D213_task2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "acbf9576",
   "metadata": {},
   "source": [
    "## Research Question"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f5a6815c",
   "metadata": {},
   "source": [
    "The research question that we hope to answer is if we can determine customer sentiment from the wording of reviews. By doing this we hope to be able to guage if a customer likes or dislikes a product.\n",
    "\n",
    "Using natural language processing, our goal is to analyize the wording within customer reviews to predict if a customer has a positive or negative feeling about a product.\n",
    "\n",
    "Our analysis will be conducted using TensorFlow neural networks"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f67b8468",
   "metadata": {},
   "source": [
    "## Data Prep"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "cd502a0d",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# Package importing\n",
    "import json\n",
    "import gzip\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.image as mpimg\n",
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.preprocessing.text import Tokenizer\n",
    "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
    "from keras import callbacks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "91431f6c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Parse for json gzip file\n",
    "def parse(path):\n",
    "  g = gzip.open(path, 'rb')\n",
    "  for l in g:\n",
    "    yield json.loads(l)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "cff15aee",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Turn our json file into data frame\n",
    "def getDF(path):\n",
    "  i = 0\n",
    "  df = {}\n",
    "  for d in parse(path):\n",
    "    df[i] = d\n",
    "    i += 1\n",
    "  return pd.DataFrame.from_dict(df, orient='index')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "74dab1ed",
   "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",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>True</td>\n",
       "      <td>01 5, 2018</td>\n",
       "      <td>A2HOI48JK8838M</td>\n",
       "      <td>B00004U9V2</td>\n",
       "      <td>{'Size:': ' 0.9 oz.'}</td>\n",
       "      <td>DB</td>\n",
       "      <td>This handcream has a beautiful fragrance. It d...</td>\n",
       "      <td>Beautiful Fragrance</td>\n",
       "      <td>1515110400</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5.0</td>\n",
       "      <td>True</td>\n",
       "      <td>04 5, 2017</td>\n",
       "      <td>A1YIPEY7HX73S7</td>\n",
       "      <td>B00004U9V2</td>\n",
       "      <td>{'Size:': ' 3.5 oz.'}</td>\n",
       "      <td>Ajaey</td>\n",
       "      <td>wonderful hand lotion, for seriously dry skin,...</td>\n",
       "      <td>wonderful hand lotion</td>\n",
       "      <td>1491350400</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5.0</td>\n",
       "      <td>True</td>\n",
       "      <td>03 27, 2017</td>\n",
       "      <td>A2QCGHIJ2TCLVP</td>\n",
       "      <td>B00004U9V2</td>\n",
       "      <td>{'Size:': ' 250 g'}</td>\n",
       "      <td>D. Jones</td>\n",
       "      <td>Best hand cream around.  Silky, thick, soaks i...</td>\n",
       "      <td>Best hand cream around</td>\n",
       "      <td>1490572800</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5.0</td>\n",
       "      <td>True</td>\n",
       "      <td>03 20, 2017</td>\n",
       "      <td>A2R4UNHFJBA6PY</td>\n",
       "      <td>B00004U9V2</td>\n",
       "      <td>{'Size:': ' 3.5 oz.'}</td>\n",
       "      <td>Amazon Customer</td>\n",
       "      <td>Thanks!!</td>\n",
       "      <td>Five Stars</td>\n",
       "      <td>1489968000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>True</td>\n",
       "      <td>02 28, 2017</td>\n",
       "      <td>A2QCGHIJ2TCLVP</td>\n",
       "      <td>B00004U9V2</td>\n",
       "      <td>{'Size:': ' 0.9 oz.'}</td>\n",
       "      <td>D. Jones</td>\n",
       "      <td>Great hand lotion.  Soaks right in and leaves ...</td>\n",
       "      <td>Great hand lotion!</td>\n",
       "      <td>1488240000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   overall  verified   reviewTime      reviewerID        asin  \\\n",
