Marco Bag of words and tf-idf pandas-2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import nltk\n",
    "import re\n",
    "import h2o\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "tmp_df = pd.read_csv('D:/marco/vr_chat-master/Marco/train_yn.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>passages</th>\n",
       "      <th>query</th>\n",
       "      <th>query_id</th>\n",
       "      <th>query_type</th>\n",
       "      <th>answers_yn</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>[Row(is_selected=0, passage_text=u'In his youn...</td>\n",
       "      <td>was ronald reagan a democrat</td>\n",
       "      <td>19700</td>\n",
       "      <td>description</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>[Row(is_selected=0, passage_text=u'So, in the ...</td>\n",
       "      <td>do game shows pay their contestants</td>\n",
       "      <td>19749</td>\n",
       "      <td>description</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>[Row(is_selected=0, passage_text=u'Even though...</td>\n",
       "      <td>does peanut butter contain cholesterol</td>\n",
       "      <td>19761</td>\n",
       "      <td>description</td>\n",
       "      <td>No</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>[Row(is_selected=0, passage_text=u'When turnin...</td>\n",
       "      <td>can you take left hand lane to turn right on a...</td>\n",
       "      <td>19762</td>\n",
       "      <td>description</td>\n",
       "      <td>No</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>[Row(is_selected=0, passage_text=u'Assault on ...</td>\n",
       "      <td>is armed assault a felony</td>\n",
       "      <td>19774</td>\n",
       "      <td>description</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            passages  \\\n",
       "0  [Row(is_selected=0, passage_text=u'In his youn...   \n",
       "1  [Row(is_selected=0, passage_text=u'So, in the ...   \n",
       "2  [Row(is_selected=0, passage_text=u'Even though...   \n",
       "3  [Row(is_selected=0, passage_text=u'When turnin...   \n",
       "4  [Row(is_selected=0, passage_text=u'Assault on ...   \n",
       "\n",
       "                                               query  query_id   query_type  \\\n",
       "0                       was ronald reagan a democrat     19700  description   \n",
       "1                do game shows pay their contestants     19749  description   \n",
       "2             does peanut butter contain cholesterol     19761  description   \n",
       "3  can you take left hand lane to turn right on a...     19762  description   \n",
       "4                          is armed assault a felony     19774  description   \n",
       "\n",
       "  answers_yn  \n",
       "0        Yes  \n",
       "1        Yes  \n",
       "2         No  \n",
       "3         No  \n",
       "4        Yes  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from nltk.corpus import stopwords \n",
    "stop_words = set(stopwords.words('english'))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def tf_cln(sent = tmp_df.iloc[0][0],cln=True):\n",
    "\n",
    "    # sent = tmp_df.iloc[0][0]\n",
    "    if(cln==True):\n",
    "        slt_sent = re.findall(r'passage_text(.*?)url',sent)\n",
    "#         print slt_sent\n",
    "    else:\n",
    "        slt_sent = sent\n",
    "        \n",
    "    token = nltk.word_tokenize(str(slt_sent))\n",
    "    filtered_sentence = [w for w in token if not w in stop_words]\n",
    "#     print filtered_sentence\n",
    "\n",
    "    for i in filtered_sentence:\n",
    "        if re.findall(r'^\\W.*',str(i)):\n",
    "    #         print re.findall(r'^\\W',i)\n",
    "            stop_words.add(i)\n",
    "        \n",
    "#     for i in re.findall(r'\\W',str(filtered_sentence)):\n",
    "#         stop_words.add(i)\n",
    "#         print i\n",
    "\n",
    "    stop_words.add('``')\n",
    "    filtered_sentence = [w for w in token if not w in stop_words]\n",
    "\n",
    "    tf = nltk.FreqDist(filtered_sentence)\n",
    "    \n",
    "    return dict(tf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "lst_tf_cln=[]\n",
    "for i in tmp_df.iloc[:,0]:\n",
    "    lst_tf_cln.append(tf_cln(i))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "lst_tf_qry = []\n",
    "for i in tmp_df.iloc[:,1]:\n",
    "    lst_tf_qry.append(tf_cln(i,False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "tmp_df = tmp_df.replace(tmp_df['query_type'].unique(),range(0,len(tmp_df['query_type'].unique())))\n",
    "tmp_df['passages'] = lst_tf_cln\n",
    "tmp_df['query'] = lst_tf_qry\n",
    "tmp_df = tmp_df.replace(['yes', 'no'], ['Yes', 'No'])\n",
