boxoffice.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Box Office"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Kaggle TMDB Box Office Prediction Competition 2019\n",
    "\n",
    "Authors: Sean Trinh, Hariharan Vijayachandran"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Part 1 - Process the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "def get_data():\n",
    "    #Get the training and testing data\n",
    "    return [pd.read_csv('./data/train.csv'), pd.read_csv('./data/test.csv')]\n",
    "\n",
    "def drop(to_drop, train, test):\n",
    "    #Drop specified attributes from the training and test data sets\n",
    "    for x in to_drop:\n",
    "        try:\n",
    "            train = train.drop(x, axis = 1)\n",
    "            if x != 'revenue':\n",
    "                test = test.drop(x, axis = 1)\n",
    "        except:\n",
    "            pass\n",
    "    return [train, test]\n",
    "\n",
    "def split_to_validate(train, fraction, random):\n",
    "    #Split training data into training and validation data\n",
    "    ret_train = train.sample(frac=fraction, random_state=random)\n",
    "    ret_test = train.drop(ret_train.index)\n",
    "    return [ret_train, ret_test]\n",
    "\n",
    "def ap(x):\n",
    "    #Convert years (ex: 86 -> 1986, 15 -> 2015)\n",
    "    if type(x) == str:\n",
    "        x = int(x.split('/')[2])\n",
    "        if x<=19:\n",
    "            return 2000+x\n",
    "        else:\n",
    "            return 1900+x\n",
    "    else:\n",
    "        return 0\n",
    "def f(x):\n",
    "    if x<0:\n",
    "        return 0\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>belongs_to_collection</th>\n",
       "      <th>budget</th>\n",
       "      <th>genres</th>\n",
       "      <th>homepage</th>\n",
       "      <th>imdb_id</th>\n",
       "      <th>original_language</th>\n",
       "      <th>original_title</th>\n",
       "      <th>overview</th>\n",
       "      <th>popularity</th>\n",
       "      <th>...</th>\n",
       "      <th>release_date</th>\n",
       "      <th>runtime</th>\n",
       "      <th>spoken_languages</th>\n",
       "      <th>status</th>\n",
       "      <th>tagline</th>\n",
       "      <th>title</th>\n",
       "      <th>Keywords</th>\n",
       "      <th>cast</th>\n",
       "      <th>crew</th>\n",
       "      <th>revenue</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>[{'id': 313576, 'name': 'Hot Tub Time Machine ...</td>\n",
       "      <td>14000000</td>\n",
       "      <td>[{'id': 35, 'name': 'Comedy'}]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>tt2637294</td>\n",
       "      <td>en</td>\n",
       "      <td>Hot Tub Time Machine 2</td>\n",
       "      <td>When Lou, who has become the \"father of the In...</td>\n",
       "      <td>6.575393</td>\n",
       "      <td>...</td>\n",
       "      <td>2/20/15</td>\n",
       "      <td>93.0</td>\n",
       "      <td>[{'iso_639_1': 'en', 'name': 'English'}]</td>\n",
       "      <td>Released</td>\n",
       "      <td>The Laws of Space and Time are About to be Vio...</td>\n",
       "      <td>Hot Tub Time Machine 2</td>\n",
       "      <td>[{'id': 4379, 'name': 'time travel'}, {'id': 9...</td>\n",
       "      <td>[{'cast_id': 4, 'character': 'Lou', 'credit_id...</td>\n",
       "      <td>[{'credit_id': '59ac067c92514107af02c8c8', 'de...</td>\n",
       "      <td>12314651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>[{'id': 107674, 'name': 'The Princess Diaries ...</td>\n",
       "      <td>40000000</td>\n",
       "      <td>[{'id': 35, 'name': 'Comedy'}, {'id': 18, 'nam...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>tt0368933</td>\n",
       "      <td>en</td>\n",
       "      <td>The Princess Diaries 2: Royal Engagement</td>\n",
       "      <td>Mia Thermopolis is now a college graduate and ...</td>\n",
       "      <td>8.248895</td>\n",
       "      <td>...</td>\n",
       "      <td>8/6/04</td>\n",
       "      <td>113.0</td>\n",
       "      <td>[{'iso_639_1': 'en', 'name': 'English'}]</td>\n",
       "      <td>Released</td>\n",
       "      <td>It can take a lifetime to find true love; she'...</td>\n",
       "      <td>The Princess Diaries 2: Royal Engagement</td>\n",
       "      <td>[{'id': 2505, 'name': 'coronation'}, {'id': 42...</td>\n",
       "      <td>[{'cast_id': 1, 'character': 'Mia Thermopolis'...</td>\n",
       "      <td>[{'credit_id': '52fe43fe9251416c7502563d', 'de...</td>\n",
       "      <td>95149435</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3300000</td>\n",
       "      <td>[{'id': 18, 'name': 'Drama'}]</td>\n",
       "      <td>http://sonyclassics.com/whiplash/</td>\n",
       "      <td>tt2582802</td>\n",
       "      <td>en</td>\n",
       "      <td>Whiplash</td>\n",
       "      <td>Under the direction of a ruthless instructor, ...</td>\n",
       "      <td>64.299990</td>\n",
       "      <td>...</td>\n",
       "      <td>10/10/14</td>\n",
       "      <td>105.0</td>\n",
       "      <td>[{'iso_639_1': 'en', 'name': 'English'}]</td>\n",
       "      <td>Released</td>\n",
       "      <td>The road to greatness can take you to the edge.</td>\n",
       "      <td>Whiplash</td>\n",
       "      <td>[{'id': 1416, 'name': 'jazz'}, {'id': 1523, 'n...</td>\n",
       "      <td>[{'cast_id': 5, 'character': 'Andrew Neimann',...</td>\n",
       "      <td>[{'credit_id': '54d5356ec3a3683ba0000039', 'de...</td>\n",
       "      <td>13092000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1200000</td>\n",
       "      <td>[{'id': 53, 'name': 'Thriller'}, {'id': 18, 'n...</td>\n",
       "      <td>http://kahaanithefilm.com/</td>\n",
       "      <td>tt1821480</td>\n",
       "      <td>hi</td>\n",
       "      <td>Kahaani</td>\n",
       "      <td>Vidya Bagchi (Vidya Balan) arrives in Kolkata ...</td>\n",
       "      <td>3.174936</td>\n",
       "      <td>...</td>\n",
       "      <td>3/9/12</td>\n",
       "      <td>122.0</td>\n",
       "      <td>[{'iso_639_1': 'en', 'name': 'English'}, {'iso...</td>\n",
       "      <td>Released</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Kahaani</td>\n",
       "      <td>[{'id': 10092, 'name': 'mystery'}, {'id': 1054...</td>\n",
       "      <td>[{'cast_id': 1, 'character': 'Vidya Bagchi', '...</td>\n",
       "      <td>[{'credit_id': '52fe48779251416c9108d6eb', 'de...</td>\n",
       "      <td>16000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>[{'id': 28, 'name': 'Action'}, {'id': 53, 'nam...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>tt1380152</td>\n",
       "      <td>ko</td>\n",
       "      <td>마린보이</td>\n",
       "      <td>Marine Boy is the story of a former national s...</td>\n",
       "      <td>1.148070</td>\n",
       "      <td>...</td>\n",
       "      <td>2/5/09</td>\n",
       "      <td>118.0</td>\n",
       "      <td>[{'iso_639_1': 'ko', 'name': '한국어/조선말'}]</td>\n",
       "      <td>Released</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Marine Boy</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[{'cast_id': 3, 'character': 'Chun-soo', 'cred...</td>\n",
       "      <td>[{'credit_id': '52fe464b9251416c75073b43', 'de...</td>\n",
       "      <td>3923970</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 23 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   id                              belongs_to_collection    budget  \\\n",
       "0   1  [{'id': 313576, 'name': 'Hot Tub Time Machine ...  14000000   \n",
       "1   2  [{'id': 107674, 'name': 'The Princess Diaries ...  40000000   \n",
       "2   3                                                NaN   3300000   \n",
       "3   4                                                NaN   1200000   \n",
       "4   5                                                NaN         0   \n",
       "\n",
       "                                              genres  \\\n",
       "0                     [{'id': 35, 'name': 'Comedy'}]   \n",
       "1  [{'id': 35, 'name': 'Comedy'}, {'id': 18, 'nam...   \n",
       "2                      [{'id': 18, 'name': 'Drama'}]   \n",
       "3  [{'id': 53, 'name': 'Thriller'}, {'id': 18, 'n...   \n",
       "4  [{'id': 28, 'name': 'Action'}, {'id': 53, 'nam...   \n",
       "\n",
       "                            homepage    imdb_id original_language  \\\n",
       "0                                NaN  tt2637294                en   \n",
       "1                                NaN  tt0368933                en   \n",
       "2  http://sonyclassics.com/whiplash/  tt2582802                en   \n",
       "3         http://kahaanithefilm.com/  tt1821480                hi   \n",
       "4                                NaN  tt1380152                ko   \n",
       "\n",
       "                             original_title  \\\n",
       "0                    Hot Tub Time Machine 2   \n",
       "1  The Princess Diaries 2: Royal Engagement   \n",
       "2                                  Whiplash   \n",
       "3                                   Kahaani   \n",
       "4                                      마린보이   \n",
       "\n",
       "                                            overview  popularity  ...  \\\n",
       "0  When Lou, who has become the \"father of the In...    6.575393  ...   \n",
       "1  Mia Thermopolis is now a college graduate and ...    8.248895  ...   \n",
       "2  Under the direction of a ruthless instructor, ...   64.299990  ...   \n",
       "3  Vidya Bagchi (Vidya Balan) arrives in Kolkata ...    3.174936  ...   \n",
       "4  Marine Boy is the story of a former national s...    1.148070  ...   \n",
       "\n",
       "  release_date runtime                                   spoken_languages  \\\n",
       "0      2/20/15    93.0           [{'iso_639_1': 'en', 'name': 'English'}]   \n",
       "1       8/6/04   113.0           [{'iso_639_1': 'en', 'name': 'English'}]   \n",
       "2     10/10/14   105.0           [{'iso_639_1': 'en', 'name': 'English'}]   \n",
       "3       3/9/12   122.0  [{'iso_639_1': 'en', 'name': 'English'}, {'iso...   \n",
       "4       2/5/09   118.0           [{'iso_639_1': 'ko', 'name': '한국어/조선말'}]   \n",
       "\n",
       "     status                                            tagline  \\\n",
       "0  Released  The Laws of Space and Time are About to be Vio...   \n",
       "1  Released  It can take a lifetime to find true love; she'...   \n",
       "2  Released    The road to greatness can take you to the edge.   \n",
       "3  Released                                                NaN   \n",
       "4  Released                                                NaN   \n",
       "\n",
       "                                      title  \\\n",
       "0                    Hot Tub Time Machine 2   \n",
       "1  The Princess Diaries 2: Royal Engagement   \n",
       "2                                  Whiplash   \n",
       "3                                   Kahaani   \n",
       "4                                Marine Boy   \n",
       "\n",
       "                                            Keywords  \\\n",
       "0  [{'id': 4379, 'name': 'time travel'}, {'id': 9...   \n",
       "1  [{'id': 2505, 'name': 'coronation'}, {'id': 42...   \n",
       "2  [{'id': 1416, 'name': 'jazz'}, {'id': 1523, 'n...   \n",
       "3  [{'id': 10092, 'name': 'mystery'}, {'id': 1054...   \n",
       "4                                                NaN   \n",
       "\n",
       "                                                cast  \\\n",
       "0  [{'cast_id': 4, 'character': 'Lou', 'credit_id...   \n",
       "1  [{'cast_id': 1, 'character': 'Mia Thermopolis'...   \n",
       "2  [{'cast_id': 5, 'character': 'Andrew Neimann',...   \n",
       "3  [{'cast_id': 1, 'character': 'Vidya Bagchi', '...   \n",
       "4  [{'cast_id': 3, 'character': 'Chun-soo', 'cred...   \n",
       "\n",
       "                                                crew   revenue  \n",
       "0  [{'credit_id': '59ac067c92514107af02c8c8', 'de...  12314651  \n",
       "1  [{'credit_id': '52fe43fe9251416c7502563d', 'de...  95149435  \n",
       "2  [{'credit_id': '54d5356ec3a3683ba0000039', 'de...  13092000  \n",
       "3  [{'credit_id': '52fe48779251416c9108d6eb', 'de...  16000000  \n",
       "4  [{'credit_id': '52fe464b9251416c75073b43', 'de...   3923970  \n",
       "\n",
       "[5 rows x 23 columns]"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''\n",
    "    Process the data.\n",
    "'''\n",
    "import pandas as pd\n",
    "\n",
    "#Get training and test data for all models\n",
    "train_all, test_all = get_data()\n",
    "cc_train, cc_test = get_data()\n",
    "date_train, date_test = get_data()\n",
    "\n",
    "train_df, test_df = get_data()\n",
    "\n",
    "# #Train all models on the same training and validation data\n",
    "training, validation = split_to_validate(train_df, 0.7, 200)\n",
    "\n",
    "train = training\n",
    "test = validation\n",
    "cc_train_s = training\n",
    "cc_test_s = validation\n",
    "date_train_s = training\n",
    "date_test_s = validation\n",
    "title_train_s = training\n",
    "title_test_s = validation\n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 3000 entries, 0 to 2999\n",
      "Data columns (total 23 columns):\n",
      "id                       3000 non-null int64\n",
      "belongs_to_collection    604 non-null object\n",
      "budget                   3000 non-null int64\n",
      "genres                   2993 non-null object\n",
      "homepage                 946 non-null object\n",
      "imdb_id                  3000 non-null object\n",
      "original_language        3000 non-null object\n",
