main.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# This is a Basic Sentiment Analysis Project | @HPD1155\n",
    "# Libraries Used\n",
    "- Tensorflow/Keras - Used for defining, building, and training the model\n",
    "- Pandas - Used for loading the dataset\n",
    "- Numpy - Used for working with arrays\n",
    "- Matplotlib - Used for plotting the data\n",
    "- Scikit-learn - Used for splitting the data into train and test sets\n",
    "- Math - Used for rounding."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here is what you run to install everything. Paste it into your terminal:\n",
    "```\n",
    "pip install tensorflow keras pandas numpy matplotlib scikit-learn\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Importing necessary libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from tensorflow.keras.preprocessing.text import Tokenizer\n",
    "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras import losses\n",
    "from sklearn.model_selection import train_test_split\n",
    "import math"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load in our dataset\n",
    "This will take out CSV filed named `data.csv` and create a dataframe out of it with **Pandas**.\n",
    "X is going to be the input data. It is getting the column named `text` from the dataframe.\n",
    "y is going to be the output data. It is getting the column named `label` from the dataframe."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('data.csv') # Dataframe\n",
    "\n",
    "X = df['text'] # Input data\n",
    "y = df['label'] # Output data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Preprocessing the data\n",
    "The neural network can't just understand plain letters like us. It needs to break them down and since it is a computer, it only knows numbers. We have to convert the text to numbers.\n",
    "\n",
    "This can be done by \"Tokenizing\" the text. This means it will break down the text into individual words inside a list. Next you can convert the list of words into a list of numbers or \"sequences\". It will be a list of numbers. Each column will be a seperate array of numbers. It would be a seperate dimension or \"section\" of the X data. `Eg: [[2, 4, 6, 3, 2, 5] [1, 6, 8, 2, 4, 5]]`. What is padding you say? Machine learning models expect all the sequences to be the same dimension or same length so each sequence has a use, there are no \"holes\" in the sequence. This means that each sequence has a \"use\" or a way that the model can interact with it. Padding essentially makes all the sequences the same length or in this case the same dimension. `Eg: [[0, 0, 0, 0, 0, 0, 2, 4, 6, 3, 2, 5] [0, 0, 0, 0, 0, 0, 1, 6, 8, 2, 4, 5]]`. They are all 12 dimensions.\n",
    "\n",
    "This section of code is going to do the following:\n",
    "- Tokenize the text (Eg. \"Hello, my name is John\" -> ['Hello', 'my', 'name', 'is', 'John'])\n",
    "- Convert the text to sequences (Eg. ['Hello', 'my', 'name', 'is', 'John'] -> [[3, 6, 8, 2, 4]])\n",
    "- pad the sequences so they are all the same dimension (Eg. [[3, 6, 8, 2, 4], [3, 6, 8, 2, 4]] -> [[0, 0, 0, 0, 0, 0, 3, 6, 8, 2, 4], [0, 0, 0, 0, 0, 0, 3, 6, 8, 2, 4]]) This is because the model expects all the sequences to be the same dimension or same length so each sequence has a use, there are no \"holes\" in the sequence.\n",
