tf_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Review the Basics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Input Neurons : Number of features used by a neural network to make predictions\n",
    "- Hidden Layers : Application of weightings to inputs to provide desired outputs\n",
    "- Output Neurons : Predictions the neural network makes\n",
    "\n",
    "Using numbers to predict another number\n",
    "\n",
    "Example : Does the size, # of bathrooms / bedrooms, etc. effect the value of a house? Based on those variables can we make predictions about the price?\n",
    "\n",
    "In Machine Learning algorithms are trained to understand relationships between features and make predictions about labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## What is a Tensor?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Numerical representation of non-numerical data. Pink has no meaning as input, but .1 Red, .075 Green, and .079 Blue does if represented as [.1, .075, .079]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How do we Improve Predicitions / Models?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Fit on more data\n",
    "- Increase hidden units : Add weightings to get desired output\n",
    "- Try different activation functions : Helps decide what data is most important\n",
    "- Try different optimization functions : Optimize by smoothing data to expected results\n",
    "- Increase learning rate : How much to change the model based on estimated error\n",
    "- Fit longer : Increase the number of times we try to fit the data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Binary & Multi-class Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Tensorflow Binary Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Binary classification refers to something being one thing or another.\n",
    "\n",
    "| Hyperparameter | Binary Classification |\n",
    "| :---: | :---: |\n",
    "| Input Layer | Shape == number of features |\n",
    "| Hidden Layers | Minimum of 1 to Unlimited |\n",
    "| Neurons in Hidden Layer | Normally 10 to 100 |\n",
    "| Output layer | 1 |\n",
    "| Hidden activation | Usually Rectified Linear Unit (ReLU) |\n",
    "| Output activation | Sigmoid |\n",
    "| Loss function | BinaryCrossentropy |\n",
    "| Optimizer | Stochastic Gradient Descent, Adam |"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "# Used to build and deploy machine learning apps\n",
    "import tensorflow as tf\n",
    "\n",
    "# Deep Learning API for creating Neural Networks (Runs on TensorFlow)\n",
    "from tensorflow import keras \n",
    "from tensorflow.keras import layers\n",
    "from tensorflow import math\n",
    "from keras.utils.vis_utils import plot_model\n",
    "\n",
    "# Scikit-learn : machine learning library\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.compose import make_column_transformer\n",
    "from sklearn.preprocessing import MinMaxScaler, OneHotEncoder"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Get Wine Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Location https://www.kaggle.com/datasets/shelvigarg/wine-quality-dataset?select=winequalityN.csv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>type</th>\n",
       "      <th>fixed acidity</th>\n",
       "      <th>volatile acidity</th>\n",
       "      <th>citric acid</th>\n",
       "      <th>residual sugar</th>\n",
       "      <th>chlorides</th>\n",
       "      <th>free sulfur dioxide</th>\n",
       "      <th>total sulfur dioxide</th>\n",
       "      <th>density</th>\n",
       "      <th>pH</th>\n",
       "      <th>sulphates</th>\n",
       "      <th>alcohol</th>\n",
       "      <th>quality</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>white</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.27</td>\n",
       "      <td>0.36</td>\n",
       "      <td>20.7</td>\n",
       "      <td>0.045</td>\n",
       "      <td>45.0</td>\n",
       "      <td>170.0</td>\n",
       "      <td>1.0010</td>\n",
       "      <td>3.00</td>\n",
       "      <td>0.45</td>\n",
       "      <td>8.8</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>white</td>\n",
       "      <td>6.3</td>\n",
       "      <td>0.30</td>\n",
       "      <td>0.34</td>\n",
       "      <td>1.6</td>\n",
       "      <td>0.049</td>\n",
       "      <td>14.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>0.9940</td>\n",
       "      <td>3.30</td>\n",
       "      <td>0.49</td>\n",
       "      <td>9.5</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>white</td>\n",
       "      <td>8.1</td>\n",
       "      <td>0.28</td>\n",
       "      <td>0.40</td>\n",
       "      <td>6.9</td>\n",
       "      <td>0.050</td>\n",
       "      <td>30.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>0.9951</td>\n",
       "      <td>3.26</td>\n",
       "      <td>0.44</td>\n",
       "      <td>10.1</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>white</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.23</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.9956</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>white</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.23</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.9956</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    type  fixed acidity  volatile acidity  citric acid  residual sugar  \\\n",