      "original_title           3000 non-null object\n",
      "overview                 2992 non-null object\n",
      "popularity               3000 non-null float64\n",
      "poster_path              2999 non-null object\n",
      "production_companies     2844 non-null object\n",
      "production_countries     2945 non-null object\n",
      "release_date             3000 non-null object\n",
      "runtime                  2998 non-null float64\n",
      "spoken_languages         2980 non-null object\n",
      "status                   3000 non-null object\n",
      "tagline                  2403 non-null object\n",
      "title                    3000 non-null object\n",
      "Keywords                 2724 non-null object\n",
      "cast                     2987 non-null object\n",
      "crew                     2984 non-null object\n",
      "revenue                  3000 non-null int64\n",
      "dtypes: float64(2), int64(3), object(18)\n",
      "memory usage: 539.1+ KB\n"
     ]
    }
   ],
   "source": [
    "train_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "#ALL ATTRIBUTES\n",
    "ALL_ATTRIBUTES = ['id','belongs_to_collection','budget','genres','homepage','imdb_id','original_language',\n",
    "                 'original_title','overview','popularity','poster_path','production_companies',\n",
    "                 'production_countries','release_date','runtime','spoken_languages', 'status', 'tagline',\n",
    "                 'title','Keywords','cast','crew','revenue']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/ipykernel_launcher.py:66: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/ipykernel_launcher.py:67: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/ipykernel_launcher.py:70: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/ipykernel_launcher.py:71: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/ipykernel_launcher.py:101: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/ipykernel_launcher.py:102: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n"
     ]
    }
   ],
   "source": [
    "to_drop = ['release_date','id','genres','homepage','original_title','title','original_language','overview',\n",
    "           'production_companies','production_countries','spoken_languages','status','tagline',\n",
    "           'belongs_to_collection', 'imdb_id','poster_path','Keywords','cast','crew']\n",
    "train_df, test_df = drop(to_drop, train_df, test_df) #testing\n",
    "train, test = drop(to_drop, train, test) #training and validation\n",
    "train_all, test_all = drop(to_drop, train_all, test_all)\n",
    "\n",
    "#cc_train, cc_test\n",
    "to_drop1 = ['budget','runtime','popularity','release_date','id','genres','homepage','original_title',\n",
    "            'title','original_language','overview','production_companies','production_countries',\n",
    "            'spoken_languages','status','tagline','belongs_to_collection', 'imdb_id','poster_path','Keywords']\n",
    "cc_train, cc_test = drop(to_drop1, cc_train, cc_test) #testing\n",
    "cc_train_s, cc_test_s = drop(to_drop1, cc_train_s, cc_test_s) #training and validation\n",
    "\n",
    "#Count the number of cast members\n",
    "cc_train_s['cast'] = cc_train_s['cast'].apply(lambda x: x.count('{') if type(x)==str else 0)\n",
    "cc_test_s['cast'] = cc_test_s['cast'].apply(lambda x: x.count('{') if type(x)==str else 0)\n",
    "cc_test['cast'] = cc_test['cast'].apply(lambda x: x.count('{') if type(x)==str else 0)\n",
    "\n",
    "#Count the number of crew members\n",
    "cc_train_s['crew'] = cc_train_s['crew'].apply(lambda x: x.count('{') if type(x)==str else 0)\n",
    "cc_test_s['crew'] = cc_test_s['crew'].apply(lambda x: x.count('{') if type(x)==str else 0)\n",
    "cc_test['crew'] = cc_test['crew'].apply(lambda x: x.count('{') if type(x)==str else 0)\n",
    "\n",
    "#on all\n",
    "cc_train['cast'] = cc_train['cast'].apply(lambda x: x.count('{') if type(x)==str else 0)\n",
    "cc_train['crew'] = cc_train['crew'].apply(lambda x: x.count('{') if type(x)==str else 0)\n",
    "\n",
    "#date_train, date_test\n",
    "to_drop2 = ['budget','runtime','popularity','id','genres','homepage','original_title','title','original_language',\n",
    "            'overview','production_companies','production_countries','spoken_languages','status','tagline',\n",
    "            'belongs_to_collection', 'imdb_id','poster_path','Keywords','cast','crew']\n",
    "date_train, date_test = drop(to_drop2, date_train, date_test) #testing\n",
    "date_train_s, date_test_s = drop(to_drop2, date_train_s, date_test_s) #training and validation\n",
    "\n",
    "#Get month and year\n",
    "date_train_s['release_month'] = date_train_s['release_date'].apply(lambda x:int(x.split('/')[0]) if type(x)==str else 0 )\n",
    "date_train_s['release_year'] = date_train_s['release_date'].apply(lambda x:int(x.split('/')[2])if type(x)==str else 0 )\n",
    "\n",
    "date_test_s['release_month'] = date_test_s['release_date'].apply(lambda x:int(x.split('/')[0])if type(x)==str else 0 )\n",
    "date_test_s['release_year'] = date_test_s['release_date'].apply(lambda x:int(x.split('/')[2]) if type(x)==str else 0)\n",
    "\n",
    "date_test['release_month'] = date_test['release_date'].apply(lambda x:int(x.split('/')[0])if type(x)==str else 0 )\n",
    "date_test['release_year'] = date_test['release_date'].apply(lambda x:int(x.split('/')[2]) if type(x)==str else 0)\n",
    "\n",
    "#Convert year \n",
    "date_train_s['release_year'] = date_train_s['release_date'].apply(ap)\n",
    "date_test_s['release_year'] = date_test_s['release_date'].apply(ap)\n",
    "date_test['release_year'] = date_test['release_date'].apply(ap)\n",
    "\n",
    "#Drop the release_date itself\n",
    "date_train_s.drop('release_date',axis = 1, inplace = True)\n",
    "date_test_s.drop('release_date',axis = 1, inplace = True)\n",
    "date_test.drop('release_date',axis = 1, inplace = True)\n",
    "\n",
    "#on all\n",
    "date_train['release_month'] = date_train['release_date'].apply(lambda x:int(x.split('/')[0]) if type(x)==str else 0 )\n",
    "date_train['release_year'] = date_train['release_date'].apply(lambda x:int(x.split('/')[2]) if type(x)==str else 0 )\n",
    "date_train['release_year'] = date_train['release_date'].apply(ap)\n",
    "date_train.drop('release_date', axis = 1, inplace = True)\n",
    "\n",
    "#make dataframe for title embedding layer \n",
    "# title_train_s = training\n",
    "# title_test_s = validation\n",
    "title_train_s = title_train_s[['revenue','title']]\n",
    "title_train_s['revenue'] = title_train_s['revenue'].apply(np.log)\n",
    "title_train_s['revenue'] = title_train_s['revenue'].apply(f)\n",
    "\n",
    "title_test_s = title_test_s[['revenue','title']]\n",
    "title_test_s['revenue'] = title_test_s['revenue'].apply(np.log)\n",
    "title_test_s['revenue'] = title_test_s['revenue'].apply(f)\n",
    "\n",
    "dictionary = set()\n",
    "def make_dict(x):\n",
    "    global dictionary\n",
    "    x = str(x).split(\" \")\n",
    "    for i in x:\n",
    "        dictionary.add(i)\n",
    "\n",
    "title_train_s['title'].apply(make_dict)\n",
    "title_test_s['title'].apply(make_dict)\n",
    "\n",
    "def set_to_dict(s):\n",
    "    s = list(s)\n",
    "    dictionary = dict()\n",
    "    for i in range(len(s)):\n",
    "        dictionary[(s[i])] = i\n",
    "    return dictionary\n",
    "\n",
    "dictionary = set_to_dict(dictionary)\n",
    "\n",
    "def create_word_vec(x):\n",
    "    global dictionary\n",
    "    x = x.split(\" \")\n",
    "    vec = [0.0]*50\n",
    "    for i in range(len(x)):\n",
    "        if x[i] in dictionary:\n",
    "            vec[i]=dictionary[x[i]]\n",
    "    return np.array(vec)\n",
    "\n",
    "title_train_s['title']= title_train_s['title'].apply(create_word_vec)\n",
    "title_test_s['title']= title_test_s['title'].apply(create_word_vec)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "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>budget</th>\n",
       "      <th>popularity</th>\n",
       "      <th>runtime</th>\n",
       "      <th>revenue</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>14000000</td>\n",
       "      <td>6.575393</td>\n",
       "      <td>93.0</td>\n",
       "      <td>12314651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>40000000</td>\n",
       "      <td>8.248895</td>\n",
       "      <td>113.0</td>\n",
       "      <td>95149435</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3300000</td>\n",
       "      <td>64.299990</td>\n",
       "      <td>105.0</td>\n",
       "      <td>13092000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1200000</td>\n",
       "      <td>3.174936</td>\n",
       "      <td>122.0</td>\n",
       "      <td>16000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>1.148070</td>\n",
       "      <td>118.0</td>\n",
       "      <td>3923970</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     budget  popularity  runtime   revenue\n",
       "0  14000000    6.575393     93.0  12314651\n",
       "1  40000000    8.248895    113.0  95149435\n",
       "2   3300000   64.299990    105.0  13092000\n",
       "3   1200000    3.174936    122.0  16000000\n",
       "4         0    1.148070    118.0   3923970"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_all.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "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>cast</th>\n",
       "      <th>crew</th>\n",
       "      <th>revenue</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>24</td>\n",
       "      <td>72</td>\n",
       "      <td>12314651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20</td>\n",
       "      <td>9</td>\n",
       "      <td>95149435</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>51</td>\n",
       "      <td>64</td>\n",
       "      <td>13092000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>16000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>3923970</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   cast  crew   revenue\n",
       "0    24    72  12314651\n",
       "1    20     9  95149435\n",
       "2    51    64  13092000\n",
       "3     7     3  16000000\n",
       "4     4     2   3923970"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cc_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>revenue</th>\n",
       "      <th>release_month</th>\n",
       "      <th>release_year</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>12314651</td>\n",
       "      <td>2</td>\n",
       "      <td>2015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>95149435</td>\n",
       "      <td>8</td>\n",
       "      <td>2004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>13092000</td>\n",
       "      <td>10</td>\n",
       "      <td>2014</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>16000000</td>\n",
       "      <td>3</td>\n",
       "      <td>2012</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3923970</td>\n",
       "      <td>2</td>\n",
       "      <td>2009</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    revenue  release_month  release_year\n",
       "0  12314651              2          2015\n",
       "1  95149435              8          2004\n",
       "2  13092000             10          2014\n",
       "3  16000000              3          2012\n",
       "4   3923970              2          2009"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "date_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2100\n",
      "900\n",
      "2100\n",
      "900\n",
      "2100\n",
      "900\n"
     ]
    }
   ],
   "source": [
    "#Print Lengths of training and validation data sets\n",
    "print(len(train))\n",
    "print(len(test))\n",
    "print(len(cc_train_s))\n",
    "print(len(cc_test_s))\n",
    "print(len(date_train_s))\n",
    "print(len(date_test_s))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "scrolled": false
   },
   "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>budget</th>\n",
       "      <th>popularity</th>\n",
       "      <th>runtime</th>\n",
       "      <th>revenue</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2585</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.657191</td>\n",
       "      <td>4.418841</td>\n",
       "      <td>14.156419</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2611</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>2.818963</td>\n",
       "      <td>4.672829</td>\n",
       "      <td>18.251453</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>219</th>\n",
       "      <td>14.771022</td>\n",
       "      <td>0.790251</td>\n",
       "      <td>4.653960</td>\n",
       "      <td>14.967980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2204</th>\n",
       "      <td>17.504390</td>\n",
       "      <td>2.382268</td>\n",
       "      <td>4.691348</td>\n",
       "      <td>17.153756</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>251</th>\n",
       "      <td>18.400478</td>\n",
       "      <td>1.985746</td>\n",
       "      <td>4.779123</td>\n",
       "      <td>16.119826</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         budget  popularity   runtime    revenue\n",
       "2585   0.000000    1.657191  4.418841  14.156419\n",
       "2611   0.000000    2.818963  4.672829  18.251453\n",
       "219   14.771022    0.790251  4.653960  14.967980\n",
       "2204  17.504390    2.382268  4.691348  17.153756\n",
       "251   18.400478    1.985746  4.779123  16.119826"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "\n",
    "#Model 1\n",
    "train = train.apply(np.log)\n",
    "train = train.applymap(f)\n",
