final-clean.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Admin\\AppData\\Local\\Temp\\ipykernel_4588\\1822130159.py:1: DtypeWarning: Columns (12) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  df = pd.read_csv('data/combined_cleaned_data/combined_with_fe.csv')\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0.1</th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>overall</th>\n",
       "      <th>verified</th>\n",
       "      <th>reviewTime</th>\n",
       "      <th>reviewerID</th>\n",
       "      <th>asin</th>\n",
       "      <th>style</th>\n",
       "      <th>reviewerName</th>\n",
       "      <th>reviewText</th>\n",
       "      <th>summary</th>\n",
       "      <th>unixReviewTime</th>\n",
       "      <th>vote</th>\n",
       "      <th>image</th>\n",
       "      <th>review_clean</th>\n",
       "      <th>sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>09 24, 2013</td>\n",
       "      <td>A3CZ890UHC8HHZ</td>\n",
       "      <td>B0009RF9DW</td>\n",
       "      <td>{'Size:': ' 6.6 oz'}</td>\n",
       "      <td>Amazon Customer</td>\n",
       "      <td>It diffuses a very mild light perfume, just wh...</td>\n",
       "      <td>What I expected</td>\n",
       "      <td>1379980800</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>wear mild perfume nice strong lotion day perfu...</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>07 12, 2014</td>\n",
       "      <td>A125TMC44CJBKK</td>\n",
       "      <td>B0012Y0ZG2</td>\n",
       "      <td>{'Size:': ' B-020'}</td>\n",
       "      <td>Mustang Shelly</td>\n",
       "      <td>All time favorite!!!! Wish they still carried ...</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>1405123200</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>time still favorite carry wish</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>12 28, 2014</td>\n",
       "      <td>A3JDSBP01M7R2R</td>\n",
       "      <td>B0012Y0ZG2</td>\n",
       "      <td>{'Size:': ' 112'}</td>\n",
       "      <td>Tim A. Brooks</td>\n",
       "      <td>One of my wife's favorites.</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>1419724800</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>one favorites wife's</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>03 3, 2014</td>\n",
       "      <td>A2HT5WENL8E5S6</td>\n",
       "      <td>B000FI4S1E</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Shasta Masser</td>\n",
       "      <td>If you have body acne this product is a must. ...</td>\n",
       "      <td>Acne Body Wash A MUST</td>\n",
       "      <td>1393804800</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>heal help within prevent already first wash mu...</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>11 7, 2014</td>\n",
       "      <td>A7LAYX9R7IMOO</td>\n",
       "      <td>B0012Y0ZG2</td>\n",
       "      <td>{'Size:': ' 123'}</td>\n",
       "      <td>andre luiz</td>\n",
       "      <td>I really is what I expected</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>1415318400</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>expect really</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0.1  Unnamed: 0  overall  verified   reviewTime      reviewerID  \\\n",
       "0             0           0        5      True  09 24, 2013  A3CZ890UHC8HHZ   \n",
       "1             1           1        5     False  07 12, 2014  A125TMC44CJBKK   \n",
       "2             2           2        5      True  12 28, 2014  A3JDSBP01M7R2R   \n",
       "3             3           3        5      True   03 3, 2014  A2HT5WENL8E5S6   \n",
       "4             4           4        5     False   11 7, 2014   A7LAYX9R7IMOO   \n",
       "\n",
       "         asin                 style     reviewerName  \\\n",
       "0  B0009RF9DW  {'Size:': ' 6.6 oz'}  Amazon Customer   \n",
       "1  B0012Y0ZG2   {'Size:': ' B-020'}   Mustang Shelly   \n",
       "2  B0012Y0ZG2     {'Size:': ' 112'}    Tim A. Brooks   \n",
       "3  B000FI4S1E                   NaN    Shasta Masser   \n",
       "4  B0012Y0ZG2     {'Size:': ' 123'}       andre luiz   \n",
       "\n",
       "                                          reviewText                summary  \\\n",
       "0  It diffuses a very mild light perfume, just wh...        What I expected   \n",
       "1  All time favorite!!!! Wish they still carried ...             Five Stars   \n",
       "2                        One of my wife's favorites.             Five Stars   \n",
       "3  If you have body acne this product is a must. ...  Acne Body Wash A MUST   \n",
       "4                        I really is what I expected             Five Stars   \n",
       "\n",
       "   unixReviewTime vote image  \\\n",
       "0      1379980800  NaN   NaN   \n",
       "1      1405123200  NaN   NaN   \n",
       "2      1419724800  NaN   NaN   \n",
       "3      1393804800  NaN   NaN   \n",
       "4      1415318400  NaN   NaN   \n",
       "\n",
       "                                        review_clean sentiment  \n",
       "0  wear mild perfume nice strong lotion day perfu...  Positive  \n",
       "1                     time still favorite carry wish  Positive  \n",
       "2                               one favorites wife's  Positive  \n",
       "3  heal help within prevent already first wash mu...  Positive  \n",
       "4                                      expect really  Positive  "
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('data/combined_cleaned_data/combined_with_fe.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop(df.columns[[0, 1]], axis = 1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>overall</th>\n",
       "      <th>verified</th>\n",
       "      <th>reviewTime</th>\n",
       "      <th>reviewerID</th>\n",
       "      <th>asin</th>\n",
       "      <th>style</th>\n",
       "      <th>reviewerName</th>\n",
       "      <th>reviewText</th>\n",
       "      <th>summary</th>\n",
       "      <th>unixReviewTime</th>\n",
       "      <th>vote</th>\n",
       "      <th>image</th>\n",
       "      <th>review_clean</th>\n",
       "      <th>sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>09 24, 2013</td>\n",
       "      <td>A3CZ890UHC8HHZ</td>\n",
       "      <td>B0009RF9DW</td>\n",
       "      <td>{'Size:': ' 6.6 oz'}</td>\n",
       "      <td>Amazon Customer</td>\n",
       "      <td>It diffuses a very mild light perfume, just wh...</td>\n",
       "      <td>What I expected</td>\n",
       "      <td>1379980800</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>wear mild perfume nice strong lotion day perfu...</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>07 12, 2014</td>\n",
       "      <td>A125TMC44CJBKK</td>\n",
       "      <td>B0012Y0ZG2</td>\n",
       "      <td>{'Size:': ' B-020'}</td>\n",
       "      <td>Mustang Shelly</td>\n",
       "      <td>All time favorite!!!! Wish they still carried ...</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>1405123200</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>time still favorite carry wish</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>12 28, 2014</td>\n",
       "      <td>A3JDSBP01M7R2R</td>\n",
       "      <td>B0012Y0ZG2</td>\n",
       "      <td>{'Size:': ' 112'}</td>\n",
       "      <td>Tim A. Brooks</td>\n",
       "      <td>One of my wife's favorites.</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>1419724800</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>one favorites wife's</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>03 3, 2014</td>\n",
       "      <td>A2HT5WENL8E5S6</td>\n",
       "      <td>B000FI4S1E</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Shasta Masser</td>\n",
       "      <td>If you have body acne this product is a must. ...</td>\n",
       "      <td>Acne Body Wash A MUST</td>\n",
       "      <td>1393804800</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>heal help within prevent already first wash mu...</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>11 7, 2014</td>\n",
       "      <td>A7LAYX9R7IMOO</td>\n",
       "      <td>B0012Y0ZG2</td>\n",
       "      <td>{'Size:': ' 123'}</td>\n",
       "      <td>andre luiz</td>\n",
       "      <td>I really is what I expected</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>1415318400</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>expect really</td>\n",
       "      <td>Positive</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   overall  verified   reviewTime      reviewerID        asin  \\\n",
       "0        5      True  09 24, 2013  A3CZ890UHC8HHZ  B0009RF9DW   \n",
       "1        5     False  07 12, 2014  A125TMC44CJBKK  B0012Y0ZG2   \n",
       "2        5      True  12 28, 2014  A3JDSBP01M7R2R  B0012Y0ZG2   \n",
       "3        5      True   03 3, 2014  A2HT5WENL8E5S6  B000FI4S1E   \n",
       "4        5     False   11 7, 2014   A7LAYX9R7IMOO  B0012Y0ZG2   \n",
       "\n",
       "                  style     reviewerName  \\\n",
       "0  {'Size:': ' 6.6 oz'}  Amazon Customer   \n",
       "1   {'Size:': ' B-020'}   Mustang Shelly   \n",
       "2     {'Size:': ' 112'}    Tim A. Brooks   \n",
       "3                   NaN    Shasta Masser   \n",
       "4     {'Size:': ' 123'}       andre luiz   \n",
       "\n",
       "                                          reviewText                summary  \\\n",
       "0  It diffuses a very mild light perfume, just wh...        What I expected   \n",
       "1  All time favorite!!!! Wish they still carried ...             Five Stars   \n",
       "2                        One of my wife's favorites.             Five Stars   \n",
       "3  If you have body acne this product is a must. ...  Acne Body Wash A MUST   \n",
       "4                        I really is what I expected             Five Stars   \n",
       "\n",
       "   unixReviewTime vote image  \\\n",
       "0      1379980800  NaN   NaN   \n",
       "1      1405123200  NaN   NaN   \n",
       "2      1419724800  NaN   NaN   \n",
       "3      1393804800  NaN   NaN   \n",
       "4      1415318400  NaN   NaN   \n",
       "\n",
       "                                        review_clean sentiment  \n",
       "0  wear mild perfume nice strong lotion day perfu...  Positive  \n",
       "1                     time still favorite carry wish  Positive  \n",
       "2                               one favorites wife's  Positive  \n",
       "3  heal help within prevent already first wash mu...  Positive  \n",
       "4                                      expect really  Positive  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "overall            int64\n",
       "verified            bool\n",
       "reviewTime        object\n",
       "reviewerID        object\n",
       "asin              object\n",
       "style             object\n",
       "reviewerName      object\n",
       "reviewText        object\n",
       "summary           object\n",
       "unixReviewTime     int64\n",
       "vote              object\n",
       "image             object\n",
       "review_clean      object\n",
       "sentiment         object\n",
       "dtype: object"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "overall               0\n",
       "verified              0\n",
       "reviewTime            0\n",
       "reviewerID            0\n",
       "asin                  0\n",
       "style             30781\n",
       "reviewerName         10\n",
       "reviewText           48\n",
       "summary              16\n",
       "unixReviewTime        0\n",
       "vote              61097\n",
       "image             71755\n",
       "review_clean        241\n",
       "sentiment             0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df. isnull(). sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.dropna(subset=['reviewText', 'review_clean'], inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "overall               0\n",
       "verified              0\n",
       "reviewTime            0\n",
       "reviewerID            0\n",
       "asin                  0\n",
       "style             30688\n",
       "reviewerName          9\n",
       "reviewText            0\n",
       "summary              15\n",
       "unixReviewTime        0\n",
       "vote              60857\n",
       "image             71525\n",
       "review_clean          0\n",
       "sentiment             0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df. isnull(). sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop(['reviewTime', 'asin', 'style', 'reviewerName', 'unixReviewTime', 'vote', 'image'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        wear mild perfume nice strong lotion day perfu...\n",
       "1                           time still favorite carry wish\n",
       "2                                     one favorites wife's\n",
       "3        heal help within prevent already first wash mu...\n",
       "4                                            expect really\n",
       "                               ...                        \n",
       "73619    think guide go acela maybe i've realize star t...\n",
       "73620                                                 good\n",
       "73621                         great quality price good job\n",
       "73622    point else sense code point good make worked g...\n",
       "73623    reason buy buy something madden great already ...\n",
       "Name: review_clean, Length: 73383, dtype: object"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['review_clean']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from textblob import Word\n",
    "import wordcloud"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "# common_words=''\n",
    "# for i in df.review_clean:\n",
    "#     i = str(i)\n",
    "#     tokens = i.split()\n",
    "#     common_words += \" \".join(tokens)+\" \"\n",
    "# wordcloud = wordcloud.WordCloud().generate(common_words)\n",
    "# plt.imshow(wordcloud, interpolation='bilinear')\n",
    "# plt.axis(\"off\")\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "# df[\"sentiment\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "# df = pd.DataFrame({'mood':['Positive', 'Negative'], 'val':[df[\"sentiment\"].value_counts()[0], df[\"sentiment\"].value_counts()[1]]})\n",
    "# ax = df.plot.bar(x='mood', y='val', rot=0, color=['green', 'red'], legend=False)\n",
    "# ax.xaxis.label.set_color('yellow')\n",
    "# ax.tick_params(axis='x', colors='yellow')\n",
    "# ax.yaxis.label.set_color('yellow')\n",
    "# ax.tick_params(axis='y', colors='yellow')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import LabelEncoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Encoded the target column\n",
    "lb=LabelEncoder()\n",
    "df['sentiment'] = lb.fit_transform(df['sentiment'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>overall</th>\n",
       "      <th>verified</th>\n",
       "      <th>reviewerID</th>\n",
       "      <th>reviewText</th>\n",
       "      <th>summary</th>\n",
       "      <th>review_clean</th>\n",
       "      <th>sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>A3CZ890UHC8HHZ</td>\n",
       "      <td>It diffuses a very mild light perfume, just wh...</td>\n",
       "      <td>What I expected</td>\n",
       "      <td>wear mild perfume nice strong lotion day perfu...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>A125TMC44CJBKK</td>\n",
       "      <td>All time favorite!!!! Wish they still carried ...</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>time still favorite carry wish</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>A3JDSBP01M7R2R</td>\n",
       "      <td>One of my wife's favorites.</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>one favorites wife's</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>A2HT5WENL8E5S6</td>\n",
       "      <td>If you have body acne this product is a must. ...</td>\n",
       "      <td>Acne Body Wash A MUST</td>\n",
       "      <td>heal help within prevent already first wash mu...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>A7LAYX9R7IMOO</td>\n",
       "      <td>I really is what I expected</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>expect really</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   overall  verified      reviewerID  \\\n",
       "0        5      True  A3CZ890UHC8HHZ   \n",
       "1        5     False  A125TMC44CJBKK   \n",
       "2        5      True  A3JDSBP01M7R2R   \n",
       "3        5      True  A2HT5WENL8E5S6   \n",
       "4        5     False   A7LAYX9R7IMOO   \n",
       "\n",
       "                                          reviewText                summary  \\\n",
       "0  It diffuses a very mild light perfume, just wh...        What I expected   \n",
       "1  All time favorite!!!! Wish they still carried ...             Five Stars   \n",
       "2                        One of my wife's favorites.             Five Stars   \n",
       "3  If you have body acne this product is a must. ...  Acne Body Wash A MUST   \n",
       "4                        I really is what I expected             Five Stars   \n",
       "\n",
       "                                        review_clean  sentiment  \n",
       "0  wear mild perfume nice strong lotion day perfu...          1  \n",
       "1                     time still favorite carry wish          1  \n",
       "2                               one favorites wife's          1  \n",
       "3  heal help within prevent already first wash mu...          1  \n",
       "4                                      expect really          1  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.preprocessing.text import Tokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nltk.tokenize import word_tokenize "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.review_clean = df.review_clean.astype(str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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>overall</th>\n",
       "      <th>verified</th>\n",
       "      <th>reviewerID</th>\n",
       "      <th>reviewText</th>\n",
       "      <th>summary</th>\n",
       "      <th>review_clean</th>\n",
       "      <th>sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>A3CZ890UHC8HHZ</td>\n",
       "      <td>It diffuses a very mild light perfume, just wh...</td>\n",
       "      <td>What I expected</td>\n",
       "      <td>wear mild perfume nice strong lotion day perfu...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>A125TMC44CJBKK</td>\n",
       "      <td>All time favorite!!!! Wish they still carried ...</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>time still favorite carry wish</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>A3JDSBP01M7R2R</td>\n",
       "      <td>One of my wife's favorites.</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>one favorites wife's</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5</td>\n",
       "      <td>True</td>\n",
       "      <td>A2HT5WENL8E5S6</td>\n",
       "      <td>If you have body acne this product is a must. ...</td>\n",