    "\n",
    "# Important Note\n",
    "We aren't preprocessing the labels/y/output data because unlike X, the y is already in a numerical format that all has the same dimension. \"1 and 0\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Tokenize the text\n",
    "tokenizer = Tokenizer()\n",
    "tokenizer.fit_on_texts(X)\n",
    "\n",
    "# Convert the text to sequences\n",
    "X_seq = tokenizer.texts_to_sequences(X)\n",
    "\n",
    "# pad the sequences so they are all the same dimension\n",
    "X_pad = pad_sequences(X_seq)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now \"visualize\" what the X data might look like the the neural network. We can do this by printing out the first 5 sequences.\n",
    "When we print out the shapes, it just states the length of each sequence. It is to show how padding the sequences makes them all the same dimension unlike keeping them only encoded."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Preview of X_pad: [[   0    0    0 ...  280  243    8]\n",
      " [   0    0    0 ...   47  543   93]\n",
      " [   0    0    0 ...  162   38  496]\n",
      " [   0    0    0 ...    1 1114  455]\n",
      " [   0    0    0 ...    3  510  652]]\n",
      "Preview of X_seq:  [[10, 219, 927, 11, 216, 119, 14, 110, 2, 655, 7567, 2383, 80, 1153, 4243, 13, 619, 8, 9, 3, 2593, 18, 93, 27, 257, 2, 1365, 14878, 3026, 95, 2, 435, 14879, 418, 1023, 10, 1775, 12, 1, 2299, 13, 29, 1, 92, 20, 1, 76, 502, 4, 1, 636, 820, 143, 10, 96, 25, 39, 2015, 53, 3, 2697, 637, 1, 636, 5, 79, 1, 2299, 38, 55, 5, 1, 220, 5, 430, 1, 1169, 18, 93, 132, 21, 39, 79, 53, 3, 1192, 637, 1, 815, 5, 104, 220, 8, 149, 1248, 3551, 5, 12, 10, 311, 34, 10, 948, 5, 39, 3176, 3027, 20, 1, 5845, 3, 180, 902, 1983, 65, 8, 1124, 16, 67, 47, 1983, 65, 8, 1124, 13, 32, 350, 4, 11, 118, 32, 350, 44, 60, 10, 345, 50, 121, 582, 12, 10, 65, 283, 173, 89, 2, 50, 3938, 863, 7040, 18361, 4834, 464, 20, 1240, 4395, 38, 24641, 2127, 15, 153, 1617, 5, 145, 54, 4396, 53, 15, 3, 283, 173, 10, 13, 321, 199, 11, 118, 6, 2, 990, 12667, 4598, 4243, 45, 29, 1, 1660, 3177, 4, 2, 8928, 14880, 24642, 1, 382, 24, 3178, 1449, 18, 100, 14881, 6, 5, 12668, 97, 144, 552, 36, 1, 525, 46, 6, 405, 58, 272, 5, 601, 644, 36, 684, 193, 5, 1, 693, 220, 1366, 15, 2, 68, 4, 24643, 3, 2, 18362, 3, 6580, 168, 983, 4, 11, 11135, 8929, 145, 5, 641, 7, 7, 582, 7, 7, 582, 16, 1, 352, 806, 14882, 24644, 4, 1826, 1036, 5063, 5846, 848, 806, 106, 291, 11, 235, 285, 9851, 565, 8, 1161, 1423, 3, 28, 6, 1881, 82, 49, 6, 28, 64, 70, 1200, 36, 110, 159, 22, 288, 6, 5, 98, 4243, 165, 49, 211, 37, 273, 5, 98, 3803, 9852, 165, 49, 18, 5063, 2384, 9, 119, 15, 385, 276, 140, 8178, 2, 18363, 3552, 8, 806, 18364, 34, 28, 68, 1073, 9, 168, 277, 197, 1826, 1036, 11, 118, 57, 5847, 4, 2, 1144, 18365, 261, 8179, 11136, 1193, 24645, 4243, 123, 1, 411, 15, 2, 24646, 4082, 138, 28, 11137, 3939, 22, 3679, 3, 285, 15, 1, 9853, 20, 1, 1719, 4, 22, 2698, 452, 394, 16, 1826, 1036, 5063, 14883, 1660, 3940, 80, 28, 8930, 1, 5550, 4, 280, 243, 8], [2889, 290, 1565, 704, 3553, 331, 41, 4, 162, 7568, 961, 14884, 782, 1436, 2128, 8931, 2485, 15, 2, 167, 52, 11138, 9, 284, 5, 67, 37, 1, 539, 24, 387, 176, 1483, 4, 11, 480, 3, 9, 88, 