    "train = train.fillna(0)\n",
    "X = train.drop('revenue', axis = 1)\n",
    "Y = train['revenue']\n",
    "\n",
    "#for all\n",
    "train_all = train_all.apply(np.log)\n",
    "train_all = train_all.applymap(f)\n",
    "train_all = train_all.fillna(0)\n",
    "X_0 = train_all.drop('revenue', axis = 1)\n",
    "Y_0 = train_all['revenue']\n",
    "\n",
    "test = test.apply(np.log)\n",
    "test = test.applymap(f)\n",
    "test = test.fillna(0)\n",
    "X_test =test.drop('revenue',axis = 1)\n",
    "Y_test = test['revenue']\n",
    "\n",
    "#Model 2\n",
    "cc_train_s['revenue'] = cc_train_s['revenue'].apply(np.log)\n",
    "cc_train_s = cc_train_s.applymap(f)\n",
    "cc_train_s = cc_train_s.fillna(0)\n",
    "X1 = cc_train_s.drop('revenue',axis = 1)\n",
    "Y1 = cc_train_s['revenue']\n",
    "\n",
    "#for all\n",
    "cc_train['revenue'] = cc_train['revenue'].apply(np.log)\n",
    "cc_train = cc_train.applymap(f)\n",
    "cc_train = cc_train.fillna(0)\n",
    "X_0_1 = cc_train.drop('revenue',axis = 1)\n",
    "Y_0_1 = cc_train['revenue']\n",
    "\n",
    "cc_test_s['revenue'] = cc_test_s['revenue'].apply(np.log)\n",
    "cc_test_s = cc_test_s.applymap(f)\n",
    "cc_test_s = cc_test_s.fillna(0)\n",
    "X1_test = cc_test_s.drop('revenue', axis = 1)\n",
    "Y1_test = cc_test_s['revenue']\n",
    "\n",
    "#Model 3\n",
    "date_train_s['revenue'] = date_train_s['revenue'].apply(np.log)\n",
    "date_train_s = date_train_s.applymap(f)\n",
    "date_train_s = date_train_s.fillna(0)\n",
    "X2 = date_train_s.drop('revenue', axis = 1)\n",
    "Y2 = date_train_s['revenue']\n",
    "\n",
    "#for all\n",
    "date_train['revenue'] = date_train['revenue'].apply(np.log)\n",
    "date_train = date_train.applymap(f)\n",
    "date_train = date_train.fillna(0)\n",
    "X_0_2 = date_train.drop('revenue', axis = 1)\n",
    "Y_0_2 = date_train['revenue']\n",
    "\n",
    "date_test_s['revenue'] = date_test_s['revenue'].apply(np.log)\n",
    "date_test_s = date_test_s.applymap(f)\n",
    "date_test_s = date_test_s.fillna(0)\n",
    "X2_test = date_test_s.drop('revenue', axis = 1)\n",
    "Y2_test = date_test_s['revenue']\n",
    "#Model 4\n",
    "# title_train_s = training\n",
    "# title_test_s = validation\n",
    "X3 = title_train_s['title']\n",
    "X3 = np.stack(X3.to_numpy(),axis = 0)\n",
    "\n",
    "Y3 = title_train_s['revenue']\n",
    "X3_test = title_test_s['title']\n",
    "Y3_test = title_test_s['revenue']\n",
    "X3_test = np.stack(X3_test.to_numpy(),axis = 0)\n",
    "#Model 1\n",
    "train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "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>cast</th>\n",
       "      <th>crew</th>\n",
       "      <th>revenue</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>24</td>\n",
       "      <td>72</td>\n",
       "      <td>16.326300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20</td>\n",
       "      <td>9</td>\n",
       "      <td>18.370959</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>51</td>\n",
       "      <td>64</td>\n",
       "      <td>16.387512</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>16.588099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>15.182614</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   cast  crew    revenue\n",
       "0    24    72  16.326300\n",
       "1    20     9  18.370959\n",
       "2    51    64  16.387512\n",
       "3     7     3  16.588099\n",
       "4     4     2  15.182614"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cc_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "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>revenue</th>\n",
       "      <th>release_month</th>\n",
       "      <th>release_year</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>16.326300</td>\n",
       "      <td>2</td>\n",
       "      <td>2015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>18.370959</td>\n",
       "      <td>8</td>\n",
       "      <td>2004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>16.387512</td>\n",
       "      <td>10</td>\n",
       "      <td>2014</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>16.588099</td>\n",
       "      <td>3</td>\n",
       "      <td>2012</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>15.182614</td>\n",
       "      <td>2</td>\n",
       "      <td>2009</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     revenue  release_month  release_year\n",
       "0  16.326300              2          2015\n",
       "1  18.370959              8          2004\n",
       "2  16.387512             10          2014\n",
       "3  16.588099              3          2012\n",
       "4  15.182614              2          2009"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "date_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import models, layers, regularizers, optimizers\n",
    "def model1(X, Y, X_test, Y_test):\n",
    "    model1 = models.Sequential()\n",
    "    model1.add(layers.Dense(356, activation='relu', input_shape=(X.shape[1],)))\n",
    "    model1.add(layers.Dropout(0.1))\n",
    "    model1.add(layers.Dense(256, activation='relu'))\n",
    "    model1.add(layers.Dense(10,activation='relu'))\n",
    "    model1.add(layers.Dense(1))\n",
    "    \n",
    "    model1.compile(optimizer=optimizers.rmsprop(lr=1e-5),loss='mse',metrics=['mean_squared_logarithmic_error'])\n",
    "    history = model1.fit(X,Y,epochs=40, batch_size = 32, validation_data=(X_test, Y_test))\n",
    "    #history = model1.fit(X, Y, epochs = 40, batch_size = 32)\n",
    "    return [model1, history]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "def model2(X1, Y1, X1_test, Y1_test): \n",
    "    model2 = models.Sequential()\n",
    "    model2.add(layers.Dense(356, activation='relu', input_shape=(X1.shape[1],)))\n",
    "    model2.add(layers.Dropout(0.2))\n",
    "    model2.add(layers.BatchNormalization())\n",
    "    model2.add(layers.Dense(256, activation='relu'))\n",
    "    model2.add(layers.Dropout(0.2))\n",
    "    model2.add(layers.Dense(128, activation='relu'))\n",
    "    model2.add(layers.Dropout(0.2))\n",
    "    model2.add(layers.Dense(10,activation='relu'))\n",
    "    model2.add(layers.Dense(1))\n",
    "\n",
    "    model2.compile(optimizer=optimizers.rmsprop(lr=1e-5),loss='mse',metrics=['mean_squared_logarithmic_error'])\n",
    "    history = model2.fit(X1, Y1, epochs=50, batch_size = 32, validation_data=(X1_test, Y1_test))\n",
    "    #history = model2.fit(X1, Y1, epochs = 50, batch_size = 32)\n",
    "    return [model2, history]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "def model3(X2, Y2, X2_test, Y2_test):\n",
    "    model3 = models.Sequential()\n",
    "    model3.add(layers.Dense(356, activation='relu', input_shape=(X2.shape[1],)))\n",
    "    model3.add(layers.Dropout(0.2))\n",
    "    model3.add(layers.Dense(256, activation='relu'))\n",
    "    model3.add(layers.Dense(10,activation='relu'))\n",
    "    model3.add(layers.Dense(1))\n",
    "\n",
    "    model3.compile(optimizer=optimizers.rmsprop(lr=1e-5),loss='mse',metrics=['mean_squared_logarithmic_error'])\n",
    "    history = model3.fit(X2, Y2, epochs=40, batch_size = 32, validation_data=(X2_test, Y2_test))\n",
    "    #history = model3.fit(X2, Y2, epochs = 50, batch_size = 32)\n",
    "    \n",
    "    return [model3, history]\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MODEL 1 - BUDGET, RUNTIME, AND POPULARITY"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 2100 samples, validate on 900 samples\n",
      "Epoch 1/40\n",
      "2100/2100 [==============================] - 1s 299us/step - loss: 254.2947 - mean_squared_logarithmic_error: 6.5742 - val_loss: 243.8921 - val_mean_squared_logarithmic_error: 5.3816\n",
      "Epoch 2/40\n",
      "2100/2100 [==============================] - 0s 64us/step - loss: 232.2865 - mean_squared_logarithmic_error: 4.6262 - val_loss: 219.9647 - val_mean_squared_logarithmic_error: 3.8003\n",
      "Epoch 3/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 208.2871 - mean_squared_logarithmic_error: 3.3352 - val_loss: 195.0492 - val_mean_squared_logarithmic_error: 2.7749\n",
      "Epoch 4/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 183.0656 - mean_squared_logarithmic_error: 2.4557 - val_loss: 167.8098 - val_mean_squared_logarithmic_error: 2.0255\n",
      "Epoch 5/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 155.9132 - mean_squared_logarithmic_error: 1.8041 - val_loss: 141.3689 - val_mean_squared_logarithmic_error: 1.5062\n",
      "Epoch 6/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 130.2368 - mean_squared_logarithmic_error: 1.3496 - val_loss: 116.9096 - val_mean_squared_logarithmic_error: 1.1271\n",
      "Epoch 7/40\n",
      "2100/2100 [==============================] - 0s 64us/step - loss: 106.4058 - mean_squared_logarithmic_error: 1.0068 - val_loss: 94.4342 - val_mean_squared_logarithmic_error: 0.8438\n",
      "Epoch 8/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 85.5508 - mean_squared_logarithmic_error: 0.7605 - val_loss: 74.9284 - val_mean_squared_logarithmic_error: 0.6374\n",
      "Epoch 9/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 66.8400 - mean_squared_logarithmic_error: 0.5736 - val_loss: 57.9192 - val_mean_squared_logarithmic_error: 0.4837\n",
      "Epoch 10/40\n",
      "2100/2100 [==============================] - 0s 70us/step - loss: 51.4782 - mean_squared_logarithmic_error: 0.4401 - val_loss: 43.5158 - val_mean_squared_logarithmic_error: 0.3685\n",
      "Epoch 11/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 38.7394 - mean_squared_logarithmic_error: 0.3379 - val_loss: 32.7780 - val_mean_squared_logarithmic_error: 0.2874\n",
      "Epoch 12/40\n",
      "2100/2100 [==============================] - 0s 64us/step - loss: 30.0041 - mean_squared_logarithmic_error: 0.2697 - val_loss: 25.9966 - val_mean_squared_logarithmic_error: 0.2334\n",
      "Epoch 13/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 25.2668 - mean_squared_logarithmic_error: 0.2263 - val_loss: 22.8422 - val_mean_squared_logarithmic_error: 0.1995\n",
      "Epoch 14/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 22.6137 - mean_squared_logarithmic_error: 0.1923 - val_loss: 20.6948 - val_mean_squared_logarithmic_error: 0.1724\n",
      "Epoch 15/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 20.4974 - mean_squared_logarithmic_error: 0.1671 - val_loss: 18.7352 - val_mean_squared_logarithmic_error: 0.1507\n",
      "Epoch 16/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 18.5310 - mean_squared_logarithmic_error: 0.1465 - val_loss: 16.8054 - val_mean_squared_logarithmic_error: 0.1311\n",
      "Epoch 17/40\n",
      "2100/2100 [==============================] - 0s 72us/step - loss: 16.6463 - mean_squared_logarithmic_error: 0.1286 - val_loss: 15.0096 - val_mean_squared_logarithmic_error: 0.1147\n",
      "Epoch 18/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 14.8983 - mean_squared_logarithmic_error: 0.1130 - val_loss: 13.3801 - val_mean_squared_logarithmic_error: 0.1013\n",
      "Epoch 19/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 13.1471 - mean_squared_logarithmic_error: 0.0996 - val_loss: 11.8716 - val_mean_squared_logarithmic_error: 0.0900\n",
      "Epoch 20/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 11.8065 - mean_squared_logarithmic_error: 0.0893 - val_loss: 10.5821 - val_mean_squared_logarithmic_error: 0.0811\n",
      "Epoch 21/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 10.5383 - mean_squared_logarithmic_error: 0.0810 - val_loss: 9.3592 - val_mean_squared_logarithmic_error: 0.0735\n",
      "Epoch 22/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 9.3700 - mean_squared_logarithmic_error: 0.0733 - val_loss: 8.3878 - val_mean_squared_logarithmic_error: 0.0679\n",
      "Epoch 23/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 8.3077 - mean_squared_logarithmic_error: 0.0676 - val_loss: 7.5083 - val_mean_squared_logarithmic_error: 0.0635\n",
      "Epoch 24/40\n",
      "2100/2100 [==============================] - 0s 71us/step - loss: 7.4951 - mean_squared_logarithmic_error: 0.0633 - val_loss: 6.8698 - val_mean_squared_logarithmic_error: 0.0605\n",
      "Epoch 25/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 6.9139 - mean_squared_logarithmic_error: 0.0608 - val_loss: 6.4090 - val_mean_squared_logarithmic_error: 0.0587\n",
      "Epoch 26/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 6.4865 - mean_squared_logarithmic_error: 0.0593 - val_loss: 6.0762 - val_mean_squared_logarithmic_error: 0.0576\n",
      "Epoch 27/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 6.2422 - mean_squared_logarithmic_error: 0.0583 - val_loss: 5.9061 - val_mean_squared_logarithmic_error: 0.0571\n",
      "Epoch 28/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 5.9832 - mean_squared_logarithmic_error: 0.0577 - val_loss: 5.8027 - val_mean_squared_logarithmic_error: 0.0570\n",
      "Epoch 29/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 5.9679 - mean_squared_logarithmic_error: 0.0580 - val_loss: 5.7358 - val_mean_squared_logarithmic_error: 0.0570\n",
      "Epoch 30/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 6.0451 - mean_squared_logarithmic_error: 0.0584 - val_loss: 5.7335 - val_mean_squared_logarithmic_error: 0.0570\n",
      "Epoch 31/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 5.9445 - mean_squared_logarithmic_error: 0.0579 - val_loss: 5.7130 - val_mean_squared_logarithmic_error: 0.0570\n",
      "Epoch 32/40\n",
      "2100/2100 [==============================] - 0s 64us/step - loss: 5.9261 - mean_squared_logarithmic_error: 0.0579 - val_loss: 5.7109 - val_mean_squared_logarithmic_error: 0.0570\n",