       "      <td>Acne Body Wash A MUST</td>\n",
       "      <td>heal help within prevent already first wash mu...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>False</td>\n",
       "      <td>A7LAYX9R7IMOO</td>\n",
       "      <td>I really is what I expected</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>expect really</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   overall  verified      reviewerID  \\\n",
       "0        5      True  A3CZ890UHC8HHZ   \n",
       "1        5     False  A125TMC44CJBKK   \n",
       "2        5      True  A3JDSBP01M7R2R   \n",
       "3        5      True  A2HT5WENL8E5S6   \n",
       "4        5     False   A7LAYX9R7IMOO   \n",
       "\n",
       "                                          reviewText                summary  \\\n",
       "0  It diffuses a very mild light perfume, just wh...        What I expected   \n",
       "1  All time favorite!!!! Wish they still carried ...             Five Stars   \n",
       "2                        One of my wife's favorites.             Five Stars   \n",
       "3  If you have body acne this product is a must. ...  Acne Body Wash A MUST   \n",
       "4                        I really is what I expected             Five Stars   \n",
       "\n",
       "                                        review_clean  sentiment  \n",
       "0  wear mild perfume nice strong lotion day perfu...          1  \n",
       "1                     time still favorite carry wish          1  \n",
       "2                               one favorites wife's          1  \n",
       "3  heal help within prevent already first wash mu...          1  \n",
       "4                                      expect really          1  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "overall          int64\n",
       "verified          bool\n",
       "reviewerID      object\n",
       "reviewText      object\n",
       "summary         object\n",
       "review_clean    object\n",
       "sentiment        int32\n",
       "dtype: object"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.review_clean = df.review_clean.astype(str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2 = df.sample(n = 30000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "corpus = df['review_clean'].tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "corpus2 = df2['review_clean'].tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['wear mild perfume nice strong lotion day perfumes shalimar want light want diffuse lotion purchasing hint',\n",
       " 'time still favorite carry wish',\n",
       " \"one favorites wife's\",\n",
       " 'heal help within prevent already first wash must body week use chest acne product difference notice back reoccurring',\n",
       " 'expect really',\n",
       " 'well great lather smell',\n",
       " 'give italy day year transform beautiful spent spa soon hard treat experience take cream find disappointed since fragrance feel must smooth like say get dear word shower live friend buy skin use one ago try completely love luxurious relax',\n",
       " \"seven product plastic fading lavendar heat amount quite scent natural quantity resealable potency sell delicate however dry decant oil smell stain come product think presentation open strike concentrate aspect generous important location overbear review enough opening seem intrinsic something careful note make act different packet mine potent sachet extract good long find spill fairly cloy really jar bag help messy manage page balance face package ding make less place relatively point queasy perfume good type guess exposed effect bit cool potpourri time sort bean blossom would lavender lavender-scented general still mask weak zipper easily fashion tend make put design i'm since contain without run revive suffocate risk tiny fill well use bit note strong loose apply give keep small review\",\n",
       " 'love good gift',\n",
       " 'product nice kind good photo gel scent doesnt light like help size attach shower star wake really people smell complement use bottle compariso morning receive small',\n",
       " 'colored hair great product wonder treat',\n",
       " 'time hair use great fine product oglivie perm',\n",
       " \"hair help favorite combine give sensation month use add reduce sponge use feel first life-span sponge scrub clean wash spongelle wet month time scrub/soap-up read life gentle shower body prolong skin one simply need rinse half do three product change get useful turn replace new spongelle's waste water start whole\",\n",
       " 'one try many buy',\n",
       " \"shampoo feel hair weigh enough dull i've disappoint somewhat natural way lather thin smooth find without fail dry straight really pleased someone usually challenge leave lather\",\n",
       " 'work good use',\n",
       " 'hair fuller use shinier great work scent natural stuff well look make stiff love mousse',\n",
       " 'product great',\n",
       " 'bought buying great butter body smell smell well also love product',\n",
       " 'product foam heavenly bath b&b particle amount feel scrub gel sponge scrub small real wash lather like help hungry luscious body make thick minor cranberry-scented also world smell lot price standard use meet seed',\n",
       " 'formula hair repair use together great way shampoo well felt love product work conditioner',\n",
       " 'prescribed doctor psoriasis make use head ointment shoulder stuff cream well someday force bad horde love go stop work medical',\n",
       " 'warns concern taste use think performance approve natural wash listerine dental browning average stain formulation come know might tooth think disinfectant next mouth high income bug suddenly wait questionable claim discretionary flavor change effect cause equally teeth anesthetic less brushing fun discoloration thing neither stain rate rinse mint-like remove toxic sense permanent sweet harmful contour use intefering palatable brown list come major want active antiseptic schedule cetylpridinium would spend wintergreen unlikely adequate particuarly artificial bottle visit potential preparation easy visit effective property side poison ingredient pass ingredient total fda risk regularly cause alternative well available label remove please chloride may moreover care rather',\n",
       " 'give except beautiful sweet power fruit fruity bit longer fragrance combination pomegranate-anise red long would little spicy stay spice perfect think brilliant absolutely hint last',\n",
       " 'come hope back scent',\n",
       " \"love can't find anywhere\",\n",
       " 'love run get aunt',\n",
       " 'effective foot high battery good use great foot already work diabetes tried battery run callus gentle travel one take convenient priority file start care operate',\n",
       " 'product age inflame crest pro taste use find breath work notice like dental dentist invigorate select product health think toward problem flavor want mouth many sensation flavor seem bitter instance sure gum unique different teeth clean troubled good provide star really rinse back list reason free rinse gear alcohol-free active version sort would issue bad fifty bottle fluoride explain special sentence ingredient mint try make plaque alcohol people base label three old wish give',\n",
       " 'shouldnt much suppose buy body cost gloss like',\n",
       " 'dime year delicious many pasha go container aroma hat quite amazon trouble break longer feel want way last like find time safari long luxury make thanks enjoy dollop small size even',\n",
       " 'nourish smell sweet bubble care skin',\n",
       " 'refund attach something please product seal would use picture like either',\n",
       " 'year stateside bring ive korea time happy find gentle use year available visit crap get every cleanser see tube',\n",
       " 'product good',\n",
       " 'get able store great find could love product wish',\n",
       " 'year peaceful nice spa several good experience delightful aroma eye clean light like block gentle eye make trip nose lavender use one also ago touch easy mask mental keep job added',\n",
       " 'luv',\n",
       " \"day year hard say cream find first i'm reacquaint best happy shower ago love part expensive discover\",\n",
       " 'well time available love theorizing would',\n",
       " \"fan i'm\",\n",
       " 'similar cheap last often store time long find sephora elsewhere',\n",
       " 'local feminine mixed scent great fragrance wash bergamot love refresh body brand husband',\n",
       " \"alaska hair spray strong good speed paul never i'll great use scent ive decide tried put mitchell nervousness smell love try medium shipping hold\",\n",
       " 'great love color pigmentation choice loose phenomenal eye fantastic set shadow',\n",
       " 'great product price shipping fast',\n",
       " 'property star smell after-use cleaning divine',\n",
       " 'cut brush remove else wood shed spa shape easier plastic side quickly curve nail much short use slightly find nails scrub clean overall light thin smooth like soft dry urban one-sided make stiff little finish really row dirt handy skin well wide get minimal anything one bristle grip garden easy',\n",
       " 'wear get nice perfume price manner like look delivered picture compliment always timely',\n",
       " 'shampoo useless hair spray believe fresh first dandruff wash money smooth like result without tried let ok aveda suffer one dry felt clear try waste receive',\n",
       " 'favorite miami',\n",
       " \"itchy hair product sister's star greasy complain right help skin\",\n",
       " 'salon variety product go much door use scent way ball recommend extra need fine product expensive rapidly name great meeting thick put little love lot go morning hint fekkai hair far par weigh give flatten wax/silicates day style first bit expand trust thin time work stiffen leave fast limp buy make volume store combination overpower condition mint condition without seem make flower always get afterwards small',\n",
       " 'one excellent product quality best really job',\n",
       " 'feel soft love product leave subtle scent fresh skin',\n",
       " 'love',\n",
       " \"attack get away good side take pet great i've think feel kill effect sponge scrub real easily like ever skeleton put little there's alive skin one determine think back cat irritate right\",\n",
       " 'love',\n",
       " 'hair clean head way cool like',\n",
       " 'love fragrance wonderful',\n",
       " 'ever excellent soap best',\n",
       " 'product gentle skin',\n",
       " 'know mixed like woman much enjoy thought enjoy like ever men result say thanks upset smell cost-(she relaxes product room young normal soap get mom\\'s point something simple want remark wife long hope package tools,blades,knives-ect)was me\"-after mother bathroom use amazon item time mom love eo would spend live thoughtful gift lavender worry still last arrive year wood walk make mother,daughter,sister invest saw say-\"i fragrance wonderful well year smell anything',\n",
       " \"feel give soap soft try lingers rich fresh tell i'd use i'll definitely pick store use different bar clothing scent way weight lather sachet ever drug leave line unusual simply love back anything surely every subtle freshness slowly\",\n",
       " 'burn away scalp throwing',\n",
       " 'favorite powder',\n",
       " 'use time happy hair manageable first',\n",
       " 'well nothing dislike work',\n",
       " 'product great',\n",
       " \"product mild price use quite somewhat scent exfoliants positive men prefer skin men,it noticeable think subtle freshness soap aggressive enough marketed cream great woodsy/green perfume bar contains therefore fragrance well although less gentle appeal little one personally face liked quality cleanser men's good use dissipates would moisturize still receptive reasonably find facial dove experience aroma fragrance body strong care\",\n",
       " 'shampoo product get seller body star notch wash top',\n",
       " 'arrive want early',\n",
       " \"shampoo can't hair local beautiful get grocery much strand together prior find work come wash detangler like beauty store highly seem beginning daughter's curly thick comb sally recommend prayer supply use break do answer conditioner charm\",\n",
       " 'great clean delicious lather smell really',\n",
       " 'wife love purchase skin product difference use see night',\n",
       " 'pricey worth product bit price best nest',\n",
       " 'product wonderful delivery happier quick',\n",
       " 'give product gumline toothbrush painful bold black much thoroughly think head way like value extra surface prefer cleaning/issues need half actually tooth might health think try sized cigarette impress get mouth clean vigorous due eventually flavor seem take big gap addition something great benefit feel gum different teeth amaze scrub feel clean sore sleek well long although provide brush gentle appearance put really anymore one like love part transformation lot miss mouthwash/half go fill tell need back make drink whole impressive regular separate help reach go bundle loud good give intend take debris use toothpaste gum energy otherwise effect first come super bit price want full sensitive almost alone separately version eating time switch gap get would mouthwash cigarette everything issue buy felt irritation also bottle area literally waterpik small attention work see even water plenty clean mainly year smoke tend diamondclean brace move fresh smoking smoother make experience head never plus massage spot whiten stick around sonicare oral overall drink awesome work toothbrush mint along mean try without wonderful end noise kit stay previously go floss well use invest say soothe difference every look care pretty give',\n",
       " 'wear year wear luxurios cream effect long fragrance alone body fact lasting subtle',\n",
       " 'effective seriously find scrub amazon best anywhere excite found get travel gentle hem see',\n",
       " \"piquant think move baby much longer scent way like smell turn bubble get spent i'd god something great please blush money golden le nose compliment love lot make hunt wrong scent gum first remind listen recreate sort tone also powder autumn hit today's freesia buy either i'm fragrance channel heavy always\",\n",
       " 'use love product regularly',\n",
       " 'kind buy christmas find use amazon aunt ross way love lotion smell could carefully year love still expensive slowly place store last',\n",
       " 'hair great love work iron stuff flat smell',\n",
       " \"local i've use longer burn smooth supply skin product think time last turn several razor cream lectric coat silicone steel surprise store decade keep sharp shave discover electric quit regular good use work amazon first want carry anyway even brand gas smoother oil blade worry run make well get look store\",\n",
       " 'track man stink stop',\n",
       " 'say nuff bare escentuals',\n",
       " 'hair nice body give thin',\n",
       " 'christmas love say rvv gel shower thanks gift give',\n",
       " \"product much wash cold hand use work nails way dry oil could simply detrimental night next hold soft get bedding favorite nail effect hand feel well seem urban slip warm glove morning rough spa good use first cream help time ensure condition garden moisturize peeling moisture moisturize see moisturizer water effective difficult smoother moisturizers fingernails oil chore soft instead weather sleep stay well they're remove bath keep household brittle\",\n",
       " 'great',\n",
       " 'worth concentrate great product thick investment',\n",
       " 'hair say great product favorite buy price',\n",
       " 'happy fine dramatic reduction cream line eye around see',\n",
       " \"wrong give perfume company good aura totally j'adore like size happy lotion purchase sell intense would actual star smell smell could bottle exactly think skeptical order original saving\",\n",
       " \"shampoo sign work problem dandruff wife's rid happy long get solution us\",\n",
       " 'hair shampoo swimmer since best unavailable got start complement one use always product',\n",
       " 'love product project best lavender perelier',\n",
       " 'need bottle big',\n",
       " 'smell great..love caddy',\n",
       " 'bath nice much milder gorgeous gel whereas bedtime bit scent light version decide floral purchase shower invigorate gift cologne part love note strong every tibetan husband',\n",
       " 'local product many online find allergic enjoy sell happy order really start one need product store',\n",
       " 'hair great love nicely product leave manageable rinse kid good smell easily',\n",
       " 'soft use woman leave love every skin',\n",
       " 'similar believe forward buy something great fragrance sell azur wife love look brand gift sorry still late hi review',\n",
       " 'cream excellent spot especially non-greasy elbow heel rough',\n",
       " 'sexy name online say find all-over scrub kill later scent ebay like romance go run order body hope oily shower slip warm product try perfect purposely amber scrub premixed',\n",
       " 'dry kind good',\n",
       " \"shampoo hair wait blonde i'll use warn newly long pureology blue highlight go one week use product new foiled aside user leave\",\n",
       " 'texture product crazy much week use rid skin bottle product think cost multiplies six night serum get fine careful make thirty feel amaze clean must better happy put retinol really one face quality clinique morning easy effect line price blue vestige even see moisturizer year oz fraction provide best absorb softer go use past always depend another low review',\n",
       " 'great product stylish work want',\n",
       " 'disappoints never gorgeous always bare color mineral must go everything buy',\n",
       " 'relax product great',\n",
       " 'much birthday amazon scent go got order relax bottle love bottle',\n",
       " 'give hair shape price go seem bleach chroma cheap price way work know well okay reflection would star combination..i pay use one also damage think price..now look keep thermique healthier',\n",
       " 'reasonably price advertise',\n",
       " 'hair great smell leave full volume',\n",
       " 'soft product nuts find long fragrance scent oh drive particular make really could skin love silky last wish',\n",
       " \"fall dime hand wear sell average need product job last soft get marketing pick problem great wal-mart must tool one innovation replace dozen rough cash exact forever reason good together promise towel see fraction big idea side rating put short i'm last mesh use save unless give\",\n",
       " \"itchy hair product sister's star greasy complain right help skin\",\n",
       " \"shampoo can't hair else soft get nothing leaf enough worth super investment manageable little buy well love expensive compare shiny\",\n",
       " 'love scent fresh',\n",
       " 'woman curve pass price love purchase gift could pay year daughter absolutely last',\n",
       " 'hair moisturize love silky product make really',\n",
       " 'love',\n",
       " 'thank',\n",
       " 'wonderful intoxicatingly',\n",
       " 'tend inch scalp quickly use find cuz size offer brand little scale well get anyway work hot pretty',\n",
