJ7KgzgfeaE29EEJTiU0IAssCDSADHyBSLwJAeEYjPJ8gTb9lteJEFh+cYIOzAUKAUQytkKgjq+6LTbCxjlphidUPI3znIEM/2owBMMBz+PZffX+C6gcBm5eKowM4eu6wvAoLpbgV+wzSrEFW/t1SsMIDMoTUcqNx3XHcAXUEIPgdMrIE4RrQ+HRkDgiH/f6jDFjXjCrubQhte1ihTL8Jqu7DVuk1Wk+PbpyCNGuRgY8h9D0IH8J3OniOTLg5W7eMDFty25+F0FRk5jPWBnB4XgPZfJB2NULLrHfoNPC/AUYh25+nC4QLkfVI3O5IUl+FrH4cwtMBAZ6xiLRrEc6uTd7DvopS3Ip9Blk+HiKL2JbeDoAZv4dnY8LI4Hf2ilukkNDzKINIY5MVgWKDlnIOJh6oup3ELxEXeI5CaGlIYzOy9Ewwq9kWGy5DP4N7CGS/hEgUW26NbuK8aRWUyn4JWfsa+N+0NnKdPRBpVyG8x2xfSV8PZReArIrKH1XSrtGIzH/GVcIysgxZdm40QzPqigp+Zln5OR/tNun/uxvKx63YJ5CRhaAvBZJ0R9ivQlyl6jkcy7cdh8gi5NbjMMvGx83qFK6eJLalHKB1RGT+y5Km8jarsmFMQo8fwvOQ/rcSrBO9nvCA0+YltA33KIQQCOFCS7sMLX8KWof5aLkfxyptKa2XorkF6nqIS+snMpv4pQlAVv1f1A1Tf4wZddU80uQ97Ksoxa1oc2RkIWb1U5jVTyEji3eNEOG5lothRxA+hOcopAwj/e9hlpyBufV4zKp7wSyD9HuxIkbsXAIBIGSVki2/zH79JmuUmGAWIcuvwgwvhPAs7F8WQctlkcwtZdyGfZSLD5H+96TWIDLfUtq2CtqwCnbZIGUwgavGtNw/CluUq0TRZkgZQVZcA6Ff2LZpJ2tfQnoORWQ9gRA78X8/kUYiy69pLOtU4oTS88BYub1QlX89MvAhot0b0O4NZO1zEJoNVNPYHRGGyDJkZBHCNSjmjFW0qeHmZswIIGIpu7JLsM/sjGJWJHVXwj0S2r1sbVLqy62DrkGI9DsbyRcXY12CkxHQV9ifajKbUmVbxkNZ3Io2Q9Y8D6GfsaxNk20RFoSmIWsntv71jE1WPe7I9gSXugQz7zEkdGXEoGFlQXqxwvR8oLWHyAIoOw/0BQ2qC0asT/uKGxHuIWjZz4N3DPF9yCaEf4+VXYah9tkEcxrMJ0j8IluAc/8k1rEQ7pGW+yN/FiJ/NlrOewj3kKTnE03AiXMSHD3sr6ulgLNX/KmupHJR9kmUxa1oE6SU4H8d+yiJINS+Dml/a51rmVXIiuutBBDhBhlBipSoa6QKqeVB6ngsC7Up5S3AdxYi7W9IKSD0HdS8EN3EbMLVYhQh9bUIZ49oqriGvdUokf53kbWvg6wAXODsHd1kTJYwiJyoT7nhM3ZAynnNWMtCtDTO2jUCRFb0ZdbwxeNCpF4Q/5rptyPL/0rje/Ai0m9qmTz7AMriVrQRkcSlUWU5ckd9znVLjY8mvoS218eQZUB0E9DcCtWP2RZFisULmU8jMv6FrJ0EJcdYcclyC02HC2JlB0pL+YqUM4ifVBMGYzmYBSArQZZA5DeSCgWMwQWeo7F1mVTegQz+0Mz1WoYQAtHuJRDZ9SonugEPZNyBcPaOP9dzCCL72ahV7sZ6ifVDtHuleVb/PoayuBVthMvyK8s4dbFFVqv4uK1okeUkdBsAEIxGYHhprCA18J6CSL8R4eiAWfsW+N+m2REoUo9mQYJwDUKmnAuBd+M3bdhR5NZo0k/DLwkDMKyvkPypsbHWbYRw9ob8n6Kt2xaCIx/hGxe/C1H9uZ5DEXlfI41SEAKhxSlspahDKW5FmyCEQKZcCLUvYfcZ3DAZpcWE5yexybUNCSnnQmShFQkBVkxy+u0IzxhrhJRRX3Nzla0PUs5HaKnRRsS/WNmNjl5g+kFKkAW07oabAdRgH8VinZeBTxGpf27Fa8ZHCC/4TkP4TmvZfEdOK0u096IUt6LNEGlXIPUlVqGiur6NHvCMQaRe3joX0TKiLopE0RjbkAhHPiJjMlIGQBo2ft1Q1DJPlhTAgJRzEek3WjVPKq6NWsLblL+T+GGCySCwNkvjfQHE29AMQ/V9mCINLaVlylSxe6IU925ObZWfDSuKyMhJo2OP9rtanGYhhAuR/bzVKTz4PSAQ3rHJh5klg+dokHcmOdgJniOjsvni6NGonzVhWF699ZCQdiUi9XLrKyPwYQOlDZZ/fEc6wTiwXhAtSR4yoeoOpKMdwnPEDsig2J1Qins3RY/oPHfDa3z18o843Q70sE63/l247a1r6dYvUfjV7odwDUa4BrfN2loaMvN+qLwdS7HFc0V4wXN0wo2y6IogvEla8Lr1U/McUqQhUi+wNjUTulniRZokwDnQqk0S/olmNWOoJ6esfgzchwPSSmNX7NGo/4K7KY9NeIGvX/mRcDCMvypAOBhh9e/ruG7MnVSWJGiEuw+i+U5G5LwDnuNB62SVTHUfHe1s47A63aRdjshKIoVaX5Sk0q5PAGqetKJkmnSzSOoiLpIiDZH1ICL9yjhzPOAc2fQy+hLkloHILf0xS89Ehmda0uirMCvvwiw9G7PiFmRkUWOJZQAZ+Bzpf8tqdtwGzVcUzUNZ3LshJZvK+OndnwkHYyMlpJSEgmE+n/gt591+Rtz5ekRn5hfz2LyumM59OjLiuCE4HE1Xi9uTkeE5EPrRqtBnVIG+BlIuhtTLEVpq3Mp0jdBXW6VWm6ubZNBqCuzsD+Et8ceJVMh8FCErohuhr4Jh14cyFVIuQKRehHDkWoeyHrPKz2JE5asFdDCSLSMQDWmMLECWjbc2j/2TsCJyDOt48Etk+o1o0c1jGfweWXkjIEAa1j+dvaDdywgtO8nrKlobpbh3Q5b+thKn29lIcQOEA2FmfDEvruJeOXcNtx1/H5FQhEhYx+l24Evz8eA3d9F94N5ZJlMGv4Xqh7Hit+ud8E8CRx4i9cLkF9M60LJNxIgVw+waEnVpxBM2gnD1RzjaWwk4tuniDnCPQKRdHrN5KrxjwfMrROYhaybWi123q5XdFEHwv0jsA4tmZFY/gvQcDUSskMKGUUH6MmT5lYict1twXUVroFwluyG+tMStrVIzfbbHA7VB/n7Mv6gsqcJfHSASihCoDlJWVM7NY+8hEm4q1nnPRNY8gX3ySgBqnq3rXCNlCBn8xspajCywX8w9KlrXpLkYVs/J2qcTjHFaCtkR3WSOm1lqQHgKsngUZtklVsnUKEI4o1Z9VGnvEPFT8mXgI+vFYhsfr0Nk4fb2ZoqdTtKKWwjhEELME0J81pYCKWDIkQPiftp707