      "Epoch 33/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 5.9149 - mean_squared_logarithmic_error: 0.0579 - val_loss: 5.7673 - val_mean_squared_logarithmic_error: 0.0570\n",
      "Epoch 34/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 5.8510 - mean_squared_logarithmic_error: 0.0576 - val_loss: 5.6746 - val_mean_squared_logarithmic_error: 0.0572\n",
      "Epoch 35/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 5.9330 - mean_squared_logarithmic_error: 0.0579 - val_loss: 5.6871 - val_mean_squared_logarithmic_error: 0.0571\n",
      "Epoch 36/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 5.8892 - mean_squared_logarithmic_error: 0.0578 - val_loss: 5.6744 - val_mean_squared_logarithmic_error: 0.0570\n",
      "Epoch 37/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 5.8889 - mean_squared_logarithmic_error: 0.0574 - val_loss: 5.7104 - val_mean_squared_logarithmic_error: 0.0569\n",
      "Epoch 38/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 5.8326 - mean_squared_logarithmic_error: 0.0571 - val_loss: 5.7028 - val_mean_squared_logarithmic_error: 0.0569\n",
      "Epoch 39/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 5.8162 - mean_squared_logarithmic_error: 0.0572 - val_loss: 5.7209 - val_mean_squared_logarithmic_error: 0.0568\n",
      "Epoch 40/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 5.8865 - mean_squared_logarithmic_error: 0.0578 - val_loss: 5.6921 - val_mean_squared_logarithmic_error: 0.0568\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_20 (Dense)             (None, 356)               1424      \n",
      "_________________________________________________________________\n",
      "dropout_10 (Dropout)         (None, 356)               0         \n",
      "_________________________________________________________________\n",
      "dense_21 (Dense)             (None, 256)               91392     \n",
      "_________________________________________________________________\n",
      "dense_22 (Dense)             (None, 10)                2570      \n",
      "_________________________________________________________________\n",
      "dense_23 (Dense)             (None, 1)                 11        \n",
      "=================================================================\n",
      "Total params: 95,397\n",
      "Trainable params: 95,397\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model1, history1 = model1(X, Y, X_test, Y_test)\n",
    "model1.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "acc = history1.history['mean_squared_logarithmic_error']\n",
    "val_acc = history1.history['val_mean_squared_logarithmic_error']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training')\n",
    "plt.plot(epochs, val_acc, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('MSLE')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "l1 = history1.history['loss']\n",
    "l2 = history1.history['val_loss']\n",
    "\n",
    "epochs = range(len(l1))\n",
    "\n",
    "plt.plot(epochs, l1, 'bo', label='Training')\n",
    "plt.plot(epochs, l2, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MODEL 2 - NUMBER OF CREW AND NUMBER OF CAST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 2100 samples, validate on 900 samples\n",
      "Epoch 1/50\n",
      "2100/2100 [==============================] - 1s 490us/step - loss: 267.3312 - mean_squared_logarithmic_error: 7.6990 - val_loss: 262.7387 - val_mean_squared_logarithmic_error: 7.6318\n",
      "Epoch 2/50\n",
      "2100/2100 [==============================] - 0s 94us/step - loss: 261.8296 - mean_squared_logarithmic_error: 7.4264 - val_loss: 259.1471 - val_mean_squared_logarithmic_error: 7.2627\n",
      "Epoch 3/50\n",
      "2100/2100 [==============================] - 0s 96us/step - loss: 256.6871 - mean_squared_logarithmic_error: 6.9869 - val_loss: 254.8586 - val_mean_squared_logarithmic_error: 6.7925\n",
      "Epoch 4/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 251.8335 - mean_squared_logarithmic_error: 6.5362 - val_loss: 250.1974 - val_mean_squared_logarithmic_error: 6.3518\n",
      "Epoch 5/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 245.7469 - mean_squared_logarithmic_error: 5.9988 - val_loss: 244.9581 - val_mean_squared_logarithmic_error: 5.9195\n",
      "Epoch 6/50\n",
      "2100/2100 [==============================] - 0s 95us/step - loss: 239.7194 - mean_squared_logarithmic_error: 5.5211 - val_loss: 239.0512 - val_mean_squared_logarithmic_error: 5.4789\n",
      "Epoch 7/50\n",
      "2100/2100 [==============================] - 0s 96us/step - loss: 232.3831 - mean_squared_logarithmic_error: 5.0355 - val_loss: 232.4878 - val_mean_squared_logarithmic_error: 5.0237\n",
      "Epoch 8/50\n",
      "2100/2100 [==============================] - 0s 94us/step - loss: 225.3931 - mean_squared_logarithmic_error: 4.5914 - val_loss: 225.1355 - val_mean_squared_logarithmic_error: 4.5596\n",
      "Epoch 9/50\n",
      "2100/2100 [==============================] - 0s 93us/step - loss: 217.6191 - mean_squared_logarithmic_error: 4.1525 - val_loss: 217.2650 - val_mean_squared_logarithmic_error: 4.1219\n",
      "Epoch 10/50\n",
      "2100/2100 [==============================] - 0s 93us/step - loss: 208.1751 - mean_squared_logarithmic_error: 3.7002 - val_loss: 209.0961 - val_mean_squared_logarithmic_error: 3.7239\n",
      "Epoch 11/50\n",
      "2100/2100 [==============================] - 0s 93us/step - loss: 199.4550 - mean_squared_logarithmic_error: 3.3213 - val_loss: 200.6874 - val_mean_squared_logarithmic_error: 3.3627\n",
      "Epoch 12/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 190.2884 - mean_squared_logarithmic_error: 2.9658 - val_loss: 192.2786 - val_mean_squared_logarithmic_error: 3.0372\n",
      "Epoch 13/50\n",
      "2100/2100 [==============================] - 0s 92us/step - loss: 180.7790 - mean_squared_logarithmic_error: 2.6389 - val_loss: 183.7075 - val_mean_squared_logarithmic_error: 2.7358\n",
      "Epoch 14/50\n",
      "2100/2100 [==============================] - 0s 94us/step - loss: 170.8107 - mean_squared_logarithmic_error: 2.3485 - val_loss: 174.9873 - val_mean_squared_logarithmic_error: 2.4533\n",
      "Epoch 15/50\n",
      "2100/2100 [==============================] - 0s 95us/step - loss: 160.8872 - mean_squared_logarithmic_error: 2.0710 - val_loss: 166.2277 - val_mean_squared_logarithmic_error: 2.1965\n",
      "Epoch 16/50\n",
      "2100/2100 [==============================] - 0s 94us/step - loss: 152.0728 - mean_squared_logarithmic_error: 1.8454 - val_loss: 157.6492 - val_mean_squared_logarithmic_error: 1.9635\n",
      "Epoch 17/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 142.1067 - mean_squared_logarithmic_error: 1.6205 - val_loss: 149.0141 - val_mean_squared_logarithmic_error: 1.7464\n",
      "Epoch 18/50\n",
      "2100/2100 [==============================] - 0s 95us/step - loss: 132.4929 - mean_squared_logarithmic_error: 1.4228 - val_loss: 140.5995 - val_mean_squared_logarithmic_error: 1.5508\n",
      "Epoch 19/50\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 123.8886 - mean_squared_logarithmic_error: 1.2498 - val_loss: 131.8801 - val_mean_squared_logarithmic_error: 1.3669\n",
      "Epoch 20/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 114.2371 - mean_squared_logarithmic_error: 1.0856 - val_loss: 123.3506 - val_mean_squared_logarithmic_error: 1.2011\n",
      "Epoch 21/50\n",
      "2100/2100 [==============================] - 0s 93us/step - loss: 105.1262 - mean_squared_logarithmic_error: 0.9445 - val_loss: 115.1024 - val_mean_squared_logarithmic_error: 1.0558\n",
      "Epoch 22/50\n",
      "2100/2100 [==============================] - 0s 98us/step - loss: 96.7597 - mean_squared_logarithmic_error: 0.8205 - val_loss: 106.5981 - val_mean_squared_logarithmic_error: 0.9204\n",
      "Epoch 23/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 88.7815 - mean_squared_logarithmic_error: 0.7087 - val_loss: 98.7204 - val_mean_squared_logarithmic_error: 0.8055\n",
      "Epoch 24/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 79.2047 - mean_squared_logarithmic_error: 0.6010 - val_loss: 90.9747 - val_mean_squared_logarithmic_error: 0.7016\n",
      "Epoch 25/50\n",
      "2100/2100 [==============================] - 0s 93us/step - loss: 70.4252 - mean_squared_logarithmic_error: 0.5019 - val_loss: 83.1815 - val_mean_squared_logarithmic_error: 0.6058\n",
      "Epoch 26/50\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 65.0348 - mean_squared_logarithmic_error: 0.4434 - val_loss: 76.2296 - val_mean_squared_logarithmic_error: 0.5281\n",
      "Epoch 27/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 58.2455 - mean_squared_logarithmic_error: 0.3739 - val_loss: 69.5209 - val_mean_squared_logarithmic_error: 0.4589\n",
      "Epoch 28/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 52.8141 - mean_squared_logarithmic_error: 0.3312 - val_loss: 63.7074 - val_mean_squared_logarithmic_error: 0.4025\n",
      "Epoch 29/50\n",
      "2100/2100 [==============================] - 0s 92us/step - loss: 48.9305 - mean_squared_logarithmic_error: 0.2971 - val_loss: 58.4984 - val_mean_squared_logarithmic_error: 0.3559\n",
      "Epoch 30/50\n",
      "2100/2100 [==============================] - 0s 94us/step - loss: 44.0028 - mean_squared_logarithmic_error: 0.2613 - val_loss: 54.0752 - val_mean_squared_logarithmic_error: 0.3169\n",
      "Epoch 31/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 39.5943 - mean_squared_logarithmic_error: 0.2261 - val_loss: 50.5666 - val_mean_squared_logarithmic_error: 0.2860\n",
      "Epoch 32/50\n",
      "2100/2100 [==============================] - 0s 90us/step - loss: 38.8143 - mean_squared_logarithmic_error: 0.2138 - val_loss: 47.4076 - val_mean_squared_logarithmic_error: 0.2602\n",
      "Epoch 33/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 36.4680 - mean_squared_logarithmic_error: 0.1973 - val_loss: 44.9004 - val_mean_squared_logarithmic_error: 0.2400\n",
      "Epoch 34/50\n",
      "2100/2100 [==============================] - 0s 90us/step - loss: 34.6927 - mean_squared_logarithmic_error: 0.1842 - val_loss: 42.7432 - val_mean_squared_logarithmic_error: 0.2249\n",
      "Epoch 35/50\n",
      "2100/2100 [==============================] - 0s 91us/step - loss: 33.3919 - mean_squared_logarithmic_error: 0.1730 - val_loss: 40.5215 - val_mean_squared_logarithmic_error: 0.2103\n",
      "Epoch 36/50\n",
      "2100/2100 [==============================] - 0s 95us/step - loss: 33.6600 - mean_squared_logarithmic_error: 0.1757 - val_loss: 38.3861 - val_mean_squared_logarithmic_error: 0.1982\n",
      "Epoch 37/50\n",
      "2100/2100 [==============================] - 0s 95us/step - loss: 31.2219 - mean_squared_logarithmic_error: 0.1604 - val_loss: 36.5561 - val_mean_squared_logarithmic_error: 0.1879\n",
      "Epoch 38/50\n",
      "2100/2100 [==============================] - 0s 90us/step - loss: 28.1317 - mean_squared_logarithmic_error: 0.1446 - val_loss: 34.7642 - val_mean_squared_logarithmic_error: 0.1778\n",
      "Epoch 39/50\n",
      "2100/2100 [==============================] - 0s 90us/step - loss: 27.0964 - mean_squared_logarithmic_error: 0.1389 - val_loss: 33.0077 - val_mean_squared_logarithmic_error: 0.1685\n",
      "Epoch 40/50\n",
      "2100/2100 [==============================] - 0s 92us/step - loss: 27.9940 - mean_squared_logarithmic_error: 0.1417 - val_loss: 31.5958 - val_mean_squared_logarithmic_error: 0.1616\n",
      "Epoch 41/50\n",
      "2100/2100 [==============================] - 0s 92us/step - loss: 25.8952 - mean_squared_logarithmic_error: 0.1323 - val_loss: 30.2659 - val_mean_squared_logarithmic_error: 0.1547\n",
      "Epoch 42/50\n",
      "2100/2100 [==============================] - 0s 96us/step - loss: 24.7767 - mean_squared_logarithmic_error: 0.1282 - val_loss: 28.8722 - val_mean_squared_logarithmic_error: 0.1475\n",
      "Epoch 43/50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2100/2100 [==============================] - 0s 90us/step - loss: 24.4807 - mean_squared_logarithmic_error: 0.1240 - val_loss: 27.7861 - val_mean_squared_logarithmic_error: 0.1425\n",
      "Epoch 44/50\n",
      "2100/2100 [==============================] - 0s 89us/step - loss: 23.3427 - mean_squared_logarithmic_error: 0.1211 - val_loss: 26.3769 - val_mean_squared_logarithmic_error: 0.1362\n",
      "Epoch 45/50\n",
      "2100/2100 [==============================] - 0s 92us/step - loss: 22.7290 - mean_squared_logarithmic_error: 0.1194 - val_loss: 25.3477 - val_mean_squared_logarithmic_error: 0.1315\n",
      "Epoch 46/50\n",
      "2100/2100 [==============================] - 0s 93us/step - loss: 22.3545 - mean_squared_logarithmic_error: 0.1150 - val_loss: 24.3924 - val_mean_squared_logarithmic_error: 0.1272\n",
      "Epoch 47/50\n",
      "2100/2100 [==============================] - 0s 90us/step - loss: 21.4148 - mean_squared_logarithmic_error: 0.1131 - val_loss: 23.4695 - val_mean_squared_logarithmic_error: 0.1231\n",
      "Epoch 48/50\n",
      "2100/2100 [==============================] - 0s 89us/step - loss: 20.6334 - mean_squared_logarithmic_error: 0.1091 - val_loss: 22.5696 - val_mean_squared_logarithmic_error: 0.1194\n",
      "Epoch 49/50\n",
      "2100/2100 [==============================] - 0s 89us/step - loss: 19.5378 - mean_squared_logarithmic_error: 0.1050 - val_loss: 21.8259 - val_mean_squared_logarithmic_error: 0.1158\n",
      "Epoch 50/50\n",
      "2100/2100 [==============================] - 0s 90us/step - loss: 20.4672 - mean_squared_logarithmic_error: 0.1080 - val_loss: 21.1966 - val_mean_squared_logarithmic_error: 0.1133\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_24 (Dense)             (None, 356)               1068      \n",