    "tmp_df = tmp_df.replace(['Yes', 'No'], ['1', '0'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "del(tmp_df['query_id'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>passages</th>\n",
       "      <th>query</th>\n",
       "      <th>query_type</th>\n",
       "      <th>answers_yn</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{u'Wilson': 1, u'United': 2, u'switched': 1, u...</td>\n",
       "      <td>{u'ronald': 1, u'reagan': 1, u'democrat': 1}</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{u'United': 1, u'show': 18, u'money': 6, u'Jeo...</td>\n",
       "      <td>{u'pay': 1, u'game': 1, u'shows': 1, u'contest...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{u'serving': 1, u'saturated': 2, u'unhealthy':...</td>\n",
       "      <td>{u'butter': 1, u'peanut': 1, u'contain': 1, u'...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{u'already': 2, u'show': 2, u'When': 5, u'mark...</td>\n",
       "      <td>{u'lane': 1, u'right': 1, u'carriageway': 1, u...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{u'sentence': 1, u'Deadly': 1, u'rob': 1, u'mi...</td>\n",
       "      <td>{u'felony': 1, u'assault': 1, u'armed': 1}</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            passages  \\\n",
       "0  {u'Wilson': 1, u'United': 2, u'switched': 1, u...   \n",
       "1  {u'United': 1, u'show': 18, u'money': 6, u'Jeo...   \n",
       "2  {u'serving': 1, u'saturated': 2, u'unhealthy':...   \n",
       "3  {u'already': 2, u'show': 2, u'When': 5, u'mark...   \n",
       "4  {u'sentence': 1, u'Deadly': 1, u'rob': 1, u'mi...   \n",
       "\n",
       "                                               query  query_type answers_yn  \n",
       "0       {u'ronald': 1, u'reagan': 1, u'democrat': 1}           0          1  \n",
       "1  {u'pay': 1, u'game': 1, u'shows': 1, u'contest...           0          1  \n",
       "2  {u'butter': 1, u'peanut': 1, u'contain': 1, u'...           0          0  \n",
       "3  {u'lane': 1, u'right': 1, u'carriageway': 1, u...           0          0  \n",
       "4         {u'felony': 1, u'assault': 1, u'armed': 1}           0          1  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checking whether there is an H2O instance running at http://localhost:54321. connected.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td>H2O cluster uptime:</td>\n",
       "<td>36 secs</td></tr>\n",
       "<tr><td>H2O cluster version:</td>\n",
       "<td>3.10.0.10</td></tr>\n",
       "<tr><td>H2O cluster version age:</td>\n",
       "<td>2 months and 4 days </td></tr>\n",
       "<tr><td>H2O cluster name:</td>\n",
       "<td>H2O_from_python_user_ggfe0u</td></tr>\n",
       "<tr><td>H2O cluster total nodes:</td>\n",
       "<td>1</td></tr>\n",
       "<tr><td>H2O cluster free memory:</td>\n",
       "<td>896 Mb</td></tr>\n",
       "<tr><td>H2O cluster total cores:</td>\n",
       "<td>2</td></tr>\n",
       "<tr><td>H2O cluster allowed cores:</td>\n",
       "<td>2</td></tr>\n",
       "<tr><td>H2O cluster status:</td>\n",
       "<td>accepting new members, healthy</td></tr>\n",
       "<tr><td>H2O connection url:</td>\n",
       "<td>http://localhost:54321</td></tr>\n",
       "<tr><td>H2O connection proxy:</td>\n",
       "<td>None</td></tr>\n",
       "<tr><td>Python version:</td>\n",
       "<td>2.7.11 final</td></tr></table></div>"
      ],
      "text/plain": [
       "--------------------------  ------------------------------\n",
       "H2O cluster uptime:         36 secs\n",
       "H2O cluster version:        3.10.0.10\n",
       "H2O cluster version age:    2 months and 4 days\n",
       "H2O cluster name:           H2O_from_python_user_ggfe0u\n",
       "H2O cluster total nodes:    1\n",
       "H2O cluster free memory:    896 Mb\n",
       "H2O cluster total cores:    2\n",
       "H2O cluster allowed cores:  2\n",
       "H2O cluster status:         accepting new members, healthy\n",
       "H2O connection url:         http://localhost:54321\n",
       "H2O connection proxy:\n",
       "Python version:             2.7.11 final\n",
       "--------------------------  ------------------------------"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "h2o.init()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "# hdf=h2o.H2OFrame(tmp_df)\n",
    "\n",
    "hdf_tf=h2o.H2OFrame(tmp_df[tmp_df.columns[0:3]])\n",
    "hdf_target = h2o.H2OFrame(pd.DataFrame(tmp_df.answers_yn))\n",
    "hdf = h2o.H2OFrame.cbind(hdf_tf,h2o.H2OFrame.asfactor(hdf_target))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# h2o.shutdown()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead>\n",