       "0      5.0      True   01 5, 2018  A2HOI48JK8838M  B00004U9V2   \n",
       "1      5.0      True   04 5, 2017  A1YIPEY7HX73S7  B00004U9V2   \n",
       "2      5.0      True  03 27, 2017  A2QCGHIJ2TCLVP  B00004U9V2   \n",
       "3      5.0      True  03 20, 2017  A2R4UNHFJBA6PY  B00004U9V2   \n",
       "4      5.0      True  02 28, 2017  A2QCGHIJ2TCLVP  B00004U9V2   \n",
       "\n",
       "                   style     reviewerName  \\\n",
       "0  {'Size:': ' 0.9 oz.'}               DB   \n",
       "1  {'Size:': ' 3.5 oz.'}            Ajaey   \n",
       "2    {'Size:': ' 250 g'}         D. Jones   \n",
       "3  {'Size:': ' 3.5 oz.'}  Amazon Customer   \n",
       "4  {'Size:': ' 0.9 oz.'}         D. Jones   \n",
       "\n",
       "                                          reviewText                 summary  \\\n",
       "0  This handcream has a beautiful fragrance. It d...     Beautiful Fragrance   \n",
       "1  wonderful hand lotion, for seriously dry skin,...   wonderful hand lotion   \n",
       "2  Best hand cream around.  Silky, thick, soaks i...  Best hand cream around   \n",
       "3                                           Thanks!!              Five Stars   \n",
       "4  Great hand lotion.  Soaks right in and leaves ...      Great hand lotion!   \n",
       "\n",
       "   unixReviewTime vote image  \n",
       "0      1515110400  NaN   NaN  \n",
       "1      1491350400  NaN   NaN  \n",
       "2      1490572800  NaN   NaN  \n",
       "3      1489968000  NaN   NaN  \n",
       "4      1488240000  NaN   NaN  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Importing data\n",
    "df = getDF('Luxury_Beauty_5.json.gz')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dd07e777",
   "metadata": {},
   "source": [
    "Looking at our data frame, we want to select the overall score and the review text columns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e8d2087f",
   "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>reviewText</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>This handcream has a beautiful fragrance. It d...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5.0</td>\n",
       "      <td>wonderful hand lotion, for seriously dry skin,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Best hand cream around.  Silky, thick, soaks i...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Thanks!!</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Great hand lotion.  Soaks right in and leaves ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   overall                                         reviewText\n",
       "0      5.0  This handcream has a beautiful fragrance. It d...\n",
       "1      5.0  wonderful hand lotion, for seriously dry skin,...\n",
       "2      5.0  Best hand cream around.  Silky, thick, soaks i...\n",
       "3      5.0                                           Thanks!!\n",
       "4      5.0  Great hand lotion.  Soaks right in and leaves ..."
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# New data frame with out selected columns\n",
    "df = df[['overall','reviewText']]\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "f0ebce26",
   "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>Score</th>\n",
       "      <th>Review</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>This handcream has a beautiful fragrance. It d...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5.0</td>\n",
       "      <td>wonderful hand lotion, for seriously dry skin,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Best hand cream around.  Silky, thick, soaks i...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Thanks!!</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Great hand lotion.  Soaks right in and leaves ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Score                                             Review\n",
       "0    5.0  This handcream has a beautiful fragrance. It d...\n",
       "1    5.0  wonderful hand lotion, for seriously dry skin,...\n",
       "2    5.0  Best hand cream around.  Silky, thick, soaks i...\n",
       "3    5.0                                           Thanks!!\n",
       "4    5.0  Great hand lotion.  Soaks right in and leaves ..."