      "_________________________________________________________________\n",
      "dropout_11 (Dropout)         (None, 356)               0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 356)               1424      \n",
      "_________________________________________________________________\n",
      "dense_25 (Dense)             (None, 256)               91392     \n",
      "_________________________________________________________________\n",
      "dropout_12 (Dropout)         (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_26 (Dense)             (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dropout_13 (Dropout)         (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_27 (Dense)             (None, 10)                1290      \n",
      "_________________________________________________________________\n",
      "dense_28 (Dense)             (None, 1)                 11        \n",
      "=================================================================\n",
      "Total params: 128,081\n",
      "Trainable params: 127,369\n",
      "Non-trainable params: 712\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model2, history2 = model2(X1, Y1, X1_test, Y1_test)\n",
    "model2.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "acc = history2.history['mean_squared_logarithmic_error']\n",
    "val_acc = history2.history['val_mean_squared_logarithmic_error']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training')\n",
    "plt.plot(epochs, val_acc, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('MSLE')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "l1 = history2.history['loss']\n",
    "l2 = history2.history['val_loss']\n",
    "\n",
    "epochs = range(len(l1))\n",
    "\n",
    "plt.plot(epochs, l1, 'bo', label='Training')\n",
    "plt.plot(epochs, l2, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MODEL 3 - RELEASE YEAR AND RELEASE MONTH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 2100 samples, validate on 900 samples\n",
      "Epoch 1/40\n",
      "2100/2100 [==============================] - 1s 358us/step - loss: 1244.1645 - mean_squared_logarithmic_error: 4.9825 - val_loss: 239.8078 - val_mean_squared_logarithmic_error: 4.9849\n",
      "Epoch 2/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 399.6031 - mean_squared_logarithmic_error: 4.4620 - val_loss: 237.2903 - val_mean_squared_logarithmic_error: 4.7867\n",
      "Epoch 3/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 273.1517 - mean_squared_logarithmic_error: 4.3948 - val_loss: 230.6617 - val_mean_squared_logarithmic_error: 4.3169\n",
      "Epoch 4/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 234.9348 - mean_squared_logarithmic_error: 3.9878 - val_loss: 219.8703 - val_mean_squared_logarithmic_error: 3.6799\n",
      "Epoch 5/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 220.1478 - mean_squared_logarithmic_error: 3.4068 - val_loss: 205.0944 - val_mean_squared_logarithmic_error: 2.9928\n",
      "Epoch 6/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 203.9532 - mean_squared_logarithmic_error: 2.7159 - val_loss: 187.8026 - val_mean_squared_logarithmic_error: 2.3740\n",
      "Epoch 7/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 181.7848 - mean_squared_logarithmic_error: 2.1319 - val_loss: 166.1575 - val_mean_squared_logarithmic_error: 1.7869\n",
      "Epoch 8/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 157.2183 - mean_squared_logarithmic_error: 1.5721 - val_loss: 143.2728 - val_mean_squared_logarithmic_error: 1.3185\n",
      "Epoch 9/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 134.2776 - mean_squared_logarithmic_error: 1.1755 - val_loss: 119.1213 - val_mean_squared_logarithmic_error: 0.9408\n",
      "Epoch 10/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 117.5001 - mean_squared_logarithmic_error: 0.8580 - val_loss: 97.1696 - val_mean_squared_logarithmic_error: 0.6728\n",
      "Epoch 11/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 89.9901 - mean_squared_logarithmic_error: 0.6197 - val_loss: 73.9984 - val_mean_squared_logarithmic_error: 0.4491\n",
      "Epoch 12/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 65.7464 - mean_squared_logarithmic_error: 0.3942 - val_loss: 50.5403 - val_mean_squared_logarithmic_error: 0.2723\n",
      "Epoch 13/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 43.3938 - mean_squared_logarithmic_error: 0.2411 - val_loss: 31.1075 - val_mean_squared_logarithmic_error: 0.1586\n",
      "Epoch 14/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 26.7926 - mean_squared_logarithmic_error: 0.1487 - val_loss: 17.7390 - val_mean_squared_logarithmic_error: 0.0980\n",
      "Epoch 15/40\n",
      "2100/2100 [==============================] - 0s 77us/step - loss: 16.1711 - mean_squared_logarithmic_error: 0.0985 - val_loss: 11.2843 - val_mean_squared_logarithmic_error: 0.0763\n",
      "Epoch 16/40\n",
      "2100/2100 [==============================] - 0s 72us/step - loss: 12.1824 - mean_squared_logarithmic_error: 0.0830 - val_loss: 9.9249 - val_mean_squared_logarithmic_error: 0.0739\n",
      "Epoch 17/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 11.3857 - mean_squared_logarithmic_error: 0.0797 - val_loss: 9.7955 - val_mean_squared_logarithmic_error: 0.0741\n",
      "Epoch 18/40\n",
      "2100/2100 [==============================] - 0s 78us/step - loss: 11.4983 - mean_squared_logarithmic_error: 0.0773 - val_loss: 9.7659 - val_mean_squared_logarithmic_error: 0.0742\n",
      "Epoch 19/40\n",
      "2100/2100 [==============================] - 0s 77us/step - loss: 11.2172 - mean_squared_logarithmic_error: 0.0768 - val_loss: 9.7684 - val_mean_squared_logarithmic_error: 0.0742\n",
      "Epoch 20/40\n",
      "2100/2100 [==============================] - 0s 76us/step - loss: 10.9067 - mean_squared_logarithmic_error: 0.0780 - val_loss: 9.7718 - val_mean_squared_logarithmic_error: 0.0742\n",
      "Epoch 21/40\n",
      "2100/2100 [==============================] - 0s 70us/step - loss: 11.0295 - mean_squared_logarithmic_error: 0.0763 - val_loss: 9.7549 - val_mean_squared_logarithmic_error: 0.0743\n",
      "Epoch 22/40\n",
      "2100/2100 [==============================] - 0s 79us/step - loss: 10.6477 - mean_squared_logarithmic_error: 0.0747 - val_loss: 9.7569 - val_mean_squared_logarithmic_error: 0.0743\n",
      "Epoch 23/40\n",
      "2100/2100 [==============================] - 0s 74us/step - loss: 10.6756 - mean_squared_logarithmic_error: 0.0748 - val_loss: 9.7748 - val_mean_squared_logarithmic_error: 0.0742\n",
      "Epoch 24/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 10.8715 - mean_squared_logarithmic_error: 0.0796 - val_loss: 9.8068 - val_mean_squared_logarithmic_error: 0.0740\n",
      "Epoch 25/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 11.2145 - mean_squared_logarithmic_error: 0.0756 - val_loss: 9.8012 - val_mean_squared_logarithmic_error: 0.0741\n",
      "Epoch 26/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 10.5876 - mean_squared_logarithmic_error: 0.0755 - val_loss: 9.8232 - val_mean_squared_logarithmic_error: 0.0740\n",
      "Epoch 27/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 10.8005 - mean_squared_logarithmic_error: 0.0752 - val_loss: 9.7894 - val_mean_squared_logarithmic_error: 0.0741\n",
      "Epoch 28/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 10.3957 - mean_squared_logarithmic_error: 0.0735 - val_loss: 9.8223 - val_mean_squared_logarithmic_error: 0.0740\n",
      "Epoch 29/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 10.8127 - mean_squared_logarithmic_error: 0.0758 - val_loss: 9.8561 - val_mean_squared_logarithmic_error: 0.0739\n",
      "Epoch 30/40\n",
      "2100/2100 [==============================] - 0s 70us/step - loss: 10.7370 - mean_squared_logarithmic_error: 0.0750 - val_loss: 9.9764 - val_mean_squared_logarithmic_error: 0.0739\n",
      "Epoch 31/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 10.5147 - mean_squared_logarithmic_error: 0.0735 - val_loss: 10.0012 - val_mean_squared_logarithmic_error: 0.0739\n",
      "Epoch 32/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 10.5750 - mean_squared_logarithmic_error: 0.0739 - val_loss: 10.1426 - val_mean_squared_logarithmic_error: 0.0740\n",
      "Epoch 33/40\n",
      "2100/2100 [==============================] - 0s 65us/step - loss: 10.3838 - mean_squared_logarithmic_error: 0.0736 - val_loss: 9.8911 - val_mean_squared_logarithmic_error: 0.0739\n",
      "Epoch 34/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 10.6893 - mean_squared_logarithmic_error: 0.0750 - val_loss: 9.8056 - val_mean_squared_logarithmic_error: 0.0740\n",
      "Epoch 35/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 10.2180 - mean_squared_logarithmic_error: 0.0733 - val_loss: 9.9477 - val_mean_squared_logarithmic_error: 0.0739\n",
      "Epoch 36/40\n",
      "2100/2100 [==============================] - 0s 67us/step - loss: 10.3168 - mean_squared_logarithmic_error: 0.0731 - val_loss: 10.0363 - val_mean_squared_logarithmic_error: 0.0739\n",
      "Epoch 37/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 10.7582 - mean_squared_logarithmic_error: 0.0755 - val_loss: 10.2990 - val_mean_squared_logarithmic_error: 0.0742\n",
      "Epoch 38/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 10.2319 - mean_squared_logarithmic_error: 0.0728 - val_loss: 10.0289 - val_mean_squared_logarithmic_error: 0.0739\n",
      "Epoch 39/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 10.4668 - mean_squared_logarithmic_error: 0.0738 - val_loss: 10.0701 - val_mean_squared_logarithmic_error: 0.0740\n",
      "Epoch 40/40\n",
      "2100/2100 [==============================] - 0s 66us/step - loss: 10.1808 - mean_squared_logarithmic_error: 0.0730 - val_loss: 10.1170 - val_mean_squared_logarithmic_error: 0.0740\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_29 (Dense)             (None, 356)               1068      \n",
      "_________________________________________________________________\n",
      "dropout_14 (Dropout)         (None, 356)               0         \n",
      "_________________________________________________________________\n",
      "dense_30 (Dense)             (None, 256)               91392     \n",
      "_________________________________________________________________\n",
      "dense_31 (Dense)             (None, 10)                2570      \n",
      "_________________________________________________________________\n",
      "dense_32 (Dense)             (None, 1)                 11        \n",
      "=================================================================\n",
      "Total params: 95,041\n",
      "Trainable params: 95,041\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model3, history3 = model3(X2, Y2, X2_test, Y2_test)\n",
    "model3.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "acc = history3.history['mean_squared_logarithmic_error']\n",
    "val_acc = history3.history['val_mean_squared_logarithmic_error']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training')\n",
    "plt.plot(epochs, val_acc, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('MSLE')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "l1 = history3.history['loss']\n",
    "l2 = history3.history['val_loss']\n",
    "\n",
    "epochs = range(len(l1))\n",
    "\n",
    "plt.plot(epochs, l1, 'bo', label='Training')\n",
    "plt.plot(epochs, l2, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MODEL 4 - TITLE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import models, layers, regularizers, optimizers\n",
    "def create_embedding_model(X,Y,X_test,Y_test):\n",
    "    model = models.Sequential()\n",
    "    model.add(layers.Embedding(len(dictionary),30, input_length = 50));\n",
    "    model.add(layers.Flatten())\n",
    "    # model.add(layers.Dense(256,activation = 'relu'))\n",
    "    # model.add(layers.Dropout(0.7))\n",
    "    model.add(layers.Dense(128,activation = 'relu'))\n",
    "    model.add(layers.Dropout(0.5))\n",
    "    model.add(layers.Dense(64,activation = 'relu'))\n",
    "    model.add(layers.Dropout(0.5))\n",
    "    model.add(layers.Dense(1))\n",
    "    model.compile(optimizer='adam',loss='mean_squared_logarithmic_error',metrics=['mean_squared_logarithmic_error'])\n",
    "    history = model.fit(X,Y,epochs=134,batch_size = 32,validation_data = (X_test,Y_test))\n",
    "    return model,history\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 2100 samples, validate on 900 samples\n",
      "Epoch 1/134\n",
      "2100/2100 [==============================] - 1s 484us/step - loss: 1.5253 - mean_squared_logarithmic_error: 1.5253 - val_loss: 0.0953 - val_mean_squared_logarithmic_error: 0.0953\n",
      "Epoch 2/134\n",
      "2100/2100 [==============================] - 0s 111us/step - loss: 0.1556 - mean_squared_logarithmic_error: 0.1556 - val_loss: 0.0769 - val_mean_squared_logarithmic_error: 0.0769\n",
      "Epoch 3/134\n",
      "2100/2100 [==============================] - 0s 119us/step - loss: 0.1448 - mean_squared_logarithmic_error: 0.1448 - val_loss: 0.0770 - val_mean_squared_logarithmic_error: 0.0770\n",
      "Epoch 4/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.1440 - mean_squared_logarithmic_error: 0.1440 - val_loss: 0.0746 - val_mean_squared_logarithmic_error: 0.0746\n",
      "Epoch 5/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.1350 - mean_squared_logarithmic_error: 0.1350 - val_loss: 0.0740 - val_mean_squared_logarithmic_error: 0.0740\n",
      "Epoch 6/134\n",