       "<tr><th>passages                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   </th><th>query                                    </th><th style=\"text-align: right;\">  query_type</th><th style=\"text-align: right;\">  answers_yn</th></tr>\n",
       "</thead>\n",
       "<tbody>\n",
       "<tr><td>{'Wilson': 1, 'United': 2, 'switched': 1, 'votes\\\\\\\\u2014the': 1, 'supported': 1, 'years': 2, 'four': 1, 'radio': 1, '1962': 2, 'Ford': 1, '1948': 1, 'voter': 1, 'candidate': 3, 'former': 3, 'Ronald': 9, 'video.duration.momentjs': 1, 'candidates': 1, 'small-town': 1, '8': 1, 'campaigned': 2, 'views': 1, 'Re-Election': 1, 'early': 2, 'party': 3, 'Illinois': 1, 'presidential': 1, 'defected': 1, 'witched': 1, 'Raised': 1, 'Republican': 6, 'Liberal': 1, 'commentator': 1, '51': 1, '50': 1, 'disillusioned': 1, 'either': 1, 'found': 1, '538': 1, 'working-class': 1, 'From': 1, 'people': 1, 'Gerald': 1, 'Wikipedia': 1, '1960s': 2, 'challenger': 1, 'economic': 1, 'election': 2, 'Listen': 1, 'even': 1, 'ultimately': 1, 'said': 1, 'fundamentally': 1, 'electoral': 1, 'conservative': 2, '525': 1, 'mate': 2, 'During': 1, 'America': 1, 'ever': 1, 'Mondale': 3, 'He': 2, 'U.S.': 5, 'Bill': 1, 'however': 1, 'free': 1, 'Democrat': 3, 'Ferraro': 2, 'Northerner': 1, 'served': 3, 'put': 1, 'path': 1, '1935-': 2, 'Nixon': 1, 'received': 1, 'Jimmy': 1, 'Clinton': 1, 'carried': 1, 'country': 1, 'changed': 2, 'became': 3, '1970s': 1, 'Yes': 2, 'think': 1, 'first': 2, 'major': 2, 'useful': 1, 'political': 3, 'When': 1, 'number': 2, 'officially': 1, 'female': 2, 'parties': 1, 'old': 1, 'Hollywood': 1, '1984': 4, '1980': 2, '1981': 2, 'trajectory': 1, 'support': 1, 'June': 1, '1989': 2, 'California': 3, 'way': 2, 'vice-presidential': 2, 'white': 1, '40th': 2, 'life': 1, 're-elected': 1, '6': 1, 'American': 2, 'dramatic': 1, 'Harry': 1, '1911': 1, 'Democratic': 6, 'November': 3, 'Carter': 1, 'grew': 1, 'politician': 1, 'presidency': 1, 'reelected': 2, 'Reagan': 15, 'elections': 1, 'governor': 3, 'recording': 1, 'time': 1, 'landslide': 3, 'age': 1, '20s': 1, 'Richard': 1, 'voters': 1, 'states': 1, 'ready': 1, 'Party': 5, 'sounding': 1, 'Walter': 3, 'States': 2, 'different': 2, 'traditionally': 1, 'us': 1, '49': 1, 'rather': 1, 'began': 1, '1967': 1, 'actor': 3, 'member': 2, '5': 1, 'largest': 1, 'President': 2, 'A': 1, 'February': 1, 'I': 1, 'Geraldine': 2, 'earlier': 1, 'running': 2, 'president': 1, 'Democrat\\\\': 1, 'desert': 1, 'younger': 1, 'Truman': 1, 'especially': 1, '1975': 1, 'later': 1, 'defeating': 3, 'deserted': 1, 'encyclopedia': 1, 'Democrats': 1, '2004': 1, 'In': 4, 'Duration': 1, 'audio': 1, '1911-2004': 1}</td><td>{'ronald': 1, 'reagan': 1, 'democrat': 1}</td><td style=\"text-align: right;\">           0</td><td style=\"text-align: right;\">           1</td></tr>\n",
       "</tbody>\n",
       "</table>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hdf.head(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "hdf_train, hdf_validate=hdf.split_frame(ratios=[0.75], seed =5010)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from h2o.estimators.naive_bayes import H2ONaiveBayesEstimator\n",
    "from h2o.estimators.gbm import H2OGradientBoostingEstimator\n",
    "from h2o.estimators.deeplearning import H2ODeepLearningEstimator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "nb =H2ONaiveBayesEstimator(seed =1)\n",
    "nb.train(x = ['query','query_type','passages'], y = 'answers_yn', training_frame=hdf_train, validation_frame=hdf_validate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "nb.auc()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gbm Model Build progress: |███████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "gbm_model = H2OGradientBoostingEstimator(seed =1,  ntrees = 170, learn_rate = .05)\n",
    "gbm_model.train(x = ['query','query_type','passages'], y = 'answers_yn', training_frame=hdf_train, validation_frame=hdf_validate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAEaCAYAAAA/lAFyAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x342072e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gbm_model.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2e88ad68>]"
      ]
     },