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Changing column names\n",
    "df = df.rename(columns={'overall': 'Score', 'reviewText': 'Review'})\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a0c7771a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Score     False\n",
       "Review     True\n",
       "dtype: bool"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check for any nulls\n",
    "df.isna().any()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9c28914f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Remove our nulls\n",
    "df = df.dropna(subset=['Review'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "3f00430f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Score     False\n",
       "Review    False\n",
       "dtype: bool"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check for any nulls again\n",
    "df.isna().any()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "47eaa667",
   "metadata": {},
   "source": [
    "We can see our dataframe no longer contains any nulls"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "04136868",
   "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>Score</th>\n",
       "      <th>Review</th>\n",
       "      <th>Sentiments</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>This handcream has a beautiful fragrance. It d...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5.0</td>\n",
       "      <td>wonderful hand lotion, for seriously dry skin,...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Best hand cream around.  Silky, thick, soaks i...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Thanks!!</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Great hand lotion.  Soaks right in and leaves ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Score                                             Review  Sentiments\n",
       "0    5.0  This handcream has a beautiful fragrance. It d...           1\n",
       "1    5.0  wonderful hand lotion, for seriously dry skin,...           1\n",
       "2    5.0  Best hand cream around.  Silky, thick, soaks i...           1\n",
       "3    5.0                                           Thanks!!           1\n",
       "4    5.0  Great hand lotion.  Soaks right in and leaves ...           1"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Creating sentiments column to gauge review 0-2 negative, 3-5 positive\n",
    "# 0 = negative, 1 = positive\n",
    "df['Sentiments'] = df.Score.apply(lambda x: 0 if x in [1, 2] else 1)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "3c9ac473",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Selecting our special characters\n",
    "unique_char = [\"!\",'\"',\"#\",\"%\",\"&\",\"'\",\"(\",\")\",\n",
    " \"*\",\"+\",\",\",\"-\",\".\",\"/\",\":\",\";\",\"<\",\n",
    " \"=\",\">\",\"?\",\"@\",\"[\",\"\\\\\",\"]\",\"^\",\"_\",\n",
    " \"`\",\"{\",\"|\",\"}\",\"~\",\"–\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "498559e4",
   "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>Score</th>\n",
       "      <th>Review</th>\n",
       "      <th>Sentiments</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>this handcream has a beautiful fragrance  it d...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5.0</td>\n",
       "      <td>wonderful hand lotion  for seriously dry skin ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5.0</td>\n",
       "      <td>best hand cream around   silky  thick  soaks i...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5.0</td>\n",
       "      <td>thanks</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>great hand lotion   soaks right in and leaves ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Score                                             Review  Sentiments\n",
       "0    5.0  this handcream has a beautiful fragrance  it d...           1\n",
       "1    5.0  wonderful hand lotion  for seriously dry skin ...           1\n",
       "2    5.0  best hand cream around   silky  thick  soaks i...           1\n",
       "3    5.0                                           thanks             1\n",
       "4    5.0  great hand lotion   soaks right in and leaves ...           1"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# replacing our special characters\n",
    "for char in unique_char:\n",
    " df['Review'] = df['Review'].str.replace(char, ' ',regex=True)\n",
    "\n",
    "# Replacing all capitals with lowercase\n",
    "df['Review'] = df['Review'].str.lower()\n",
    "\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "a70c3e00",
   "metadata": {},
   "outputs": [],
   "source": [
    "# exporting our prepped data\n",
    "df.to_csv('cleaned_task2.csv', index = False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "75cbc1ab",
   "metadata": {},
   "source": [