      "2100/2100 [==============================] - 0s 110us/step - loss: 0.1325 - mean_squared_logarithmic_error: 0.1325 - val_loss: 0.0750 - val_mean_squared_logarithmic_error: 0.0750\n",
      "Epoch 7/134\n",
      "2100/2100 [==============================] - 0s 127us/step - loss: 0.1218 - mean_squared_logarithmic_error: 0.1218 - val_loss: 0.0749 - val_mean_squared_logarithmic_error: 0.0749\n",
      "Epoch 8/134\n",
      "2100/2100 [==============================] - 0s 124us/step - loss: 0.1257 - mean_squared_logarithmic_error: 0.1257 - val_loss: 0.0752 - val_mean_squared_logarithmic_error: 0.0752\n",
      "Epoch 9/134\n",
      "2100/2100 [==============================] - 0s 113us/step - loss: 0.1160 - mean_squared_logarithmic_error: 0.1160 - val_loss: 0.0734 - val_mean_squared_logarithmic_error: 0.0734\n",
      "Epoch 10/134\n",
      "2100/2100 [==============================] - 0s 115us/step - loss: 0.1085 - mean_squared_logarithmic_error: 0.1085 - val_loss: 0.0733 - val_mean_squared_logarithmic_error: 0.0733\n",
      "Epoch 11/134\n",
      "2100/2100 [==============================] - 0s 113us/step - loss: 0.1037 - mean_squared_logarithmic_error: 0.1037 - val_loss: 0.0742 - val_mean_squared_logarithmic_error: 0.0742\n",
      "Epoch 12/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.1000 - mean_squared_logarithmic_error: 0.1000 - val_loss: 0.0754 - val_mean_squared_logarithmic_error: 0.0754\n",
      "Epoch 13/134\n",
      "2100/2100 [==============================] - 0s 115us/step - loss: 0.0906 - mean_squared_logarithmic_error: 0.0906 - val_loss: 0.0770 - val_mean_squared_logarithmic_error: 0.0770\n",
      "Epoch 14/134\n",
      "2100/2100 [==============================] - 0s 106us/step - loss: 0.0826 - mean_squared_logarithmic_error: 0.0826 - val_loss: 0.0796 - val_mean_squared_logarithmic_error: 0.0796\n",
      "Epoch 15/134\n",
      "2100/2100 [==============================] - 0s 121us/step - loss: 0.0833 - mean_squared_logarithmic_error: 0.0833 - val_loss: 0.0818 - val_mean_squared_logarithmic_error: 0.0818\n",
      "Epoch 16/134\n",
      "2100/2100 [==============================] - 0s 106us/step - loss: 0.0765 - mean_squared_logarithmic_error: 0.0765 - val_loss: 0.0811 - val_mean_squared_logarithmic_error: 0.0811\n",
      "Epoch 17/134\n",
      "2100/2100 [==============================] - 0s 104us/step - loss: 0.0730 - mean_squared_logarithmic_error: 0.0730 - val_loss: 0.0829 - val_mean_squared_logarithmic_error: 0.0829\n",
      "Epoch 18/134\n",
      "2100/2100 [==============================] - 0s 105us/step - loss: 0.0653 - mean_squared_logarithmic_error: 0.0653 - val_loss: 0.0832 - val_mean_squared_logarithmic_error: 0.0832\n",
      "Epoch 19/134\n",
      "2100/2100 [==============================] - 0s 111us/step - loss: 0.0637 - mean_squared_logarithmic_error: 0.0637 - val_loss: 0.0835 - val_mean_squared_logarithmic_error: 0.0835\n",
      "Epoch 20/134\n",
      "2100/2100 [==============================] - 0s 124us/step - loss: 0.0632 - mean_squared_logarithmic_error: 0.0632 - val_loss: 0.0841 - val_mean_squared_logarithmic_error: 0.0841\n",
      "Epoch 21/134\n",
      "2100/2100 [==============================] - 0s 107us/step - loss: 0.0622 - mean_squared_logarithmic_error: 0.0622 - val_loss: 0.0830 - val_mean_squared_logarithmic_error: 0.0830\n",
      "Epoch 22/134\n",
      "2100/2100 [==============================] - 0s 107us/step - loss: 0.0574 - mean_squared_logarithmic_error: 0.0574 - val_loss: 0.0845 - val_mean_squared_logarithmic_error: 0.0845\n",
      "Epoch 23/134\n",
      "2100/2100 [==============================] - 0s 112us/step - loss: 0.0558 - mean_squared_logarithmic_error: 0.0558 - val_loss: 0.0822 - val_mean_squared_logarithmic_error: 0.0822\n",
      "Epoch 24/134\n",
      "2100/2100 [==============================] - 0s 112us/step - loss: 0.0547 - mean_squared_logarithmic_error: 0.0547 - val_loss: 0.0828 - val_mean_squared_logarithmic_error: 0.0828\n",
      "Epoch 25/134\n",
      "2100/2100 [==============================] - 0s 116us/step - loss: 0.0571 - mean_squared_logarithmic_error: 0.0571 - val_loss: 0.0853 - val_mean_squared_logarithmic_error: 0.0853\n",
      "Epoch 26/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0563 - mean_squared_logarithmic_error: 0.0563 - val_loss: 0.0849 - val_mean_squared_logarithmic_error: 0.0849\n",
      "Epoch 27/134\n",
      "2100/2100 [==============================] - 0s 121us/step - loss: 0.0542 - mean_squared_logarithmic_error: 0.0542 - val_loss: 0.0831 - val_mean_squared_logarithmic_error: 0.0831\n",
      "Epoch 28/134\n",
      "2100/2100 [==============================] - 0s 111us/step - loss: 0.0558 - mean_squared_logarithmic_error: 0.0558 - val_loss: 0.0843 - val_mean_squared_logarithmic_error: 0.0843\n",
      "Epoch 29/134\n",
      "2100/2100 [==============================] - 0s 106us/step - loss: 0.0560 - mean_squared_logarithmic_error: 0.0560 - val_loss: 0.0842 - val_mean_squared_logarithmic_error: 0.0842\n",
      "Epoch 30/134\n",
      "2100/2100 [==============================] - 0s 111us/step - loss: 0.0530 - mean_squared_logarithmic_error: 0.0530 - val_loss: 0.0841 - val_mean_squared_logarithmic_error: 0.0841\n",
      "Epoch 31/134\n",
      "2100/2100 [==============================] - 0s 122us/step - loss: 0.0536 - mean_squared_logarithmic_error: 0.0536 - val_loss: 0.0829 - val_mean_squared_logarithmic_error: 0.0829\n",
      "Epoch 32/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0548 - mean_squared_logarithmic_error: 0.0548 - val_loss: 0.0828 - val_mean_squared_logarithmic_error: 0.0828\n",
      "Epoch 33/134\n",
      "2100/2100 [==============================] - 0s 104us/step - loss: 0.0547 - mean_squared_logarithmic_error: 0.0547 - val_loss: 0.0834 - val_mean_squared_logarithmic_error: 0.0834\n",
      "Epoch 34/134\n",
      "2100/2100 [==============================] - 0s 107us/step - loss: 0.0511 - mean_squared_logarithmic_error: 0.0511 - val_loss: 0.0829 - val_mean_squared_logarithmic_error: 0.0829\n",
      "Epoch 35/134\n",
      "2100/2100 [==============================] - 0s 106us/step - loss: 0.0523 - mean_squared_logarithmic_error: 0.0523 - val_loss: 0.0821 - val_mean_squared_logarithmic_error: 0.0821\n",
      "Epoch 36/134\n",
      "2100/2100 [==============================] - 0s 107us/step - loss: 0.0557 - mean_squared_logarithmic_error: 0.0557 - val_loss: 0.0832 - val_mean_squared_logarithmic_error: 0.0832\n",
      "Epoch 37/134\n",
      "2100/2100 [==============================] - 0s 104us/step - loss: 0.0530 - mean_squared_logarithmic_error: 0.0530 - val_loss: 0.0835 - val_mean_squared_logarithmic_error: 0.0835\n",
      "Epoch 38/134\n",
      "2100/2100 [==============================] - 0s 108us/step - loss: 0.0545 - mean_squared_logarithmic_error: 0.0545 - val_loss: 0.0842 - val_mean_squared_logarithmic_error: 0.0842\n",
      "Epoch 39/134\n",
      "2100/2100 [==============================] - 0s 123us/step - loss: 0.0545 - mean_squared_logarithmic_error: 0.0545 - val_loss: 0.0840 - val_mean_squared_logarithmic_error: 0.0840\n",
      "Epoch 40/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0500 - mean_squared_logarithmic_error: 0.0500 - val_loss: 0.0834 - val_mean_squared_logarithmic_error: 0.0834\n",
      "Epoch 41/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0518 - mean_squared_logarithmic_error: 0.0518 - val_loss: 0.0839 - val_mean_squared_logarithmic_error: 0.0839\n",
      "Epoch 42/134\n",
      "2100/2100 [==============================] - 0s 109us/step - loss: 0.0511 - mean_squared_logarithmic_error: 0.0511 - val_loss: 0.0858 - val_mean_squared_logarithmic_error: 0.0858\n",
      "Epoch 43/134\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2100/2100 [==============================] - 0s 99us/step - loss: 0.0524 - mean_squared_logarithmic_error: 0.0524 - val_loss: 0.0864 - val_mean_squared_logarithmic_error: 0.0864\n",
      "Epoch 44/134\n",
      "2100/2100 [==============================] - 0s 99us/step - loss: 0.0557 - mean_squared_logarithmic_error: 0.0557 - val_loss: 0.0845 - val_mean_squared_logarithmic_error: 0.0845\n",
      "Epoch 45/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0513 - mean_squared_logarithmic_error: 0.0513 - val_loss: 0.0839 - val_mean_squared_logarithmic_error: 0.0839\n",
      "Epoch 46/134\n",
      "2100/2100 [==============================] - 0s 130us/step - loss: 0.0514 - mean_squared_logarithmic_error: 0.0514 - val_loss: 0.0836 - val_mean_squared_logarithmic_error: 0.0836\n",
      "Epoch 47/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0486 - mean_squared_logarithmic_error: 0.0486 - val_loss: 0.0846 - val_mean_squared_logarithmic_error: 0.0846\n",
      "Epoch 48/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0507 - mean_squared_logarithmic_error: 0.0507 - val_loss: 0.0859 - val_mean_squared_logarithmic_error: 0.0859\n",
      "Epoch 49/134\n",
      "2100/2100 [==============================] - 0s 106us/step - loss: 0.0524 - mean_squared_logarithmic_error: 0.0524 - val_loss: 0.0857 - val_mean_squared_logarithmic_error: 0.0857\n",
      "Epoch 50/134\n",
      "2100/2100 [==============================] - 0s 125us/step - loss: 0.0495 - mean_squared_logarithmic_error: 0.0495 - val_loss: 0.0838 - val_mean_squared_logarithmic_error: 0.0838\n",
      "Epoch 51/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0479 - mean_squared_logarithmic_error: 0.0479 - val_loss: 0.0837 - val_mean_squared_logarithmic_error: 0.0837\n",
      "Epoch 52/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0456 - mean_squared_logarithmic_error: 0.0456 - val_loss: 0.0842 - val_mean_squared_logarithmic_error: 0.0842\n",
      "Epoch 53/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0500 - mean_squared_logarithmic_error: 0.0500 - val_loss: 0.0851 - val_mean_squared_logarithmic_error: 0.0851\n",
      "Epoch 54/134\n",
      "2100/2100 [==============================] - 0s 99us/step - loss: 0.0487 - mean_squared_logarithmic_error: 0.0487 - val_loss: 0.0867 - val_mean_squared_logarithmic_error: 0.0867\n",
      "Epoch 55/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0488 - mean_squared_logarithmic_error: 0.0488 - val_loss: 0.0845 - val_mean_squared_logarithmic_error: 0.0845\n",
      "Epoch 56/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0468 - mean_squared_logarithmic_error: 0.0468 - val_loss: 0.0844 - val_mean_squared_logarithmic_error: 0.0844\n",
      "Epoch 57/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0484 - mean_squared_logarithmic_error: 0.0484 - val_loss: 0.0839 - val_mean_squared_logarithmic_error: 0.0839\n",
      "Epoch 58/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0479 - mean_squared_logarithmic_error: 0.0479 - val_loss: 0.0827 - val_mean_squared_logarithmic_error: 0.0827\n",
      "Epoch 59/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0453 - mean_squared_logarithmic_error: 0.0453 - val_loss: 0.0839 - val_mean_squared_logarithmic_error: 0.0839\n",
      "Epoch 60/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0477 - mean_squared_logarithmic_error: 0.0477 - val_loss: 0.0861 - val_mean_squared_logarithmic_error: 0.0861\n",
      "Epoch 61/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0484 - mean_squared_logarithmic_error: 0.0484 - val_loss: 0.0840 - val_mean_squared_logarithmic_error: 0.0840\n",
      "Epoch 62/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0482 - mean_squared_logarithmic_error: 0.0482 - val_loss: 0.0830 - val_mean_squared_logarithmic_error: 0.0830\n",
      "Epoch 63/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0478 - mean_squared_logarithmic_error: 0.0478 - val_loss: 0.0829 - val_mean_squared_logarithmic_error: 0.0829\n",
      "Epoch 64/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0472 - mean_squared_logarithmic_error: 0.0472 - val_loss: 0.0855 - val_mean_squared_logarithmic_error: 0.0855\n",
      "Epoch 65/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0460 - mean_squared_logarithmic_error: 0.0460 - val_loss: 0.0846 - val_mean_squared_logarithmic_error: 0.0846\n",
      "Epoch 66/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0464 - mean_squared_logarithmic_error: 0.0464 - val_loss: 0.0876 - val_mean_squared_logarithmic_error: 0.0876\n",
      "Epoch 67/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0463 - mean_squared_logarithmic_error: 0.0463 - val_loss: 0.0836 - val_mean_squared_logarithmic_error: 0.0836\n",
      "Epoch 68/134\n",
      "2100/2100 [==============================] - 0s 99us/step - loss: 0.0467 - mean_squared_logarithmic_error: 0.0467 - val_loss: 0.0836 - val_mean_squared_logarithmic_error: 0.0836\n",
      "Epoch 69/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0440 - mean_squared_logarithmic_error: 0.0440 - val_loss: 0.0855 - val_mean_squared_logarithmic_error: 0.0855\n",
      "Epoch 70/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0462 - mean_squared_logarithmic_error: 0.0462 - val_loss: 0.0842 - val_mean_squared_logarithmic_error: 0.0842\n",
      "Epoch 71/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0469 - mean_squared_logarithmic_error: 0.0469 - val_loss: 0.0836 - val_mean_squared_logarithmic_error: 0.0836\n",
      "Epoch 72/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0460 - mean_squared_logarithmic_error: 0.0460 - val_loss: 0.0827 - val_mean_squared_logarithmic_error: 0.0827\n",
      "Epoch 73/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0445 - mean_squared_logarithmic_error: 0.0445 - val_loss: 0.0919 - val_mean_squared_logarithmic_error: 0.0919\n",
      "Epoch 74/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0470 - mean_squared_logarithmic_error: 0.0470 - val_loss: 0.0873 - val_mean_squared_logarithmic_error: 0.0873\n",