26, 205, 47, 543, 5, 94, 43, 33, 122, 1201, 5, 3680, 53, 3, 141, 46, 179, 710, 8, 11, 350, 4, 61, 2633, 33, 24, 24647, 8, 4244, 16, 2, 24648, 4599, 35, 1291, 94, 199, 85, 1, 24649, 5, 864, 47, 656, 247, 6, 29, 44, 10, 471, 1450, 1, 782, 14885, 15, 939, 46, 6, 153, 73, 319, 582, 1, 4244, 2437, 346, 44, 14886, 92, 85, 1, 17, 51, 42, 408, 1934, 132, 33, 24, 387, 4245, 93, 886, 2330, 93, 4245, 173, 93, 886, 2330, 173, 11, 6, 14887, 89, 30, 83, 35, 86, 388, 1, 4600, 187, 3, 9, 259, 8, 61, 2533, 4, 9, 287, 1, 144, 187, 88, 26, 61, 349, 1108, 43, 46, 6, 27, 55, 54, 391, 155, 320, 5, 505, 16, 11, 355, 24650, 8, 1, 98, 249, 975, 66, 33, 245, 63, 25, 2, 412, 4, 147, 2385, 61, 3453, 1506, 4835, 66, 47, 543, 93], [11, 17, 13, 34, 837, 416, 3, 518, 10, 1367, 2129, 1037, 218, 139, 1, 17, 1, 810, 8, 1, 17, 24, 2634, 3, 55, 150, 1, 107, 6, 205, 31, 48, 2090, 9, 6, 100, 212, 4246, 3, 31, 48, 760, 645, 43, 23, 24, 1720, 38, 917, 23, 88, 7041, 1, 500, 8, 1, 17, 3, 26, 752, 18, 43, 23, 24, 162, 38, 1170, 11, 6, 2, 19, 12, 88, 1827, 128, 678, 3, 26, 2817, 15, 29, 1, 3554, 60, 10, 227, 783, 24, 837, 229, 3, 691, 23, 69, 48, 7569, 5, 1, 944, 8, 11, 17, 308, 50, 49, 3, 1, 204, 13, 50, 1222, 80, 9, 13, 37, 33, 69, 273, 5, 1661, 53, 1, 1125, 18, 9, 155, 1260, 31, 29, 20, 2, 2220, 4, 328, 162, 162, 110, 322, 328, 110, 475, 1, 111, 8, 11, 17, 6, 2, 420, 6581, 602, 6, 1024, 8, 203, 18, 15, 11, 241, 4, 17, 8, 48, 4, 1, 52, 633, 135, 40, 111, 13, 1109, 51, 59, 89, 48, 4, 40, 976, 5848, 10, 694, 1038, 41, 1424, 10, 101, 1, 248, 151, 44, 11, 17, 6, 417, 1, 880, 16, 443, 1, 171, 120, 40, 2130, 516, 41, 5, 26, 1, 402, 1, 437, 949, 6, 24651, 82, 40, 558, 516, 438, 3, 437, 10, 101, 12, 13, 950, 413, 3, 82, 24652, 6581, 602, 5849, 69, 29, 40, 5849, 31, 369, 110, 3346, 30, 5288, 12, 13, 176, 645, 5, 34, 5, 24653, 9, 29, 2, 1222, 17, 18, 842, 43, 337, 162, 38, 496], [1, 90, 205, 151, 44, 14888, 881, 7042, 6, 1, 152, 18366, 77, 12669, 14, 873, 8180, 9854, 1, 278, 106, 6, 2, 270, 15, 2, 200, 509, 620, 722, 45, 1008, 8, 2, 1365, 1527, 3, 35, 856, 14889, 5, 444, 8, 14888, 60, 13, 22, 347, 493, 15, 40, 200, 509, 18367, 406, 6, 50, 24654, 59, 6, 24655, 14890, 4397, 1955, 3, 634, 1984, 3, 8181, 15, 9, 3, 2486, 2763, 30, 516, 3, 42, 29, 8, 40, 406, 1, 19, 87, 394, 602, 14891, 4836, 14, 2950, 2, 129, 35, 1003, 40, 51, 40, 521, 2179, 193, 557, 85, 14888, 35, 543, 5, 490, 2, 5551, 8, 493, 3, 35, 984, 40, 186, 29, 1, 56, 6200, 477, 40, 1594, 547, 4, 585, 8, 22, 9855, 602, 6, 1, 1154, 340, 8, 8932, 199, 143, 6582, 16, 22, 166, 15, 1299, 1855, 9856, 3, 18368, 18369, 4836, 2221, 16, 442, 4398, 1882, 4, 544, 3, 1, 2386, 3, 237, 130, 28, 285, 2, 12670, 106, 209, 2, 1401, 129, 18, 4, 265, 28, 160, 84, 205, 3, 7043, 2331, 28, 6, 1, 3681, 24656, 5, 390, 18367, 106, 36, 5064, 246, 7, 7, 873, 8180, 9854, 6, 2, 2818, 1797, 59, 301, 5, 1, 169, 493, 60, 6, 2, 6583, 266, 2, 241, 4, 329, 1544, 14889, 2, 493, 39, 37, 1749, 319, 14, 2950, 553, 39, 14, 28, 6, 8, 2, 92, 39, 37, 250, 319, 40, 121, 397, 6, 14892, 14, 121, 939, 24, 18, 6584, 6201, 3, 12671, 31, 210, 100, 22, 334, 341, 5, 14893, 3, 51, 28, 3804, 84, 28, 1566, 5552, 4601, 5, 14893, 28, 274, 5, 369, 3, 8180, 9854, 1003, 811, 3, 1402, 3, 24657, 46, 6, 2, 295, 4, 266, 8, 1, 19, 55, 150, 1, 266, 6, 8, 2, 295, 1749, 83, 1113, 8, 1, 693, 7044, 3, 