       " 'shampoo soap spend perfume woman great use sicily bar yummy remind time men would fun intoxicate smell also love think every make',\n",
       " 'absolute love find favorite body hard wash',\n",
       " 'well also gel entire nice shower smell fantastic make bathroom lather',\n",
       " 'great describe smell',\n",
       " 'mom nice like',\n",
       " \"case nice hung i'll storage use scent way clothes grandma memory scent could half usually starch provence room last least get lightly pick something buy undertone clean iron happy although left refresh line little smells bring suggest one entire go back place spray stain let try leaf spray go good spritz fabulous use perfectly pillow pre outside truly first laundry hour bit light remember would sometimes bed lavender winter also hankies bottle brightens inside day de fresh experience watch i'm fragrance take musty wonderful smell sleep overwhelming spritz get give small\",\n",
       " 'wear day hat scent discontinue buy find stock price fragrance clean azur long find deliver heavy site one well like another product try lot promise crisp keep right night',\n",
       " 'nice body scent clean wash',\n",
       " \"favorite probably comany thyme i'm stock since fragrance time source find run thrill glad available production filigree new stop\",\n",
       " 'know summer hard scent find best wash ever lotion cologne one get discover',\n",
       " 'it...thanks produce company make much amazon sell sell wonderful someone product slow see thank',\n",
       " 'feel make feel totally clean pleasant like without look slip shower even pretty',\n",
       " 'moisturizer fabulous',\n",
       " 'else product use much psoriasis sure cream oganic review topical work bit want awesome ive like help organic improve im really go steroid use someone dermatologist bottle product see still may even receive injection whole',\n",
       " 'use never highly something product recommend amazing work fantastic keep use like',\n",
       " 'feel time amazon wash like store decide lazy really buy expensive advise',\n",
       " 'favorite miami',\n",
       " 'shampoo difference dandruff watery see much',\n",
       " 'great product stylish work want',\n",
       " \"shampoo hair leave good great i've amaze feel find really smells suitable product finally healthy\",\n",
       " 'love value gel shower good',\n",
       " 'hair soft large buy scalp use vegan quite bit come size love long dry make help daughter conditioner wish',\n",
       " \"can't local beat price store\",\n",
       " 'item describe',\n",
       " 'product great',\n",
       " 'finf love product hard',\n",
       " 'absolute love find favorite body hard wash',\n",
       " 'hair live protect use matrix time excellent fortify beach biolage specially go conditioner',\n",
       " 'finf love product hard',\n",
       " 'hair tingle great make clean scalp job',\n",
       " 'love scent favorite',\n",
       " 'shampoo ever try would best foolish use',\n",
       " \"concern much use i've shot burn like however dry stain tooth aftertaste chronic get mouth important bad sensation i'd something benefit non-alcohol-based act provide taste really rinse remained deal point far help sweet rinse worth lack mouthwashes time pleased issue potential tasting concern fresh side make plus i'm since mint without alcohol unpleasant overly may rather\",\n",
       " 'product excellent',\n",
       " 'worth love soap love make dollar smell husband',\n",
       " \"can't glow able good make use use find locally price lotion love\",\n",
       " 'favorite',\n",
       " 'great summery service thanks light smell',\n",
       " 'soft spray good mist fragrance femine like highly lotion would recommend body overpower smell exactly',\n",
       " 'favorite discontinue summertime disappointed find amazon fragrance thrill bedbathandbeyond product',\n",
       " \"shampoo many go best way try ever long wonderful little hesitate buy sulfate-free product and...i've conditioner\",\n",
       " \"spray breezy scent fresh enough secondly airy great tweens perfume first i'm come describe scent yummy oh light well without stong bargain overpower teen smell could well sugar cookie someone also citrus bottle love get strong perfect offer even bother justice small wish\",\n",
       " 'advertised',\n",
       " 'love good gift',\n",
       " 'hair beautiful product dark shampoos baby use much blonde detanglers lovely natural adorable old curl along help deep purchased son manageable naturally curly beginning dry weight start dry year use bouncy daughter conditioner research product three',\n",
       " 'wife love purchase skin product difference use see night',\n",
       " 'love',\n",
       " 'well great lather smell',\n",
       " 'love',\n",
       " 'soft use woman leave love every skin',\n",
       " 'get product high price much price want way love lotion however little handle think shipping',\n",
       " 'definitely great performs smell buy',\n",
       " 'wanted,because bring please grandson old store walmart year fine back',\n",
       " 'dime year delicious many pasha go container aroma hat quite amazon trouble break longer feel want way last like find time safari long luxury make thanks enjoy dollop small size even',\n",
       " 'enchant good orchid gel order carry shower smell love keep',\n",
       " 'get let addict nice popsugar rich good month instruction great first long since lather last like box hang ring dry smell give really one say love product new us even exfoliation receive',\n",
       " 'size perfect',\n",
       " \"spray nothing spray lightly instruction large fragrance iron like time long put overpower sheet there's suggest lavender use bottle linen water relax pretty last\",\n",
       " \"hair year residue make great amount full without stickiness take's smell use complaint work small hold\",\n",
       " 'product excellent',\n",
       " 'good go swears kort way wife long value little skin irritation prevention product us',\n",
       " 'hair ever product best',\n",
       " 'product nice kind good photo gel scent doesnt light like help size attach shower star wake really people smell complement use bottle compariso morning receive small',\n",
       " 'power also great love refresh travel scent',\n",
       " 'wrong soap large de many scents...that reason never pre bring bar head scent...your cracking want therefore size suitcase purchase purchase would brand risk travel go break say spin provence dozen apart',\n",
       " 'hair property good use amaze feel suprised cleanse all-organic product praise',\n",
       " 'one soap use bottle sight wealthy unless',\n",
       " 'also sulfate soap great nice free overwhelming smell',\n",
       " 'shampoo soap aftershave product favorite bath de nice wash lingers perfume enough leaf pre short use gel fragrance two sensitive lather surprise like smooth time lotion delicate shower arrives attractive french dispenser lavender scent skin square love silky provence may irritate pump clear overwhelm easily',\n",
       " 'wear day hat scent discontinue buy find stock price fragrance clean azur long find deliver heavy site one well like another product try lot promise crisp keep right night',\n",
       " 'price product great',\n",
       " 'great scent sulfate',\n",
       " 'shampoo hair bonus help nice pack good give choose chemically color longer type scent natural condition dye brand body really shine base brunette love keep deal',\n",
       " 'also cream love body buy smell cleanse',\n",
       " 'love run get aunt',\n",
       " 'hair nice give wild light mousse gray old control',\n",
       " \"product crest crowd much advantage taste quite pro-heath like market figure think user mouth thing marketing seem definitely buy well long nearly neither rinse one mother gimmick pass got would pro-health particular wintergreen buy fan lasting husband cheesey i'm yet along tried unpleasant older\",\n",
       " 'happy product',\n",
       " \"shampoo can't hair fall grow province probably psoriasis amazon longer without polytar available stop back thank\",\n",
       " 'sticky excellent great love product price',\n",
       " 'shampoo hair good option problem use feel work dandruff shoulder clean easily time try long would dry deal one also lot snowy look possible',\n",
       " \"verbena favorite de rich choose pre i've bar quite bar scent mine soap last lather last like time long refreshing order lavender linden provence rise include whole\",\n",
       " 'replacement term discontinue great find redken long really need product wish',\n",
       " \"soap unwrap favorite de rich treat treat select use pre i've long-lasting bar scent want lather wonderful thick remain bouquet provence rise\",\n",
       " 'get wish could',\n",
       " 'soap soft review bob kelp leave whim favorite w/kelp good chlorine moisturize use trout longer scent best light soap sea last swimming like tried counteract dry pleased spice skin winter salt citrus try husband',\n",
       " 'great clean delicious lather smell really',\n",
       " 'product foam heavenly bath b&b particle amount feel scrub gel sponge scrub small real wash lather like help hungry luscious body make thick minor cranberry-scented also world smell lot price standard use meet seed',\n",
       " 'love fragrance',\n",
       " 'know tight hair soft clean get product nice act exfoliant soon plastic irritated random definitely use scrub consider rid section certain smooth instead red physical especially blemish dry flaky neither skin microbeads need face walnut product piece still heal try leave shell care',\n",
       " 'shampoo know hair year restore fall look shampoos reduce buy scalp use psoriasis discontinue costly dryness helpful small real like come help find doublicate line dry use bottle three still six',\n",
       " 'look old opposite make',\n",
       " 'buy',\n",
       " 'luv',\n",
       " 'use love product regularly',\n",
       " 'lingers fresh lightly fragrance scent clean love lotion purchase purchasing hour gift skin daughter bergamont',\n",
       " 'love scent fresh',\n",
       " 'pamper hard make great find locally purchase would recommend extra gift bottle product',\n",
       " \"favorite sweet good drugstore fruity find summery shop price scent yummy cheap size mom love refresh get shower body glad hook smell that's gel handsoap despite small family brand last\",\n",
       " 'shampoo give pack ever sample extra best use',\n",
       " 'worth smell buy amaze',\n",
       " 'time great condition product exactly perfect describe price thank receive',\n",
       " 'disney cheap world fabulous',\n",
       " \"else relief hair effective look beautiful fall product nothing wash go due scalp save definitely truly great use head hurt i'm real match two chemical highly try organic look would cure dry gentle recommend smell buy one past save literally issue product conditioner add\",\n",
       " 'luv',\n",
       " \"discontinue great find leave product i'm glad what's wish\",\n",
       " 'love great son scent',\n",
       " 'great love work',\n",
       " 'product excellent',\n",
       " 'excellent quality shipping fast',\n",
       " 'hair make def awesome look treatment shinny recommend reinforce everybody healthy',\n",
       " \"local i've use longer burn smooth supply skin product think time last turn several razor cream lectric coat silicone steel surprise store decade keep sharp shave discover electric quit regular good use work amazon first want carry anyway even brand gas smoother oil blade worry run make well get look store\",\n",
       " 'soap get baths kinda become lame okay without get liquid',\n",
       " 'shampoo hair clean product scalp thoroughly lather addition great use stuff scent add improve gentle well use shine health lather',\n",
       " 'worth concentrate great product thick investment',\n",
       " 'love husband',\n",
       " 'love smell',\n",
       " 'shampoo email product nice wed company bag buy definitely something fit thought want join natural customer cruise purchase recommend service little tube recieved gift bill go explore also touch thank guest',\n",
       " \"know soft help able person make repurchase different saw old smooth balm sell time know look sell unfortunately condition recommend would hope lip chilitis stay buy anymore store's product lip stop every last\",\n",
       " 'product company together gel fragrance two layer butter especially shower body stay make lavender skin perlier love actually anything even',\n",
       " 'love',\n",
       " 'year stash hotel use bring find gel longer scent various glad friend ask french bath/shower ago available love u.s back discover finally fell last',\n",
       " 'purchase process antaeus bath gel product shower promise chanel receive quick easy',\n",
       " 'everyday great use work',\n",
       " 'get wash exception single allergic red body thing occasionally use one lot perfect rash complain',\n",
       " 'ear smile sister',\n",
       " 'smell nice',\n",
       " 'refund attach something please product seal would use picture like either',\n",
       " 'far able find best happy purchase look buy love product prell finally conditioner',\n",
       " 'awesome',\n",
       " 'love product songelle good',\n",
       " 'year make lotion fav go anymore wish',\n",
       " 'great',\n",
       " 'bubble clean bath delicious rinse moisturizing...i great use smell well shave even easily',\n",
       " 'mountain arrive fresh promotes great wash time antibacterial dial describe smell also love product',\n",
       " 'great gel bedroom make smell buy well need love thank',\n",
       " 'regular product hard strong due great find feels gel longer fragrance scent wash like body smell love last',\n",
       " 'sad shampoo gouge discontinue conditioner price awesome old know sell maker include people product',\n",
       " 'shampoo love disney world',\n",
       " 'hair help away coarse take first since like get dry gift use product dullness soften',\n",
       " \"ect second horrible sunbeam i've rem ever consistant drug clump product advertise job normal elbow brush bind norelco's high favorite razor poor grease seem great make lectric well little micro decade smell one love mark shave need quality success make popular remington good use worth braun screen time everything inexpensive still work clean paste shaver pipe best corner hair/stubble cause every clog store\",\n",
       " 'soft leaf bath residue germany either recommend find quite amazon longer dryness best store time soak irritation excellent tub carry would body exist lived buy skin relaxed also always usually love camomile product carry herbaflor thank leave liquid',\n",
       " 'birthday love receive others much',\n",
       " \"surprise soap year product mom's raisin experience use bravo truly realize array first unique bake wash minute work oatmeal like beauty time try nutmeg wide love shower particular body smell start skin use delighted one cookie actually industry every stop care\",\n",
       " 'use love product regularly',\n",
       " 'big wrap bubble confidance love nice good it...smells deliver box....buy fast....wrapped',\n",
       " 'effective high sport clean wash way classic old body smell spice endurance love get product pure',\n",
       " 'shampoo remove hair build-up feeling soft leave hard home thing good much feel clean though purchase would dry really use well also manage product work neutrogena water conditioner easy',\n",
       " 'shampoo hair basically soft texture wear volume scalp treat use wavy never slightly color break super iron straighten calm would recommend straight one use bottle write entire try product medium look make conditioner argon review',\n",
       " 'ever love smell body best wash',\n",
       " 'buy great',\n",
       " \"soap get oz difficult bathroom wonderfully problem spring exaggerate i'm bar way decide happy wonderful splurge smell buy go irish strong smelling bedroom...so back open spoil\",\n",
       " 'awesome',\n",
       " 'avon product favorite',\n",
       " \"get good showel something great i've get gel amazon bring item price best full like go time musk ever lotion get body smell smell pay compliment egyptian smell back keep store\",\n",
       " \"give rosewood texture cetyl sorbate concentration away rub cinnamal wash daily leaf honey go list acetate lactobacillus particle maybe i've use quite peel work fruit complex find grit/solids/beads scent acid ci wash less spf red linalool result stearate soybean chemical level dig oil smell geranium could skin need reaction usually might actually product guess usual think spread one expensive last normally aloe turned chlorophyllin-copper lactic can't get plant clearing strong sensation enough clay beta-carotene seem change definitely cause sure careful something cleanly test enzyme color breakout papaya barbadensis minute note scrub feel phenoxyethanol pumpkin routine extract bergamot subscription long provide pogostemon box although acetic peel huge antibacterial faster digestive see thing pineapple really papain maculatum one drying personally face green lot new need cablin limonene place right low palmitate citrusy kind easy go good cetearyl exist guess glyceryl probably worth grit couple gum tocopheryl first super bit want sensitive extract learn help time especially get longtime peeling damage lavender also sd sun tingle irritation exfoliator potassium sort juice mask thinner exfoliators work even see exfoliate water leave patch physical day basically stash wood acne-prone grabbed property facial make ingredient avoid i'm oil polyethylene around since pretty combination forget basis two move like vitamin beauty orange floral though ferment oily pectin dehydroxanthan cinnamomum introduction sorbic alcohol bromelain salicylic use chlorophyll/copper past cinnamon get every may cassia retinyl portion\",\n",
       " \"discontinue great find leave product i'm glad what's wish\",\n",
       " 'shadow box love favorite come eye even instruction',\n",
       " 'also cream great take fine puffiness line tighten eye conceals makeup calibra skin',\n",
       " \"cut hair shampoo phyto laugh work much blonde great i've fry switch rehab stuff kerastase go red tried chemical charge davines that's help year stylist tell product literally see entire back conditioner theorie\",\n",
       " 'use since wash run ever trip face love product',\n",
       " 'good either nails price way cheap like daily look hope picture buy hold',\n",
       " 'wanted,because bring please grandson old store walmart year fine back',\n",
       " 'look second side class consider amazing bit like purchase science messy use year daughter still new',\n",
       " 'time hair use great fine product oglivie perm',\n",
       " 'year get lauder use probably never use estee time without foundation glide brand top could skin compliment another product',\n",
       " 'time excellent product good thank receive',\n",
       " 'gift perfect',\n",
       " 'favorite',\n",
       " 'love',\n",
       " 'wear year wear luxurios cream effect long fragrance alone body fact lasting subtle',\n",
       " 'shampoo great',\n",