    "To continue our analysis, we must tokenize the stings of words. This will convert words in the sentence string into their own string. By doing this we will be able to run analysis on each word, instead of the main string. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "ee1ff8f8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Splitting data into 80/20 training and testing\n",
    "split = round(len(df)*0.8)\n",
    "train_reviews = df['Review'][:split]\n",
    "train_label = df['Sentiments'][:split]\n",
    "test_reviews = df['Review'][split:]\n",
    "test_label = df['Sentiments'][split:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "f17fe92f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Setting up the split after tokenization\n",
    "training_sentences = []\n",
    "training_labels = []\n",
    "testing_sentences = []\n",
    "testing_labels = []\n",
    "for row in train_reviews:\n",
    "    training_sentences.append(str(row))\n",
    "for row in train_label:\n",
    "    training_labels.append(row)\n",
    "for row in test_reviews:\n",
    "    testing_sentences.append(str(row))\n",
    "for row in test_label:\n",
    "    testing_labels.append(row)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "1f7f01b4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Training at 50000 words\n",
    "vocab_size = 50000\n",
    "# 16 dimensions\n",
    "embedding_dim = 16\n",
    "# Max length of 120 words per review\n",
    "max_length = 120\n",
    "# Truncating after 120 \n",
    "trunc_type = 'post'\n",
    "oov_tok = '<OOV>'\n",
    "# Padding after 120\n",
    "padding_type = 'post'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "689fe742",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Starting the tokenizer\n",
    "tokenizer = Tokenizer(num_words=vocab_size, oov_token=oov_tok)\n",
    "# Fitting the tokenizer\n",
    "tokenizer.fit_on_texts(training_sentences)\n",
    "# Getting our word index\n",
    "word_index = tokenizer.word_index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "5368b17f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Setting sequences\n",
    "sequences = tokenizer.texts_to_sequences(training_sentences)\n",
    "# Setting padding\n",
    "padded = pad_sequences(sequences, maxlen=max_length, truncating=trunc_type)\n",
    "testing_sentences = tokenizer.texts_to_sequences(testing_sentences)\n",
    "testing_padded = pad_sequences(testing_sentences, maxlen=max_length)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7ce6dc3f",
   "metadata": {},
   "source": [
    "## Network Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "386d9ff7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating model\n",
    "model = tf.keras.Sequential([\n",
    "    tf.keras.layers.Embedding(vocab_size, embedding_dim, input_length=max_length),\n",
    "    tf.keras.layers.GlobalAveragePooling1D(),\n",
    "    tf.keras.layers.Dense(10, activation='relu'),\n",
    "    tf.keras.layers.Dense(6, activation='relu'),\n",
    "    tf.keras.layers.Dense(1, activation='relu')\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5fc4ac5d",
   "metadata": {},
   "source": [
    "Using TensorFlow, we can create our Keras model. Our embedding layer was previously set up with our chosen parameters. Our GlobalAveragePooling1D is set to output a 2D object and our third layer is set to relu activation with 10. The next two layers are set to 6 and 1 with relu activation respectively. By using relu, we are running linear activation, I am running this as we are working with binary 1 and 0 for our outputs. This will provide high accuracy with fewer tests needed to run.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "173f4b51",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Compiling the model\n",
    "model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "618dbc8a",
   "metadata": {},
   "source": [
    "Using binary_crossentropy for our analysis allows us to better analyze our binary output from our Keras model. Combining this loss function with the adam optimizer allows us to run a very efficient analysis with little requirement for extra memory. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "86553ba6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding (Embedding)        (None, 120, 16)           800000    \n",
      "_________________________________________________________________\n",
      "global_average_pooling1d (Gl (None, 16)                0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 10)                170       \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 6)                 66        \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 7         \n",
      "=================================================================\n",
      "Total params: 800,243\n",
      "Trainable params: 800,243\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "fae8b894",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Training the model\n",
    "training_final = np.array(training_labels)\n",
    "testing_final = np.array(testing_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "54492754",