303, 732, 280, 3, 8180, 24658, 14889, 3805, 6, 1211, 2050, 6, 113, 64, 24659, 130, 246, 36, 1, 191, 484, 3, 91, 8182, 8180, 9854, 190, 21, 5, 1025, 1, 3179, 59, 6, 8, 7, 7, 133, 372, 543, 3, 8180, 9854, 156, 3454, 849, 15, 9, 8, 1, 115, 597, 92, 18, 9, 543, 3, 59, 189, 406, 1, 2819, 18, 59, 182, 59, 274, 5, 1368, 2, 9857, 6, 1012, 18, 211, 58, 12672, 796, 59, 453, 34, 14894, 59, 24660, 3, 274, 5, 2, 1721, 173, 1484, 2387, 32, 9858, 45, 75, 14895, 40, 5, 7045, 59, 596, 2635, 3, 733, 15, 11, 18, 51, 59, 1045, 21, 63, 5, 2951, 1, 9857, 18, 5, 141, 5, 1, 1261, 5, 399, 84, 34, 12, 568, 6, 376, 4, 18370, 3, 9, 4602, 40, 29, 123, 173, 59, 453, 12673, 3, 1984, 3, 59, 58, 1110, 596, 12674, 8, 2300, 3, 9, 221, 409, 72, 12, 7, 7, 18366, 77, 12669, 410, 40, 29, 8, 11, 536, 6202, 3, 24661, 214, 873, 8933, 24662, 195, 26, 2, 50, 49, 164, 43, 32, 291, 4, 1, 9859, 4, 602, 14891, 4836, 11139, 3267, 8183, 16, 84, 12, 6, 1009, 2180, 5, 1300, 18371, 4, 1202, 8934, 4399, 81, 267, 108, 1353, 2181, 4083, 24663, 1856, 3108, 3, 24664, 3553, 25, 4603, 84, 41, 14, 27, 4, 1, 918, 2699, 4, 22, 9860, 1114, 1249, 8184, 18, 11, 6, 21, 14, 1134, 2, 19, 14, 154, 4, 76, 3028, 827, 620, 166, 198, 109, 134, 14, 18372, 4247, 12675, 18368, 18369, 4836, 3, 1, 18373, 209, 18374, 4400, 1299, 1855, 9856, 9, 195, 945, 940, 14, 18371, 553, 14, 2, 241, 4, 18375, 1985, 156, 444, 130, 1292, 3, 93, 6585, 3, 197, 2091, 516, 85, 133, 4, 2, 660, 3, 48, 56, 95, 12, 2, 730, 24665, 1857, 8, 392, 1956, 18, 12, 5850, 21, 63, 190, 1618, 3, 24666, 1, 19, 14896, 3, 2131, 91, 6586, 1858, 1, 124, 3, 93, 326, 18376, 41, 15, 58, 295, 4, 32, 300, 46, 24, 87, 4248, 8, 1, 201, 8180, 9854, 306, 5851, 4837, 15, 40, 509, 3, 160, 74, 2534, 29, 364, 43, 59, 6, 8185, 16, 2222, 37, 3555, 8, 1545, 8186, 24667, 3555, 1301, 60, 227, 1339, 40, 6587, 3, 3941, 2534, 12, 215, 133, 12, 6, 2764, 1546, 11, 6, 32, 205, 19, 448, 2, 1194, 27, 18, 27, 12, 856, 27, 8935, 3, 18377, 95, 1485, 27, 139, 32, 985, 24668, 7, 7, 32, 3029, 6588, 4, 1, 169, 753, 19, 1486, 1320, 31, 6589, 1859, 24669, 1321, 12, 45, 229, 199, 30, 3028, 977, 8, 1, 1114, 455], [51, 10, 81, 358, 44, 5289, 238, 18378, 10, 155, 548, 73, 18, 10, 192, 15, 1, 199, 83, 1, 199, 1883, 1, 199, 204, 3, 1860, 23, 96, 31, 224, 98, 2, 6590, 17, 44, 1, 3109, 5289, 303, 190, 5, 26, 767, 44, 18, 24670, 1126, 42, 34, 4084, 24671, 42, 2, 655, 24672, 4, 47, 9, 6, 157, 20, 8, 18379, 34, 373, 126, 4, 265, 42, 21, 29, 44, 24673, 18, 23, 140, 548, 52, 36, 2, 17, 211, 110, 3455, 14, 1, 3556, 5289, 17, 7, 7, 3, 1046, 36, 29, 1, 2594, 480, 9, 64, 6, 2, 74, 17, 1507, 111, 3, 2, 252, 355, 107, 4838, 48, 4, 1, 152, 25, 2016, 8, 76, 99, 12, 33, 24, 158, 177, 1024, 34, 9, 64, 189, 26, 329, 880, 1884, 7, 7, 86, 1212, 142, 5289, 238, 18378, 910, 23, 24, 2952, 217, 36, 48, 445, 493, 8, 986, 2595, 3, 174, 5, 832, 5, 1, 191, 484, 95, 23, 1828, 369, 93, 23, 227, 158, 161, 9, 705, 3, 510, 652]]\n",
      "Shape of X_seq 1 and X_seq 2: 387 197\n",
      "Shape of X_pad 1 and X_pad 2: 2156 2156\n"
     ]
    }
   ],
   "source": [
    "print(\"Preview of X_pad:\", X_pad[:5])\n",
    "print('Preview of X_seq: ', X_seq[:5])\n",
    "# This is to show why we pad the sequences\n",
    "print(\"Shape of X_seq 1 and X_seq 2:\", len(X_seq[0]), len(X_seq[1]))\n",