      "Epoch 75/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0477 - mean_squared_logarithmic_error: 0.0477 - val_loss: 0.0845 - val_mean_squared_logarithmic_error: 0.0845\n",
      "Epoch 76/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0439 - mean_squared_logarithmic_error: 0.0439 - val_loss: 0.0844 - val_mean_squared_logarithmic_error: 0.0844\n",
      "Epoch 77/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0474 - mean_squared_logarithmic_error: 0.0474 - val_loss: 0.0837 - val_mean_squared_logarithmic_error: 0.0837\n",
      "Epoch 78/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0418 - mean_squared_logarithmic_error: 0.0418 - val_loss: 0.0836 - val_mean_squared_logarithmic_error: 0.0836\n",
      "Epoch 79/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0418 - mean_squared_logarithmic_error: 0.0418 - val_loss: 0.0864 - val_mean_squared_logarithmic_error: 0.0864\n",
      "Epoch 80/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0455 - mean_squared_logarithmic_error: 0.0455 - val_loss: 0.0891 - val_mean_squared_logarithmic_error: 0.0891\n",
      "Epoch 81/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0448 - mean_squared_logarithmic_error: 0.0448 - val_loss: 0.0842 - val_mean_squared_logarithmic_error: 0.0842\n",
      "Epoch 82/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0464 - mean_squared_logarithmic_error: 0.0464 - val_loss: 0.0838 - val_mean_squared_logarithmic_error: 0.0838\n",
      "Epoch 83/134\n",
      "2100/2100 [==============================] - 0s 99us/step - loss: 0.0465 - mean_squared_logarithmic_error: 0.0465 - val_loss: 0.0844 - val_mean_squared_logarithmic_error: 0.0844\n",
      "Epoch 84/134\n",
      "2100/2100 [==============================] - 0s 98us/step - loss: 0.0422 - mean_squared_logarithmic_error: 0.0422 - val_loss: 0.0858 - val_mean_squared_logarithmic_error: 0.0858\n",
      "Epoch 85/134\n",
      "2100/2100 [==============================] - 0s 99us/step - loss: 0.0418 - mean_squared_logarithmic_error: 0.0418 - val_loss: 0.0845 - val_mean_squared_logarithmic_error: 0.0845\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 86/134\n",
      "2100/2100 [==============================] - 0s 99us/step - loss: 0.0414 - mean_squared_logarithmic_error: 0.0414 - val_loss: 0.0862 - val_mean_squared_logarithmic_error: 0.0862\n",
      "Epoch 87/134\n",
      "2100/2100 [==============================] - 0s 99us/step - loss: 0.0421 - mean_squared_logarithmic_error: 0.0421 - val_loss: 0.0842 - val_mean_squared_logarithmic_error: 0.0842\n",
      "Epoch 88/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0434 - mean_squared_logarithmic_error: 0.0434 - val_loss: 0.0838 - val_mean_squared_logarithmic_error: 0.0838\n",
      "Epoch 89/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0412 - mean_squared_logarithmic_error: 0.0412 - val_loss: 0.0832 - val_mean_squared_logarithmic_error: 0.0832\n",
      "Epoch 90/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0415 - mean_squared_logarithmic_error: 0.0415 - val_loss: 0.0851 - val_mean_squared_logarithmic_error: 0.0851\n",
      "Epoch 91/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0456 - mean_squared_logarithmic_error: 0.0456 - val_loss: 0.0839 - val_mean_squared_logarithmic_error: 0.0839\n",
      "Epoch 92/134\n",
      "2100/2100 [==============================] - 0s 107us/step - loss: 0.0400 - mean_squared_logarithmic_error: 0.0400 - val_loss: 0.0835 - val_mean_squared_logarithmic_error: 0.0835\n",
      "Epoch 93/134\n",
      "2100/2100 [==============================] - 0s 120us/step - loss: 0.0416 - mean_squared_logarithmic_error: 0.0416 - val_loss: 0.0850 - val_mean_squared_logarithmic_error: 0.0850\n",
      "Epoch 94/134\n",
      "2100/2100 [==============================] - 0s 116us/step - loss: 0.0428 - mean_squared_logarithmic_error: 0.0428 - val_loss: 0.0844 - val_mean_squared_logarithmic_error: 0.0844\n",
      "Epoch 95/134\n",
      "2100/2100 [==============================] - 0s 112us/step - loss: 0.0424 - mean_squared_logarithmic_error: 0.0424 - val_loss: 0.0838 - val_mean_squared_logarithmic_error: 0.0838\n",
      "Epoch 96/134\n",
      "2100/2100 [==============================] - 0s 113us/step - loss: 0.0493 - mean_squared_logarithmic_error: 0.0493 - val_loss: 0.0833 - val_mean_squared_logarithmic_error: 0.0833\n",
      "Epoch 97/134\n",
      "2100/2100 [==============================] - 0s 117us/step - loss: 0.0419 - mean_squared_logarithmic_error: 0.0419 - val_loss: 0.0834 - val_mean_squared_logarithmic_error: 0.0834\n",
      "Epoch 98/134\n",
      "2100/2100 [==============================] - 0s 105us/step - loss: 0.0416 - mean_squared_logarithmic_error: 0.0416 - val_loss: 0.0826 - val_mean_squared_logarithmic_error: 0.0826\n",
      "Epoch 99/134\n",
      "2100/2100 [==============================] - 0s 121us/step - loss: 0.0410 - mean_squared_logarithmic_error: 0.0410 - val_loss: 0.0823 - val_mean_squared_logarithmic_error: 0.0823\n",
      "Epoch 100/134\n",
      "2100/2100 [==============================] - 0s 120us/step - loss: 0.0394 - mean_squared_logarithmic_error: 0.0394 - val_loss: 0.0830 - val_mean_squared_logarithmic_error: 0.0830\n",
      "Epoch 101/134\n",
      "2100/2100 [==============================] - 0s 114us/step - loss: 0.0439 - mean_squared_logarithmic_error: 0.0439 - val_loss: 0.0815 - val_mean_squared_logarithmic_error: 0.0815\n",
      "Epoch 102/134\n",
      "2100/2100 [==============================] - 0s 109us/step - loss: 0.0376 - mean_squared_logarithmic_error: 0.0376 - val_loss: 0.0818 - val_mean_squared_logarithmic_error: 0.0818\n",
      "Epoch 103/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0414 - mean_squared_logarithmic_error: 0.0414 - val_loss: 0.0826 - val_mean_squared_logarithmic_error: 0.0826\n",
      "Epoch 104/134\n",
      "2100/2100 [==============================] - 0s 104us/step - loss: 0.0425 - mean_squared_logarithmic_error: 0.0425 - val_loss: 0.0822 - val_mean_squared_logarithmic_error: 0.0822\n",
      "Epoch 105/134\n",
      "2100/2100 [==============================] - 0s 117us/step - loss: 0.0386 - mean_squared_logarithmic_error: 0.0386 - val_loss: 0.0826 - val_mean_squared_logarithmic_error: 0.0826\n",
      "Epoch 106/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0382 - mean_squared_logarithmic_error: 0.0382 - val_loss: 0.0826 - val_mean_squared_logarithmic_error: 0.0826\n",
      "Epoch 107/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0405 - mean_squared_logarithmic_error: 0.0405 - val_loss: 0.0833 - val_mean_squared_logarithmic_error: 0.0833\n",
      "Epoch 108/134\n",
      "2100/2100 [==============================] - 0s 104us/step - loss: 0.0389 - mean_squared_logarithmic_error: 0.0389 - val_loss: 0.0863 - val_mean_squared_logarithmic_error: 0.0863\n",
      "Epoch 109/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0372 - mean_squared_logarithmic_error: 0.0372 - val_loss: 0.0839 - val_mean_squared_logarithmic_error: 0.0839\n",
      "Epoch 110/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0375 - mean_squared_logarithmic_error: 0.0375 - val_loss: 0.0834 - val_mean_squared_logarithmic_error: 0.0834\n",
      "Epoch 111/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0385 - mean_squared_logarithmic_error: 0.0385 - val_loss: 0.0841 - val_mean_squared_logarithmic_error: 0.0841\n",
      "Epoch 112/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0362 - mean_squared_logarithmic_error: 0.0362 - val_loss: 0.0825 - val_mean_squared_logarithmic_error: 0.0825\n",
      "Epoch 113/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0398 - mean_squared_logarithmic_error: 0.0398 - val_loss: 0.0832 - val_mean_squared_logarithmic_error: 0.0832\n",
      "Epoch 114/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0403 - mean_squared_logarithmic_error: 0.0403 - val_loss: 0.0816 - val_mean_squared_logarithmic_error: 0.0816\n",
      "Epoch 115/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0387 - mean_squared_logarithmic_error: 0.0387 - val_loss: 0.0853 - val_mean_squared_logarithmic_error: 0.0853\n",
      "Epoch 116/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0383 - mean_squared_logarithmic_error: 0.0383 - val_loss: 0.0834 - val_mean_squared_logarithmic_error: 0.0834\n",
      "Epoch 117/134\n",
      "2100/2100 [==============================] - 0s 116us/step - loss: 0.0365 - mean_squared_logarithmic_error: 0.0365 - val_loss: 0.0824 - val_mean_squared_logarithmic_error: 0.0824\n",
      "Epoch 118/134\n",
      "2100/2100 [==============================] - 0s 122us/step - loss: 0.0352 - mean_squared_logarithmic_error: 0.0352 - val_loss: 0.0825 - val_mean_squared_logarithmic_error: 0.0825\n",
      "Epoch 119/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0371 - mean_squared_logarithmic_error: 0.0371 - val_loss: 0.0828 - val_mean_squared_logarithmic_error: 0.0828\n",
      "Epoch 120/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0340 - mean_squared_logarithmic_error: 0.0340 - val_loss: 0.0817 - val_mean_squared_logarithmic_error: 0.0817\n",
      "Epoch 121/134\n",
      "2100/2100 [==============================] - 0s 109us/step - loss: 0.0343 - mean_squared_logarithmic_error: 0.0343 - val_loss: 0.0835 - val_mean_squared_logarithmic_error: 0.0835\n",
      "Epoch 122/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0357 - mean_squared_logarithmic_error: 0.0357 - val_loss: 0.0818 - val_mean_squared_logarithmic_error: 0.0818\n",
      "Epoch 123/134\n",
      "2100/2100 [==============================] - 0s 103us/step - loss: 0.0354 - mean_squared_logarithmic_error: 0.0354 - val_loss: 0.0824 - val_mean_squared_logarithmic_error: 0.0824\n",
      "Epoch 124/134\n",
      "2100/2100 [==============================] - 0s 105us/step - loss: 0.0334 - mean_squared_logarithmic_error: 0.0334 - val_loss: 0.0850 - val_mean_squared_logarithmic_error: 0.0850\n",
      "Epoch 125/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0340 - mean_squared_logarithmic_error: 0.0340 - val_loss: 0.0816 - val_mean_squared_logarithmic_error: 0.0816\n",
      "Epoch 126/134\n",
      "2100/2100 [==============================] - 0s 101us/step - loss: 0.0349 - mean_squared_logarithmic_error: 0.0349 - val_loss: 0.0824 - val_mean_squared_logarithmic_error: 0.0824\n",
      "Epoch 127/134\n",
      "2100/2100 [==============================] - 0s 108us/step - loss: 0.0354 - mean_squared_logarithmic_error: 0.0354 - val_loss: 0.0812 - val_mean_squared_logarithmic_error: 0.0812\n",
      "Epoch 128/134\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0326 - mean_squared_logarithmic_error: 0.0326 - val_loss: 0.0820 - val_mean_squared_logarithmic_error: 0.0820\n",
      "Epoch 129/134\n",
      "2100/2100 [==============================] - 0s 106us/step - loss: 0.0329 - mean_squared_logarithmic_error: 0.0329 - val_loss: 0.0832 - val_mean_squared_logarithmic_error: 0.0832\n",
      "Epoch 130/134\n",
      "2100/2100 [==============================] - 0s 105us/step - loss: 0.0345 - mean_squared_logarithmic_error: 0.0345 - val_loss: 0.0820 - val_mean_squared_logarithmic_error: 0.0820\n",
      "Epoch 131/134\n",
      "2100/2100 [==============================] - 0s 105us/step - loss: 0.0331 - mean_squared_logarithmic_error: 0.0331 - val_loss: 0.0836 - val_mean_squared_logarithmic_error: 0.0836\n",
      "Epoch 132/134\n",
      "2100/2100 [==============================] - 0s 102us/step - loss: 0.0329 - mean_squared_logarithmic_error: 0.0329 - val_loss: 0.0830 - val_mean_squared_logarithmic_error: 0.0830\n",
      "Epoch 133/134\n",
      "2100/2100 [==============================] - 0s 100us/step - loss: 0.0337 - mean_squared_logarithmic_error: 0.0337 - val_loss: 0.0858 - val_mean_squared_logarithmic_error: 0.0858\n",
      "Epoch 134/134\n",
      "2100/2100 [==============================] - 0s 99us/step - loss: 0.0295 - mean_squared_logarithmic_error: 0.0295 - val_loss: 0.0844 - val_mean_squared_logarithmic_error: 0.0844\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_3 (Embedding)      (None, 50, 30)            109860    \n",
      "_________________________________________________________________\n",
      "flatten_3 (Flatten)          (None, 1500)              0         \n",
      "_________________________________________________________________\n",
      "dense_33 (Dense)             (None, 128)               192128    \n",
      "_________________________________________________________________\n",
      "dropout_15 (Dropout)         (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_34 (Dense)             (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dropout_16 (Dropout)         (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_35 (Dense)             (None, 1)                 65        \n",
      "=================================================================\n",
      "Total params: 310,309\n",
      "Trainable params: 310,309\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model4, history4 = create_embedding_model(X3,Y3,X3_test,Y3_test)\n",
    "model4.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "acc = history4.history['mean_squared_logarithmic_error']\n",
    "val_acc = history4.history['val_mean_squared_logarithmic_error']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training')\n",
    "plt.plot(epochs, val_acc, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('MSLE')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "l1 = history4.history['loss']\n",
    "l2 = history4.history['val_loss']\n",
    "\n",
    "epochs = range(len(l1))\n",
    "\n",
    "plt.plot(epochs, l1, 'bo', label='Training')\n",