       "0  white            7.0              0.27         0.36            20.7   \n",
       "1  white            6.3              0.30         0.34             1.6   \n",
       "2  white            8.1              0.28         0.40             6.9   \n",
       "3  white            7.2              0.23         0.32             8.5   \n",
       "4  white            7.2              0.23         0.32             8.5   \n",
       "\n",
       "   chlorides  free sulfur dioxide  total sulfur dioxide  density    pH  \\\n",
       "0      0.045                 45.0                 170.0   1.0010  3.00   \n",
       "1      0.049                 14.0                 132.0   0.9940  3.30   \n",
       "2      0.050                 30.0                  97.0   0.9951  3.26   \n",
       "3      0.058                 47.0                 186.0   0.9956  3.19   \n",
       "4      0.058                 47.0                 186.0   0.9956  3.19   \n",
       "\n",
       "   sulphates  alcohol  quality  \n",
       "0       0.45      8.8        6  \n",
       "1       0.49      9.5        6  \n",
       "2       0.44     10.1        6  \n",
       "3       0.40      9.9        6  \n",
       "4       0.40      9.9        6  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "wine_data = pd.read_csv(\"https://raw.githubusercontent.com/derekbanas/tensorflow/main/winequalityN.csv\")\n",
    "wine_data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Delete Rows with Unknowns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "type                     0\n",
       "fixed acidity           10\n",
       "volatile acidity         8\n",
       "citric acid              3\n",
       "residual sugar           2\n",
       "chlorides                2\n",
       "free sulfur dioxide      0\n",
       "total sulfur dioxide     0\n",
       "density                  0\n",
       "pH                       9\n",
       "sulphates                4\n",
       "alcohol                  0\n",
       "quality                  0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check for nans in columns\n",
    "wine_data.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "type                    0\n",
       "fixed acidity           0\n",
       "volatile acidity        0\n",
       "citric acid             0\n",
       "residual sugar          0\n",
       "chlorides               0\n",
       "free sulfur dioxide     0\n",
       "total sulfur dioxide    0\n",
       "density                 0\n",
       "pH                      0\n",
       "sulphates               0\n",
       "alcohol                 0\n",
       "quality                 0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# This deletes all of them\n",
    "wine_data = wine_data.dropna()\n",
    "wine_data.isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Rename Column"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>white_wine</th>\n",
       "      <th>fixed acidity</th>\n",
       "      <th>volatile acidity</th>\n",
       "      <th>citric acid</th>\n",
       "      <th>residual sugar</th>\n",
       "      <th>chlorides</th>\n",
       "      <th>free sulfur dioxide</th>\n",
       "      <th>total sulfur dioxide</th>\n",
       "      <th>density</th>\n",
       "      <th>pH</th>\n",
       "      <th>sulphates</th>\n",
       "      <th>alcohol</th>\n",
       "      <th>quality</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>white</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.270</td>\n",
       "      <td>0.36</td>\n",
       "      <td>20.7</td>\n",
       "      <td>0.045</td>\n",
       "      <td>45.0</td>\n",
       "      <td>170.0</td>\n",
       "      <td>1.00100</td>\n",
       "      <td>3.00</td>\n",
       "      <td>0.45</td>\n",
       "      <td>8.8</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>white</td>\n",
       "      <td>6.3</td>\n",
       "      <td>0.300</td>\n",
       "      <td>0.34</td>\n",
       "      <td>1.6</td>\n",
       "      <td>0.049</td>\n",
       "      <td>14.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>0.99400</td>\n",
       "      <td>3.30</td>\n",
       "      <td>0.49</td>\n",
       "      <td>9.5</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>white</td>\n",
       "      <td>8.1</td>\n",
       "      <td>0.280</td>\n",
       "      <td>0.40</td>\n",
       "      <td>6.9</td>\n",
       "      <td>0.050</td>\n",
       "      <td>30.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>0.99510</td>\n",
       "      <td>3.26</td>\n",
       "      <td>0.44</td>\n",
       "      <td>10.1</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>white</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.230</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.99560</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>white</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.230</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.99560</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6491</th>\n",
       "      <td>red</td>\n",
       "      <td>6.8</td>\n",
       "      <td>0.620</td>\n",
       "      <td>0.08</td>\n",
       "      <td>1.9</td>\n",
       "      <td>0.068</td>\n",