    "print(\"Shape of X_pad 1 and X_pad 2:\", len(X_pad[0]), len(X_pad[1]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we will split our data into train and test sets. This is so that we have sufficient data to train the model and evaluate it's performance on unseen data. We will use 80% of the data for training and 20% for testing as shown by the parameter `test_size` which takes 20% of the data as test data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X_pad, y, test_size=0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define the model\n",
    "We will now be defining the model with the following architecture and layers.\n",
    "In a nutshell, here is the job of each layer:\n",
    "- **Embedding Layer:** This layer takes in an integer matrix of size (input_dim, output_dim) as input and produces an output matrix of size (input_dim, output_dim) as output. This layer is used to learn word vectors.\n",
    "- **GlobalAveragePooling1D:** This layer takes in a list of vectors and returns a vector with the average of the list of vectors.\n",
    "- **Dropout:** This layer randomly sets input units to 0 with a frequency of rate at each step during training time, which helps prevent overfitting.\n",
    "- **Dense:** This layer has 32 units which are used to compute an output and uses the relu activation function.\n",
    "- **Dense:** This layer has 16 units which are used to compute an output and uses the relu activation function.\n",
    "- **Dense:** This layer has 1 unit which is used to compute an output and uses the sigmoid activation function to output a value between 0 and 1 or the probability of the input being true/positive.\n",
    "\n",
    "## Activation functions in a nutshell\n",
    "- **Relu:** Rectified Linear Unit | This is used to find relationships in the data that are not linear\n",
    "- **Sigmoid:** Sigmoid Activation Function | This is used to calculate the probability of the input being true/positive being from 0 to 1. (Eg. 0.7 output is 70% probability of the input being true/positive)\n",
    "\n",
    "## Units or the first hyperparameter of the layers\n",
    "The first hyperparameter you see after each layer is the unit size or amount of neurons in place. The more you have, the more complex relationships your model can learn. This also means that more training time.\n",
    "\n",
    "- The first layer known as the input layer is used to take in the input data. In this case, it is the embedding layer. It's job is to learn the word vectors and pass it to the layers that learn patters.\n",
    "- The first actual \"learning\" layer is the first Dense layer. This will learn the first essential relationships. \n",
    "- This is next passed down to the second Dense layer. These are known as **hidden layers** which are used to learn more complex relationships. \n",
    "- The final layer known as the **output layer** is used to output a value between 0 and 1 or the probability of the input being true/positive."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.Sequential([\n",
    "    layers.Embedding(input_dim=len(tokenizer.word_index) + 1, output_dim=32),\n",
    "    layers.GlobalAveragePooling1D(),\n",
    "    layers.Dropout(0.2),\n",
    "    layers.Dense(32, activation='relu'),\n",
    "    layers.Dense(16, activation='relu'),\n",