    "plt.plot(epochs, l2, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "# print(model4.predict(df))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "ENSEMBLE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 2100 samples, validate on 900 samples\n",
      "Epoch 1/40\n",
      "2100/2100 [==============================] - 1s 406us/step - loss: 157.7333 - mean_squared_logarithmic_error: 1.7126 - val_loss: 125.1355 - val_mean_squared_logarithmic_error: 1.0341\n",
      "Epoch 2/40\n",
      "2100/2100 [==============================] - 0s 71us/step - loss: 100.1049 - mean_squared_logarithmic_error: 0.7369 - val_loss: 78.1642 - val_mean_squared_logarithmic_error: 0.4907\n",
      "Epoch 3/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 57.7449 - mean_squared_logarithmic_error: 0.3333 - val_loss: 43.1486 - val_mean_squared_logarithmic_error: 0.2284\n",
      "Epoch 4/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 27.3811 - mean_squared_logarithmic_error: 0.1395 - val_loss: 20.3901 - val_mean_squared_logarithmic_error: 0.1092\n",
      "Epoch 5/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 10.4857 - mean_squared_logarithmic_error: 0.0629 - val_loss: 9.9977 - val_mean_squared_logarithmic_error: 0.0713\n",
      "Epoch 6/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 5.5663 - mean_squared_logarithmic_error: 0.0472 - val_loss: 8.8202 - val_mean_squared_logarithmic_error: 0.0694\n",
      "Epoch 7/40\n",
      "2100/2100 [==============================] - 0s 70us/step - loss: 4.6829 - mean_squared_logarithmic_error: 0.0434 - val_loss: 8.6712 - val_mean_squared_logarithmic_error: 0.0694\n",
      "Epoch 8/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 4.2586 - mean_squared_logarithmic_error: 0.0410 - val_loss: 8.5801 - val_mean_squared_logarithmic_error: 0.0689\n",
      "Epoch 9/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 3.6353 - mean_squared_logarithmic_error: 0.0369 - val_loss: 8.5320 - val_mean_squared_logarithmic_error: 0.0685\n",
      "Epoch 10/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 3.2570 - mean_squared_logarithmic_error: 0.0348 - val_loss: 8.4207 - val_mean_squared_logarithmic_error: 0.0680\n",
      "Epoch 11/40\n",
      "2100/2100 [==============================] - 0s 70us/step - loss: 2.7744 - mean_squared_logarithmic_error: 0.0308 - val_loss: 8.5554 - val_mean_squared_logarithmic_error: 0.0678\n",
      "Epoch 12/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 2.5192 - mean_squared_logarithmic_error: 0.0287 - val_loss: 8.6402 - val_mean_squared_logarithmic_error: 0.0682\n",
      "Epoch 13/40\n",
      "2100/2100 [==============================] - 0s 79us/step - loss: 2.2284 - mean_squared_logarithmic_error: 0.0256 - val_loss: 8.7314 - val_mean_squared_logarithmic_error: 0.0688\n",
      "Epoch 14/40\n",
      "2100/2100 [==============================] - 0s 72us/step - loss: 2.0987 - mean_squared_logarithmic_error: 0.0245 - val_loss: 9.1185 - val_mean_squared_logarithmic_error: 0.0694\n",
      "Epoch 15/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 1.9209 - mean_squared_logarithmic_error: 0.0224 - val_loss: 9.3488 - val_mean_squared_logarithmic_error: 0.0703\n",
      "Epoch 16/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 1.6775 - mean_squared_logarithmic_error: 0.0199 - val_loss: 9.5276 - val_mean_squared_logarithmic_error: 0.0710\n",
      "Epoch 17/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 1.5870 - mean_squared_logarithmic_error: 0.0182 - val_loss: 9.7342 - val_mean_squared_logarithmic_error: 0.0718\n",
      "Epoch 18/40\n",
      "2100/2100 [==============================] - 0s 77us/step - loss: 1.5320 - mean_squared_logarithmic_error: 0.0175 - val_loss: 10.0066 - val_mean_squared_logarithmic_error: 0.0726\n",
      "Epoch 19/40\n",
      "2100/2100 [==============================] - 0s 78us/step - loss: 1.3919 - mean_squared_logarithmic_error: 0.0157 - val_loss: 10.1400 - val_mean_squared_logarithmic_error: 0.0733\n",
      "Epoch 20/40\n",
      "2100/2100 [==============================] - 0s 71us/step - loss: 1.4206 - mean_squared_logarithmic_error: 0.0147 - val_loss: 10.4180 - val_mean_squared_logarithmic_error: 0.0742\n",
      "Epoch 21/40\n",
      "2100/2100 [==============================] - 0s 79us/step - loss: 1.2779 - mean_squared_logarithmic_error: 0.0139 - val_loss: 10.3573 - val_mean_squared_logarithmic_error: 0.0744\n",
      "Epoch 22/40\n",
      "2100/2100 [==============================] - 0s 76us/step - loss: 1.2690 - mean_squared_logarithmic_error: 0.0134 - val_loss: 10.4177 - val_mean_squared_logarithmic_error: 0.0748\n",
      "Epoch 23/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 1.2873 - mean_squared_logarithmic_error: 0.0132 - val_loss: 10.4088 - val_mean_squared_logarithmic_error: 0.0750\n",
      "Epoch 24/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 1.2039 - mean_squared_logarithmic_error: 0.0119 - val_loss: 10.5620 - val_mean_squared_logarithmic_error: 0.0756\n",
      "Epoch 25/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 1.2109 - mean_squared_logarithmic_error: 0.0114 - val_loss: 10.8627 - val_mean_squared_logarithmic_error: 0.0765\n",
      "Epoch 26/40\n",
      "2100/2100 [==============================] - 0s 75us/step - loss: 1.1362 - mean_squared_logarithmic_error: 0.0111 - val_loss: 10.7244 - val_mean_squared_logarithmic_error: 0.0763\n",
      "Epoch 27/40\n",
      "2100/2100 [==============================] - 0s 81us/step - loss: 1.0953 - mean_squared_logarithmic_error: 0.0103 - val_loss: 10.5537 - val_mean_squared_logarithmic_error: 0.0762\n",
      "Epoch 28/40\n",
      "2100/2100 [==============================] - 0s 73us/step - loss: 1.1122 - mean_squared_logarithmic_error: 0.0098 - val_loss: 10.5759 - val_mean_squared_logarithmic_error: 0.0763\n",
      "Epoch 29/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 1.1119 - mean_squared_logarithmic_error: 0.0097 - val_loss: 10.8483 - val_mean_squared_logarithmic_error: 0.0770\n",
      "Epoch 30/40\n",
      "2100/2100 [==============================] - 0s 68us/step - loss: 1.1257 - mean_squared_logarithmic_error: 0.0089 - val_loss: 10.9721 - val_mean_squared_logarithmic_error: 0.0776\n",
      "Epoch 31/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 1.0346 - mean_squared_logarithmic_error: 0.0081 - val_loss: 10.8932 - val_mean_squared_logarithmic_error: 0.0774\n",
      "Epoch 32/40\n",
      "2100/2100 [==============================] - 0s 80us/step - loss: 1.0337 - mean_squared_logarithmic_error: 0.0086 - val_loss: 10.8142 - val_mean_squared_logarithmic_error: 0.0772\n",
      "Epoch 33/40\n",
      "2100/2100 [==============================] - 0s 79us/step - loss: 1.0439 - mean_squared_logarithmic_error: 0.0085 - val_loss: 11.0450 - val_mean_squared_logarithmic_error: 0.0780\n",
      "Epoch 34/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 0.9819 - mean_squared_logarithmic_error: 0.0078 - val_loss: 10.9461 - val_mean_squared_logarithmic_error: 0.0777\n",
      "Epoch 35/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 0.9949 - mean_squared_logarithmic_error: 0.0071 - val_loss: 11.1512 - val_mean_squared_logarithmic_error: 0.0783\n",
      "Epoch 36/40\n",
      "2100/2100 [==============================] - 0s 70us/step - loss: 0.9631 - mean_squared_logarithmic_error: 0.0073 - val_loss: 10.9424 - val_mean_squared_logarithmic_error: 0.0778\n",
      "Epoch 37/40\n",
      "2100/2100 [==============================] - 0s 69us/step - loss: 1.0029 - mean_squared_logarithmic_error: 0.0074 - val_loss: 10.9464 - val_mean_squared_logarithmic_error: 0.0779\n",
      "Epoch 38/40\n",
      "2100/2100 [==============================] - 0s 74us/step - loss: 0.9467 - mean_squared_logarithmic_error: 0.0067 - val_loss: 10.9646 - val_mean_squared_logarithmic_error: 0.0780\n",
      "Epoch 39/40\n",
      "2100/2100 [==============================] - 0s 71us/step - loss: 0.9694 - mean_squared_logarithmic_error: 0.0064 - val_loss: 10.8989 - val_mean_squared_logarithmic_error: 0.0778\n",
      "Epoch 40/40\n",
      "2100/2100 [==============================] - 0s 83us/step - loss: 0.9014 - mean_squared_logarithmic_error: 0.0059 - val_loss: 10.9922 - val_mean_squared_logarithmic_error: 0.0781\n"
     ]
    }
   ],
   "source": [
    "#MODEL ONE\n",
    "test_df = test_df.apply(np.log)\n",
    "test_df = test_df.applymap(f)\n",
    "test_df = test_df.fillna(0)\n",
    "\n",
    "test_all = test_all.apply(np.log)\n",
    "test_all = test_all.applymap(f)\n",
    "test_all = test_all.fillna(0)\n",
    "result_X_train_1 = model1.predict(X)\n",
    "#result_X_train_1 = model1.predict(X_0)\n",
    "result_X_validation_1 = model1.predict(X_test) #Get prediction\n",
    "\n",
    "#MODEL TWO\n",
    "cc_test = cc_test.applymap(f)\n",
    "cc_test = cc_test.fillna(0)\n",
    "result_X_train_2 = model2.predict(X1)\n",
    "#result_X_train_2 = model2.predict(X_0_1)\n",
    "result_X_validation_2 = model2.predict(X1_test) #Get prediction\n",
    "\n",
    "#MODEL THREE\n",
    "date_test = date_test.applymap(f)\n",
    "date_test = date_test.fillna(0)\n",
    "result_X_train_3 = model3.predict(X2)\n",
    "#result_X_train_3 = model3.predict(X_0_2)\n",
    "result_X_validation_3 = model3.predict(X2_test) #Get prediction\n",
    "\n",
    "#MODEL FOUR\n",
    "title_test = pd.read_csv('./data/test.csv')\n",
    "title_test = title_test['title']\n",
    "title_test = title_test.apply(lambda x:create_word_vec(str(x)))\n",
    "title_test = np.stack(title_test.to_numpy(),axis = 0)\n",
    "result_X_train_4 = model4.predict(X3)\n",
    "result_X_validation_4 = model4.predict(X3_test)\n",
    "\n",
    "\n",
    "#Create data frames for ensemble model training using predictions from all models\n",
    "training_data_ensemble = pd.DataFrame({'result1':result_X_train_1.reshape(result_X_train_1.shape[0]),\n",
    "                                       'result2':result_X_train_2.reshape(result_X_train_2.shape[0]),\n",
    "                                       'result3':result_X_train_3.reshape(result_X_train_3.shape[0]),\n",
    "                                       'result4':result_X_train_4.reshape(result_X_train_4.shape[0])})\n",
    "training_Y_ensemble = Y\n",
    "#training_Y_ensemble = Y_0\n",
    "\n",
    "validation_data_ensemble = pd.DataFrame({'result1':result_X_validation_1.reshape(result_X_validation_1.shape[0]),\n",
    "                                         'result2':result_X_validation_2.reshape(result_X_validation_2.shape[0]),\n",
    "                                         'result3':result_X_validation_3.reshape(result_X_validation_3.shape[0]),\n",
    "                                         'result4':result_X_validation_4.reshape(result_X_validation_4.shape[0])})\n",
    "validation_Y_ensemble = Y_test\n",
    "\n",
    "#Create the Ensemble model\n",
    "model_e = models.Sequential()\n",
    "model_e.add(layers.Dense(356, activation='relu', input_shape=(training_data_ensemble.shape[1],)))\n",
    "model_e.add(layers.Dropout(0.1))\n",
    "model_e.add(layers.Dense(256, activation='relu'))\n",
    "model_e.add(layers.Dense(10,activation='relu'))\n",
    "model_e.add(layers.Dense(1))    \n",
    "model_e.compile(optimizer=optimizers.rmsprop(lr=1e-5),loss='mse',metrics=['mean_squared_logarithmic_error'])\n",
    "history_e = model_e.fit(training_data_ensemble, training_Y_ensemble, epochs=40, batch_size = 32,\n",
    "                         validation_data=(validation_data_ensemble, validation_Y_ensemble))\n",
    "\n",
    "#history_e = model_e.fit(training_data_ensemble, training_Y_ensemble, epochs=40, batch_size = 32)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "acc = history_e.history['mean_squared_logarithmic_error']\n",
    "val_acc = history_e.history['val_mean_squared_logarithmic_error']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training')\n",
    "plt.plot(epochs, val_acc, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('MSLE')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "l1 = history_e.history['loss']\n",
    "l2 = history_e.history['val_loss']\n",
    "\n",
    "epochs = range(len(l1))\n",
    "\n",
    "plt.plot(epochs, l1, 'bo', label='Training')\n",
    "plt.plot(epochs, l2, 'r', label='Validation')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "GET PREDICTION USING ENSEMBLE METHOD"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Get test predictions\n",
    "test_m1_prediction = model1.predict(test_df)\n",
    "#test_m1_prediction = model1.predict(test_all)\n",
    "test_m2_prediction = model2.predict(cc_test)\n",
    "test_m3_prediction = model3.predict(date_test)\n",
    "test_m4_prediction = model4.predict(title_test)\n",
    "test_data_ensemble = pd.DataFrame({'result1':test_m1_prediction.reshape(test_m1_prediction.shape[0]),\n",
    "                                   'result2':test_m2_prediction.reshape(test_m2_prediction.shape[0]),\n",
    "                                   'result3':test_m3_prediction.reshape(test_m3_prediction.shape[0]),\n",
    "                                   'result4':test_m4_prediction.reshape(test_m4_prediction.shape[0])})\n",
    "test_ensemble_prediction = model_e.predict(test_data_ensemble)\n",
    "test_ensemble_prediction = np.exp(test_ensemble_prediction)\n",
    "\n",
    "#Generate the submission CSV\n",
    "output_df = pd.read_csv('./data/test.csv')\n",
    "for x in ALL_ATTRIBUTES:\n",
    "    try:\n",
    "        if x != 'revenue'and x != 'id':\n",
    "            output_df = output_df.drop(x, axis=1)\n",
    "    except:\n",
    "        pass\n",
    "\n",
    "output_df['revenue'] = test_ensemble_prediction\n",
    "output_df.to_csv(\"./submission.csv\",index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}