       "      <td>28.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.99651</td>\n",
       "      <td>3.42</td>\n",
       "      <td>0.82</td>\n",
       "      <td>9.5</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6492</th>\n",
       "      <td>red</td>\n",
       "      <td>6.2</td>\n",
       "      <td>0.600</td>\n",
       "      <td>0.08</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.090</td>\n",
       "      <td>32.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0.99490</td>\n",
       "      <td>3.45</td>\n",
       "      <td>0.58</td>\n",
       "      <td>10.5</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6494</th>\n",
       "      <td>red</td>\n",
       "      <td>6.3</td>\n",
       "      <td>0.510</td>\n",
       "      <td>0.13</td>\n",
       "      <td>2.3</td>\n",
       "      <td>0.076</td>\n",
       "      <td>29.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>0.99574</td>\n",
       "      <td>3.42</td>\n",
       "      <td>0.75</td>\n",
       "      <td>11.0</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6495</th>\n",
       "      <td>red</td>\n",
       "      <td>5.9</td>\n",
       "      <td>0.645</td>\n",
       "      <td>0.12</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.075</td>\n",
       "      <td>32.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0.99547</td>\n",
       "      <td>3.57</td>\n",
       "      <td>0.71</td>\n",
       "      <td>10.2</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6496</th>\n",
       "      <td>red</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.310</td>\n",
       "      <td>0.47</td>\n",
       "      <td>3.6</td>\n",
       "      <td>0.067</td>\n",
       "      <td>18.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>0.99549</td>\n",
       "      <td>3.39</td>\n",
       "      <td>0.66</td>\n",
       "      <td>11.0</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6463 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     white_wine  fixed acidity  volatile acidity  citric acid  residual sugar  \\\n",
       "0         white            7.0             0.270         0.36            20.7   \n",
       "1         white            6.3             0.300         0.34             1.6   \n",
       "2         white            8.1             0.280         0.40             6.9   \n",
       "3         white            7.2             0.230         0.32             8.5   \n",
       "4         white            7.2             0.230         0.32             8.5   \n",
       "...         ...            ...               ...          ...             ...   \n",
       "6491        red            6.8             0.620         0.08             1.9   \n",
       "6492        red            6.2             0.600         0.08             2.0   \n",
       "6494        red            6.3             0.510         0.13             2.3   \n",
       "6495        red            5.9             0.645         0.12             2.0   \n",
       "6496        red            6.0             0.310         0.47             3.6   \n",
       "\n",
       "      chlorides  free sulfur dioxide  total sulfur dioxide  density    pH  \\\n",
       "0         0.045                 45.0                 170.0  1.00100  3.00   \n",
       "1         0.049                 14.0                 132.0  0.99400  3.30   \n",
       "2         0.050                 30.0                  97.0  0.99510  3.26   \n",
       "3         0.058                 47.0                 186.0  0.99560  3.19   \n",
       "4         0.058                 47.0                 186.0  0.99560  3.19   \n",
       "...         ...                  ...                   ...      ...   ...   \n",
       "6491      0.068                 28.0                  38.0  0.99651  3.42   \n",
       "6492      0.090                 32.0                  44.0  0.99490  3.45   \n",
       "6494      0.076                 29.0                  40.0  0.99574  3.42   \n",
       "6495      0.075                 32.0                  44.0  0.99547  3.57   \n",
       "6496      0.067                 18.0                  42.0  0.99549  3.39   \n",
       "\n",
       "      sulphates  alcohol  quality  \n",
       "0          0.45      8.8        6  \n",
       "1          0.49      9.5        6  \n",
       "2          0.44     10.1        6  \n",
       "3          0.40      9.9        6  \n",
       "4          0.40      9.9        6  \n",
       "...         ...      ...      ...  \n",
       "6491       0.82      9.5        6  \n",
       "6492       0.58     10.5        5  \n",
       "6494       0.75     11.0        6  \n",
       "6495       0.71     10.2        5  \n",
       "6496       0.66     11.0        6  \n",
       "\n",
       "[6463 rows x 13 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "wine_data.rename(columns={\"type\": \"white_wine\"},\n",
    "                inplace=True)\n",
    "wine_data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Change white_wine Based on condition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>white_wine</th>\n",
       "      <th>fixed acidity</th>\n",
       "      <th>volatile acidity</th>\n",
       "      <th>citric acid</th>\n",
       "      <th>residual sugar</th>\n",
       "      <th>chlorides</th>\n",
       "      <th>free sulfur dioxide</th>\n",