     "execution_count": 118,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXkAAAEACAYAAABWLgY0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xd8VFX6x/HPAUSQJghSQid0iKhAUBFDUSlSbBRdBdui\nARdddXUta1xxLcuuBZafggoiJYmANFFRISBIL4EkRIrUgAwgVVpIzu+Pm2VjJGQSJnNnJt/365WX\nmZlzZx7vK3ny8NxzzzHWWkREJDQVczsAEREpPEryIiIhTEleRCSEKcmLiIQwJXkRkRCmJC8iEsLy\nTPLGmI+MMfuMMesvMOY9Y8xmY8w6Y0wr34YoIiIF5U0lPw64NbcXjTHdgAbW2obAYOB9H8UmIiIX\nKc8kb61dDBy6wJDewISsscuBCsaYqr4JT0RELoYvevJhwK5sj9OynhMREZfpwquISAgr4YP3SANq\nZXtcM+u53zHGaKEcEZECsNaaghznbZI3WV/nMwsYAsQZY9oBh621+3J7Iy2I5oiJiSEmJsbtMAKC\nzsX/5OdcWGvZ8ssW1u9bz5q9a0g5kELK/hR2HN5BoysaEV4pnKplqlKjXA1qlq9JWPkwwsqFEVY+\njMtLXV64/yM+EGo/F9bC4sUwdCisXw/jxsH990OxnP0UjweioyElxRkUGYkxBcrvgBdJ3hgzGYgC\nrjDG7AReBkoC1lo7xlo71xjT3RizBfgVeKDA0YjIb1hr2XpoK+t+XseBEwfYuH8jifsS2XV0F3uO\n7aHyZZVpVa0V11a/lnta3EOzKs1oeEVDShYv6XboksOqVdChA/TuDV9/DdWq5RhgLcTHw7BhMGgQ\nTJwIpUpd9OfmmeSttfd4MWboRUciUgSdzTzLz8d/ZvfR3aQdTWPZ7mX8+es/s+PIDnYe2cn2w9sp\nWbwkbcPaUuWyKjSo2ICejXtSp0IdapSrQZmSZdz+XxAv/PwzfP45dO4MM2acZ0D26n3mTIiM9Nln\n+6InLwUQFRXldggBoyicizMZZ9h2aBv7T+xn15Fd/LDrBxbvWkyyJ5nKl1U+10opHV6a6mWr065m\nO+pUqEPtCrWpVrbaRf1zPViF0s9FixZOFf9+zruICql6z874s0dujLHqyUuoO5Nxhm9/+pY1e9ew\n9dBWtv6ylcR9iVxZ5kquLHPluSTevnZ7rq52NZeWuNTtkKWQnD4NCxZAnz5w6BCULp3txfP03nNj\njCn0C68ikouzmWdZs3cNC7YtYP72+SzdtZSIqhF0qNOBG2rdwP0R9xNRNYIrLrvC7VDFD44ehSlT\nYMUK579XXOH891yC90P1np0qeZF8OpF+giRPEkt2LmH+9vl8v+N7alWoRae6nehYryM31bmJiqUr\nuh2m+FFaGrRrB7t3O4+vvhoeeQTatoVrroFz3bZ8VO/ZXUwlryQvkouT6SfZfXQ3u47uYsfhHXy1\n9SvW7l3LrqO7aHxFY9rVbEenep2IqhvFlWWudDtc8bNNm2D+fEhMhLFjncQ+a5ZTuZfMObkpZ/Ue\nE5Ov6l1JXsQHrLVM3zid2ORYluxcwsGTBwkrF0atCrWoVb4WUXWjuK7mdTS6ohGXFL/E7XDFJb/+\nCt27w5Yt0KwZNG4M99zjVPK/m/MOBa7es1OSFymgs5lnWbprKXM3z2XO5jkAPH3d03So04E6l9eh\nmNHKH/JbTzwB+/fDhAlQvPgFBl5k9Z6dkryIFw6dPMT6fetJPZBK6oFUNh7YyIq0FdS9vC7dG3an\ne8PuRIZFUrzYhX5zpahr3hwmTYJWF9o5wwfVe3aaXSOSi1/P/MrS3UuZljKN2ORYmlVpRtPKTWlS\nuQmd63fmmurXUKNcDbfDlCAxYgScOAH16uUywM8zZ7yhJC8h42T6SZI8SSTuS2Ti+oms2rOK9Mx0\nWtdoza0NbiV1SCpVy2qrA8m/rVvh73+H775z5r1XqHCeQYV41+rFUJKXoJVpM1mycwlTkqawYPuC\ncwtzXVXtKgZfO5hZA2ZRrmS5Inm3qFy8s2fhlVecC6wbNkCNGrByJVSvnmNgAFbv2aknL0Hl6Omj\nLNy+kG9/+pbPUz+nQqkKDGgxgB4Ne9C0SlMtzCU+8+abMGcOPPYYVKrkzHe/MudMWR/33nOjC68S\n0o6dPsb0jdOZkjSFJbuWEBkWSed6nbmt0W20rNrS7fAkRDVrBp98Am3anOdFH86c8YYuvErISc9I\nZ9nuZXyW8hkT10/kxjo38kCrB5jWd5pWXpRCN24c7NwJl59v2f0A7b3nRkleAkamzWRG6gwmb5jM\nd9u+o97l9ejZqCeJjyZSq0KtvN9AxEemToXhwyE8PNuTAd57z42SvASE1XtW88jsRyhmijG07VBG\ndR9FtbI5d1UQKXzjx8O6dfDee7msORME1Xt2SvLiqm2HtjFm9Rg+Xvcx/7rlX9zb8l7NhhFXbNkC\nDz7obM03fDg0aEDQVu/ZKcmLX1lrSdyXyIzUGcxIncGeY3u4vcntrBu8jurlcs5NE/GP/fuhYUPo\n1s1ZIrhePYK6es9OSV4KlbWWTQc3sWD7AhbvXMyiHYsoUawEtze5nZHdRnJ9reu1jIC47vhxqFsX\n5s7lt9X7wIFBWb1npymU4nPWWpbtXkZ8cjyzN83mdMZpOtfrTPva7Wlfuz2Nr2islowEjA8+cDb1\n2LcPNi7Mqt6Tk53mfIBU75onLwHh1NlTxCbFMnLFSI6cOsLAqwbSs3FPrqp6lZK6BKTTp6FpU/jr\nc5b+xeIp92JW9f7KKwFVvSvJi6uOnT7GpA2TeO3712hWpRl/avsnujXspmV6JaBZC506wYEUD6si\no7l0c2BV79npZijxu0ybyeo9qxm3bhyxSbFE1Y0i7q44rq91vduhiXjllRjLdTvjGc4wijUdCPHB\n3XvPjZK85Mumg5v4z4r/EJscS8VSFRnQYgAbHttAWPkwt0MTyZO1zlTJzz/w0OY/0dxaK5lis4J3\n5ow3lOQlT+kZ6Xyx+Qs+WP0Bq/es5uFrHmbpQ0upX7G+26GJ5EvdOpbbTsTz2vFhHL19ICXGhWb1\nnp168nJB3/30HYPnDKZa2Wo8fM3D9Gvej9KXlHY7LJF8O7rFw7yG0dzVJHB777lRT1587mzmWZ78\n6klmbZrF6O6j6dGoh9shiRTI9m2Wcd3ieezHYdh6A2Ft6Ffv2amSl99ZvWc1z89/HmstU/tOpfyl\n5d0OSSTfli51eu83fx5NM5K5YtZ4St0UPNV7dhdTyWuOm5yzMm0lt3x6C33i+tC1QVfm3DNHCV6C\n0t49lvc7xvFCfASV2zWANWuDNsFfLFXyAsDSXUvpFduL1zu/zv1X3a8dliQonTkDC+I8lHs2muqH\nkqmXMD6oeu+5USUvBbbv+D4en/s4vWJ7MaHPBB6+5mEleAkqp045yxL06G4ZWCqO1g9FcFnLBlTY\nujYkEvzF0oXXImzj/o10n9ydPo37sPShpYRXCs/7IJEAkpnpLDWTEO/hy3rR1KuXTMnJM7lCyf0c\nVfJF1KIdi4j6JIqXb3qZt7u+rQQvQScuDhrUt1RbGMem0hE07t6Aksmq3nNST76Isdby4ZoPeWH+\nC0y+czJd6ndxOySRfLMWrg7z8EWdaGocTsYE2bz3/NI8efHKwRMHGTxnMJt/2czCQQtpWqWp2yGJ\n5NuG9Za4O+P5dv8wKvQfiPlH0Zr3nl9K8iEs02by1ZavSNiewMIdC0nyJPHHa/7IxDsmUqqEfikk\nCHk8pPeJ5vEjyVRaNJNi14Vu9e4rXvXkjTFdjTGpxphNxphnz/N6eWPMLGPMOmPMBmPMIJ9HKvk2\naf0kHv/yccpcUoY3u7zJgWcO8HbXt5XgJfhYi2dUHCcaRpB0sgGzYtYqwXspz568MaYYsAnoDOwB\nVgL9rbWp2cb8FShvrf2rMaYy8CNQ1Vp7Nsd7qSfvJ2cyztBvaj9uqHUDT1//tNvhiBSItbA/2cPh\ne6LJ2JDMZ93Hk9kmkgcfhNq13Y7Ofwp7nnxbYLO1doe1Nh2IBXrnGGOBclnflwMO5kzw4h+nzp5i\nQuIEWo9pTUZmBg9e/aDbIYnk2/Hj0K+vM+/dRkSQcroBnq/W8rcvIomJKVoJ/mJ505MPA3Zle7wb\nJ/FnNwqYZYzZA5QF+vkmPMmvB2c+SNqxNF7t+Cq9GvfStnsSFM6ehZMnnX1W58+H/7zs4ePS0bSq\nm0zxCTPpE8IzZwqbry683gqstdZ2MsY0AL4xxkRYa4/nHBgTE3Pu+6ioKKKionwUgnyz9RuW7l5K\ncnQyl11ymdvhiHjl1Cno0wcWLoTq1SyDK8Wz8swwSt4/EF4pmjNnEhISSEhI8Ml7edOTbwfEWGu7\nZj1+DrDW2jezjZkDvG6tXZL1+DvgWWvtqhzvpZ58IUg7msbE9RN5d/m7jOk5htsa3eZ2SCLnpKXB\njBmwe7fztWcPHDgAv/wCBw86VXyVKs6aM43eiYbk4FvvvbAV9jz5lUC4MaYOsBfoDwzIMWYH0AVY\nYoypCjQCfipIQOKdE+knmJE6g08SP2Fl2krubHonU/tO1R6rEjASE51cPXEi3HYbhIdDly5Qo4aT\n1CtVgiuugMtKW8xn8XDXMBg40DmgCFbvhSXPJG+tzTDGDAXm4Vyo/chau9EYM9h52Y4BhgPjjTHr\nsw77i7X2l0KLugg7dPIQby55kzGrx9A2rC2DrhrEjH4ztFuTBIxvv4VnnnGq9IEDYflyqJ/bTpEe\nDwzMqt5nhvZeq27RsgZB4vTZ07y7/F1G/DCCPk368FKHl6hVoZbbYYmcc/o0vPEGvPUW/POf8Oij\nUCy3+XvWQnw8DMuq3l95RdX7BWhZgxCX+HMi931+H3Uur8OiBxbRpHITt0MS+Z1Zs2D6dEhKgnr1\nLjDQ43GWjlT17hdahTLAjVk9hi6fduGp655iVv9ZSvASkLZsgRdfhAcfvECCt9ZZOjIiAho0gLVa\nMdIf1K4JYIk/J9Lxk44sfWgpjSs3djsckfP69VeoUweGD3daNOeVvXrXzJl8085QIcZay6T1k+jy\naRfev+19JXgJWLNnwzXXwI035pLgVb27Tj35AJPkSWLo3KEcOX2EL+/9ktY1Wrsdksg5e/Y4s2WW\nL4fvv3fmu7/zDnTtep7B6r0HBFXyAWT+tvl0/KQjdze7m1WPrFKCl4CxcCE0bgwtW8LYsVC6NLz0\nkjMXvls3+M3qGareA4p68gHCWku3Sd3o27yvFhWTgJGeDjExMG6ck9y7d8+R0HNS771QqCcfAp6e\n9zQHTx5kQIucNxOL+F9mptOGadIE1q1zivEePS6Q4FW9Byz15APApPWTmLtlLj88+IPuXBVXjR/v\nJHWPx+m5T53qRa5W7z2gqZJ32ZmMM/x53p+ZdMckKpau6HY4UoSdOQOvvQbFi0PbtvDxx3nka1Xv\nQUE9eRedSD9B38/6UqZkGeLuinM7HCmCTp1yqvelS53KPSLCWW2gZMk8DlTv3a/Ukw9Ch08d5taJ\nt1KxdEUm3j7R7XCkCLEWduxw5rg/9ZRzE9P11ztLEkyfnkeCV/UedNSTd8GOwzvoOaUnHet25O2u\nb1PM6G+tFK4zZ2DMGCc/r18PZcs6eToiAiZNgptu8uJN1HsPSkryfrZk5xLu+uwunr3hWYZFDtP2\nfFLofv4Z2reHhg3hb3+Dq6+GypXz8QY5V4zUeu9BRT15P5qQOIGn5z3NJ30+oVvDbm6HI0XEoEFO\nUh8xogAHq/ceEC6mJ68k7ydvLXmL0StHM/feuTSr0sztcKQIKVsWtm1zdmPymtZ7DyhaTz6AWWt5\n7tvnmLN5DosfXEzN8jXdDkmKkPR0yMiAy/Kzr7t67yFFV/wK0Yn0E/zh8z+QsCOBRYMWKcGLX40d\nC9WqOTNnLr3UiwM0cyYkqZIvJD8d+ok74u6gZdWWLBi4gMsuyU8pJVJwBw86uzS9/jqMHg39+nlx\nkKr3kKVKvhB8teUrrvvoOh66+iEm9JmgBC+F5tQpWLHCuSb67LPQubOzafYXXzjz3++6K483UPUe\n8nTh1YfOZJxh+KLhfLjmQ2LviqVDnQ5uhyQhbOdOZ1XIEiWgRQto2tTJ1Z07e9mD18yZoKELrwEg\n7WgafeL6cGWZK1n9x9VUL1fd7ZAkhCUmwt13w8MPwzPP5LH8b06a916kKMn7QMr+FLpO7MpjrR/j\nufbP6QYnKTQHDzo99tRUZ+PsXPdUzY1670WOevIX6ccDP3LzpzfzWqfX+OuNf1WCl0KTnAxdusBV\nVznz3vOV4NV7L7LUk78ImTaTVu+3Ymjbofzx2j+6HY6EqDNnnJkyo0bB3//uJPd81RLqvQc9rULp\nktikWMqULMMj1zzidigSopYvh2uvhVWrnOL7scfykeBVvQvqyReYtZZXF73KqG6j1KKRQrF+vbPl\n3qhRTh++wNW7eu9Fmir5Avrmp28oZorRqV4nt0OREDVyJDz5JPTvr+pdCk6VfAHM2zqPP0z/A+N6\nj1MVL4XCWpgyBbZsycdBqt7lPFTJ58PJ9JP84/t/cN/n9zG933R6NOrhdkgSgvbudaZHtmwJVat6\ncYCqd7kAVfJe2rh/Iz2n9KRVtVYsfWgp9SvWdzskCTHffQfPP+9U79dd59yvlOc/FFW9Sx6U5L2w\nas8qekzuwVtd3mJgq4FuhyMh4vhx2L0bjh6Fdeuc6n3kSOdO1mJ5/Rtbd62KlzRPPg9pR9OI/DCS\nkd1GcnvT290OR0LI7bc7UyOrVYPy5aFnT3jiCS8O1Lz3Ikdr1xSSTQc30X9qf6LbRCvBi0/t3w+L\nF8PSpRAe7uVBqt6lAJTkz+P4mePnVpN8/sbnebLdk26HJCFk+3Z44AG49958JHj13qWAvJpdY4zp\naoxJNcZsMsY8m8uYKGPMWmNMkjFmgW/D9J9pKdNoMqoJe4/vZcNjG/jzdX/WNEnxCWvhvfegdWu4\n9Vb45z+9PEgzZ+Qi5FnJG2OKAaOAzsAeYKUxZqa1NjXbmArAf4BbrLVpxpjKhRVwYTmTcYZn5j3D\nnM1ziLsrjhtq3+B2SBJCrIU33nC6LUuWQOPGXhyk6l18wJt2TVtgs7V2B4AxJhboDaRmG3MPMM1a\nmwZgrT3g60AL0+FTh7lt8m1UKl2J1X9czeWlLnc7JAkRiYkwZgx8+aUzYyYhAWrmtdWveu/iQ94k\n+TBgV7bHu3ESf3aNgEuy2jRlgfestZ/6JsTCdez0MbpN6kbr6q15t9u7FDO6P0x8Y+FCZzrkE084\nhXiLFpr3Lv7nqwuvJYBrgE5AGWCpMWaptTY/N2W7InpuNC2qtOC9bu+p9y4+sXMn/OUvTlvmk0+g\nWzcvDlL1LoXEmySfBtTO9rhm1nPZ7QYOWGtPAaeMMYuAq4DfJfmYmJhz30dFRREVFZW/iH3oy81f\nMm/rPFKHpCrBi09kZECrVvCnP8FHH0GZMl4cpOpdckhISCAhIcEn75XnzVDGmOLAjzgXXvcCK4AB\n1tqN2cY0AUYCXYFLgeVAP2ttSo73Cpiboay11H6nNhNvn8hNdW9yOxwJEbt2ORdVT5zwYnDO6v2V\nV1S9y3kV6s1Q1toMY8xQYB7OlMuPrLUbjTGDnZftGGttqjHma2A9kAGMyZngA83inYspWbwkHep0\ncDsUCREbNzqbejx73knGOah6Fz8pUssaHD51mJHLRxKbHMv+X/czvNNwbdsnPvHss07//dFHnUXG\nSpbMZaCqdykALWvghWkp03jsi8fo3rA7H/f6mDZhbTSTRnxi2TL49FOnkq9Y8QIDVb2LC4pEkv9i\n0xc8+fWTfHHPF7QJa+N2OBJC0tNh8GD4178ukOA1c0ZcVCSS/NvL3mbELSOU4OWiZGQ40yKXLYOk\nJOcrNRW6d3e26DsvVe/ispDvV/x44EcS9yUSUTXC7VAkiKWkQI0azo1Ne/dCVBS8/76Tw6dOPc9N\nTlpzRgJESFfyp86e4uZPb+atLm/RpHITt8ORIHX2LCxYAO3aOcV4nlS9SwAJ6Uo+LimOZlWa8cDV\nD7gdigSpvn2hbFkYNQq6ds1jsKp3CUAhWclnZGbwn5X/4fXFr/Nxr4/dDkeC0OzZ8I9/OPutbt0K\nYWF5HKDqXQJUyCX5jMwMesf25tiZY8zoN4PImvplk7xZC6tXwxdfOF9HjjhLA/fqBcWL53FwfLyz\njoFmzkgACrkkn7A9gZ8O/UTio4lcUvwSt8ORAHfkCHz2mdOO+fVX6N3bSe7t21/ghqb/8nhgyBBn\nmo2qdwlQIdOTP5Nxhhfnv0i/qf14+aaXleDlgtascZYBrl0b5s6FN9+EH3+EESOgUycvEnx8vNN7\nr19fvXcJaCFRyf9304/LS11O4qOJhJXPq4EqRdH27c4mHgkJMHkyvPSSs6HHBe9SzUnVuwSZoK/k\nrbU8POthmldpzqwBs5Tg5byWL4c2bZzlf4sXhw0bYOjQfCZ4Ve8ShIK+kh+zegxbD21l0h2TtBaN\nnNe8eXDvvTB+PPToUYA3UPUuQSyos+KqPat4ccGLxN4Zy6UlLnU7HAlAy5Y5Cf7zzwuY4FW9S5AL\nykreWktsUixPfP0EH/b8kMaVG7sdkgSgn392Lq5+9JEzWyZfVL1LiAjKSn7QzEG8seQNPu/3Ob2b\n9HY7HAlAy5bBzTc767v36pXPg1W9SwgJuk1Dth/eTusxrdn15C5KX1LaR5FJqKlRA154wbkJ1evt\ne7NX7+PHK7lLwLiYTUOCqpK31vLEV0/waOtHleDlvBITnaV/ixeHBx7IR4JX9S4hKqh68mPXjGXn\nkZ3E3RXndigSoEaMgCZNYPp0L1cXUO9dQlzQVPK7j+7mhfkvMPGOiZpJI+e1di0sXAg33eRlglf1\nLkVA0FTyf/ryTwxpM4RmVZq5HYoEoLQ0eOYZuP126Nkzj8HZq/cZM5yF4kVCVFBU8jNTZ5K8P5nn\n2j/ndigSgCZMcAry1q3h1Veh2IV+qnNW70rwEuICvpLf+stWHv/ycT7p8wmlSmgJV/mt115z5sEv\nXgxNm15goKp3KaICupI/fOowN396M89c/wwd63V0OxwJIKdPOxdXX3sN5s/PI8GrepciLGAreWst\ng+cMpmt4Vx6PfNztcCSAjB0Lzz3n5O2xY6Fu3VwGqnoXCbwkb63l3eXvMnrlaE6kn2DZw8vcDkkC\nyIoV8OKLsGSJM1UyV9l3a/r0U+3WJEVWwN3x+sbiN4hNimVsz7G0rtEa4/XdLFIUdO/uzKB55JFc\nBmSv3seNU/UuISFk7nhNO5rGP3/4J7MHzKZNWBsleDnn9Gn4179g3Tq4775cBqn3LvI7AdWuiUuO\n444md1CrQi23Q5EAsmED9OkDzZvDd9+dp/Oi3rtIrgKqkj+beZaKpfOzVY+EsrNnnWUKOnZ05r/P\nmnWeWTSq3kUuKGAq+UybyYq0FbS8sqXboUgA2L7dWQu+fHln674GDXIMUPUu4pWAqeTnbZ1H6oFU\nnrr+KbdDkQDw5z/DLbfAt9+eJ8GrehfxWsBU8smeZNrVbEfZkmXdDkVckpkJS5fC5MlO9T5pUo6l\nglW9i+RbQFTyy3Yv49/