    "    layers.Dense(1, activation='sigmoid')\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "At this point, we will now compile the model. This means that we will specify the loss function, the optimizer, and the metrics we are interested in evaluating the model.\n",
    "What we will use:\n",
    "- **Loss:** Binary-Crossentropy\n",
    "- **Optimizer:** Adam\n",
    "- **Metrics:** Accuracy\n",
    "\n",
    "# What you should know\n",
    "- **Loss functions: ** These are used to calculate the loss of a model or how good the model is performing. The lower it is, the better. Don't let it get too low or the model will start overfitting.\n",
    "- **Optimizer:** This is used to adjust the weights of the model in order to minimize the loss. Weights are how the model learns by adjusting the values of the parameters and how much strength or \"weight\" each parameter has.\n",
    "- **Metrics:** These are used to evaluate the performance of the model. It shows an easier way to read loss."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we will train the model on our data. `X_train` and `y_train`\n",
    "We will be using 10 epochs to train the model which means that we will be training the model for 10 iterations with a default batch size of 32.\n",
    "You will see that as we train it, the **loss** decreases and the **accuracy** increases as it learns and get's better."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "125/125 [==============================] - 3s 18ms/step - loss: 0.6933 - accuracy: 0.4935\n",
      "Epoch 2/10\n",
      "125/125 [==============================] - 2s 17ms/step - loss: 0.6932 - accuracy: 0.5148\n",
      "Epoch 3/10\n",
      "125/125 [==============================] - 2s 17ms/step - loss: 0.6916 - accuracy: 0.5275\n",
      "Epoch 4/10\n",
      "125/125 [==============================] - 2s 17ms/step - loss: 0.6821 - accuracy: 0.6025\n",
      "Epoch 5/10\n",
      "125/125 [==============================] - 2s 16ms/step - loss: 0.6131 - accuracy: 0.7115\n",
      "Epoch 6/10\n",
      "125/125 [==============================] - 2s 17ms/step - loss: 0.4643 - accuracy: 0.7970\n",
      "Epoch 7/10\n",
      "125/125 [==============================] - 2s 17ms/step - loss: 0.3366 - accuracy: 0.8702\n",
      "Epoch 8/10\n",
      "125/125 [==============================] - 2s 17ms/step - loss: 0.2428 - accuracy: 0.9162\n",
      "Epoch 9/10\n",
      "125/125 [==============================] - 2s 16ms/step - loss: 0.1932 - accuracy: 0.9325\n",
      "Epoch 10/10\n",
      "125/125 [==============================] - 2s 16ms/step - loss: 0.1600 - accuracy: 0.9455\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X_train, y_train, epochs=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets now save our model. This way, we don't have to retrain the model every time we want to use it.\n",
    "Our file will be called `model.keras`\n",
    "\n",
    "You can check `runsaved.py` to see how to load and run the model!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('model.keras')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can evalute how well the model performs on unseen data. This will run the test data through the model and compare the results to the actual values.\n",