       " 'favorite powder',\n",
       " 'one ever best use ive',\n",
       " \"feel much bar longer enjoy chemical smooth thanks average skin might grow last soap high i'd something slightly clean must less additionally rate paraben sticky toxic leaf animal type immensely price light linger test masculine irritating still contain earthy initially fragrance mill yet floral excellent seem harsh please overwhelm film non\",\n",
       " 'value great',\n",
       " 'work good use',\n",
       " 'apart cheap come couple mine scrub really thin start',\n",
       " 'time arrive order exactly describe',\n",
       " 'hair keep great shelf work walgreens thin thats wish',\n",
       " \"shampoo forever kind go good make scalp crazy seem cranky...it's symptom randomly like hat retaliates find start product eczema conditioner last\",\n",
       " 'product company together gel fragrance two layer butter especially shower body stay make lavender skin perlier love actually anything even',\n",
       " 'get wish could',\n",
       " 'year love product use',\n",
       " 'shampoo shine love body frizz',\n",
       " 'also sulfate soap great nice free overwhelming smell',\n",
       " 'hair fuller use shinier great work scent natural stuff well look make stiff love mousse',\n",
       " 'well worth pre soap expense de bar provence fantastic intoxicating soap smell large',\n",
       " 'give italy day year transform beautiful spent spa soon hard treat experience take cream find disappointed since fragrance feel must smooth like say get dear word shower live friend buy skin use one ago try completely love luxurious relax',\n",
       " 'brush year soap apply aftershave get shaver razor tea use much never pre-shave cold comfortable use lectric double willingly way rinse lather prep without burn wonderful edge close get part green product shave shave preparation water add electric',\n",
       " 'product bath it...especially pricey worth day gel brown bit long especially huge stressful shower molton travel fan revive cempaka',\n",
       " 'well great lather smell',\n",
       " 'love',\n",
       " 'excellent',\n",
       " 'hair great smell leave full volume',\n",
       " 'shampoo burt organc vacation oz pack enough good preservative great since bee downside chemical two wash like weekend regulation testify would body reaction pesticide travel test fact week one tsa animal conditioner bottle ounce make small enviroment',\n",
       " 'product great',\n",
       " 'year get lauder use probably never use estee time without foundation glide brand top could skin compliment another product',\n",
       " \"can't get perfume hard person wonder find scent dang paris department store hilton make anymore rare compliment always love product stand\",\n",
       " 'soft leaf bath residue germany either recommend find quite amazon longer dryness best store time soak irritation excellent tub carry would body exist lived buy skin relaxed also always usually love camomile product carry herbaflor thank leave liquid',\n",
       " 'bottom hair cap riche mixed professionnel gold incredible home promotes chrome fusio-dose pink made-to-measure minute feel program best light stuff kerastase orange protection ever treatment condition match line systeme dry fantastic make everything buy also shine booster love green weak healthy shiny',\n",
       " 'one soap use bottle sight wealthy unless',\n",
       " \"favorite sweet good drugstore fruity find summery shop price scent yummy cheap size mom love refresh get shower body glad hook smell that's gel handsoap despite small family brand last\",\n",
       " 'ever excellent soap best',\n",
       " 'nice',\n",
       " \"near problem happy amazon reason i'm fox fix nowhere sell\",\n",
       " 'thank',\n",
       " \"de treat pre i've scent soap try refresh linden love provence\",\n",
       " 'similar believe forward buy something great fragrance sell azur wife love look brand gift sorry still late hi review',\n",
       " 'norelco know give slight smoothly treat move much i\\'ll washed lotion maybe use thought odorless star wash extremely wet like paw\\'s dry linger hour smell start skin market hear product think time noooo least impressed the&nbsp;<a razor thing something lectric third hand razor exceptionally good lotion seem eat star really fact face love shave shave electric assessment unpleasent reason glide disappear kind home good work comfortable metallic first bit want time get would actual check buy ago ha close suppose sensotouch figure even brand data-hook=\"product-link-linked plenty today\\'s class=\"a-link-normal recently year perspiration drag purpose buy cat plus shaver never definite resent blue</a>&nbsp;and since blade williams try without though end force well awful shave odor absolutely evidently',\n",
       " 'clean clear deep time refresh sinus also love peppermint--it',\n",
       " 'fine',\n",
       " 'time awesome use step gift use dad stuff',\n",
       " 'nice expect',\n",
       " 'product excellent',\n",
       " 'year stash hotel use bring find gel longer scent various glad friend ask french bath/shower ago available love u.s back discover finally fell last',\n",
       " 'use result silky product receive smooth',\n",
       " 'good prodauct',\n",
       " 'shampoo appreciate airbnb price waist guests little product perfect provide right',\n",
       " 'much great feel stuff brand make buy skin love cost wish',\n",
       " 'arrived much great sooner product--has visible really skin also expect love seller difference product make',\n",
       " \"product wash black much i've amount quite pattern long dollar scent wash smooth nutmeg shower dry opposite chemical-smelling skin realm usually product try subtle cucumber job subdue normally impressed soap clean soft several many residue climate open certainly company innate addition great gel bar dryness ensues clean routine fragrance exceptionally ability well preference brand little really rinse fact one entire need make right point similar exceptional eucalyptus leave compare men's rather awards price lather magazine decide pepper outstanding would spend various fantastic overpower standard truth felt also bottle silky area even brand leave receive remove year recently towel humidity experience case need aroma fragrance over-foam sage last try seem body secondary overwhelming use well past citrus see give small pretty\",\n",
       " \"can't without love shampoo...i favorite product...i\",\n",
       " 'dandruff skin worse irritate make rest',\n",
       " 'great product much..it smell like',\n",
       " 'good stuff',\n",
       " 'well hair difference notice lather',\n",
       " \"shampoo hair day get away wash strong buy twice take used reason build i've use find outrage star buildup bit price clean way root two thin without long oily shinny dry chemical fantastic start really start one week research breaking lot end every made\",\n",
       " 'provide refreshing freshens amaze good moisturization job smell skin',\n",
       " \"can't get perfume hard person wonder find scent dang paris department store hilton make anymore rare compliment always love product stand\",\n",
       " 'nourish smell sweet bubble care skin',\n",
       " 'need excellent though edible powder quality little competitive price fresh amla',\n",
       " 'know plantlife give gentles read lovely week bar breath natural way notice smooth like herbal intense delicate private dry true dry prefer rope,&nbsp;<a skin depend rope traditional might provence six lot finger soap soft ahhh many artisanal cardamom,&nbsp;<a bar,<a take definitely something great bar hand want soap enrich potent fragrance deep go size long well find breathe sandalwood cloy word nose french absinthe one love pure back right plantlife--just gram shea let unobtrusive foot yes sweet soon able scent consider large cardamom day--that pre together collection ago,&nbsp;<a light want lather butter particularly moisture also aromatherapy trade arm data-hook=\"product-link-linked easily class=\"a-link-normal oz de sandalwood</a disrespect buy choose rating gram</a hop moisturizers review fragrance combination come interestingly unscented wonderful grasp people happen heavy well use leg say notice may look small give',\n",
       " 'finger feel quickly oil amazing greasy light last absorb oily cuticle apricot moisture smell one essie great leave',\n",
       " 'hair product folks talk seem use saw like growth le grey frizzy really product slow noticed',\n",
       " 'arrived child crest kid thing make colgate grit toothpaste use make different standup teeth best scream pump like kick good taste spit generally go well also right tube counter five stand',\n",
       " 'shouldnt much suppose buy body cost gloss like',\n",
       " 'reasonable happy available say find love price enough',\n",
       " 'well great lather smell',\n",
       " \"scent i've amaze sensitive dislike take get smell skin love lot care\",\n",
       " 'arrive tried shampoo help box still itchy open leak',\n",
       " 'day hair soft get leaf bleach break use continue thin comb shower use actually love product world change every healthy',\n",
       " 'work great',\n",
       " 'magic mixed gotta work cum black water really chicken voodoo',\n",
       " \"shampoo hair buy big i've best stop try ever order loss difference see\",\n",
       " 'product good quick use feel regularly always satisfy completely delivery look deal',\n",
       " \"hair perform product many dull together i've make use somewhat best without tried smooth find weight well test c-straight animal product perfect mop control frizz conditioner\",\n",
       " 'hair several favorite ton side make type heat matter use slightly amount color somewhat since month beauty add anyone...no oily line would recommend dry launch sally moisture supply heavy buy use one ago shine fine product right back',\n",
       " 'thank',\n",
       " \"favorite probably comany thyme i'm stock since fragrance time source find run thrill glad available production filigree new stop\",\n",
       " \"stain soap beautiful oz high bathroom make i've washcloth zum bar bedroom potently bar color best price milk ever find unfortunately gentle luxury put wonderful make white smell item scent gift skin smell quality formula goat's\",\n",
       " 'fall wash crest use bacteria complex natural antibiotic smooth however recommend quaternary prefer generally practice could health think astringent normal problem mouth absence category strong treat flavor everyday gum teeth refresh safe rinse show balance regular cation harmful good gingivitis use alcohol-free beneficial regard disturb effect pleasant active prevent antiseptic time minty pro-health particular mouthwash would also fluoride ammonium agent species short helpful oral basis mint ingredient microbiology cetylpyridinium body infection plaque people interfere cause use research period microorganisms chloride proliferate healthy small',\n",
       " 'shampoo age product shampoo contribute use way immediately order come product young madera whenever kerastase less found thick loss love lot back right hair ship believe truly first awhile early time switch would shampoot introduced year old tried wonderful people use well told u.s every may',\n",
       " 'shampoo hair texture far use best thick dry someone shine crave back give conditioner',\n",
       " 'bath nice much milder gorgeous gel whereas bedtime bit scent light version decide floral purchase shower invigorate gift cologne part love note strong every tibetan husband',\n",
       " 'time love come',\n",
       " 'pressurize move fill power much force use quite head tongue shot way like dental model dentist floss need think six time many stream tip gum different teeth clean handle little one model brace turn whatever predecessor increase wearer spray speed debris function couple beneficial comfortable crowns,etc performs gum irrigate come reservoir buy produce etc also pressure still area waterpik maintain cleaner water attach user easy effective remove year current tip never around since evolve vary eliminate device specific plaque disease people alternative floss pay washer individual past always tell decay pretty control',\n",
       " \"shampoo can't without live\",\n",
       " \"wrong beautiful silver photo gorgeous amazon bit flake queen order tiny goodness d/c'd...thank bad exactly aubergine swatch plum receive fast\",\n",
       " 'buyer use google',\n",
       " 'soft product nuts find long fragrance scent oh drive particular make really could skin love silky last wish',\n",
       " 'hand shampoo/conditioner best',\n",
       " \"wood nice northern fir's great oil scent comb pair well beard even make\",\n",
       " 'ensoleillante light like long lotion eau moisturize body clarins year love try absolutely make wish',\n",
       " 'concentrate fragrant moisturizes body wash beautifully lather',\n",
       " 'local feminine mixed scent great fragrance wash bergamot love refresh body brand husband',\n",
       " \"least fifteen year soap nice de make pre buying lovely i've item scent mine milk light love box would really gift dented come always loose provence quality shipping edge\",\n",
       " \"discontinue great find leave product i'm glad what's wish\",\n",
       " 'hair look love product make feel way',\n",
       " 'new number',\n",
       " 'hair nice give wild light mousse gray old control',\n",
       " \"pro i've nails notice treatment service loyal traditional last get struggle-some pastel favorite thing professionally sure level different range star drama satin application essie tell original possible point reapplied far chip fabulous shiny ready bit reapplying almost butter changer would bottle release experience milky forget two without call stay go apply every may rout small overflow give dreamy stellar hard gala negative much base prevent think dishwasher leveled sea like dry level could week usually might confuse can't perform nothing wear within deserve careful jelly color coat color long finish one love new wonder compare contour nude day self amazon time call top wider game also still anyway suppose weak tend choice moment since classic yet layer try though relate impossible girl patient there's essies use goal always see notice give horribly second nice remotely magic week use work longer customer buy--only reapply couture however recommend hour need others manicure ridiculously enough i'd opaque change nail slightly gel third equally soooo initial rarely go less find fairly intermittently brand consistency little polish discover aside spite improvement regular sense able favorably good take worn anytime platinum get thicker bold fix satisfied brand fast ballet chip overtime i'm often work mistake come shade look screw throughout well say difference look perceive setter redos chance find quite smooth admittedly depend astound rebranded perfect shade less-than-stellar wear impressed brush strong hue take feel iron line discoloration grow that's change issue need technique currently drip-free reason follow exception yup resist super price want skill test sort calm bed silk coat confirm even revlon day wore swear never say review rebranding instead frustrate tiny past get keep beautifully polish pretty\",\n",
       " \"bubble bath plug amaze load scent wash water,it time suds long too.this wonderful body put little make smell relaxes skin warm,running bonus,if wash's softens bathtub add last\",\n",
       " 'year leave kid buy great awesome box leave ago still smell deal',\n",
       " 'pricey love wish',\n",
       " 'feel soap soft leave nice rinse large big quite bar greasy nicely like good without sandalwood really smell skin well smelly laters fragrant',\n",
       " 'treatment best keratin',\n",
       " 'goutal lotion annick lovely favorite line hadrien de eau',\n",
       " 'great product delivery fast',\n",
       " \"hair recently soft dark lovely great feel wash happy purchase manageable daughter's curly make smell use product\",\n",
       " 'feel sign proof cheap age shred reviewer stuck much use amount quite find work glue glide reminded notice like dental dentist outperform recommend flossing toy start product get anyone waxy happen wire change definitely tired something goretex different teeth suppose unlike little minor really one sew package back unstuck original show try free good gingivitis use probably breakable early hate get would restock dog standard sample etc ago felt shred still finally even start convince easily bottom year smoother brace current bracket around maneuver hole try run life wire regularly floss use well told remove strong noticed every store stand',\n",
       " 'favorite marked price scent cannot store order sell would retail anymore bulk personal care justify',\n",
       " 'shampoo day hair serum get nice lingers air coarse use minute frizz wash iron time result eliminate brother naturally recommend dry flat smell magnificent heat compliment usually leave',\n",
       " 'love product',\n",
       " 'product great',\n",
       " 'purchase great product fragrance',\n",
       " 'much birthday amazon scent go got order relax bottle love bottle',\n",
       " 'price expect',\n",
       " 'nice away cause air hand use smooth like shower dry comfortably true foil roughness perfect hold finger conveniently enough great caddy hand hang urban little hook love rough right foot spa perfectly scratch pumice fit top damage sometimes work easy bottom stone stone difficult inner without make allows patch use loop ribbon',\n",
       " 'shampoo compliment frequently get beautiful hair great shiny',\n",
       " 'hubby want',\n",
       " 'day shea feel soap leave nice residue type much feel different since fragrance mill clean scent enrich yet reminds version like butter without long grandmother present overpower standard french knew bad skin fact well always squeeky actually soothe save grow lasting liked room subscribe guest add',\n",
       " \"wow soap year oz de many due hubby large pre use very...very i'm oil bar scent enjoy light soap certain size,smell olive purchase lavender fact product provence size\",\n",
       " 'moisturizes greasy rich feel creamy',\n",
       " \"shampoo can't hair get product produce purpose woman say lovely unique fragrance best wife men line market general shine product characterizes\",\n",
       " 'arrive condition product promise good',\n",
       " \"itchy hair product sister's star greasy complain right help skin\",\n",
       " 'time hair use great fine product oglivie perm',\n",
       " 'shampoo excellent year find product favorite several difficult become',\n",
       " 'awesome',\n",
       " 'return detectable scent jasmine item barely',\n",
       " 'feel hair soft product favorite great clean ive try always lot leave care',\n",
       " 'gel favorite shower',\n",
       " 'shampoo fruity great scent wash lather yet like body masculine really thumb',\n",
       " 'time satisfied product new brand item come',\n",
       " 'work usual seem',\n",
       " 'week time amount use head first dandruff small two nicely clear less son messy get eczema trace',\n",