L/s1DVz/kdijigp9+chYXq1sXBg+G6tXhhx+gdPYtA1S9ixSI65V8RmYG\nfT/ry4ibR3BdrevcDkf8yFr47DN4/HFng4+5c6FFixyDVL2LXBTXk/wPu36gYumKDGg5wO1QxI+2\nboV77nFm0EyblstG27prVeSiedWuMcZ0NcakGmM2GWOevcC4NsaYdGPMHd4GEJsUy51N7/R2uISI\nLl3g3nth5crzJHiPx5la8/LLTvX+5ptK8CIFlGeSN8YUA0YBtwLNgQHGmN+tGpI17g3ga28//JeT\nvzAlaQqPXJPbPeoSitLSYM8eZ2mC3639rt67iE95065pC2y21u4AMMbEAr2B1BzjHgemAm28/fC0\no2lUL1ed6uWqe3uIBLlt26BrV+fmpt9cWFXvXaRQeNOuCQN2ZXu8O+u5c4wxNYA+1tr/A7xecGbZ\n7mW6+akIOXQIOnd2cvlf/pLtBVXvIoXGVxde3wGy9+q9SvSbf9lMw0oNfRSCBLLMTGdhsd69nWup\ngKp3ET/wJsmnAbWzPa6Z9Vx2rYFY4ywbWRnoZoxJt9bOyvlmMTExgDN18sP9H/L9K98XJG4JImfP\nwtChcPgwvPVW1pOaOSOSq4SEBBISEnzyXnmuJ2+MKQ78CHQG9gIrgAHW2o25jB8HzLbWTj/Pa+fW\nk/9m6zf8LeFvLH1o6cX9H0jAe/VV+OYbmD0bKpzWeu8i+VWo68lbazOAocA8IBmItdZuNMYMNsb8\n8XyH5PWeJ9NP8o/F/6BD7Q75DliCR0YGvPMOjBzpLFVQ4Wv13kX8zZWdocavG8/kDZP58t4vKV6s\nuN8+X/yrf3/YuxfGvemh/r9UvYsUVNDtDLV6z2o61OmgBB+iDh2Cv/0NvvoKPh8QT/0+qt5F3OL3\nZQ0OnzrM5KTJrB281t8fLYXov62Z2bNh9WoY2M1D2g1DKPOuZs6IuMnvlfzolaPp1bgXtSvUznuw\nBI3t22H4cGc1yQOj4xm1KIIyLVS9i7jN75X8tkPbaFujrb8/VgrZlClw780eeozXvHeRQOL3Sv77\nnd/T/Mrm/v5YKUTDh8PW1+N5e7567yKBxpWefJ0Kdfz9sVJYPB7avDWEJ65I4pJ4Ve8igcbvlXz9\nivWpVaGWvz9WCkPWmjO7S9Zn5wxV7yKByO+V/PM3Pk8xExC7DkpBZVtzZu3LM3j6+XYkVXU7KBE5\nH79n2xtq3eDvjxRfyqreD1Wqz+t913Lry+347DMIC8v7UBHxP79X8hVLV/T3R4ovZFXvmRuS+KTP\nDB78oB2DB8OiRdDkd1vIiEigUN9E8hYfT3qzCOZvr0+dg2uZtrsdK1fC++8rwYsEOtc38pYA5vFw\ndvAQ0tcm8VjlGdC8HfMnQ0NtASASNJTk5ZzMTFi1CubNg1Kz4hm45k9MMAOZ2uJTmrQqxauvQs2a\nbkcpIvnhyiqUElg8Hhg92vlqXNHDKDuEer8mcfBf46h1VztKqBQQcVXQrUIpgaNlS2jUCH78Eda/\nGM/3RyJ6+RKfAAAMcUlEQVS46vb6lN+6lnr9leBFgp0q+SLOGMjY66HY41rvXSRQqZKXAlmyBIZe\nGU+xVlpzRiRUqZIvok5s97AycggRJomKM1S9iwQyVfKSL0kvx/NreARHKtWjdKqqd5FQpstqRciG\n7zwcHzSEyj8nkfrGTHo9Hel2SCJSyFTJFwHffw//aBVP1VsiKB5en+p713KjErxIkaCefAj74Qd4\n8ykPj24YQrtySZSJH0/JG5XcRYKNevLyO++9BxN7xTMlJYJbH6tPxW1rleBFiiBV8iHoQIqHH64e\nwq01k7h08niIVHIXCWaq5IWMDPj6K8u7N8ST0SKCitfW59LktUrwIkWcKvkgl5kJsbHwzvMeXj8a\nzdUlkyk+YTwVblFyFwkVquSLkAMH4OOP4e67nXVnypaxpLwcx+KjEXR+pAGVtq9VgheRc1TJB4E9\ne+Dzz2H6dGcp4FtugV694OowD03ei6bEj8kwfrxaMyIhSpV8CBs3Dpo3h2XLYOhQ2LsXPou33Fcy\njhb3RFCicQNnzRkleBE5D93xGsDmz4e//tVJ8I0bZz3p8cD90ZCcDDNnKrmLyAWpkg9QBw7AgAHw\n4YdZCd5aiIuDiAhooOpdRLyjSj7AbN8O77wDEybAgw9Cjx441Xu0qncRyT9V8gHi2DF4/nm49loo\nVQoSE2HEPy0mXtW7iBScKnmXHTwIH3wAo0ZBly6wfj2EheFU73dHQ0qKqncRKTBV8i77wx+cAv3L\nL50WTViNbL338HBYs0YJXkQKTJW8y44dgxdegKuu4n+9d1XvIuIjXlXyxpiuxphUY8wmY8yz53n9\nHmNMYtbXYmNMS9+HGsKsqncRKRx5VvLGmGLAKKAzsAdYaYyZaa1NzTbsJ6CDtfaIMaYrMBbQnnJ5\nOHEC8Hho8lI0eFS9i4jveVPJtwU2W2t3WGvTgVigd/YB1tpl1tojWQ+XAWG+DTP0HD1iGd4yjjk7\nI6jQWtW7iBQOb3ryYcCubI934yT+3DwMfHkxQYU6u8/D5nbRPHYwhQoJMzHtlNxFpHD49MKrMaYj\n8ADQPrcxMTEx576PiooiKirKlyEEtPQzlpXPxNP4/4axvuIgbl03EVO/lNthiUiASUhIICEhwSfv\nlecqlMaYdkCMtbZr1uPnAGutfTPHuAhgGtDVWrs1l/cqcqtQbtvmXFNNXeThzvnRtCyewraXxnHj\n05GU0NwmEfFCYa9CuRIIN8bUMcaUBPoDs3IEUBsnwd+XW4IvitauhevaWarMj+P/foigwwPh1D24\nho7PKcGLiH/kmWqstRnGmKHAPJw/Ch9ZazcaYwY7L9sxwEtAJWC0McYA6dbaC/XtQ97evfCn/h6W\n14mmzu4U+HompXVhVUT8TJuG+FhiIjwxzFJvZTzvMoyyQwdhXolxFqQRESmAi2nXqGngI9bCpEnw\n2jAPc+tGU6d2CsXGa967iLhLa9f4yIdjLav+Esd6E0G9m8Mptlbz3kXEfWrX+MChHz2kdoom/EwK\nVeaMU3IXEZ/SHq9usZaF0XGcaRrBwYrhnPpB1buIBBb15AvK4yH9kWiunJ3CySkzua2fkruIBB5V\n8vmVtWKkjYhg7bFwbqu+hrpK8CISoFTJ50fWeu82JYXoGjNJSo9kxlduByUikjtV8t7IVr1vMeHc\nUGoNK0wks2dDS62cLyIBTJV8XrKq9xOrUri/5Ey2/RTJs8/BXXdBMf2JFJEApzSVm2y7NZ2uHU6d\ng2sYNDqSVaugb18leBEJDponfz7Z91odNw5PvUhatHCeFhHxN82T95Vs1bttEM7OGWv4+nAkY8a4\nHZiISMGokv+vrOr91JoURjQfx7+XRHLppdCsGTRtCjfeCP36uR2kiBRFWqDsYlgL8fEwbBi7Og/i\nuqMTGXJ9KZI+gBo13A5OROTiFO0kn6337hk7kx4vRDJiFPTv73ZgIiK+UTR78tZyekIcZ5tHsNWE\n0/XKNTS4J5I771RLRkRCS5Gp5BMSYNQoOLbVw9CNzoqRL9WcyZHDkdzeF2Z/A5dc4naUIiK+VSQu\nvP78M/Tobrm/dDyPpQ7jzIBBXPZWDMUu025NIhL4dOE1F6dPw7vvwsdveJheLZqmp1Iwc2dSUssB\ni0gREdI9+XaRzrz3pOIRNOsVjlmj9d5FpGgJ2XZNepqH2bWi6dM4hWLjtVuTiAQv3fGazdEjlkVD\n4jhcJ4KMeuGwWtW7iBRdIdOT37gR4kZ6aPF/0bQtm8JP/57JXY9HYgr0t09EJDSERJJ/9x3LD0/G\n80GpYRwdMIjaH0+kdinNnBERCf6evMfDtGrRdKmRQoVp6r2LSOgpcj15a2HFcqf3frJRBFtsOJdt\nVO9dRCSnoGvXbNgA/37OQ+950bS5LIU3r5/JwfBISpR1OzIRkcATFO2ajAyYPRvee9fSODGet84M\n4/SAQVQeGQPqvYtIiAvpO17Xr3cWDat7mYfxJaKpVTUFM34m5dSaERHJU8BV8keOwJIlTvUOMPo/\nlj+UjOfeFcNg0CCIiVH1LiJFSkhU8hkZ8NRTzlozALfdBhVOe/j7pmgiiqfAzJm6sCoikk+uz645\nftzJ3x07wtSpsGoV2EzL7D/EMXF9BG36h3NpsmbOiIgUhGuVvLXQuDFs3gy1akH37vD++9Cssgfu\ndnZrUvUuInJxXEvy993n7L539iwUL85v9lpl0CCYOFG9dxGRi+RKkk9MhC++gOXLsxJ8tr1WVb2L\niPiOVz15Y0xXY0yqMWaTMebZXMa8Z4zZbIxZZ4xpldt7JSc7G2VHR0Ojhs5670REQHg4aL13ERGf\nyjPJG2OKAaOAW4HmwABjTJMcY7oBDay1DYHBwPu5vd/o0dCsGTz7gAfT92545RWnen/jjSLVnklI\nSHA7hIChc/E/Ohf/o3PhG95U8m2BzdbaHdbadCAW6J1jTG9gAoC1djlQwRhT9XxvNmWyZWiVOMq3\nL9rVu36A/0fn4n90Lv5H58I3vOnJhwG7sj3ejZP4LzQmLeu5fTnfbFX9u6m3SL13ERF/8Ps8+fo3\na69VERF/yXNZA2