    "As you can see, this will probably output somewhere around 84% accuracy or 0.84 in decimal."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "32/32 [==============================] - 0s 3ms/step - loss: 0.3403 - accuracy: 0.8640\n",
      "Loss: 0.340265691280365\n",
      "Accuracy: 0.8640000224113464\n"
     ]
    }
   ],
   "source": [
    "loss, accuracy = model.evaluate(X_test, y_test)\n",
    "print(\"Loss:\", loss)\n",
    "print(\"Accuracy:\", accuracy)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see, our accuracy is 84%, but it was 95% when training. This might be caused by overfitting. You can add a validation set to avoid this.\n",
    "Now lets graph the loss and accuracy of our model when it was training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'Epoch')"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# This will be our training loss\n",
    "plt.plot(history.history['loss'])\n",
    "plt.title('Model loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'Epoch')"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# This will be our accuracy\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.title('Model accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see, as we trained the model, the loss decreased and the accuracy increased. This is a good sign.\n",
    "Lets now try getting the AI to predict whether a sentence is positive or negative from a user input.\n",
    "We also want to predict on test data just to see how the outputs might looks like.\n",
    "\n",
    "# How to tell what your AI friend is saying\n",
    "- **0.00:** It is 0% confident that the sentence is positive\n",
    "- **0.50:** It is 50% confident that the sentence is positive\n",
    "- **1.00:** It is 100% confident that the sentence is positive"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 1/32 [..............................] - ETA: 0s"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "32/32 [==============================] - 0s 3ms/step\n",
      "[[0.00615509]\n",
      " [0.00586868]\n",
      " [0.00681347]\n",
      " [0.276614  ]\n",
      " [0.41824052]]\n",
      "User sentence: This HPD1155 guy is awesome! He makes everything so cool and great. His walkthrough was great too!\n",
      "1/1 [==============================] - 0s 13ms/step\n",
      "The probability of the sentence being positive is:  1.0000\n"
     ]
    }
   ],
   "source": [
    "# Predict on test data\n",
    "prediction = model.predict(X_test)\n",
    "print(prediction[:5])\n",
    "\n",
    "# New data\n",
    "raw_input = input(\"Enter a sentence: \")\n",
    "print('User sentence:', raw_input)\n",
    "encoded_input = tokenizer.texts_to_sequences([raw_input])\n",
    "padded_input = pad_sequences(encoded_input)\n",
    "\n",
    "prediction = model.predict(padded_input)\n",
    "print(\"The probability of the sentence being positive is: \", \"{:.{precision}f}\".format(float(prediction[0][0]), precision=4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "That is it for this project. Thank you guys!!\n",
    "\n",
    "# Contact\n",
    "You can contact me through discord @encrypted.dll"
   ]
  }
 ],
 "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": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}