ycOP4Y23M/vfMzesQ+wy8UCPPLJ3tpA+dECkRWg6yyapgUH4ysvAVZdR+y9M+YJacjzbKY4UJoiOyJIDJAbHtBOqM/TVjixgoSptQ7uiAy6/nZG1y7MTqEf7XaiBn1fOdGYWzLsEZoWPVVOpC8L70+ETBLILKYuPcjXJZbSbFLaI7FfS2wtK0E2ZuRUrJhZREFS9Zj6E0XOnK5Xdz0yt/wpHhiFLg3xcPgQ/tx8Kn2xXfWLCggWGtvhQWqgxQu3dCyG2gmoUCIdx74iAt6/o3Tcy/hjpPvZ9nM5nY2bwYiJeFpaZQhy6+0UvDransEQF8erZPRYDlXf8j9EdxjseKvtzUG2JFNORdkv1KXZCKlCdp+ScyL1qauXxpVy0nc5szZx6qbnT8VfCcTv4pgnBeRSEG4h4KjU/wxmEmUEFC0FUn5uIUQXYCTgPuAG9pUor2Mud8v5LHLnqe8uAJN03B5nIx/8AJOuHRswnmHnjaaR6fk8vZ9H7J81ioyctMZd9UJ/OGiI+NuNLbvnofb67L1jXtTPeR2bvtU4nAwzPWH3UXB0o2EA5ZymfXlPOb/uJg73rmeg09pg4pvKWdD7SQaxzk7wTPWSju3/eSPREusrkC4GlTTC7wWrQcd3D52BxFaJlJfi6z6N4SnYfmUk3khhCHwJabwQfBz65CWFy181XCuD5FSLys15W8Q+AprI7M+Hit5ySwj1qoWgBe8JyIcXZGhKdiGN4p24DqgCbkVbUWym5OPA38H0ttQlr2O5bNW8Y8/PkDIv12hBGrgmWtexeV2cswFiRMi9h/ei3s+vDnp6409/3BevfMd+5MSDj/r4KTXainfTppC4bJNdUobrGzvUCDMI395lv8WvdjqES4i9UpkaJrlQqirC+IDLQuRcTey4mriFokSDtBXQj3FLc1KqJnIjvuQ6y4C7pFgVlitxmQNdlX0Er4czEKoeaqeTG5rXTzUvVxECrgPsXo/Aqb/Y6h5BOuFtu0FIcC5PyL9VnD2sr449LXWc8AELQ+RPdFKX3ePRKZeArWvUBcjL3yAlViVdKSOotVpUnELIU4GiqWUc4QQRyYYNwGYANCtm+oTB/DaP96NUdrbCPlDvHTrWxx93mFoWuvtD2fnZ3LrpKt54MKnkKYkEorg9roQmsY/3r+R1IzELoXW4MuXfyDkt1d4kVCE5TNXMeDgvq16TaGlQs4HEPwc6f8YiFgWo+80q3aIoxNEEli2jrzY38O/NiONPhnciIx7kTVPRV01NnKIjKhCj3fNaLLPdiEBBzj7gpYFWhrCdwa4xyCEwPS/a9N8WIBIRWS/UNf7UuR+jIwstaJbtE7gOiBGIWvp1yG9xyH971lVFt0jEL7TEZqy4XYlyVjcY4BThRAnYjn8MoQQb0opY4rsSiknAhMBRowYoSL0gUXT428JVJfVULa5gtxOreu+OOyMg+g3ug9fvvw965dtpMfgbhx/6dG067BzYm5DgfjKTghBKNCStO2mEcIdt8CRSLkAGfwG2wgOkboTCvZLyxoO/UDcrElZA+m3WmVkky7vaoC+FNF+YYyylTKyPTyyoRwyiKx5CZH5j7qjwtUfXP3jXkW4+seMV+x6mlTcUsrbgNsAohb3TQ2VtsIel8cVd7PQNCUeX7x6zTtGXpccLrz77DZZuykOOnk4G1cWEQk1dk3oEYP9RyToeNJGCPdQZOr4aKXCbWnxXhDO6Cd/g68e98HN6E/pBEdPMFYRP5XdCeHpTcvpPcZyX9Q8YVUwTKo2SQTLEq8XZaKvIn50i241h8BSxFa0SgS09sr1sQeh4rjbkLHnH4bTZe/P3X9EL9KzW9hxpA2RUlJVVk2gpmU1pMddfSJun5uGOsCT4uH0a0/cKe4aO7T0a6y0eN+51oZl2lWIvO9tC14JLRPSJgD28fKxgz1WKr1rdFMDrevG+5NzdAAtH+E5CC1nMiL9ZpIK5XN0QzQMDRQuEtdDcSLD8zFLTkFuPQK59Vjk1iMwA583fT3FbkGzMiellD8BP7WJJHshF9x1Jj9/PJOK4ioiIWvjyeHUcPs8XPf8hKTWME2TRdOXUb65gh4H7NemBaZ+/ngmL9w8ia3rS5FSMuCg/bn6mfH0GJT8nkVOx2ye/OV+Hr7kGVb/vg6ny4HQBGff/EfOu/30NpM9GYRrACLznuTGpl6F1DpAzdNgFsUfKE0r3d4oTLCaAZ5DEa4DkaHvon7u+orVi8i4J9bilVUkZXHblcd19Ir6zO1evm7wHIosu5CYaBFzM1TeholA853Y9HUVuxTRFgVjRowYIWfP3kuTPZpJVVk1Hzz6Gd+9MRU9ojP65OH86dbT6Niz6RKty2ev5u5xD+KvDiCEQI8Y9B3Ri3s+upmMdq27OTT1/V956KKnG/mgfek+npvzIJ17d2z2mmWby6mp8NOxZz4ud6KEkd0bs/o/Vk0RO0UqMqKbmPF8+z5IvQgt3Yqilfo6ZPUDEJoCmOAchMi4BeEeFTNLhqYjK65qOp09ZQJaRuPejDI0BVl+NbF+bqcVRujcH8JT7NcTeYj86cptsgsQQsyRUia14aJcJW1M0eotbFqzhbTsVAYd2o/jLz06KaVdsbWSm8feQ+mmcgLVQatCYCDM0hkruOuUB1pVRiklz173qu3GYcgf4s1732/Ruu06ZNOtX+c9RmlLqSONjVY4YD1E6l+spBca3ocnatXGUdoiGzLuRKRdv/2Qszta9vOI9osR7Zeg5X7QSGkDVliflsTLMvCh/aU9RyDavQbu0ZacIgN85yJyP4LwrPjryRKkkeALIwFS34BZcQPm5sGYmwdgll0Yv7SAYodQRaZsqK3yozk0fKmJa4aUb6lg+azV+NK8DDq0Hw5nrD/70+e+ZuLNkwgHI0hTsnZhITO+mMsl//cnzrju5ITXv+uPDxKobhxdoIcNVs9fx6rf19J7aI+W3WADNq8rpqaiYYKGhWmYzPxibtJrVZVW8/nEb5n73UIyc9M58bJjOHDs4N3agpNSImtfgdrnQFqbfdI9HJFxH8LZFaFlQe5HyOpnIPipZWG7h1mbmLXPR+fY4B6KlnKW7ammamILoUG7N5FbjyZxfe/4mbjCPQzR7o1Gx2XChB8J+gJwdkooX6NZxiZk6WnRZslRN1D4N2TpBdDuFYQ71pCUZhnS/4715SHSEClng+cYhNi7q1i2Fkpx12POt/N57obX2LDcsjj6jerNlU9eSp9hPWPG6RGdJ654ke/fnobL7QQJDpeDW16/itEnDQcsN8ELN74ek8UopSTkD/PK7W9z2Omjye/WIH4YCIciXDvmTgqXxE9PF0Kwat66VlPcDqeDRB4zzZHch9m6xeu5/vC7CAcjdQk4M76Yy5jTRnPL61fttspb1jwRdYXUU5DhmVayTN6XCK2d9ZN5F2TetX1e4LOEzy2uQk8S4chBpl4GtU/GH+Tsh5RG8xSeo3M0CibOdfV1ya8VRdY8EY1Db7gpGkRW3YPI3V7iSOqrkaXnxLiYZGQOuIZB9sRWaSK9t6NcJVFmfzOfu8c9RMHiDRi6gaEbLP5lOTcccTdrF8VuPD13w2v8+M50IsEI/qoA/uoA1WU13HvOo6yevw6An979Je61TFPyw2T78LAp7/7ClnXFCYvVaw6NzNzW83Hndckhr4t9o1aHy8HhZx2S1Dr/Ous/1FbUxmRNBmtD/PzRDKZ/NBPDMJj2wW/ccdL93HDEP/jvI59QXb4jLb12HGnWQO3LNLZqTZB+ZO1b8Se7RxE/29EHnuN2WD6Ren7iWiyReciKK5Gy6Ro4dXgOS3TSSuhpLsHviGv962tjinnJihujlnk9F5P0Q3gO0m/v+lHEslsobsMw+GHydK499E4u6X8t/xn/LOuXb9ypMjx77Stxfbyv3jm57vfaKj9fvfyDbUZkJBjhnQc+AqwEG7uaIQB6WGf57DVcffDtnOj7E6fnXsLzN75OTUUtP0yeFjf2extCCEYcN6Q5t9fketc9PwFPirvBcXC5nRx93qFNrrFu8Xq2FpbYWqDB2hAfPfE5tx33fzx08dPM/HIeC6ctZdLd/+WSvtdQtCZB44G2JjI/QaW9EIS+QxqbkbUvY1b/Bxn8zup0AwhHPvjOpHHYoBO0bITvlB0WT2jZiJxPrUgRW4JWpmfwy+TXTDkfK2U+Dt7jmyWjRaLww21NGCw/uFVV0M4wCUCgsWtH0ZhdrrhN0+SfZzzCYxOeZ8kvy9mwfBPfvD6FK4bfwtzvF+4UGapKqylaW2x7TkrJ3G+3b7AULt2Iy2P/h26akiW/rgAsN4sv3d5H7vI4mfG/2SybsZJISKe6rIZPnvmKq0bfhh5OnPjhcju57c1rmPbBDN558GOmfzQjbinX5jD0qEE8/P09DD68f11BOCkhEtb5+9h/MvHvbyT8CijfUoEjTsw6wPrlG1ny28qYl1IoEKa6rIYHL3pqh+VvMaKJJCizyopzrn4Mal9AVt6ELPkD0thsTc/4B6RdDiIdK7HYbfW2zPkAobVOzLpwdkNk/Se6vg0ygPQn+DKwWY+0a4jtPK9Zv2f8y/LpNxf3oSSOUY928ZEViUvSmhXNv/Y+yC53Jv388Szmfb8w5g/aNExC/hD3n/c4726a2CpFiUKBEJMf+JjPnv+G6vIa2nfL5ZQrjuOM609GaIJEzkqhbffNprdLS6go09tZSTUjjh9Kdn4mIX8Y09hujQhNoIeNRkpQD+uUbChl/+E98aZ6bK1uzaFxzbOXcf/5TyBNSTgYxu1z4/F5ePj7u3e4w03/0X3I65KDy+2sy3w0IgZGxOB/z31Nv1G9OfxM+0JV3Qd2jfuFoTk0wsGIbQ0T05SsmL2G0qJycjruglZYrqHEL13qAXMLMTVCpB+MELL8b4jcD6363Wl/Q6ZOsPpNivRWU9gxmFGFJ+OkwzdT4WlpE5DuEdEuPAXg7ItIvdhKf28BIv06ZHiaTfiiF5F+x/b9DUf3BL5/oSoOJskut7g/f+GbuK6BSDDCkl9W7PA1DN3gusPuYvL9H1K5tQpTNylaU8zEm99gXNZFLJiyhO6D7ZNMNE1wyB+3N2Pt0qcjHXvYh/O5PE7Gnm/5Dx0OB49OvZf9R/TC43OTmpmCJ8VN5z4d8aTaZ8SFAmEKlm4gr2suTnfsO9WT4uGM607i2etexV8VIFATxNBNAtVBKooruXnsP5u0vMPBMJ+98C1XH3QbE4beyKt3TaZkUylfvfojlx94E2d1vIwf3/nZNl09WBuqcwPZkd0+i0PGjcTtbWxNuTzOhBucTreT6rJd4+sWwgUZ/6KxNevBsmvsnqkB+iqkvn2DTwgnwtG+bZQ2gKtfghrczqi/vXkI9zC07CfQcj9Gy3qwxUobQDh7IdpNBtdI6hpPOHohsp9GeI/aPk5LA99Z2H89eBBpjeujKxqzyy3uqtIEf7ACaitb0k8vli9e+p5Vc9fangvUBLn//CeY8NCfeXHZmzFWoaYJfBk+Lr733Jg5t0++jusPs6IntmVEgvWCeOWOyRQs2cD1L1xOTsdsnvr1fgqXbaRo9WZyu+Tw3RtT+ODx+KnFpm7y5C/38eItb/Ldm1PRwzrtOmbx53+cRaAmiGna+xJDgRAzPp/LmHGxf8BrFxbw6l3vMO/7hdamoRB1XwAblhfx34c+QXNqhANNR0DEcydt46aX/8Z9tY8x97sFOJyOui+VWyZdzafPfs2cb+bb37Nh0qHHrivKr/uzj/wAACAASURBVPlOQjrykNVPgL7Yal3m2A8iCZLIhBP09eDsvVNkFFo7pO8UCHxGo+JRwoVIvXSnyJEI4eqPyHkLKQMgjZh+mTHjMm5D4rfuRbiwvngEIvPfCNfgnSrznsouV9zDjhnMukWFRGx8u5GQTp/hPW1mNUZKSfmWCjSHRlZeZsy5phJIwoEw3781lcem/ouXb3+b+T8uRnNqHHLqCC6977xGFnaPQd14ZenjPH/TJH6a/HOd28M0JKah89O7v5CVl8H4B6xaXN36daZz7w7cdvz/sfjXFUjT3i3jcGpkt8+ktKic61+4nGufuww9rDP3uwVM+ud7rJq3Nu7ccCDM+uWb6n43TZOvXv6Bp695uYEFvX1+3UsnklxEQjzf/ja8KR7u/fRWNq4qYvnMVaRmpnDgMQcgBKyau4Z53y+McRsBeFLcnHLFH/CmtKTFVush3KMQOZafWIamIcuvImGDA6mDc+c2XxYZ91idc4KfgfBg1cdOQWQ9jnAm001n5yCEL2GXNyFciMwHkGk3QOR3q0Kje6RV4VGRFLtccY+7+gQ+fe6bRorb43Nz2BkHJeX3nPH5HJ659lVKNpYhpaRrv05c88xlDBrTD9M0Kd/ctP9vxew19BnWkwe+ujMpudt1yKa4oMR2wy7kD/HJM19x4T1n4/Za/zNO+e8vLJ2xMiZUriGGbjLvh0VcMezvHP2nMdzw0hW8fPvbfPzUl40UXkPcXjf53awNoBVzVnP3uIcoLSqPq+hbQjwfdkM69+5YlyJfsqmMa8fcQXVpTcw9OF0OEIKjzj2U8f8+v9VkbA1k7YskTnrRwNkLsZOs7W0I4UZkPYA0bop+GWRG62fvco9nixCOfHD8YVeLsUeyyxV3buccHvnhbv7vnMesyASng0gowpHnjuHa5y5rcv7ML+dx79mPxoTyrV1QyK3H3cujU/5FzwP2ayJTzMLON9sU6xYlKCwkBFs3lNYpsM9f/K7JML9tREIRfpg8nbLNFcz66vek5ghNMGbcyGiq/D/xV7Wsul8i4jVISMT9f3qckg1ljV48QtO45tnx/9/eeUdHVXV9+Dn3Tk1PgFATaui9gxSlCAKCoAgq9v7aUBQVxc6nWLDga31RbCgiIIhI79J7b6EmtNBDytTz/TFJTMhMMglpg+dZK2sx9945d88k/O45++xC73u7FZV5RYczr4JRgBaFiPy8ZGzxgtDLg5539yTF1U2pCzd4WnR9t288B7cd4eKZZGo2iSW8fJhf7/1ixEQf8dd2JoyaxNi5o4mpV5Wju33HhetG3a9Y5csJjQrx6YN32Z0c3HqE3Wv2U6tpbIGF1GFz+i3aulHnrT9exGw1M+2jP3H66fooKNXiClZo6tTR0+xZt9/rasFhczD3m8VlU7gN1cF+zNdJiPoNoVcqUZMUiuyUCeEGTxJIraYF89OlXEjhWLzv5I2tS3cC8NC7d/LmkHFe3RQGo05EdDh3vzE017n8GPBYbya+8kuuZBxNE7il5N17/wtS4na7CYkIxmg2eI3YuFKadW1Ik86eiIBty3fl6Y4pLGariTtf9V53wxenE89iNHtvXgweYS+LiOAHkfbN5HaXGMDUHq2AdTwUiqImMJ1jGWi6luf+UWbRp/b9WjHif48QXj4Uc5AZ3aAhNEFkxXCGvjiQL7e8T2R0uO+BfDDg8d40aFcXS8g/oU1mqwm3lLgcLtKSPWF7tlQ7509dwOXM209dGAwmnTot/tnALVc5MkfceVFxx8s3+4zh9kXlWhXz9IvHNii+2uJXgjB3ymikYCarIqAIAkPNjEQYhaJ0KTMz7sJgDbFSv10cO/7eneucpokcoXHdbutM11s7cmj7UQBqNI4pcGLP/s0H+d/zP7J5yQ40TdD2hpY8/ul9JO49wdyJi0lPScdgMrBp4bZcM2uX043RYqRSzWjOJJ5FM2ikXUrP88HjD0Jo3PjoPxs8fR7qyeLJKwvlj/aFyWqizQ0tCvy+yOhw2vZpwdrZm3KETYInLn3oC7n7Q5YVtJDHkJYBkP4XUiYjTG0yGvEG9FxHcZUQ8H+Fj4+/D2uIJccsU9M1giOCuW/MbVnHnA4nSKjdrAa1m9UosGjvWbefpzuNZsP8rbgcLhw2JytnrOPJDi8R26Aqr08fydh5r+Byun26Q3Rd446XbubLLe/TuldzTPmE1/lDm97NqVTjnxjoBu3iGPBYL8xB5qz2YQbjlT2fhYB1czazYvoatizdgcvlvw/9uW8fo07LmliCzRhMOpYgM0azkfv/73aadW10RXYVN8JQDRHyIFroMwhzZyXaijJDQM+4Aeo0r8n4NW/z3SuTWT93M5qu0WlQO+56dTDRsRXYtnwXX4yYyL6NBxFC0LxbYx4dd0+B08M/ffIb0i+bxUopSbuUzoRRk3j1N08XkgpVoxCa8BqG57Q7WT9/C1VqV2TNnxv9Dq/Li2PxJ3Ide3DsnbTr24oZn/5FUsJZKtWswMqZ67ClFM73bU938N2rkzFbTUgJZquRlyc/45fwBocF8fGKt9i9dj/bV+wmKNTCNQPb5oq1VygU/nNVty7bvHg7L/d720s7Lgv/XfsOMfX887GmpaQzMPIeXE7vM02j2cjstEmAZ2Y+4rpXvVYPBE/YnsGk43ZKn+MVBHOwmVnJP+Z5TfyWQzzdZbTXxgyFxRJs5qutH/hM/1coFAVDtS7L4NMnJ3gNFUxPsfHt6F/8Hyifh1v22XW9NnW4+el+mINMXjcJpVviSHcWiWgDBIfl34m8RuMYTOaizUpz2l1M/9iTui+lZMuSHXw18ge+eWkS+zYeKNJ7KRSKnFy1wn3xbDKJ+3K7EcAjnutmb/J7LGuIleoNq/k837Bj3Ryv733zNt5b+Brd7ujsd/eYwmCyGLnhgR75XqfrOo+Pvw+z1bt4W0N8t2jzFaHidDjZvmIP6ak2nun6Ci/f+A6/fTCTyWN/5+kuoxlz+0c+66ooFIor46oV7nwpYMTcfz6615Om7YX4zQc5e+JcjmMN2sXR98Ee+YqXuZA1OsxBZqrVrcKQkQNyHE9PtXFw+xHOHM9pz7VDruGV356lVtPqCCEwmAxcd1snxs4bzWvTR/LBktepXLsilhALQWFWzEEmYhtW8yn2ABEVw/nquR/Yuz6e9JR0pPSUabWl2lk1cz0zPvW/uL9CofCfgN+c9EVYVCgx9apwcFvu9GVNE1m9If2l0TX1MJgMXrMSbWkOpn74Jw+OHZZ17MjuREb1+b88w/0iosMZ+sJNrPlzI4n7jnM6MXdq+OVUqlmRiOgwet1zHT3v6oLZ6hF+l9PF/178iT8+n4emazgdTuq2rMXz3z+B0WJk7/p4QiKC+WzDWIQQWT/Z+W7vePas209Swlli6lUhtkFVbo99xGuqviXYTN+HevD2HR973WS1pdr4bdwsBj7ZN8/Po1AoCk5ACvepI0n8+dUCjuxJpEajGPo+2IPyVXP3THx8/P2M6jMmx0ahEAJLiCVXqdb8OLD1MJrmfYHitDtZPnV1DuH+7tXJecZSm6xGBj/bn5uH9+Omx2/gxtA78xVtgEo1K/DegldzHf/40a9Y8OPyHPHSO1bu4b6Gw0HgCT2UYLQYGTVpOC275y6fKYSgfts46merDPvKb8/ywvVv4rQ7swqBWYLNdB7Ungbt4vIsYnX2uOpmolAUBwEn3EunrOLdez5Futw47E7WzNrAlPdm8vLkZ2jfL+csummXhoyd9wpfj/yBnav3IoSg7Q0tePDdOwtcd0PTtHwb+GZn/dzNeYpa1bgq9H2oO+BpoOBy+pcKv+PvPbmOnTl+jvnfL/W6GshshebMjC1PTuOVAWP5fMNYv6JqGravy4SdH/H7+L/YumwnEdFh9H+0F617NcfldOXpwy9fLcqvz6RQKApGQAn3uVMXePeeT3PU4shMdnlr6DgmJ35FcHhwjvc06liPj1a8hdvt9uoe8JeaTWMxWYyebMfLMJqNuYpU5Zfgc2z/Ce6o/h+e/voRFvywDLfLv7BMoyn3r2zH37sLFKXitDv49f2ZjPj6UVIupPDTmGks+nkF9jQ7dVrU4P7/u4N6bf4pWVqhWrkcq4lMDEYDfR7swawv5+eqj2IOMjNk5E1+26RQKPwnoDYnF01a7vOcEIIlk1f6PK9pWqFFGzxC/MR/H8y1WacbdMLKh3LTEzfkON5lcPusWinesKXaSLmQylu3jmP1LP9i3nWDTpdbc9cLSbmQml/EYg5cTjc7/t5D4v7jDI15hCnvz+RM4lmSz15i08LtPNnxJaZ97LtLT3buf/sOmnZugDnIjKYJdKOOyeJ5kPV9KP+Il38DUko2xycyf8NeDhw/U9rmKK4CAmrGfepwks/Kd+kpNs4cO+f1XFHRdXAHQiKC+OalSezfdAiTxci1Qzpy35jbCYsKzXHtsNG3sPCn5f7NhP0QXaEJQqNCuPv1IbnO1WtT29+PkEV4uVDeGPwB6V5WEG6Xm69H/kCngW2Jjq2Q5zgms5G357zMnnX7WT93C7pRp+OANsTWL5sFpEqavQlJPPXZDJJT0xFC4HK5qRdTgXGPDiAyJP8YfIXCGwEl3LWa1cASYvEqNtZQS56x1kVFq57NaNWzWY5jx+JPsHXpSkKjQmh2bSN0g+5p9luESanRMeX5ZNUYoirl7ghUo3EsQWFWv2t+W4LNdBncni+f+8HnNU6ni4U/Lee2Fwf5NWa9NnVyuFcUcDElnQfGTeFSWs5N6h2HT/L4+Gn8+MLtV7QKLGoupqQzdcU2Fm3ah9Gg079DI25oWx/zFda6URQ9AfUb6XprR7589vvLW6UihKecaseb2nh9X3GRdimNt4Z+yOZF27OEWjfqPDrubj5/ZmKu2iaFRdM1Wvdq5lW0weMGeuqLh3j/3v96LXBlMBmyNiktwWZa9mhK7eY18xYNCWdPXigS+/+tzFi1A6eXglxOl5uDJ86y/dAJmtQs2CZ5cXHibDJ3jp3EpTQbtoxN7t1HT/Hz4k18++wQgiyqH2RZIl8ftxAiRgixWAixSwixQwjxVEkY5g1LkJn3Fr5KRHQ41lArJosRa6iFqMqRvL/oNYymK6+2VxDeHvYJmxZtx57uIPViGqnJaSSfvcR7931G8tmUIruP0WSg3yM5e/OdOpLE0T2JWa6YbkM78eJPw6lSuyIGo45u1GnZownvL36Nm5/uS93WtWl9fTNe+OFJOtzYmtcGvZcl5t7QdI2G7ev6PK/Iny3xx0j38R1LKdlz9FQJW+SbMZMWcP5SWpZoA6TbnRw+eY5v564rRcsU3vBnxu0ERkgpNwohQoENQoj5UsqdxWybV2o3q8EvCV+yft4WThw8RZU6lWjZo0mBy7ReKaeOJLFh3hYcXpJPiqpBr6ZrGM0Gbh81iDrNawKwc/VePrj/M04cPIVm0DGaDNzz5lD6P9qLzoPa0WlgW1IupGIwGbI6p2ev4rfo5+WeGi4+imBlEhIZTKdBbfO8RpE35cOD0YSnG9Ll6JpGRCF93BdS0pm9ZhdHks5Ro2IUfdrWJzTId9mC/EhJt7N2zxFcXv5u7U4X0//ezmMDrin0+IqiJ1/hllIeB45n/DtZCLELqAqUinCDJ7qiXZ+WpXV7AA5sPYLB5LstV14IIfKMCc/EGmLh/UWvUaeFR7QP7zzK8z3fyJbJ6CAd+CrDV93/0V4IIQiJCPY6npSSr0f+mK9ol68axbilb5T4CuZqY1CnJsxcuYN0h/dZd6fGtbwez4vVuw4z4os/kEjS7U4sJgOf/L6CT/5zE63q+r/HczTpPOcvpRFqtaBrAk1ogPcEsJT0omvKoSgaCuTjFkLUAFoAa7ycewh4CCA2NrYITCvbhJUPRRaiiFLNJrHUbBLLkskr882UdDldOYo8/fjmb16jamypNiaO/oW+D/bIMwTx/KkLXDiT7PO8ZtAYt/QNGravW6Y2zQKVutUqcNf1rfl+/npsDidSgkHXMOga7zzQF4uXmPy8uJiSzogv/iDN/s9kIdMV89RnvzPvnYfy9UXvPHySl779i6OnzmetBHQt7/yGutWifZ5TlA5+/+UIIUKAqcBwKeXFy89LKb8CvgJPPe4is7CM0qBdHCGRwV4TcgxGHTTxT7ZiBuYgE09+9iBRlSJYPWsDaclpecZf6wads8fPUbtZDQA2LdyG24cbxmF3ciz+RJ7ZkCaLEZnHwyIoxEqjDvV8G6QoMI/060CHhtWZvHgzx85epGH1itx2XQtiKkQUeKw56/cgfYQqSQnzN+5jQEffzS0ST1/goQ+nkHpZGzmXWyKERJA7EMpiMvDojQXrNaoofvwSbiGEEY9o/ySlnFa8JgUGQghe/e1ZnuvxhqeOR8Z/Bkuwma63diQo1MKfXy3wRJtkXP/0lw/T+Jr6AHyycgyfDZ/IpoXbfLpNHDYHsQ3+Wf4aLb5dF26nK89KfgDB4cHUbxvHjpW7cz0wDEadbnd0zvdzKwpOs1pVaFbryjvDH00673OzM83uIPF03lFA3y/YgM2H2yZzNSAAU0b4n5SSkUOuo32D6ldkd0E5df4SG/clYDIaaF8/tsxFtBw7c4EfF25k3e6jhAVbGNylKT1b1UX3UcuoOMhXuIVnDTUB2CWlHFf8JgUO9drU4dvdH/PH53PZsmQHEdHh3PjI9bTo3gQhBMNGD2bX6r2YrCYad6qfw2dcvWEMY+eNZu+GeJ7uPDqXr9xoMtC0a0MqVv8nAeb6u65lygczvYb8Va5VMd9kGYCnvniIp655CVuqLavrvNFsIKx8GMNG31LYr0JRAtSoGInVZCDNi3gHmY3EVvTM4pMuXGLP0SRCg8w0qVEZLcPdtmrnIa8bkJk4XW5u6dKUrk1qYdA1WtSpmiXiJYHT5WbMpAX8tXa35yEiBC63m2du6cotnZuWmB15sfXAcf7zyVQcTheOjNXr7iMnmbV6Fx8/NqDExNuf38o1wJ3ANiHE5oxjo6SUs4vPrMChXOVI7nnDe6XBsHKhPsvHpqfamDdxCfN/WEpU5UhOJ5zBYDZm+c3rtKjJSz8/neM9g5/tz6KfV3Dm2Nks8dY0gclq4pn/PeqXvTUaxfDFpvf4+e3prJm1Ad2o0+32Tgwe0Z/w8mH+fmxFKdCrdT0+nLrM6zld0+jcuBajJsxm0eb9mAw6bikJtpgY+2BfmteuSlA+XZCEgBCLiWsa18zXlmNnLrBq52E0IejUuCYVIkIK9Zmy8/nMv5m7fg92pwt7tozjcb8tJaZ8OO1KeOZ/OVJKXpwwO5erKc3uZNP+BOas20Pfdg1KxJaruudkWSXlYipPtB/FqSOns0q/Gi1GTGYjQ1+4idbXN8+KJLmcS+dT+O2Dmcz7bil2m4MW3Roz7JXBVG9Q/FmjitJnc3wiT376O24pSbU5CDIb0TWNz54cxE8LN7J4S3wud4jVbGTK6DtZvu0gH05bht1LFUnw+LMnjLiVBrG++4i63ZK3Ji1g9ppdaEKA8Bwbcm1zhg/qXOhNbZvDSbfnviDN5j1Kq3ntKnzzbO5yD+B5iPy0cBOb9icQFRrE4K7N6NKkVpFvsO85eor7PvjVp41Na1Vm4nMFKxednYL0nAyozMmrhR9en8KJgydzuDwc6Q5cDhc7V+1l6PMDfb43JCKYe968jXvevK0kTFWUMZrXrsrcsQ+xcOM+Ek9fICY6gm7N40hOS2fR5v05ZqqZOJ0ufly4kacGdmbmqh3sPZqE67IJm8mg0715nTxFG+C7eeuYs253rvtMWrSRSpGh3NatRaE+18lzyXk2pdp/7LTX4xv3JfDEf3/Hmc11sWn/Ma5tVpu37u1dpOJ9MdWG7qOVH3ji60sKJdylwNyJi736qd0uN+vmbMaWZsvqbKNQXI7VZKRf+4Y5jm3Ym4TJoHsVbofLzca9CZiNBiaMGMKvSzfz44INnLvkqW1TPjyEB25oy8BrcjfXyI7bLflu/gavG6Qut2Tc1KUM7NSkwGGOAOHBVpx5RDyFe0kwcrndjPz6z1wz4DS7gyVb4lm+/SBdmhQ8Vt4XcVXL+1yt6JqgdQHi6K8UJdylgLcQwkyE8FQ6VMKtyI+lW+P5bOZK4o+fwWTQcfgQFSArS9NiMnBXz9bc1dOvFXkOUmx2Um2+k7dcbsm8DXvo38F3SKIvwoMttIyryro9R3NtoAogtmIk5y6l5aiouPXAcdLt3t0WaXYHU5ZuKVLhjgix0rddA2av3Z3LHWU06NzZo+DfaWEJqHrcVws1G/tOUAqJCCY06so3ehRXN9NWbOPFCbPZl3gat9uTRXm5+yMTq8nI4K7NvJ4rCFaTMd8e23PW7fZ7vMtzEl67qxdRoUGYDDllSQIb9iYw8NWJHDn1T+nmiynpebpCMlcURckLt3Xjhrb1MRl0QiwmgsxGKoQHM/7xgcRGFzw2v7CoGXcpcO9bt/HG4PdzpZ5bgs3c+epgn70tS5KUi6msmrmeS+dTaNihLnVbFbzmt6LwpNkc7Es8TZDZSO0q5XIIlM3hZNxvS33GdGdHCGhQPZrrml15yV2DrtEyrhprduduwJ2J5kNIk9NsHD11jrAgC0u3HuC7ees5fTGF0CAzQ7o244Eb2hEdEcKwHq34ZHruhil2pwuHy8XoiXP4bqRnfyeuWgWfcelGXadVXNG7Loy6zivDevLEgGvYk5BEiNVMw9iKWSGXJYUS7lKg7Q0teHz8/Xw+fCIIEAicDidDXxhIv4evz/f9xc2Cn5bx0UNfoulaRtq9Rp3mNRjz54u5WsMpiha3W/LFrJX8uHAjuqbhdksiQ628eU9vWtTxZMVuiT/m96ablLDj0Ek2xScWiZC9dmdP+rw0wWv+ptmoEx0ZwqDXJnL6QgpVy4dzd6/WbNyXyB+rdmDQddJsDiQyKwEsOdXGDws2siX+OH3a1efzP1b6jDWXEvYcTSLp/CXKhQXz/pQlPjOJDQaNodc1v+LP64vI0KAST0zKjgoHLEXsNgc7V+7B5XTRsENdrGWgI8reDfE80/WVXKsBo8lAy55NeeuPF0vJsn8Hn/+xkh8W5N4AtJgM/PTiHdSsFMXqXYcZ+dUsLqXnXSwsO/Vjopk06o58r0tOTef3lTtYujUei9FI/44N6dY8DkO2ptBf/7ma/81Zg8P5z2aiUdewmIw4XK4ctmdGYeSV+JP5+cxGQ76RGcEWE98+O4QDJ87w+vfzc9RtyUQTggkjBtOsdmB1YVLhgAGCyWyk+XWNS9uMHPz63kyvFQ8ddiebFm4jKeEMFaqVKwXLrn7S7A5+9CLaAA6ni2/mrOXNe3rTtGZlnAUscLYvMYk0mwOr2XvZhGXbDjB++griL+uJuW7PUb6qsJrvn78tK/X8wb7tqVI+nC9nrSLh9AWsJgOxFSPZn3gG12V25SfYmaTbnT4jNnKO56ZqhXDembzIq2gDmIw6RsOVS5vL7Wb5toPM37AXKSU9WtWlS0ZWaWmjhFuRg/gth3zWEzeajSTsPaaEu5g4ePyMz/0Nl1uyfm8CAEEWE/f2bsOEv9b6JXZZ+PCu/LJkE59MX+H9geFyceDEWbqP/IJPHhtIm3oxAPRt14C+7RoweclmPpy6jH0JSVxpGXpfBbQy0YTA4XBx7TOf5ekqsjtcLN0aT/2Y6EL7ntPsDh7+8Dfij5/JCjdctu0AsdGRTBhxq88HYElR+o8ORZkiOqa8z3NOh5NyVaJK0JqrD7vDycZ9CWzYl5BrYy3IbMo1Y8153oiUkp8XbeL7eRsK1LCjcY3KWL3UV09Jt/PxNO+inR2bw8Xwz2Zw8tw/ZYF3HDrBx9OWY3e6rli0g8xG6latQF6ue7eUuKTE4XJ7jVfPft0PCzbw4oTZPn3g+fHlrFXszVilZJJqc3Dg+BnGz1hRqDGLEiXcihwMGt4XS3DuGHKhCarVraK6t18BU5dvpfvIL3nqsxkM/2wG3Z/7gl+Xbsk6X71iJNE+an5YTAZu7tyUX5duYfyMFaSk27MyBcEzmY4KDUITItfE2mIy8Nyt13odd+3uI34v/Z0uF5OXbM56/cOCDdic+Ue25IfJoFOrcjnG3HsDIRYzxmz2GDUtz2xFX6TbnazYfpAVOw4Wyqapy7d5Xc3YnS5m/L3dr0YoxYkSbkUO2t7Qgj4PdMccZMpaZlqCzYSXD+OVKSNK2brAZdHm/Xzw21JS0u1ZP6k2Bx9NXcb8DXsBT+nfN+/pjdVsJLtWCSDUaqZb8zp8/sdK77NjIYiJDqdDw+pEhQWhawJd0+jQsDoTnxtKw+reU9k92Yr+iZDD5eaHBRuYvNQj3gdOnM2znrwvDJrAoGtoQqBrAofTxZ6jp3jooync2aMVt3ZtRpVyYcRGR/BAn3Z++8kvJ83uYOqyrQV+n8vtJiWPjV+bw5XnjL8kUD5uRQ6EEDz64b10H9aFud8u5sLpZFp0a0y3OzpjDS58X8N/O+N/9+6OSHc4GT9jBT1beRozN6lZmf8+PohHP5ma5UqRwMXUdO4a+7NPwZBSsiX+eNZrq8lIVKiVt+7pTWRokE+7WsZVxe70f6PT5ZZ8PG05Bk2jenQk8cdOF0i846qU4/burejSpAa3vvkj5y6lIvE8FM4mp/HN3LXcfX1rZr11P+DZlP181qo8xzRoms/N2sIk4eiaRnRECKfOX/J6PiLEiimj09S5S2nMXrOLY2cuUKdKeXq1rlci9cOVcCu8UrdVbZV0U0Q4XW6OnDzn83zi6YvYHc6s2te/LNmE05VToG0OF05XGlL6J7Jpdgcnz7sYM2kh7z98o8/rFm+Jx+Uq2Owx3e7ks5kref/hfvy946BfiUCZuKTEYjQwY+VOLqbZcvnG0+1OJs5dz7DurQi2mDAadEKtJpK9tOzLxJdomww6bevH+G1bdu7v3ZYPpy3zGpZ5X++2CCFYsmU/o775Cyk9SVFWk5EPpy3n8ycH0ahGpULd11+Uq0ShKGZ0TWDMMkq1bwAAEytJREFUoxeox3XgOe9yu1m0eb9X94DL7S7Q7NbpcrNi+0FSfSz7N+5LYNxvSwu1sZhmc1AxMpRH+3XwhN9l+KV9ZU5mcuD4Wd74cR7/+2sNDh+rB4OuseXAsazXjw/olOeYQpA1A84xjkFjSNfCJeHc0qUpN3dqgsmgYzUZsJqMmAw6Azo04rbrmnP6QgqjJvxFut2ZtTJKszu4lGbj8U+n+/xsRYWacSsUxYwQgl6t6zF73W5cl1XA0zVBj5Z1s/YTHE53npEQEs+sz+5wZTX7zQtNE1xMtXldvn8zd12BZsvZcbrdTFq4CZNR5427e3P01DnOp6QTYjbx3fz1PjvbA147+GTH5nDk6CQzuGsz3puyxGf1QF0IOjepyfJtBzEZdVxuSWSIlXcf7FvoBg9CCEYMvpY7e7Zm5Y5DSCnp2KgGFSNDAZixaofP79/h9IQj9mhZt1D39gcl3ApFCTB8UGfW7TnKuUup2DKiFcxGnfBgK0/f3CXrOovJQOWoMBLPeO8fWatyOT54+EYmzFnL6l2HMRt00h1OzlxM9Xp9ut1Jv5cnUKdKOZ6+uUuOLjK7Dp8o9Odxud38vGQTUnoaNTSMrcinTwzE7ZYs336AnUdOFXpsp0tSPyZnZ/leresxe80ur9uotaqU572HbuTMxRT2JZ4mIsRKvWoViqQWd3RECDddkztJ7tCJsz73G+xOF8fO5OqnXqQo4VYoSoDI0CAmj76Tacu38ddajwD1blOfWzo3IfSyWtOPD+jI6z/OzzUbNhl0+ratT7mwIN64u1fW8V8Wb/LU7fAxAXdLyd7E0/xn/DReuq0Hgzo34fjZi5zLo7xwfmSfbKbZHGw/dJxPpq8gIek8+4+d8f1GPxACRnw5k27N64CEI0nnqRQZSpDVRFq6I8dM12Iy8OLQbgCUCwumXFjJ1NKpXbkcZqOe9RDOjsmoE1OheCsFqlolCkUZ5NelWxj/uyfRw+V2Y3M4Meg6JoOO0+Xils5NeWpQZ76YtYpv5qzze1xNCFZ89Bj9X/mW0xdSitRmi9GAW7oLFKXiL2ajjpTQqEZFdh4+hcvtpnXdajw+oJPPUMfi5FxyKn1fnuDV1RQZamXO2w9i1H3va3ijILVKlHArFGUUu8PJ4i3xvPb93FwzO4vJQP2YaLbEH/MzCvsfbu3alClLtxb4ffmhaxomg5avD9to0HE6XYW6v9GgM+vN+4qkOfGVsnrXYUZ88Qfg2ZgMMhsxGXW+eOoW6larUODxlHArFFcJYyYt4Pe/txc6CaUkMegaRt23cGcmBV3XvA6t68YwZtKCQt0nxGJi4sih1Kpc+jVzUtLtzN+wl5PnkqlRKYrrmtXOCussKKo6oEJxleCtlVdJ0KZeDNsOHs/hCrCYDLSKq8aqnYe9RlQ4XW6fkR8Wo4Hnh1xHt5ZxhFrNvPTNXwj8zdnMyaV0Ow+Om8Ls/3sAcyFFsqgItpi8bl4WN0q4FYoyTHAJZOFdjiYED/dtT7rDyffz13P01AVioyO4+/rWaJpg9a7D+SpudlG2mo30aBFH/46NsiI9NscnXpGrxuZwsnDTPvq0bXAFowQuSrgVijLM4C7NePfXxV43wfKasWpCIGV+hVK9ExUaRPPaVdE0QceGNXKcy6vrTHYMBp3q0RFUCA/h1mub0aVJrRzheSFWM5Dse4B8SLU52HP0lBJuhUJR9ujbrgF/rN7BriOnssRb1wQmg4Hbu7fgp4Ubc4m6Udf4+LGbEMCEOevYceh4vhuGmeOajQY++s8An3WshRB5PzEyrwO+HH6Lzzopg7s0ZdzU3CnlnnuAQdfzzD40GXQqhJf+BiV4Nibnrt/DtgPHqRARQv8ODalSLrxY76mEW6EowxgNniiFWat3MmX5VlLSbLSuG8Pd17cmNjqSFnWq8tmMlexJOIXFaKRHqzieublLVmx4uwbVWb7tAC9OmE2qLXfHGJNB55pGNbA7XTSvU4VB1zTJsyhVt+Z1mLZ8m8/uM5lIyLPY0oCOjZmzfg87D5/MEm+jrmE06NxzfWvCgq1UKx/O8M9nePWbC0GZmG0fOH6GBz74FZvTRZrNgVHXmThvHU/f3KXQ6fb+oKJKFIqrHJfbzR1vT8qV7Wc2GmhTL4ZPHrvJ77GklDzx6e9eG0FkomuC7i3ieOeBvnmO5XC5mLN2N1OXb+NSup2ODatze7eWVIoKzbpm/oa9vPLdXFxuz8anQdfQNY3Rw3qUunBLKen38gROnE3OtQAxGw18N3JogcICVTigQnEVk5pu5/Cpc4QHW/xekl9KszH2l8XM37g3yw0y8JrGPDWwc4HD1xwuF9/OWcvXs9d4jXgRAqa+ejc1KhZNt6SEpPNMWbaV+ONnqFkxisFdmxEbXbyZif6waX8iT3w63etKRtcE/Ts0YvSwnn6Pp8IBFYqrEKfLzUfTljF1+TYMuobT5aJGxSjG3HdDvjHNIVYzb97bm1F3dGfTvkRmrd7Jmt1HGfHlH9zerSUdGlbP9Z6EpPNM+GstK3YcxKBr3NCmAXf1bEVEiJXuLeL4du56XO7cs26TQWfRpv3c17ttkXzuahUictRzKSscP+u7HonLLTly6nyx3VsJt0IRILz980L+Wrcbm8NJ5iRvb0IS9743memv3UNUmG/ftN3h5I/VO/lm7lpOnPFEc0g8PtqN+xK5uXMTnrmla9b1+xNPc8/7v2CzO7Nm1T8t2sjstbv4edQd7Dx80mdLMZvDxfq9CUUm3GWV6tGRPvdoDbpGXDXf/VuvFL+EWwjRG/gY0IH/SSnfKTaLFApFLs5eTOXPNbtyVaSTeGKapyzbwsP9OgBw8lwy385dx6LN+xHAdc3qsPnAMQ6fPOu1KFKa3cFvy7bSu039rLofb01aQGp6TheAw+niXHIqX89eQ7sGsT6r7wkB5fN4iPhCSsnBE2dJdziJq1q+wLU+SpqG1StSJSqMgyfP5gqRNGgaQ69tUWz3zle4hRA68F+gJ5AArBNCzJRS7iw2qxQKRQ62HTqO0aB7LSVqd7pYvv0gD/frQELSeYa9M4lUmyMrGmPK8i35NmCwO13MXLWDhtUrcjElnV2HT3q9zuFyM3vtboYP6pyrKXEmZqOBQZ2bFOTjsXzbQUZ9MztHr8cuTWry/sP9/W5mXNIIIRj/+EAe/HAK55JTsTmcmAwGpJS8eW/vYvXD+zPjbgvsl1IeyDD2F2AAoIRboSghrCZjnudDMkLv3puyhEtp9hwp6f7EH7il5Gyyp6a3zenMs5Z1SroNTdN4+/4+PPf1LBxOV5Y7xWoy0L9DI5rXrpr/TTPYdvA4T38+I1ca/bJtB7lz7CR+HjXM77FKmkpRocx4/V5W7TzE3sQkIkOD6N7Ck9ZfnPgj3FWBo9leJwDtiscchULhjRZxVX0mxVhNRgZ2aoLT5WblDu91RPLDajLSpq6nP2P5sGAiQqw+m+W6peTj6csZcUtXfh41jEmLNrLt0Amiw4MZcm0L2jeILdC93/11sU+b9xxNYvuhEzQu5h6OV4KmCa5pXJNrGtcsuXv6cY23v5Zc37IQ4iEhxHohxPqkpKQrt0yhUGRh1HXeuKsXFqOB7JNhi8lAk5qV6N4iLqMnZcFFWwgwmwz0adcg47XgyZs6+ewfKSX8tmwraTYH1StG8uJt3Zn04h189J+b6NCweoE7z+w5krdezFm3u0Dj/RvwR7gTgOytkqsBxy6/SEr5lZSytZSydYUKBa9Fq1Ao8qZrs9pMePZWrmtWhwrhwdSuUo5nbu7C+CcGYtA1zEYDNSpGFmhMg65Rs1I5vn12SI6CVn3aNcBq8r0gN+gaR5OKJtxN1/MW+rLq4y5N/HGVrAPihBA1gURgKHB7sVqlUCi80iC2Iu8/fKPP88Nv7sLIr2blatZr1DWsZhNOlwtNCOxOF12b1uL+G9r5zO6Ljgzl4ImzXs85nC7Cgy1ezxWUrk1rM2/DXq/nhIDrW9W74ntIKVm54xDT/95OcpqNDg2qM7BTkyL7DCVNvsItpXQKIR4H5uIJB/xGSrmj2C1TKBQFplPjmrxxTy/enbyElHQbEggLsvDibd3o3LgWu4+eJM3mpF5MhYwKfb4Zcm0zPpq2PFchKE0I6larkNXx/Ep5YWg3lm494DWFvnVczBW3JnO7Jc//bxYrdx4mLSMAftuB40yct55vnxtCzUpFk+FZkqiUd4XiKsTtlhw5dQ4hBLHREV79zgdPnOWrP1ezZtdhjAadvu0acvf1rbNmoQ6XiyfGT2fboRNZgmc2GrCYPHU4YqML5pbJizPJKbzw9Z9s3J+IlJ4VwuAuzRh+c5crdpX8tXY3b/00P1eFRCEgrkp5fnn5zisav6hQtUoUCkWe7Dh0goc++g2b3ZkV0WE06JQLDeLnl4ZlibfT5WbJlv1M/3s7qel2OjWpxc2dmhARYi0Wu6SUOF1ujIaiS765a+wkth/yHpduNhqY/PKwIn0IFRZVq0ShUOTJGz/Oz5pFZ+JwujhzMYWJ89bx1MDOgGdjsEfLuvRoWbdE7BJCFKloA5xJTvN5zmjQOJecViaEuyCo7VqF4l9G0vlLHD55zus5h8vNrNVXV25d4+oVfYY22h0uqhcwEqcsoIRbofiXYXM4fSbzAF7T6gOZe3u3xeRlFm82GujVul6xuX2KEyXcCsW/jMrlwnym0GtC0LZejNdzgUr9mGjevLc3wRYTwRYTQRYTJoNO58Y1GXV799I2r1AoH7dC8S9D1zSeuKmT1ybEJqPOw307FOv99yYksWl/IkFmI12b1iasBGKpu7eIo3PjmqzZfYSUdDtNa1Uu9r6QxYkSboXiX8hN1zRGSsn4GX9nRJa4qVo+nNHDelKnavHUkU6zOXj6ixlsiT8OeLrE/N/PCxkx+Fpu6dy0WO6ZHZPRQOcmtYr9PiWBEm6F4l/KwE5N6N+xEQlJFzAZdSpHhRXr/d74cR6b449hv6wm+LgpS6lVKYqWcdWK9f5XE8rHrVD8i9E1jeoVI4tdtM9fSmPx5vhcog2Q7nDyzdx1xXr/qw0l3AqFothJOH3Ba2RHJvHHTpegNYGPEm6FQlHslA8LzjPMsEJ4SAlaE/go4VYoFMVOpahQGlav6LXBsNVkZFiPlqVgVeCihFuhUJQI/3dfH8qFBWM1e2LINSGwmAxc36ouPUsopf5qQUWVKBSKEqFSVCi/v34v8zbsYeXOQ4RZLfRr35CmtSqXtmkBhxJuhUJRYlgymgn379CotE0JaJSrRKFQKAIMJdwKhUIRYCjhVigUigBDCbdCoVAEGEq4FQqFIsAolp6TQogk4HARD1seCMS8WGV3yROotiu7S5ayZnd1KWUFfy4sFuEuDoQQ6/1tpFmWUHaXPIFqu7K7ZAlUu0G5ShQKhSLgUMKtUCgUAUYgCfdXpW1AIVF2lzyBaruyu2QJVLsDx8etUCgUCg+BNONWKBQKBQEm3EKIN4UQW4UQm4UQ84QQVUrbJn8QQrwnhNidYft0IUREadvkD0KIwUKIHUIItxCizO++CyF6CyH2CCH2CyFeKG17/EUI8Y0Q4pQQYntp21IQhBAxQojFQohdGX8nT5W2Tf4ghLAIIdYKIbZk2P16adtUUALKVSKECJNSXsz495NAQynlI6VsVr4IIa4HFkkpnUKIsQBSyudL2ax8EUI0ANzAl8CzUsr1pWyST4QQOrAX6AkkAOuA26SUO0vVMD8QQnQBLgHfSykbl7Y9/iKEqAxUllJuFEKEAhuAm8r6dy6EEECwlPKSEMIIrACeklKuLmXT/CagZtyZop1BMBAQTx0p5TwppTPj5WogINpZSyl3SSn3lLYdftIW2C+lPCCltAO/AANK2Sa/kFIuA86Wth0FRUp5XEq5MePfycAuoGrpWpU/0sOljJfGjJ+A0JJMAkq4AYQQY4QQR4E7gFdK255CcB/wV2kbcRVSFTia7XUCASAiVwtCiBpAC2BN6VriH0IIXQixGTgFzJdSBoTdmZQ54RZCLBBCbPfyMwBASvmSlDIG+Al4vHSt/Yf87M645iXAicf2MoE/dgcIuZsZBtgsKlARQoQAU4Hhl62KyyxSSpeUsjme1W9bIUTAuKigDHbAkVL28PPSScCfwKvFaI7f5Ge3EOJuoB/QXZahjYUCfN9lnQQgJtvrasCxUrLlX0OGj3gq8JOUclpp21NQpJTnhRBLgN5AwGwOl7kZd14IIeKyvewP7C4tWwqCEKI38DzQX0qZWtr2XKWsA+KEEDWFECZgKDCzlG26qsnY5JsA7JJSjitte/xFCFEhM7JLCGEFehAgWpJJoEWVTAXq4Yl0OAw8IqVMLF2r8kcIsR8wA2cyDq0OkGiYgcB4oAJwHtgspexVulb5RgjRB/gI0IFvpJRjStkkvxBC/Axci6da3UngVSnlhFI1yg+EEJ2A5cA2PP8nAUZJKWeXnlX5I4RoCnyH5+9EA36VUr5RulYVjIASboVCoVAEmKtEoVAoFEq4FQqFIuBQwq1QKBQBhhJuhUKhCDCUcCsUCkWAoYRboVAoAgwl3AqFQhFgKOFWKBSKAOP/AXzG9nAgotG5AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "est = KMeans(4)  # 4 clusters\n",
    "est.fit(X)\n",
    "y_kmeans = est.predict(X)\n",
    "plt.scatter(X[:, 0], X[:, 1], c=y_kmeans, s=50);"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}