NMOyDGWts16/FzgLXWvpltzPvAAmttXNbjVOAma+2+HO/lvzUURERCSGEua7AS\nCDfG1AH2Av2BATnGzAKGAHFZfxQO50zwFxOkiIgUTJ5J3lqbYYwZCszDae98ZK3daIwZ7Lxsx1hr\n5xpjuhtjtgC/Ag8UbtgiIuINv65CKSIi/lUoF159efNUsMvrXBhj7jHGJGZ9LTbGtHQjTn/w5uci\na1wbY0y6MeYOf8bnT17+jkQZY9YaY5KMMQv8HaO/ePE7Ut4YMysrV2wwxgxyIcxCZ4z5yBizzxiz\n/gJj8p83rbU+/cL5w7EFqANcAqwDmuQY0w34Iuv7SGCZr+MIhC8vz0U7oELW912L8rnINu47YA5w\nh9txu/hzUQFIBsKyHld2O24Xz8Vfgdf/ex6Ag0AJt2MvhHPRHmgFrM/l9QLlzcKo5H1681SQy/Nc\nWGuXWWuPZD1chnN/QSjy5ucC4HFgKuDxZ3B+5s25uAeYZq1NA7DWHvBzjP7izbmwQLms78sBB621\nZ/0Yo19YaxcDhy4wpEB5szCS/PlunsqZuHK7eSrUeHMusnsY+LJQI3JPnufCGFMD6GOt/T8glGdi\nefNz0QioZIxZYIxZaYy5z2/R+Zc352IU0MwYswdIBIb5KbZAU6C8GTA7QxV1xpiOOLOS2rsdi4ve\nAbL3ZEM50eelBHAN0AkoAyw1xiy11m5xNyxX3AqstdZ2MsY0AL4xxkRYa4+7HVgwKIwknwbUzva4\nZtZzOcfUymNMKPDmXGCMiQDGAF2ttRf651ow8+ZctAZijTEGp/fazRiTbq2d5acY/cWbc7EbOGCt\nPQWcMsYsAq7C6V+HEm/OxQPA6wDW2q3GmG1AE2CVXyIMHAXKm4XRrjl385QxpiTOzVM5f0lnAffD\nuTtqz3vzVAjI81wYY2oD04D7rLVbXYjRX/I8F9ba+llf9XD68tEhmODBu9+RmUB7Y0xxY8xlOBfa\nNvo5Tn/w5lzsALoAZPWgGwE/+TVK/zHk/i/YAuVNn1fyVjdPnePNuQBeAioBo7Mq2HRrbc4F4IKe\nl+fiN4f4PUg/8fJ3JNUY8zWwHsgAxlhrU1wMu1B4+XMxHBifbWrhX6y1v7gUcqExxkwGooArjDE7\ngZeBklxk3tTNUCIiIczvq1CKiIj/KMmLiIQwJXkRkRCmJC8iEsKU5EVEQpiSvIhICFOSFxEJYUry\nIiIh7P8BiFqi9/9/KHEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2e9e0e48>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "roc_valid = gbm_model.roc(valid=True)\n",
    "roc_train= gbm_model.roc()\n",
    "\n",
    "plt.plot(roc_valid[0],roc_valid[1])\n",
    "plt.plot(roc_train[0],roc_train[1])\n",
    "plt.plot(np.arange(0.0,1.1,0.1),np.arange(0.0,1.1,0.1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "deeplearning Model Build progress: |██████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "dl = H2ODeepLearningEstimator(seed =1,epochs =15,hidden=[20,20], model_id='dl1')\n",
    "dl.train(x = ['query','query_type','passages'], y = 'answers_yn', training_frame=hdf_train, validation_frame=hdf_validate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEZCAYAAACEkhK6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2e62c080>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "dl.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x467f5550>]"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXkAAAEACAYAAABWLgY0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMpiKvIhIBvv/Qx8ea1YeeOwAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x228a3b38>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "roc_valid = dl.roc(valid=True)\n",
    "roc_train= dl.roc()\n",
    "\n",
    "plt.plot(roc_valid[0],roc_valid[1])\n",
    "plt.plot(roc_train[0],roc_train[1])\n",
    "plt.plot(np.arange(0.0,1.1,0.1),np.arange(0.0,1.1,0.1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Reterieve best result\n",
    "\n",
    "dl = H2ODeepLearningEstimator(seed =1,epochs =10,hidden=[20,20], model_id='dl1')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "H2OResponseError",
     "evalue": "Server error water.exceptions.H2OKeyNotFoundArgumentException:\n  Error: Object 'dl1' not found in function: predict for argument: model\n  Request: POST /4/Predictions/models/dl1/frames/py_23_sid_9d1e\n",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mH2OResponseError\u001b[0m                          Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-132-643cef721f0c>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mprediction\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mdl\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[1;33m(\u001b[0m \u001b[0mhdf_validate\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'query'\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m'query_type'\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m'passages'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      2\u001b[0m \u001b[0mpdf_pred\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mprediction\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mas_data_frame\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\DATO\\Anaconda2\\envs\\gl-env\\lib\\site-packages\\h2o\\model\\model_base.pyc\u001b[0m in \u001b[0;36mpredict\u001b[1;34m(self, test_data)\u001b[0m\n\u001b[0;32m    146\u001b[0m         \"\"\"\n\u001b[0;32m    147\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtest_data\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mh2o\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mH2OFrame\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m \u001b[1;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"test_data must be an instance of H2OFrame\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 148\u001b[1;33m         j = H2OJob(h2o.api(\"POST /4/Predictions/models/%s/frames/%s\" % (self.model_id, test_data.frame_id)),\n\u001b[0m\u001b[0;32m    149\u001b[0m                    self._model_json['algo'] + \" prediction\")\n\u001b[0;32m    150\u001b[0m         \u001b[0mj\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpoll\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\DATO\\Anaconda2\\envs\\gl-env\\lib\\site-packages\\h2o\\h2o.pyc\u001b[0m in \u001b[0;36mapi\u001b[1;34m(endpoint, data, json, filename, save_to)\u001b[0m\n\u001b[0;32m     82\u001b[0m     \u001b[1;31m# type checks are performed in H2OConnection class\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     83\u001b[0m     \u001b[0m_check_connection\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 84\u001b[1;33m     \u001b[1;32mreturn\u001b[0m \u001b[0mh2oconn\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrequest\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mendpoint\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdata\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mdata\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mjson\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mjson\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mfilename\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mfilename\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msave_to\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0msave_to\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     85\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     86\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\DATO\\Anaconda2\\envs\\gl-env\\lib\\site-packages\\h2o\\backend\\connection.pyc\u001b[0m in \u001b[0;36mrequest\u001b[1;34m(self, endpoint, data, json, filename, save_to)\u001b[0m\n\u001b[0;32m    257\u001b[0m                                     auth=self._auth, verify=self._verify_ssl_cert, proxies=self._proxies)\n\u001b[0;32m    258\u001b[0m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_log_end_transaction\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mstart_time\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mresp\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 259\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_process_response\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mresp\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msave_to\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    260\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    261\u001b[0m         \u001b[1;32mexcept\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mrequests\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mexceptions\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mConnectionError\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mrequests\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mexceptions\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mHTTPError\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\DATO\\Anaconda2\\envs\\gl-env\\lib\\site-packages\\h2o\\backend\\connection.pyc\u001b[0m in \u001b[0;36m_process_response\u001b[1;34m(response, save_to)\u001b[0m\n\u001b[0;32m    584\u001b[0m         \u001b[1;31m# Client errors (400 = \"Bad Request\", 404 = \"Not Found\", 412 = \"Precondition Failed\")\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    585\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mstatus_code\u001b[0m \u001b[1;32min\u001b[0m \u001b[1;33m{\u001b[0m\u001b[1;36m400\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m404\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m412\u001b[0m\u001b[1;33m}\u001b[0m \u001b[1;32mand\u001b[0m \u001b[0misinstance\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mH2OErrorV3\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mH2OModelBuilderErrorV3\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 586\u001b[1;33m             \u001b[1;32mraise\u001b[0m \u001b[0mH2OResponseError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    587\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    588\u001b[0m         \u001b[1;31m# Server errors (notably 500 = \"Server Error\")\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mH2OResponseError\u001b[0m: Server error water.exceptions.H2OKeyNotFoundArgumentException:\n  Error: Object 'dl1' not found in function: predict for argument: model\n  Request: POST /4/Predictions/models/dl1/frames/py_23_sid_9d1e\n"
     ]
    }
   ],
   "source": [
    "prediction = dl.predict(hdf_validate[['query','query_type','passages']])\n",
    "pdf_pred=prediction.as_data_frame()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}