       " \"favorite sweet good drugstore fruity find summery shop price scent yummy cheap size mom love refresh get shower body glad hook smell that's gel handsoap despite small family brand last\",\n",
       " 'shampoo hair nothing several go reviewer banana something wonder use review feel trust like though rub greesy fishy read shower im scent agree noticed water',\n",
       " 'know hard speedy win xmas gel kudos live price renew brother get shower thanks request buy town another perfect jp escada delivery last give',\n",
       " 'hair silver keep regular yellow reorder look silvery basis rather',\n",
       " \"shampoo hair look full feel work i'm soft happy well conditioner\",\n",
       " \"beautiful summer nice sweetness sweet good thing fresh i'd choice scent light foody call slight nearly really that's one lemony anticipate subtle hint rather\",\n",
       " 'favorite forever',\n",
       " 'satisfaction great product price shipping fast',\n",
       " 'shampoo hair oz ml gel adidas active ok shower body thank start',\n",
       " 'good',\n",
       " 'nice',\n",
       " 'good either nails price way cheap like daily look hope picture buy hold',\n",
       " \"date expect i'm item receive pleased\",\n",
       " 'fall pick soap quote thing woman pricey buck mean best wash want highly ever recommend shower bueller course one smelling ferris',\n",
       " 'love',\n",
       " 'shampoo hair far nothing nice good excessive cause use dryness feel condition separately time however star give well condition shine part save product look provide healthy conditioner fancy',\n",
       " 'great soft want silky leaf moisturize body hour wash must buy skin',\n",
       " 'hair silver keep regular yellow reorder look silvery basis rather',\n",
       " 'hair else coconut consider versatile flavor something great use taste fry cook like fry especially would however salads dish skin also bake product note meat/chicken/veggies one mainly',\n",
       " 'treatment best keratin',\n",
       " 'fave decade',\n",
       " 'love smell',\n",
       " 'day use every good product----i',\n",
       " 'good',\n",
       " 'heal help within prevent already first wash must body week use chest acne product difference notice back reoccurring',\n",
       " \"can't year wear find color bit like store sparkle little anymore lot\",\n",
       " 'wife love purchase skin product difference use see night',\n",
       " 'shouldnt much suppose buy body cost gloss like',\n",
       " 'product thing scalp work fragrance pleasant comparison heavy love product flaky',\n",
       " 'arrive condition product promise good',\n",
       " 'fill much pop effort use refill smooth like neatly dentist recommend extra stay holster floss come half sooo traditional useful open mouth flip ton pick definitely water/half sure feel afterward different teeth basin feel clean pain cord leave dial little messy discomfort one love turn make place right high coiled string regular set worth wether full honestly compartmnt help time investment mouthwash also lid waterpik even water easy save kid brace buy head never best total run behind alternative floss use well anything keep store',\n",
       " 'price product good',\n",
       " 'good thank',\n",
       " 'time try ever make love first thymes hope since continue scent absolutely',\n",
       " \"hair help favorite combine give sensation month use add reduce sponge use feel first life-span sponge scrub clean wash spongelle wet month time scrub/soap-up read life gentle shower body prolong skin one simply need rinse half do three product change get useful turn replace new spongelle's waste water start whole\",\n",
       " 'star',\n",
       " 'need excellent though edible powder quality little competitive price fresh amla',\n",
       " 'give sexy able moon disappoint fragrance sparkle happy buy love try',\n",
       " 'day favorite nice color brown would recommend friend sparkle week',\n",
       " \"plastic fabric i'll hand like recommend hear actually might product usual action conditioner wear tight big something cream later minute hand drawback deep size glove prepping put glove face application deal right hair regular kind type take worked otherwise assume sort would line moisturize bed mask non-absorbent rest willing glob push process around often rub soaked struggle soak sleep snug outrageously well wrist always keep\",\n",
       " \"shampoo hair soft nice much use feel i'm feel like soft dry make always medication\",\n",
       " 'expect really',\n",
       " 'good',\n",
       " 'get let addict nice popsugar rich good month instruction great first long since lather last like box hang ring dry smell give really one say love product new us even exfoliation receive',\n",
       " 'appreciate receiver gift',\n",
       " 'great product service fast',\n",
       " 'hair soft large buy scalp use vegan quite bit come size love long dry make help daughter conditioner wish',\n",
       " 'hair leave reason good probably great make want size manageable curly complaint little smell buy need stop unknown lather shiny',\n",
       " 'hair soft large buy scalp use vegan quite bit come size love long dry make help daughter conditioner wish',\n",
       " 'product amazing',\n",
       " 'price smell great awesome',\n",
       " 'thank',\n",
       " 'shampoo hair vacation kid wash lake year love perfect every conditioner',\n",
       " 'product great',\n",
       " 'thank',\n",
       " 'year love use',\n",
       " 'love',\n",
       " 'soap soft high leave scent pricey worth incredibly long wide spectrum skin last every quality penny fragrant',\n",
       " 'arrive perfect condition',\n",
       " 'quick much anywhere europe smooth pity well thers market available shave',\n",
       " 'shampoo hair know product earthy together lose maybe feel work feel less thicker faster smell well unpleasant grow even',\n",
       " 'shampoo hair bleach blonde best set smooth make fine love healthy conditioner',\n",
       " 'absolute love find favorite body hard wash',\n",
       " 'holiday find love gel shower fragrance london',\n",
       " 'great product delivery fast',\n",
       " \"weird month week i've amount work natural result spend eyelid could start skin half product think can't itchy get curler allow within lenght dna mine money accidentally line star lahses light improvement eyelash good direct lash use price month get give irritation also still even see hit rest make one's short i'm overall two though eye plateau use three difference notice look justify\",\n",
       " \"shampoo nice hard scalp much use like understand bottle...we thru conditioner turn finger conditioner.....hope conditioner.....i possibly feel happy less put conditioner..........wow grow push make hair large company couple again...the update cool month help strengthen thicker felt also bottle still work young menopause buy make silk.....my never i'm stick mint stiff rinse\",\n",
       " 'price product good',\n",
       " 'bee product help delicious hard good find feel savannah shower dry really smell skin',\n",
       " 'love absolutely',\n",
       " 'year bath online never discontinue think love body glad look',\n",
       " \"shampoo arrive kept know let send favorite nice enough supplier care take definitely wrap brief come know stuff anywhere almost tightly ever purchase peace mind apology top broke there's buy bottle damage replace shipping\",\n",
       " 'organic great gel good natural smell',\n",
       " 'avon product favorite',\n",
       " 'know discontinue still old whey best scent spice',\n",
       " 'product heavenly wash never break afraid time order fail caress body smell of.....the receive',\n",
       " 'day let beautiful plus feels spring amaze describe scent wash shower smooth like body smell skin botanical silky see thereafter garden',\n",
       " 'soft beautiful use lotion basis smooth daily ever one compliment love',\n",
       " 'well nothing dislike work',\n",
       " 'favorite fresh moisturize collection feel thyme scent clean like azur shower skin use always product',\n",
       " 'purchase great product fragrance',\n",
       " 'product bath it...especially pricey worth day gel brown bit long especially huge stressful shower molton travel fan revive cempaka',\n",
       " 'effective seriously find scrub amazon best anywhere excite found get travel gentle hem see',\n",
       " 'use since wash run ever trip face love product',\n",
       " 'discontinue would scent go postal',\n",
       " 'good',\n",
       " 'product awesome',\n",
       " 'love shampoo',\n",
       " 'seal bottle tamper seams water extremely like runny',\n",
       " 'soft spray good mist fragrance femine like highly lotion would recommend body overpower smell exactly',\n",
       " 'similar cheap last often store time long find sephora elsewhere',\n",
       " 'look second side class consider amazing bit like purchase science messy use year daughter still new',\n",
       " \"give feeling holder nice power fill leak clearly i'll additional use find tongue performance i've shot long order i.e wide appliance glad true level handy stale could compact inch agree ouch product try equation state others glass counter unit can't useless auto-retractable get attachment several anyone pour i'd sure simple hesitant teeth clean unit trade-off size less range read find gentle dial keep discomfort fact touch silent scraper electric point compare low beyond believe manual bathroom good give set type couple fit big bit neat price want sale outstanding bet would actual reservoir half-way everything anyway waterpik small figure friday clean even water leave clean easy hose basically wow sound adjustment shaver never review run discount toothbrush range without keep device wonderful noise there's ask toss past benefit give keep pretty\",\n",
       " 'know mild expiration use fruit pump like come protection order seal maximum product think squeeze bubble get cavity colgate seem flavor definitely slightly gel feel less date leave fun safe little really messy love operate pump tube easy fruity toothpaste bit safety time minty blue still leave plenty bottom kid broken seal though use year',\n",
       " 'get nice clothe small scent overall last time long drawer put mesh bag messy well b/c work',\n",
       " 'leaf negative calmness scent enjoy wash bath come provence soap thing great bar hand feel wash soap tub little french face part gram add shea help scent single pre enhance want butter luxury also see even de make delightful ingredient aroma sizes fragrance mill mint wonderful body soak gifts bit say difference',\n",
       " 'favorite fresh moisturize collection feel thyme scent clean like azur shower skin use always product',\n",
       " 'great providing product product wonderful thank biolage',\n",
       " 'fall pick soap quote thing woman pricey buck mean best wash want highly ever recommend shower bueller course one smelling ferris',\n",
       " 'hair cap nice enough comfy shoulder provide keep print shower dry inch also past fine lot room lot pretty cute',\n",
       " 'feel soap nice good definitely milky great scent enjoy like sandalwood really smell skin foam',\n",
       " 'lessen nice hard wash use scent like smell cologne skin fine product job hold soap get strong expect great slightly feel color hand clean soap really smell grain leaf mention perfume good use would white water darker purify easy dove round describe lessens body shape use strong',\n",
       " 'fall beautiful mild sweet nice good fresh umm fruity use scrub packaging yummy daily time suds creamy smell also overly perfect',\n",
       " 'year hard use find gel amazon fragrance match way happy find shower star love five',\n",
       " \"love can't get unfortunately scent anymore\",\n",
       " \"hair year residue make great amount full without stickiness take's smell use complaint work small hold\",\n",
       " 'value great',\n",
       " 'hair product use gel naturelle try find curly glad reaction begin well always product work mousse',\n",
       " 'great use feel eye neck perspire sting skin face product shaving arm',\n",
       " 'ever excellent soap best',\n",
       " 'amnazon soft product leave lauder great use money estee smooth wel;l buy skin product save work place',\n",
       " 'intrigue soap good anyone home thing online clean store comment find order smell town usually use',\n",
       " 'sock foot length hard well...also enough make comfy find best wash cotton thin boot thick stay ankle anymore try hot period',\n",
       " 'much definitely buy smell need love lasted lather easily',\n",
       " 'gel--love great shower scent',\n",
       " \"look hold nice rub plastic retinoid explore wet topic youthful appear delicate however shower dry skin week brush want surgeon great careful simple obviously caffeine brand affect that's one new light whole doctor truly seek let's light want well-made flaw improve talk creme work even exfoliate moisturizer form purpose reduce idea cellulite hang massive overall episode though follow reveals well phenomenal biology keep\",\n",
       " 'arrived much great sooner product--has visible really skin also expect love seller difference product make',\n",
       " \"nice good buy bee stuff happy smell daughter always quality burt's\",\n",
       " 'problem beneficial use amazon complete transaction dental hopefully product health advertise easy',\n",
       " 'one excellent product quality best really job',\n",
       " \"perfume-y bath favorite family's use great find amazon longer feel clean wash excite without purchase shower wonderful body smell relax carry work store relax\",\n",
       " 'time try ever make love first thymes hope since continue scent absolutely',\n",
       " 'product great',\n",
       " 'moisturizes greasy rich feel creamy',\n",
       " 'say great wash time bodywash would recommend dry body smell skin us husband',\n",
       " 'christmas love look arrived son day package present invite variety efficient',\n",
       " 'relic hair man finer provide northern craftsmanship appreciate great use life comb what-so-ever thanks thing fir issue appreciates',\n",
       " 'quit henkel product enough buy bionutrients worth something great make find work amazon longer store try adult unfortunately say find nourishes hope really skin well response love product lot old work',\n",
       " 'get able month type seem never everyday use time scrub rid advice sensitive certain therefore cause life oily really friend smell start skin one bad irritation face try pretty',\n",
       " 'look similar perfectly style school although long fit lab coat give pretty',\n",
       " 'work great',\n",
       " 'soap soft high leave scent pricey worth incredibly long wide spectrum skin last every quality penny fragrant',\n",
       " 'excellent great coverage...love enough large',\n",
       " 'ribbon difficult disappear olay use find wash store dry body hope really skin well get',\n",
       " 'hair great love work iron stuff flat smell',\n",
       " 'fine',\n",
       " 'shampoo use love stuff',\n",
       " 'thank',\n",
       " 'favorite way relax',\n",
       " 'find love hard',\n",
       " \"shampoo dab hbl look product finger use amount work wash value pea dry masque...i smell stylist shine fine product cost size can't residual lightly change recommend clean washing masque daily remain naturally hydrate apply that's occasionally need make hair fly crown leaf live quickly first high volumizing residual aways towel ago end area rich fresh tip never there....basically seal without oily use year apply begin healthy small brittle shiny\",\n",
       " 'well hair please beautiful love clean condition smell shiny',\n",
       " 'shampoo hair use love enjoy wonder conditioner',\n",
       " 'shampoo hair clean product scalp thoroughly lather addition great use stuff scent add improve gentle well use shine health lather',\n",
       " 'discontinue bad',\n",
       " \"can't speak swirl small angled plus they'd use price set clean glitter come red size highly brush handle get line recommend mostly one use well perfect personal beginner hold professional\",\n",
       " 'favorite scent migraine make nicer either somewhat bit fragrance fake fresher thing smell lavender one sickening strong even brand soothes',\n",
       " \"cut hair insures cape durable fit husband's neck inexpensive velcro need proper\",\n",
       " 'time love lemon favorite fragrance smell',\n",
       " 'love scent fresh',\n",
       " 'love philosophy great say awesome niece smell set buy',\n",
       " 'one pricey organic though great nonetheless love product product little',\n",
       " \"cut hair insures cape durable fit husband's neck inexpensive velcro need proper\",\n",
       " \"know crest weird consequently protect taste use breath none wash red recommend dentist hour chemical pesticide prefer propylparaben mimic tooth product action can't mouth folk sensation seem nail definitely something color contain find cutex put really non-pro-health that's fight whatever enamel good give estrogen use first accustomed active time minty pro-health tingly mouthwash remover also tastebuds husband liquid fresh side ingredient last old cetylpyridinium fungus there's alcohol use unpleasant after-taste chloride insist give clear concoction floride polish pretty\",\n",
       " \"shampoo hair wait blonde i'll use warn newly long pureology blue highlight go one week use product new foiled aside user leave\",\n",
       " 'feel product bath leave nice never cream moisturize great use want condition line wonderful body make smell buy go skin also perlier another product absolutely',\n",
       " \"shampoo hair year leave go buy contain coarse phosphate definitely something great feel first i'm want worry smooth like beauty sally's purchase supply that's ago need love silky even conditioner\",\n",
       " 'effective seriously find scrub amazon best anywhere excite found get travel gentle hem see',\n",
       " \"know wear soap wood man enough best-smelling woman make big sure i've i'm bar hand around scent want way although ever though sandalwood smell wrap say manly even\",\n",
       " 'arrive want early',\n",
       " 'problem arrive naturaly sound definitely great vendor wife would safe buy always liked product prell conditioner',\n",
       " \"bubble bath plug amaze load scent wash water,it time suds long too.this wonderful body put little make smell relaxes skin warm,running bonus,if wash's softens bathtub add last\",\n",
       " 'love great scent smell boyfriend',\n",
       " 'love love would expensive',\n",
       " 'local nice mild reason find fragrance cannot wash old like shelf body retail really spice unknown foxcrest',\n",
       " \"shampoo hair day get away wash strong buy twice take used reason build i've use find outrage star buildup bit price clean way root two thin without long oily shinny dry chemical fantastic start really start one week research breaking lot end every made\",\n",
       " 'shampoo know hair nevertheless reduce use dandruff doesnt high help dry less get rating',\n",
       " 'love',\n",
       " 'hair make good really smell',\n",
       " 'far favorite great come since fragrance size pre-packed gift ribbon perfect',\n",