   "metadata": {},
   "source": [
    "## Model Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "56faad95",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 27412 samples, validate on 6853 samples\n",
      "Epoch 1/25\n",
      "27412/27412 [==============================] - 12s 449us/sample - loss: 0.7161 - accuracy: 0.7107 - val_loss: 0.2384 - val_accuracy: 0.9259\n",
      "Epoch 2/25\n",
      "27412/27412 [==============================] - 11s 392us/sample - loss: 0.1985 - accuracy: 0.9243 - val_loss: 0.2491 - val_accuracy: 0.9260\n",
      "Epoch 3/25\n",
      "27412/27412 [==============================] - 12s 450us/sample - loss: 0.1650 - accuracy: 0.9270 - val_loss: 0.2658 - val_accuracy: 0.9273\n",
      "Epoch 4/25\n",
      "27412/27412 [==============================] - 11s 419us/sample - loss: 0.1399 - accuracy: 0.9370 - val_loss: 0.3563 - val_accuracy: 0.9302\n",
      "Epoch 5/25\n",
      "27412/27412 [==============================] - 12s 428us/sample - loss: 0.1225 - accuracy: 0.9466 - val_loss: 0.3539 - val_accuracy: 0.9262\n",
      "Epoch 6/25\n",
      "27412/27412 [==============================] - 12s 427us/sample - loss: 0.0986 - accuracy: 0.9599 - val_loss: 0.4092 - val_accuracy: 0.9257\n",
      "Epoch 7/25\n",
      "27412/27412 [==============================] - 11s 412us/sample - loss: 0.0987 - accuracy: 0.9602 - val_loss: 0.4238 - val_accuracy: 0.9221\n",
      "Epoch 8/25\n",
      "27412/27412 [==============================] - 12s 433us/sample - loss: 0.0757 - accuracy: 0.9706 - val_loss: 0.4946 - val_accuracy: 0.9235\n",
      "Epoch 9/25\n",
      "27412/27412 [==============================] - 13s 479us/sample - loss: 0.1174 - accuracy: 0.9695 - val_loss: 0.4841 - val_accuracy: 0.9234\n",
      "Epoch 10/25\n",
      "27412/27412 [==============================] - 12s 450us/sample - loss: 0.1011 - accuracy: 0.9662 - val_loss: 0.6166 - val_accuracy: 0.9211\n",
      "Epoch 11/25\n",
      "27412/27412 [==============================] - 12s 421us/sample - loss: 0.0623 - accuracy: 0.9796 - val_loss: 0.6856 - val_accuracy: 0.9215\n",
      "Epoch 12/25\n",
      "27412/27412 [==============================] - 11s 401us/sample - loss: 0.0598 - accuracy: 0.9829 - val_loss: 0.7459 - val_accuracy: 0.9187\n",
      "Epoch 13/25\n",
      "27412/27412 [==============================] - 11s 418us/sample - loss: 0.0485 - accuracy: 0.9868 - val_loss: 0.7883 - val_accuracy: 0.9187\n",
      "Epoch 14/25\n",
      "27412/27412 [==============================] - 13s 477us/sample - loss: 0.0455 - accuracy: 0.9888 - val_loss: 0.8272 - val_accuracy: 0.9161\n",
      "Epoch 15/25\n",
      "27412/27412 [==============================] - 12s 426us/sample - loss: 0.6235 - accuracy: 0.9277 - val_loss: 0.6712 - val_accuracy: 0.9162\n",
      "Epoch 16/25\n",
      "27412/27412 [==============================] - 11s 412us/sample - loss: 0.0592 - accuracy: 0.9834 - val_loss: 0.7228 - val_accuracy: 0.9170\n",
      "Epoch 17/25\n",
      "27412/27412 [==============================] - 11s 412us/sample - loss: 0.0481 - accuracy: 0.9870 - val_loss: 0.8182 - val_accuracy: 0.9108\n",
      "Epoch 18/25\n",
      "27412/27412 [==============================] - 11s 405us/sample - loss: 0.0454 - accuracy: 0.9899 - val_loss: 0.8730 - val_accuracy: 0.9129\n",
      "Epoch 19/25\n",
      "27412/27412 [==============================] - 11s 400us/sample - loss: 0.0413 - accuracy: 0.9917 - val_loss: 0.9034 - val_accuracy: 0.9104\n",
      "Epoch 20/25\n",
      "27412/27412 [==============================] - 11s 405us/sample - loss: 0.3418 - accuracy: 0.9669 - val_loss: 0.7899 - val_accuracy: 0.9190\n",
      "Epoch 21/25\n",
      "27412/27412 [==============================] - 11s 403us/sample - loss: 0.0416 - accuracy: 0.9921 - val_loss: 0.8915 - val_accuracy: 0.9130\n",
      "Epoch 22/25\n",
      "27412/27412 [==============================] - 11s 403us/sample - loss: 0.0374 - accuracy: 0.9932 - val_loss: 0.9027 - val_accuracy: 0.9123\n",
      "Epoch 23/25\n",
      "27412/27412 [==============================] - 11s 410us/sample - loss: 0.0366 - accuracy: 0.9938 - val_loss: 0.9047 - val_accuracy: 0.9167\n",
      "Epoch 24/25\n",
      "27412/27412 [==============================] - 11s 396us/sample - loss: 0.0354 - accuracy: 0.9945 - val_loss: 0.9466 - val_accuracy: 0.9124\n",
      "Epoch 25/25\n",
      "27412/27412 [==============================] - 11s 395us/sample - loss: 0.0345 - accuracy: 0.9949 - val_loss: 1.0271 - val_accuracy: 0.9095\n"
     ]
    }
   ],
   "source": [
    "# Running the model for 25 epochs\n",
    "model1 = model.fit(padded, training_final, epochs=25, validation_data=(testing_padded,\n",
    "testing_final))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "d4119f16",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 27412 samples, validate on 6853 samples\n",
      "Epoch 1/25\n",
      "27412/27412 [==============================] - 11s 399us/sample - loss: 0.0374 - accuracy: 0.9949 - val_loss: 1.0323 - val_accuracy: 0.9095\n",
      "Epoch 2/25\n",
      "27412/27412 [==============================] - 11s 405us/sample - loss: 0.0364 - accuracy: 0.9950 - val_loss: 1.0899 - val_accuracy: 0.9079\n",
      "Epoch 3/25\n",
      "27412/27412 [==============================] - 11s 394us/sample - loss: 0.1101 - accuracy: 0.9899 - val_loss: 1.0562 - val_accuracy: 0.9108\n",
      "Epoch 4/25\n",
      "27412/27412 [==============================] - 11s 391us/sample - loss: 0.0336 - accuracy: 0.9961 - val_loss: 1.1165 - val_accuracy: 0.9087\n",
      "Epoch 5/25\n",
      "27412/27412 [==============================] - 11s 395us/sample - loss: 0.0353 - accuracy: 0.9959 - val_loss: 1.1333 - val_accuracy: 0.9063\n",
      "Epoch 6/25\n",
      "27412/27412 [==============================] - 11s 402us/sample - loss: 0.0357 - accuracy: 0.9961 - val_loss: 1.1785 - val_accuracy: 0.9028\n"
     ]
    }
   ],
   "source": [
    "# Early stopping to prevent overfitting of our model\n",
    "earlystopping = callbacks.EarlyStopping(monitor =\"val_loss\", \n",
    "                                        mode =\"min\", patience = 5, \n",
    "                                        restore_best_weights = True)\n",
    "\n",
    "# Running model again\n",
    "model2 = model.fit(padded, training_final, epochs=25, validation_data=(testing_padded,\n",
    "                                                                       testing_final),\n",
    "                                                                       callbacks = [earlystopping])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4f946483",
   "metadata": {},
   "source": [
    "We can see our model now stops at 6, preventing overfitting.\n",
    "\n",
    "We can see that our accuracy stops at .9961 and our val_accuracy is .9028. Using this we can argue that our model has a prediction model that is 90% correct in predicting customer sentiment based on words in a review. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "d1a3e823",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# epoch vs Mean Absolute Error\n",
    "plt.plot(model1.history['loss'], label='MAE (training data)')\n",
    "plt.title('MAE vs epoch')\n",
    "plt.ylabel('MAE value')\n",
    "plt.xlabel('No. epoch')\n",
    "plt.legend(loc=\"upper left\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a4a0745",
   "metadata": {},
   "source": [
    "After 6 epochs we begin to drop, showing that 6 is our optimal number of tests to run"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "748415e0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(model1.history['accuracy'], label='Accuracy')\n",
    "plt.title('accuracy vs epoch')\n",
    "plt.ylabel('accuracy value')\n",
    "plt.xlabel('No. epoch')\n",
    "plt.legend(loc=\"upper left\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "f614b2da",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "acc = model1.history['accuracy']\n",
    "val_acc = model1.history['val_accuracy']\n",
    "loss = model1.history['loss']\n",
    "val_loss = model1.history['val_loss']\n",
    "epochs=range(len(acc))\n",
    "plt.plot(epochs, acc, 'r', 'Training Accuracy')\n",
    "plt.plot(epochs, val_acc, 'b', 'Validation Accuracy')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.figure()\n",
    "plt.plot(epochs, loss, 'r', 'Training Loss')\n",
    "plt.plot(epochs, val_loss, 'b', 'Validation Loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.figure()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bae050f9",
   "metadata": {},
   "source": [
    "## Summary and Recommendations"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1a13173c",
   "metadata": {},
   "source": [
    "Our final prediction model had an output with 90% accuracy in predicting customer sentiment. This is a good outcome and we can use that to argue that our model works and will predict customer sentiment with 90% accuracy. Fine-tuning the network allows us to zero in on the parameters that will present the most accurate and fine-tuned model. Setting up our stop point allows us to prevent overfitting, even with an increase in accuracy. The tradeoff of this is that we get the best model that we can, without needed to overfit the model. \n",
    "\n",
    "\n",
    "For our stakeholders, the accuracy of our model is a key indication that they can trust our model. Using our analysis they can better predict customer sentiment based on the predictive outcomes of our model. Based on word usage in a review, our company can predict what a customer may rate a product and take a course of action based on the prediction. Rolling our prediction model into an overall customer retention strategy can help to prevent churn or better engineer products that customers may rate positively."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ee212628",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}