       " 'favorite scent migraine make nicer either somewhat bit fragrance fake fresher thing smell lavender one sickening strong even brand soothes',\n",
       " 'time love come',\n",
       " 'recently sense salad amazing bit yummy food wash eat highly refresh recommend body wake course citrus every morning whole',\n",
       " 'give smoothly help bag go undereye darkness eye therefore ive cause time long huge however genetic he ago also fan irritation product accept husband use keep control',\n",
       " 'hair soft best stuff iron highly ever less recommend make frizzy flat use silky',\n",
       " 'hair leave favorite fragrance brand combine love silky bvlgari conditioner',\n",
       " 'nice',\n",
       " 'dandruff skin worse irritate make rest',\n",
       " 'product amazing',\n",
       " 'itchy soap get generous rose nice easy side much big amount think onto come bar hand enjoy fragrance best way overall feel wet like find shower moisturize smell make dry really smell skin spin lot rise job time',\n",
       " 'good thank',\n",
       " 'thank',\n",
       " 'good need price cheap would recommend friend smell suitable close satisfied product family',\n",
       " 'time come love always excellent soap third order husband',\n",
       " 'know hair great look smell work good really yet',\n",
       " 'work usual seem',\n",
       " 'year love product use',\n",
       " 'diaper year try overnight protect believer moisturizers much lotion magic become quite different bit vaselineand burn like good highly chapped moist crack petroleum dry recommend minor top cynicbut skin well etc product aquaphor lips heal rash ointment work jelly-type make dissolve',\n",
       " 'soap leave nice overt big say scent softer wife love skin fan',\n",
       " 'love love would expensive',\n",
       " 'never great get always puppy pup clean smell',\n",
       " 'texture scent hop creamy lavender',\n",
       " 'know nice invite much use must-have way like curl shower smart smell curtain product provence spread closet everyone high many definitely granddaughter absolute great slightly clean iron blossom line thing touch quality make sprayed hair adore spray exception bathroom home scent fabulous washing pre light lavender also linens even water de catch fresh never delightful bring freshen fragrance try dampen drawer throughout wonderful girl rug use anything use absolutely pretty',\n",
       " 'day perfect essence pleasant',\n",
       " 'assortment nice please',\n",
       " 'good thank',\n",
       " \"remove can't problem get mouth product second live crest go many pro afford able warns suggest cavity clean use first teeth anywhere want excited pain like okay decide read dentist thanks people web that's could cause rinse staining near go use label multiprotection health product search\",\n",
       " 'local nice mild reason find fragrance cannot wash old like shelf body retail really spice unknown foxcrest',\n",
       " 'product awesome',\n",
       " 'also cream love body buy smell cleanse',\n",
       " \"get good showel something great i've get gel amazon bring item price best full like go time musk ever lotion get body smell smell pay compliment egyptian smell back keep store\",\n",
       " \"normally drop get bath difficult able mitt large fabric amount color work i'm wash lather soapy like soft washcloth accidentally would extra boucle use well work pouf easy\",\n",
       " 'far bath favorite anyone quite gel fragrance last time would recommend product',\n",
       " 'fall philosophy great yummy yet lip smell goody also love another gloss',\n",
       " 'ever love smell body best wash',\n",
       " 'expensive lovely',\n",
       " 'shampoo fruity great scent wash lather yet like body masculine really thumb',\n",
       " 'product amazing',\n",
       " 'actually love early come',\n",
       " 'soft incredible cream hand massage coordinate smell skin love try make',\n",
       " 'shampoo itchy hair hinder except nope scalp banana either psoriasis thickness head pour money linger like help smell smell neither help artificial pure difference water make absolutely lathers waste banana liquid',\n",
       " 'love use wash locs husband',\n",
       " \"store can't find product best make pantene anymore\",\n",
       " \"shampoo nice hard scalp much use like understand bottle...we thru conditioner turn finger conditioner.....hope conditioner.....i possibly feel happy less put conditioner..........wow grow push make hair large company couple again...the update cool month help strengthen thicker felt also bottle still work young menopause buy make silk.....my never i'm stick mint stiff rinse\",\n",
       " 'hair great leave clean conditioner',\n",
       " 'soft beautiful use lotion basis smooth daily ever one compliment love',\n",
       " 'hair soft clean love product leaf quickly',\n",
       " \"day summer favorite difficult discontinue find i'm scent clean pin like unfortunately would end smell warm one come use winter love world back wish whole\",\n",
       " \"surprise soap year product mom's raisin experience use bravo truly realize array first unique bake wash minute work oatmeal like beauty time try nutmeg wide love shower particular body smell start skin use delighted one cookie actually industry every stop care\",\n",
       " \"shampoo can't without live\",\n",
       " 'feel product bath leave nice never cream moisturize great use want condition line wonderful body make smell buy go skin also perlier another product absolutely',\n",
       " 'hair great love work iron stuff flat smell',\n",
       " 'use love product regularly',\n",
       " 'haur great work look like',\n",
       " 'shampoo hair flare irritated psoriasis wash happy less replace lot swap every cleaner',\n",
       " 'give soft puffiness noticing go use change recommend cream week use eye firm line around greasy want wrinkle cream highly long diminish eye make little way calibra start skin use week also promise fine lot new try look reduces young',\n",
       " 'shampoo well hair clean good lather smell',\n",
       " 'favorite miami',\n",
       " 'clean leave nice sent hard enough great find wash like seem body place cologne near well dry skin bottle lot retailer expensive lather lucky',\n",
       " 'luv',\n",
       " 'well hair difference notice lather',\n",
       " 'like interesting nice body drops scent enough candy smell whole',\n",
       " 'shampoo product get seller body star notch wash top',\n",
       " 'day hair soft get leaf bleach break use continue thin comb shower use actually love product world change every healthy',\n",
       " 'christmas time outstanding pack deliver ask well always like product girl',\n",
       " 'give product nice hygienist painful protect person use find like listerine dentist however ideal could minority loathe might product low-threshold try think action mouth claim plan seem great act feel clean routine consult tingle really course part love go morning enamel endorse bonus necessarily use excite return would bed also bottle suppose miss provide plan harm heavily genuinely excitingly deviate always strong may family replace',\n",
       " 'luxurious gel fragrant good lather',\n",
       " 'salon online anti-frizz great find work sell anymore well shine product look deal',\n",
       " 'work great',\n",
       " 'soft elastic go pricey definitely worth feel way long make little skin love amazingly',\n",
       " 'available love still conditioner wish',\n",
       " \"body wife's wash go\",\n",
       " 'complains product great',\n",
       " 'arrive condition product promise good',\n",
       " 'amount generous product packaging face buy future exelent use',\n",
       " 'buy great',\n",
       " 'favorite way relax',\n",
       " 'great product much..it smell like',\n",
       " 'shampoo love thank quickly come',\n",
       " 'nice expect',\n",
       " \"one favorites wife's\",\n",
       " 'shampoo great',\n",
       " 'day use love perfume recommend last it.it',\n",
       " 'love scent favorite',\n",
       " 'great scent sulfate',\n",
       " 'good recommend use already dandruff amazing pleasant solve result get smell issue well time',\n",
       " 'beautiful go definitely great amazon burn order get carying thanks smell product tan',\n",
       " \"eucalyptus fresh great i've lather scent wash way sage natural like tried pepper partial body dry overpower smell skin well citrus product one irritate\",\n",
       " 'product smell great',\n",
       " 'hair help away coarse take first since like get dry gift use product dullness soften',\n",
       " 'wife love great product',\n",
       " 'clip look summer nice type everyday pattern different hesitant accessory mine clip order glad little wait',\n",
       " 'work smell great',\n",
       " 'soap forward plastic association benefit sample enjoy milk soap old cover memory especially moisturize star b/w smell brings use open look bath film',\n",
       " 'recommend',\n",
       " \"can't use hard find i've gel amaze become wash last lather month ever though shower body smell faint anymore bad one bottle love strong\",\n",
       " 'get wash exception single allergic red body thing occasionally use one lot perfect rash complain',\n",
       " 'shampoo love use picky peppermint husband',\n",
       " 'spray perfume give put add clean light lotion little smell bring compliment love lot',\n",
       " 'love',\n",
       " 'definitely purchase great seller would describe price shipping fast',\n",
       " 'feel minimize week line eye around greasy fragrance feel firm notice like help look looker make could start skin fine face product look family comment make leave young improvement',\n",
       " 'fashion get bath wait use till find condition seasonal body glad item year love product seller perfect autumn timely receive',\n",
       " 'one big relaxed feel head shampoos chemical good inventor really bottle.thanks pretty',\n",
       " \"piquant think move baby much longer scent way like smell turn bubble get spent i'd god something great please blush money golden le nose compliment love lot make hunt wrong scent gum first remind listen recreate sort tone also powder autumn hit today's freesia buy either i'm fragrance channel heavy always\",\n",
       " 'discontinue would scent go postal',\n",
       " 'excellent great coverage...love enough large',\n",
       " 'give away much quite tongue bacteria aquatic dental dentist lexus shoot placque regulate cleanse near world health job turn mouth enough flow pick projectile sure gum eye teeth considerably enter godsend although read enter little place love new pump whole nozzle compare flossers let low yugo reach good instruction reduce first ready come want friend gift phase partnership also inexpensive pressure decent small cleaner water start receive jet make reach never best oral anywhere excellent floss thorough',\n",
       " \"hair product length good lock scalp bother moisturize great i'm feel fragrance clean two sensitive like fragrance color though keep weigh mid use fine look even keep thinning\",\n",
       " 'shampoo hair year follow bad appear make hoping day bay star dandruff super hardly wash excited organic result struggle unfortunately keep dry make really past selenium anti-dandruff difference search atleast look see keep review',\n",
       " 'suds moisturize leave favorite nice body subtle scent wash skin',\n",
       " \"shampoo hair soft leave certainly volume baby plus never i'll recommend i've wave already clean natural add growth excellent gentle friend uhuru loss finish really pleased buy bottle fine silky stop even develop\",\n",
       " 'shampoo summer leave kind consider fresh single malt woman grapefruit problem bar scotch macallan vendor cool like perhaps would really smell well love package smell expensive morning meet',\n",
       " 'thank',\n",
       " 'deep well musk day feel scent clean lather smell incredible...almost',\n",
       " \"day year hard say cream find first i'm reacquaint best happy shower ago love part expensive discover\",\n",
       " 'thin/water good use much week maybe gel lather nicely thin like would return expect smell bottle go liquid',\n",
       " 'hair short superb long get curly straight daughter say product work',\n",
       " 'hair beautiful product dark shampoos baby use much blonde detanglers lovely natural adorable old curl along help deep purchased son manageable naturally curly beginning dry weight start dry year use bouncy daughter conditioner research product three',\n",
       " 'fragrance nice fruity',\n",
       " 'fave decade',\n",
       " 'shampoo hair vacation kid wash lake year love perfect every conditioner',\n",
       " 'question get wear buy use constantly scent light pleasant lotion shave',\n",
       " 'hair lovely love curly make',\n",
       " 'lovin',\n",
       " 'love',\n",
       " \"else relief hair effective look beautiful fall product nothing wash go due scalp save definitely truly great use head hurt i'm real match two chemical highly try organic look would cure dry gentle recommend smell buy one past save literally issue product conditioner add\",\n",
       " 'love',\n",
       " 'year container store great dent come order box use love product package right',\n",
       " \"plastic fabric i'll hand like recommend hear actually might product usual action conditioner wear tight big something cream later minute hand drawback deep size glove prepping put glove face application deal right hair regular kind type take worked otherwise assume sort would line moisturize bed mask non-absorbent rest willing glob push process around often rub soaked struggle soak sleep snug outrageously well wrist always keep\",\n",
       " 'look delicious prank something great wash blood like highly taste recommend body friend little smell also love pull product funny',\n",
       " 'give hair shape price go seem bleach chroma cheap price way work know well okay reflection would star combination..i pay use one also damage think price..now look keep thermique healthier',\n",
       " 'recently sense salad amazing bit yummy food wash eat highly refresh recommend body wake course citrus every morning whole',\n",
       " 'well hair difference notice lather',\n",
       " \"kasia ginko product nice hard air enamor woman read cosmetic much use fruit scent become wash chemical sea like ever delicate recommend shower dry smell generally scent cleanse skin parabens simply chamomile carrot world search cucumber aloe soap soft aggressive biloba nothing get important thing enough tea feel totally feel know clean pumpkin lately extract happy lotion find pour refresh eat brand safe little thing consider course dead cancer face love green paraben bombard trace beat morning slowly possible spirulina try low free bathroom good animal strange preservative online use large anytime first light cream almost cool organic test especially everything level also bottle silky still even count water book fancy gelee forward today&iexcl;s never ingredient appear i'm since best becomes last orange look collapse body lumps make always example uncomfortable cosmetic old drank keep stacked toxicity\",\n",
       " 'wear day warms adore spray office projection much lovely hand use fragrance nicely body heavy softness bomb perfect powder night',\n",
       " \"shampoo sad nice shampoo stay stranger maybe use i've wash like recommend hour smell skin week four product room time impact conservatively great longevity mine clean really smell one compliment entire increase hair similar perfume use hotel alone month ritz-carlton bottle loofa still even day mexico make gotten two run body use three\",\n",
       " 'use time happy hair manageable first',\n",
       " 'wrong thing scent read everyday use first description wash intend musk order without lotion order body smell prefer white item carefully one actually funny',\n",
       " 'product excellent',\n",
       " 'scrubber duty good take foot smooth callus heavy touch file care finish',\n",
       " 'buy great',\n",
       " 'hair nice body give thin',\n",
       " 'buy',\n",
       " 'shampoo tell love friend',\n",
       " \"shampoo sign work problem dandruff wife's rid happy long get solution us\",\n",
       " \"hair year product calofornia good scalp i'm hawaii whether brand live use love quality\",\n",
       " 'man go never saturday night fragrance scent rid awesome axe get really woods warm blend leather perfect friday wish',\n",
       " 'option several nice crest fresh flavor much change different clean non-mint mint smooth like refresh pro-health flavor line invigorate wintergreen need actually usual',\n",
       " \"day summer favorite difficult discontinue find i'm scent clean pin like unfortunately would end smell warm one come use winter love world back wish whole\",\n",
       " 'sense give great use already doesnt awesome doubt organic run chemical make smell skin produce foam love product lot morning irritate',\n",
       " 'far loud agin best light like smell buy one love think mousse',\n",
       " 'fairy mix bubble britney bath perfume spray scent spritz collection amaze goddess oh like tub spear soak god one come love smell vanilla',\n",
       " 'it...thanks produce company make much amazon sell sell wonderful someone product slow see thank',\n",
       " 'problem arrive naturaly sound definitely great vendor wife would safe buy always liked product prell conditioner',\n",
       " 'expensive lovely',\n",
       " \"countertop wall hard read prevent use onto star average customer regard like come wide adult mess spray dentist market could leakage try suggested one job area apart unit mouth close mirror many orthodontia thing i'd take tip teeth begin huge shut put nearly effectiveness rate one manage love model turn ruin original make possible whole coordinate spray easy bathroom wallpaper direct used child first amazon come update month time hate fantastic peeling worried usage dump still waterpik even water functional cleaning easily fast remove year brace buy make eye-opening rating review okay without unbelievable son device make use well research available note family last small wish\",\n",
       " 'shampoo hair nothing several go reviewer banana something wonder use review feel trust like though rub greesy fishy read shower im scent agree noticed water',\n",
       " 'time try ever make love first thymes hope since continue scent absolutely',\n",
       " 'price smell great awesome',\n",
       " 'use felt improve water scent like skin',\n",
       " 'shampoo useless hair spray believe fresh first dandruff wash money smooth like result without tried let ok aveda suffer one dry felt clear try waste receive',\n",
       " 'fragrance nice',\n",
       " 'hair anyone fabulous use terribly recommend dry course product miss',\n",
       " 'similar believe forward buy something great fragrance sell azur wife love look brand gift sorry still late hi review',\n",
       " 'effective high sport clean wash way classic old body smell spice endurance love get product pure',\n",
       " \"relief feel get mom lanolin yes gold good month go baby use much greasy teethe way old like breast highly long however recommend little bed need bottle actually throw new rash we'd look morning\",\n",
       " 'shampoo hair bubble get hop reason buy good much unable maybe scrub dandruff bit feel clean rid like help better would however thick little smell begin winter decent afterwards matter',\n",
       " 'fine',\n",
       " 'product gentle skin',\n",
       " 'best cream ever',\n",
       " \"shampoo can't hair local beautiful get grocery much strand together prior find work come wash detangler like beauty store highly seem beginning daughter's curly thick comb sally recommend prayer supply use break do answer conditioner charm\",\n",
       " 'foot curve smoother great han use much great foot spot performs type price palm well reasonable without thus glide contour shower fit comfortably surface cover spit cheese remarkable well miss product tool eggs job rough area remove',\n",
       " 'smell nice',\n",
       " 'always great product masculine tobs scent pleasant',\n",
       " 'favorite forever',\n",
       " 'sense give great use already doesnt awesome doubt organic run chemical make smell skin produce foam love product lot morning irritate',\n",
       " 'shampoo give feel hair soft product fresh type definitely feels clean root would recommend star near general might right think even five',\n",
       " 'popular mouth mild crest free pro gingivitis germ quick much taste great use breath freshen mouthwash describe wash cool burn like good kill common recommend brand plaque alcohol wintergreen harsh cause bad another health brushing taste leave',\n",
       " 'price product great',\n",
       " 'great packaging good squeeze smell easy',\n",
       " 'feel time amazon wash like store decide lazy really buy expensive advise',\n",
       " \"help hair great fine i've year three difference make really used\",\n",
       " 'shampoo hair far nothing nice good excessive cause use dryness feel condition separately time however star give well condition shine part save product look provide healthy conditioner fancy',\n",
       " 'expensive lovely',\n",
       " 'cut hair look grows get soft recently leave nice courser shape horrible volume use much manage recommend miserable great use different hugely conditioner wash matrix ran like good left travel thick split terrific little smell hairdresser could fine conditioner usually product end back look lose absolutely layer',\n",
       " 'product awesome',\n",
       " 'shampoo also hair love leave leave nice clean.lt smell',\n",
       " 'dab beautiful seem protective unique come scent like time long refil extra little pouch handy last',\n",
       " 'hermes product price best',\n",
       " \"least mouth far peroxide thing claim go take i'll re-moisten use breath feel rinse way like word dry make bad one rinse drink usually afterwards water job hydrogen\",\n",
       " 'shampoo day hair serum get nice lingers air coarse use minute frizz wash iron time result eliminate brother naturally recommend dry flat smell magnificent heat compliment usually leave',\n",
       " 'love scent',\n",
       " \"hair i'd ordinary amazon color clip find handle design thick glad claw comfortably buy love strong new bear wait grip easily\",\n",
       " 'mom nice like',\n",
       " 'work smell great',\n",
       " 'shampoo organic top outstanding',\n",
       " 'hair fuller use shinier great work scent natural stuff well look make stiff love mousse',\n",
       " 'soap soft review bob kelp leave whim favorite w/kelp good chlorine moisturize use trout longer scent best light soap sea last swimming like tried counteract dry pleased spice skin winter salt citrus try husband',\n",
       " 'shampoo hair year recent likely buy go gel sell best happy well purchse conditioner',\n",
       " 'bottom hair cap riche mixed professionnel gold incredible home promotes chrome fusio-dose pink made-to-measure minute feel program best light stuff kerastase orange protection ever treatment condition match line systeme dry fantastic make everything buy also shine booster love green weak healthy shiny',\n",
       " \"hair look large shame rich lovely bit color well-made look daughter's expect dimension think look\",\n",
       " 'arrive clean nice good much paint great think first hand lather smoooooth linger like nicely refresh dude shower really smell primer macho might product even fast',\n",
       " 'house use clothing scent like warmer sock recommend dry smell could know half electronic handful last high drawer great making later amazing underwear potent long put little smell bag love entire quality original place warm type work probably use item bit light want candle almost potpourri time highly purchase would blossom top buy lavender etc ago also still brilliant inside plan wax arrive incredible bag idea drawer smell make well ceramic cheesecloth',\n",
       " 'year hard use find gel amazon fragrance match way happy find shower star love five',\n",
       " 'product thing scalp work fragrance pleasant comparison heavy love product flaky',\n",
       " \"can't without love shampoo...i favorite product...i\",\n",
       " 'well pleased',\n",
       " 'price fragrance go love find always product timely receive matter',\n",
       " \"day year hard say cream find first i'm reacquaint best happy shower ago love part expensive discover\",\n",
       " 'p.s product lanolin good cream avoid want way awesome stretch nearly go love product mark',\n",
       " 'shampoo hair use choice helpful best time wife long loss worry product',\n",
       " \"hair year residue make great amount full without stickiness take's smell use complaint work small hold\",\n",
       " 'dandruff skin worse irritate make rest',\n",
       " \"give nice over-priced good i'd steal large either gel big sponge hand bite feel natural sea nicely palm hold size five less shower star little make hold really buy skin exactly still expensive easy\",\n",
       " \"stain soap beautiful oz high bathroom make i've washcloth zum bar bedroom potently bar color best price milk ever find unfortunately gentle luxury put wonderful make white smell item scent gift skin smell quality ship formula goat's\",\n",
       " 'good',\n",
       " \"relief least concentrated apply get mouth day treatment sensitivity take bring work piece teeth longer bit minute desire sensitive meant help time mention seem level direction one past actually tooth they're effectively portion might period leave small strip whole\",\n",
       " 'never great get always puppy pup clean smell',\n",
       " 'know give get good feel try order would skin tell love product back',\n",
       " 'shampoo hair get many good use never use reviews clean cannot lather ive understand ever worst mostly could buy felt water',\n",
       " 'love drop',\n",
       " 'many great feel amaze body item repurchase smell love product absolutely time',\n",
       " 'wrong thing scent read everyday use first description wash intend musk order without lotion order body smell prefer white item carefully one actually funny',\n",
       " 'aroma awesome',\n",
       " 'great business amazon price easy',\n",
       " 'wow disappoint thought gel totally awesome seal would however shower torn rate accord bottle try package hype receive review',\n",
       " 'nicer much work best perm wonderful could buy ogilvie foam still product wish',\n",
       " 'difficult hard anytime cream find fragrance signature obtain lotion would body moire buy site come available easily',\n",
       " 'fall pick soap quote thing woman pricey buck mean best wash want highly ever recommend shower bueller course one smelling ferris',\n",
       " 'product great',\n",
       " 'love great scent smell boyfriend',\n",
       " 'lovin',\n",
       " 'love shampoo',\n",
       " 'moisturizer fabulous',\n",
       " 'totally natural smell bad',\n",
       " 'longlasting scent pleasant extremely pleased',\n",
       " \"perfume-y bath favorite family's use great find amazon longer feel clean wash excite without purchase shower wonderful body smell relax carry work store relax\",\n",
       " 'hair shampoo swimmer since best unavailable got start complement one use always product',\n",
       " 'good need price cheap would recommend friend smell suitable close satisfied product family',\n",
       " \"shampoo can't store local find unfortunately love try really like\",\n",
       " 'get wish could',\n",
       " 'look delicious prank something great wash blood like highly taste recommend body friend little smell also love pull product funny',\n",
       " 'know mixed like woman much enjoy thought enjoy like ever men result say thanks upset smell cost-(she relaxes product room young normal soap get mom\\'s point something simple want remark wife long hope package tools,blades,knives-ect)was me\"-after mother bathroom use amazon item time mom love eo would spend live thoughtful gift lavender worry still last arrive year wood walk make mother,daughter,sister invest saw say-\"i fragrance wonderful well year smell anything',\n",
       " \"shampoo dab hbl look product finger use amount work wash value pea dry masque...i smell stylist shine fine product cost size can't residual lightly change recommend clean washing masque daily remain naturally hydrate apply that's occasionally need make hair fly crown leaf live quickly first high volumizing residual aways towel ago end area rich fresh tip never there....basically seal without oily use year apply begin healthy small brittle shiny\",\n",
       " \"know cross sum baby apparently hand use think sigh small acid wash extremely washcloth result forehead cloth delicate recommend level fiercely could skin week properly half take guys product aggravate time counter night get important several we've upper thing enough happen problem sure breakouts make scrub eye clean begin path teen read demonstration smack really one include face part shave go change cleanser make right improvement wrong reason good take probably appointment day considerable suspect watch time highly get backstory particularly everything cancel irritation somehow finally see even young easy else result-delay idea make complexion spot initially since work without mention seem boy direction assume salicylic use dermatologist get show another see look clear\",\n",
       " 'time great condition product exactly perfect describe price thank receive',\n",
       " 'wife love great product',\n",
       " 'product nice',\n",
       " 'shampoo get able enough hand disappoint head consistency water hold',\n",
       " 'ever love smell body best wash',\n",
       " 'shampoo hair basically soft texture wear volume scalp treat use wavy never slightly color break super iron straighten calm would recommend straight one use bottle write entire try product medium look make conditioner argon review',\n",
       " 'know fracture manage cold work adult dentist turn tooth suggest jalapeno alarm strip least get want several sensitivity reason cream popsicle fun thing really one love lot new make oops nope use karma help chomp would sometimes also even see sensi-stop year wore ice chip forward brace hop repair microwave around come though bite people previously use look',\n",
       " 'replacement term discontinue great find redken long really need product wish',\n",
       " 'love use wash locs husband',\n",
       " 'favorite miami',\n",
       " 'love',\n",
       " 'wanted,because bring please grandson old store walmart year fine back',\n",
       " \"hair similar go bad make experience instruction stick short amount head i'm two scam time tremendous result beware consistency pace return week one need bottle foam product try buyer matter suggest shaved see electrocuted water junk trash\",\n",
       " \"hair product length good lock scalp bother moisturize great i'm feel fragrance clean two sensitive like fragrance color though keep weigh mid use fine look even keep thinning\",\n",
       " \"day high something stuff find without hawk difficulty fantastic etc faux son's actually product paraben quality expensive wish hold\",\n",
       " 'always expect use',\n",
       " \"can't local beat price store\",\n",
       " 'shampoo love still make axe wish',\n",
       " 'hair itch hard move good scalp crazy change feel around doesnt clean lather like smell still',\n",
       " 'exactly want thank',\n",
       " 'wife always cream get hard us',\n",
       " 'recommend',\n",
       " 'bueno muy',\n",
       " 'great look everywhere gift',\n",
       " 'great love product moisturize scent wash',\n",
       " 'expensive lovely',\n",
       " 'prescribed doctor psoriasis make use head ointment shoulder stuff cream well someday force bad horde love go stop work medical',\n",
       " 'also cream love body buy smell cleanse',\n",
       " \"shampoo arrive hair forgotten good i'd online use great find i've locally since grandson time ever order well use product visit package\",\n",
       " 'hair keep great shelf work walgreens thin thats wish',\n",
       " \"soap shin muscle fat large much use probably hand bar scent lather wash sensitive notice scratchy find puff cloth shower would skip dried there's bag skin well near irritation nylon leg love botanical low lather rough uncomfortably\",\n",
       " 'feel minimize week line eye around greasy fragrance feel firm notice like help look looker make could start skin fine face product look family comment make leave young improvement',\n",
       " \"problem i've year scalp take flaky care\",\n",
       " 'problem expectation gel musky full sensitive like go without shower overpower thanks cologne skin gel ship fast',\n",
       " 'wonderful intoxicatingly',\n",
       " 'product great',\n",
       " 'estimate arrive hesitation get excelent important product good much definitely big buying something surprise way save money like quantity time interested size especially recomend expect everything bottle product exactly seller promised delivery deal',\n",
       " 'add feel without love oily dry stuff glow moisture skin',\n",
       " 'effective high sport clean wash way classic old body smell spice endurance love get product pure',\n",
       " 'year stash hotel use bring find gel longer scent various glad friend ask french bath/shower ago available love u.s back discover finally fell last',\n",
       " 'feel soap get soft leaf de fresh pre fragrance clean soap love highly everyone recommend skin provence absolutely matter',\n",
       " 'leaf negative calmness scent enjoy wash bath come provence soap thing great bar hand feel wash soap tub little french face part gram add shea help scent single pre enhance want butter luxury also see even de make delightful ingredient aroma sizes fragrance mill mint wonderful body soak gifts bit say difference',\n",
       " 'wife love great product',\n",
       " 'hair property poor large anti lather dandruff come lather pleasant vague like zero require quantity smell bad also bottle awful waste water easy',\n",
       " 'love fragrance',\n",
       " 'love granddaughter fragrance',\n",
       " \"else itchy ridiculous good buy dove make cream find use amazon dove's oil longer price since wash anywhere lotion manufacture body dry could skin pair go relieve\",\n",
       " 'ensoleillante light like long lotion eau moisturize body clarins year love try absolutely make wish',\n",
       " 'prescribed doctor psoriasis make use head ointment shoulder stuff cream well someday force bad horde love go stop work medical',\n",
       " 'soft beautiful use lotion basis smooth daily ever one compliment love',\n",
       " 'shampoo hair flare irritated psoriasis wash happy less replace lot swap every cleaner',\n",
       " 'always freshen travel good',\n",
       " 'shampoo use hair love yellow years...keeps gray',\n",
       " 'love shower take',\n",
       " 'itchy soap get generous rose nice easy side much big amount think onto come bar hand enjoy fragrance best way overall feel wet like find shower moisturize smell make dry really smell skin spin lot rise job time',\n",
       " 'lingers fresh lightly fragrance scent clean love lotion purchase purchasing hour gift skin daughter bergamont',\n",
       " 'unfortunately love music scent best make anymore',\n",
       " 'love mom',\n",
       " 'nice away heat use hair-straightener recommend travel compact actually product perfect gauge great make sleek daily glove although avid rate one complaint love right hair option quickly become come digital zero highly plate also thinner honest temperature receive easy user grab buy case exchange serve discount run heat adjust straightener store review',\n",
       " 'know product many company good nicely andalou like purchase wonderful smell skin well quality work lather',\n",
       " 'hair ever product best',\n",
       " 'shampoo great',\n",
       " 'man go never saturday night fragrance scent rid awesome axe get really woods warm blend leather perfect friday wish',\n",
       " 'shea feel soap soft leave spot consider enough use bottom-line bar price pleasure costco butter purchase tub huge gentle luxury patchoule girley im guilty brick skin sudzing also hippy love product subtle',\n",
       " 'bee bath buy julep acquaint effect fragrance wash mint last savannah like time long would recommend body invigorate originally one well always become lather',\n",
       " 'know tight hair soft clean get product nice act exfoliant soon plastic irritated random definitely use scrub consider rid section certain smooth instead red physical especially blemish dry flaky neither skin microbeads need face walnut product piece still heal try leave shell care',\n",
       " \"product breezy bubble pop bright quite breath bury scent sea salty citrus-y smell edge skin salt provence think open glass can't soap soft get pour enough dare take gingerbread bar feel long tingle luscious nose warm creamy picture touch humid tickle let leave spray sweet brine pre rock sip sweetish outcropping want chop lather iced get would particular white lime-y also bottle ocean's day root de rich fresh ginger make fish--think fragrance ocean brisk okay effervescence zingy make brace\",\n",
       " \"wrong give perfume company good aura totally j'adore like size happy lotion purchase sell intense would actual star smell smell could bottle exactly think skeptical order original saving\",\n",
       " \"shampoo forever kind go good make scalp crazy seem cranky...it's symptom randomly like hat retaliates find start product eczema conditioner last\",\n",
       " 'love drop',\n",
       " 'actually love early come',\n",
       " 'give know soap want forever staple large worth never lovely make sizes best present light lather chirstmas aromas purchase purchase complaint wonderful well one four always strong cost others absolutely receive five everyone last',\n",
       " \"know female head like shower ax share need guy detailer job get thing market roommate feel color different scrub wife find showerhead place tool daughter right scrubber foot likely bathroom want awhile besides etc rougher loofa area clean flip basically girlfriend side hang elbow axe tough scheme end body ask do another life's afterthought\",\n",
       " 'soft dry hair less frizz blow',\n",
       " \"nice lively concern use clothing scent remain way like clothes potency recommend ironing scent add come noticeable product guess others get lend poly loosely thing enough sprayer wonder great iron eliminate long happen qc consider french bag one term understate added odd impart spritz leak company's steam fabric safety steamer perhaps would lavender bottle linen lid steam weak bed water receive inner idea modicum make never couch appear since saw fragrance overall garment seal well another notice shipping rather review\",\n",
       " 'get product eye(s beauticontrol ph time burn balance accidentally sting happen use love',\n",
       " 'time available need bottle still hope want next',\n",
       " \"shampoo hair company fuller give volume great hide amazon want thin stop selling sally's thicker hour thanks buy\",\n",
       " 'shampoo difference dandruff watery see much',\n",
       " 'product bath it...especially pricey worth day gel brown bit long especially huge stressful shower molton travel fan revive cempaka',\n",
       " 'spot hand sponge price sea life shower extend well decent well-drained keep small hold easily exfoliates',\n",
       " 'one favorite',\n",
       " 'shampoo itchy know get many good bought irritate burn scalp use state reviews description two almost honestly red time immediately gentle return one bottle',\n",
       " 'great son delivery price us',\n",
       " 'well appreciate expect delivery really item rapid',\n",
       " 'find favorite hard',\n",
       " 'feel clean soap yet wonderful dry really skin use love leave',\n",
       " 'artificial great lovely except product work color',\n",
       " \"shampoo water get bad poor good scalp much use review feel shampoo's lather try start use agree product think expensive quality start improvement\",\n",
       " 'shadow box love favorite come eye even instruction',\n",
       " 'love',\n",
       " 'affordable thank wish',\n",
       " 'nice',\n",
       " \"can't year low wait use available get love absolutely make\",\n",
       " 'love procuct',\n",
       " \"fall shampoo give significantly baby i've find wash extremely ever result foundation mineral week elizabeth perfect expensive penny makeup last least brush soft high bristle take arden fullness wash good less brand-new nearly really one tell quality angle apply coverage reason forever good single use worn worth also fan powder work even bristle unbelievably easy press day except year side vary never get i'm word diehard though bare wonderful well use note loose every absolutely care\",\n",
       " 'pricey worth product bit price best nest',\n",
       " 'get able store great find could love product wish',\n",
       " 'goutal lotion annick lovely favorite line hadrien de eau',\n",
       " 'good',\n",
       " 'also cream great take fine puffiness line tighten eye conceals makeup calibra skin',\n",
       " 'smell nice',\n",
       " \"wear day year get city wear much never weirdly drink complexion first continue dave hide two time try highly look buster's get recommend make ticket redness game ago culver powder every even coffee\",\n",
       " \"heal look get similar foot vaseline mistakenly fraction neosporin use split much recommend crack cut use i've quite feel think hand price greasy since feel like good time would worst wonderful doctor favor half soothe product aquaphor eczema heal outbreak ointment work breakout time severe\",\n",
       " 'help hair fine/thin texture thickness product good',\n",
       " 'shea soap favorite forever use bars france bar amaze mill soap butter clarisonic french traditional face love provence quad giant last',\n",
       " \"else jan good month lash procyde use something everyday use review work gonna i'm longer want there's factor marini noticeable lash conditioner try difference back expensive intervention revitalash even eyelash md\",\n",
       " \"mild plastic crazy much use work smooth like dry bent travel emory uneven brush thing nail great simple daily provide ok little charge bend really one quality file whole similar soon free foot/heel consider maintenance use probably bit thin hotel flimsy item foam area separation work edge buff idea piece i'm board basic toe kit heavy well pretty\",\n",
       " 'product heavenly wash never break afraid time order fail caress body smell of.....the receive',\n",
       " 'soft elastic go pricey definitely worth feel way long make little skin love amazingly',\n",
       " 'assortment nice please',\n",
       " 'love use wash locs husband',\n",
       " 'great',\n",
       " 'feel product bath leave nice never cream moisturize great use want condition line wonderful body make smell buy go skin also perlier another product absolutely',\n",
       " \"shampoo hair stimulate great i've whether cannot growth condition start use tell leave conditioner\",\n",
       " 'love',\n",
       " 'love run get aunt',\n",
       " 'great',\n",
       " 'always great product masculine tobs scent pleasant',\n",
       " 'shampoo product le grande use great francisco also love family conditioner',\n",
       " \"much shampoo&body wash love bottle product perfect cheesy title definitely sure philosophy wrap read box friend write picture daughter love sm back sign make beautiful reuse mother recommend home bit adorable mom message heart flimsy me..grr send buy nevertheless christmas moment i'm sadly around bow yet box)the look there's give give\",\n",
       " 'hair several favorite ton side make type heat matter use slightly amount color somewhat since month beauty add anyone...no oily line would recommend dry launch sally moisture supply heavy buy use one ago shine fine product right back',\n",
       " 'shampoo hair bleach blonde best set smooth make fine love healthy conditioner',\n",
       " 'shampoo condo buy rental price along size lotion oceanfront gift perfect welcome',\n",
       " 'fall philosophy great yummy yet lip smell goody also love another gloss',\n",
       " 'product good quick use feel regularly always satisfy completely delivery look deal',\n",
       " 'squeeze nice hard kid lie pink large toothpaste find great teeth around clean mint like time involve long taste brush fun fantastic tube force make use one felt half fan green love bubblegum tube squeeze last',\n",
       " 'cologne different price',\n",
       " 'love either body dry wash smell skin',\n",
       " \"shampoo can't without live\",\n",
       " 'cream excellent spot especially non-greasy elbow heel rough',\n",
       " \"hold holder hygienist power much base wash hand i've use die dishwasher rubber charge way flush tethered dental model extra recommend hour recommend floss powerful maneuverable valve/plug week need tuck tooth battery unit last brush get several anyone lightly tire problem gum largely teeth amaze feel clean recharge routine unit store find handle brush dial food charge really one flosser replace love quality light make add high leaf session set gum machine amazon first design super want fail month rechargable get coil ago material still lid waterpik see even water clean hose easy remove year recede tank buy make fresh side tip never plug either review sonicare two last old slipping battery behind warranty wonderful tube floss use year well base do every starting use rout store wish feature control\",\n",
       " 'great',\n",
       " 'get let addict nice popsugar rich good month instruction great first long since lather last like box hang ring dry smell give really one say love product new us even exfoliation receive',\n",
       " \"love can't get unfortunately scent anymore\",\n",
       " 'shampoo one great buy',\n",
       " 'luv',\n",
       " 'know heal apply finger house able clearly pretty-this big definitely amazing greasy best want wash hand old look swap goop put wipe hardcore etc minor-scrapes flower heal ointment state anything lace make',\n",
       " 'texture get favorite perfume absolute find set want two lotion wonderful body smell everything usually love bottle',\n",
       " 'use felt improve water scent like skin',\n",
       " 'shampoo email product nice wed company bag buy definitely something fit thought want join natural customer cruise purchase recommend service little tube recieved gift bill go explore also touch thank guest',\n",
       " 'shampoo shine love body frizz',\n",
       " 'spray perfume give put add clean light lotion little smell bring compliment love lot',\n",
       " 'day hair cap get nice good barely great head overall wash want long shower thick fit comfortably stay well perfect damp every even edge keep water place job',\n",
       " 'product foam heavenly bath b&b particle amount feel scrub gel sponge scrub small real wash lather like help hungry luscious body make thick minor cranberry-scented also world smell lot price standard use meet seed',\n",
       " 'shampoo great fine love volume hair....gives',\n",
       " 'shampoo summer leave kind consider fresh single malt woman grapefruit problem bar scotch macallan vendor cool like perhaps would really smell well love package smell expensive morning meet',\n",
       " 'year make lotion fav go anymore wish',\n",
       " 'love',\n",
       " ...]"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corpus"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "list"
      ]
     },
     "execution_count": 118,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "type(corpus)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False    73074\n",
       "True       225\n",
       "dtype: int64"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features = (df.review_clean.str.split(expand=True).stack().value_counts() > 1500).value_counts()\n",
    "features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "607"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "tfidf = TfidfVectorizer(max_features=1500)       # taking 500 unique words by highest frequency"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# X = tfidf.fit_transform(corpus).toarray()\n",
    "# y = df.iloc[:, -1].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = tfidf.fit_transform(corpus).toarray()\n",
    "y = df.iloc[:, -1].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.        , 0.        , 0.        , ..., 0.        , 0.69096545,\n",
       "        0.        ],\n",
       "       [0.        , 0.        , 0.47647222, ..., 0.        , 0.        ,\n",
       "        0.        ],\n",
       "       [0.        , 0.        , 0.        , ..., 0.        , 0.        ,\n",
       "        0.        ],\n",
       "       ...,\n",
       "       [0.        , 0.        , 0.        , ..., 0.        , 0.        ,\n",
       "        0.65897468],\n",
       "       [0.        , 0.        , 0.        , ..., 0.        , 0.        ,\n",
       "        0.        ],\n",
       "       [0.        , 0.        , 0.        , ..., 0.        , 0.        ,\n",
       "        0.        ]])"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 1, 1, ..., 1, 1, 1])"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "float64\n",
      "int32\n"
     ]
    }
   ],
   "source": [
    "print(X.dtype)\n",
    "print(y.dtype)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'numpy.ndarray'>\n",
      "<class 'numpy.ndarray'>\n"
     ]
    }
   ],
   "source": [
    "print(type(X))\n",
    "print(type(y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7500000\n",
      "30000\n"
     ]
    }
   ],
   "source": [
    "print(X.size)\n",
    "print(y.size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from sklearn.model_selection import train_test_split\n",
    "# X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.25, random_state = 42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Embedding, LSTM, SpatialDropout1D\n",
    "# from tensorflow.keras import Sequential\n",
    "# from tensorflow.keras.layers import Dense, Embedding, LSTM, SpatialDropout1D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_features = features[1].item()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "int"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "type(max_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.preprocessing.sequence import pad_sequences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_features = 2000\n",
    "tokenizer = Tokenizer(num_words=max_features, split=' ')\n",
    "tokenizer.fit_on_texts(df['review_clean'].values)\n",
    "X = tokenizer.texts_to_sequences(df['review_clean'].values)\n",
    "X = pad_sequences(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " embedding (Embedding)       (None, 661, 128)          256000    \n",
      "                                                                 \n",
      " spatial_dropout1d (SpatialD  (None, 661, 128)         0         \n",
      " ropout1D)                                                       \n",
      "                                                                 \n",
      " lstm (LSTM)                 (None, 196)               254800    \n",
      "                                                                 \n",
      " dense (Dense)               (None, 2)                 394       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 511,194\n",
      "Trainable params: 511,194\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "embed_dim = 128\n",
    "lstm_out = 196\n",
    "\n",
    "model = Sequential()\n",
    "model.add(Embedding(max_features, embed_dim,input_length = X.shape[1]))\n",
    "model.add(SpatialDropout1D(0.4))\n",
    "model.add(LSTM(lstm_out, dropout=0.2))\n",
    "model.add(Dense(2,activation='softmax'))\n",
    "model.compile(loss = 'categorical_crossentropy', optimizer='adam',metrics = ['accuracy'])\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(49166, 661) (49166, 2)\n",
      "(24217, 661) (24217, 2)\n"
     ]
    }
   ],
   "source": [
    "Y = pd.get_dummies(df['sentiment']).values\n",
    "X_train, X_test, Y_train, Y_test = train_test_split(X,Y, test_size = 0.33, random_state = 42)\n",
    "print(X_train.shape,Y_train.shape)\n",
    "print(X_test.shape,Y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(73383, 2)"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.get_dummies(df['sentiment']).values.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(73383, 661)"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "tf.config.run_functions_eagerly(True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "1537/1537 - 92s - loss: 0.3286 - accuracy: 0.8661 - 92s/epoch - 60ms/step\n",
      "Epoch 2/10\n",
      "1537/1537 - 91s - loss: 0.2705 - accuracy: 0.8884 - 91s/epoch - 59ms/step\n",
      "Epoch 3/10\n",
      "1537/1537 - 90s - loss: 0.2559 - accuracy: 0.8956 - 90s/epoch - 58ms/step\n",
      "Epoch 4/10\n",
      "1537/1537 - 90s - loss: 0.2434 - accuracy: 0.9007 - 90s/epoch - 59ms/step\n",
      "Epoch 5/10\n",
      "1537/1537 - 90s - loss: 0.2322 - accuracy: 0.9053 - 90s/epoch - 59ms/step\n",
      "Epoch 6/10\n",
      "1537/1537 - 90s - loss: 0.2226 - accuracy: 0.9106 - 90s/epoch - 59ms/step\n",
      "Epoch 7/10\n",
      "1537/1537 - 91s - loss: 0.2116 - accuracy: 0.9151 - 91s/epoch - 59ms/step\n",
      "Epoch 8/10\n",
      "1537/1537 - 90s - loss: 0.2016 - accuracy: 0.9196 - 90s/epoch - 59ms/step\n",
      "Epoch 9/10\n",
      "1537/1537 - 91s - loss: 0.1909 - accuracy: 0.9250 - 91s/epoch - 59ms/step\n",
      "Epoch 10/10\n",
      "1537/1537 - 91s - loss: 0.1809 - accuracy: 0.9296 - 91s/epoch - 59ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x28debcbaa30>"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batch_size = 32\n",
    "model.fit(X_train, Y_train, epochs = 10, batch_size=batch_size, verbose = 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('model2.hdf5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "new_model = tf.keras.models.load_model('model2.hdf5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "loss = [0.3286, 0.2705, 0.2559, 0.2434, 0.2322, 0.2226, 0.2116, 0.2016, 0.1909, 0.1809]\n",
    "acc = [0.8661, 0.8884, 0.8956, 0.9007, 0.9053, 0.9106, 0.9151, 0.9196, 0.9250, 0.9296]\n",
    "epoch = [i for i in range(1,11)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(epoch, acc, label=\"Acc\")\n",
    "plt.plot(epoch, loss, label=\"Loss\")\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Value\")\n",
    "plt.title('Loss and Acc')\n",
    "plt.legend(loc=\"right\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "664/664 - 29s - loss: 0.3170 - accuracy: 0.8862 - 29s/epoch - 44ms/step\n",
      "score: 0.32\n",
      "acc: 0.89\n"
     ]
    }
   ],
   "source": [
    "validation_size = 1500\n",
    "batch_size = 32\n",
    "X_validate = X_test[-validation_size:]\n",
    "Y_validate = Y_test[-validation_size:]\n",
    "X_test = X_test[:-validation_size]\n",
    "Y_test = Y_test[:-validation_size]\n",
    "score,acc = new_model.evaluate(X_test, Y_test, verbose = 2, batch_size = batch_size)\n",
    "print(\"score: %.2f\" % (score))\n",
    "print(\"acc: %.2f\" % (acc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 - 0s - 305ms/epoch - 305ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 37ms/epoch - 37ms/step\n",
      "1/1 - 0s - 38ms/epoch - 38ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 38ms/epoch - 38ms/step\n",
      "1/1 - 0s - 37ms/epoch - 37ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 37ms/epoch - 37ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 37ms/epoch - 37ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 37ms/epoch - 37ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 38ms/epoch - 38ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 37ms/epoch - 37ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 35ms/epoch - 35ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 37ms/epoch - 37ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 30ms/epoch - 30ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 34ms/epoch - 34ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "1/1 - 0s - 36ms/epoch - 36ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 33ms/epoch - 33ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 32ms/epoch - 32ms/step\n",
      "1/1 - 0s - 31ms/epoch - 31ms/step\n",
      "pos_acc 95.02806736166801 %\n",
      "neg_acc 42.68774703557312 %\n"
     ]
    }
   ],
   "source": [
    "pos_cnt, neg_cnt, pos_correct, neg_correct = 0, 0, 0, 0\n",
    "for x in range(len(X_validate)):\n",
    "    \n",
    "    result = new_model.predict(X_validate[x].reshape(1,X_test.shape[1]),batch_size=1,verbose = 2)[0]\n",
    "   \n",
    "    if np.argmax(result) == np.argmax(Y_validate[x]):\n",
    "        if np.argmax(Y_validate[x]) == 0:\n",
    "            neg_correct += 1\n",
    "        else:\n",
    "            pos_correct += 1\n",
    "       \n",
    "    if np.argmax(Y_validate[x]) == 0:\n",
    "        neg_cnt += 1\n",
    "    else:\n",
    "        pos_cnt += 1\n",
    "\n",
    "\n",
    "\n",
    "print(\"pos_acc\", pos_correct/pos_cnt*100, \"%\")\n",
    "print(\"neg_acc\", neg_correct/neg_cnt*100, \"%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "13768953aedf00ec6bad39a037e4d4688640d7d465f42e95b481cfa8be0fcd5a"
  },
  "kernelspec": {
   "display_name": "capstone",
   "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": "3.8.0"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}