       "      <th>total sulfur dioxide</th>\n",
       "      <th>density</th>\n",
       "      <th>pH</th>\n",
       "      <th>sulphates</th>\n",
       "      <th>alcohol</th>\n",
       "      <th>quality</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.270</td>\n",
       "      <td>0.36</td>\n",
       "      <td>20.7</td>\n",
       "      <td>0.045</td>\n",
       "      <td>45.0</td>\n",
       "      <td>170.0</td>\n",
       "      <td>1.00100</td>\n",
       "      <td>3.00</td>\n",
       "      <td>0.45</td>\n",
       "      <td>8.8</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>6.3</td>\n",
       "      <td>0.300</td>\n",
       "      <td>0.34</td>\n",
       "      <td>1.6</td>\n",
       "      <td>0.049</td>\n",
       "      <td>14.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>0.99400</td>\n",
       "      <td>3.30</td>\n",
       "      <td>0.49</td>\n",
       "      <td>9.5</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>8.1</td>\n",
       "      <td>0.280</td>\n",
       "      <td>0.40</td>\n",
       "      <td>6.9</td>\n",
       "      <td>0.050</td>\n",
       "      <td>30.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>0.99510</td>\n",
       "      <td>3.26</td>\n",
       "      <td>0.44</td>\n",
       "      <td>10.1</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.230</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.99560</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.230</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.99560</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6491</th>\n",
       "      <td>0</td>\n",
       "      <td>6.8</td>\n",
       "      <td>0.620</td>\n",
       "      <td>0.08</td>\n",
       "      <td>1.9</td>\n",
       "      <td>0.068</td>\n",
       "      <td>28.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.99651</td>\n",
       "      <td>3.42</td>\n",
       "      <td>0.82</td>\n",
       "      <td>9.5</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6492</th>\n",
       "      <td>0</td>\n",
       "      <td>6.2</td>\n",
       "      <td>0.600</td>\n",
       "      <td>0.08</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.090</td>\n",
       "      <td>32.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0.99490</td>\n",
       "      <td>3.45</td>\n",
       "      <td>0.58</td>\n",
       "      <td>10.5</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6494</th>\n",
       "      <td>0</td>\n",
       "      <td>6.3</td>\n",
       "      <td>0.510</td>\n",
       "      <td>0.13</td>\n",
       "      <td>2.3</td>\n",
       "      <td>0.076</td>\n",
       "      <td>29.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>0.99574</td>\n",
       "      <td>3.42</td>\n",
       "      <td>0.75</td>\n",
       "      <td>11.0</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6495</th>\n",
       "      <td>0</td>\n",
       "      <td>5.9</td>\n",
       "      <td>0.645</td>\n",
       "      <td>0.12</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.075</td>\n",
       "      <td>32.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0.99547</td>\n",
       "      <td>3.57</td>\n",
       "      <td>0.71</td>\n",
       "      <td>10.2</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6496</th>\n",
       "      <td>0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.310</td>\n",
       "      <td>0.47</td>\n",
       "      <td>3.6</td>\n",
       "      <td>0.067</td>\n",
       "      <td>18.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>0.99549</td>\n",
       "      <td>3.39</td>\n",
       "      <td>0.66</td>\n",
       "      <td>11.0</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6463 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     white_wine  fixed acidity  volatile acidity  citric acid  residual sugar  \\\n",
       "0             1            7.0             0.270         0.36            20.7   \n",
       "1             1            6.3             0.300         0.34             1.6   \n",
       "2             1            8.1             0.280         0.40             6.9   \n",
       "3             1            7.2             0.230         0.32             8.5   \n",
       "4             1            7.2             0.230         0.32             8.5   \n",
       "...         ...            ...               ...          ...             ...   \n",
       "6491          0            6.8             0.620         0.08             1.9   \n",
       "6492          0            6.2             0.600         0.08             2.0   \n",
       "6494          0            6.3             0.510         0.13             2.3   \n",
       "6495          0            5.9             0.645         0.12             2.0   \n",
       "6496          0            6.0             0.310         0.47             3.6   \n",
       "\n",
       "      chlorides  free sulfur dioxide  total sulfur dioxide  density    pH  \\\n",
       "0         0.045                 45.0                 170.0  1.00100  3.00   \n",
       "1         0.049                 14.0                 132.0  0.99400  3.30   \n",
       "2         0.050                 30.0                  97.0  0.99510  3.26   \n",
       "3         0.058                 47.0                 186.0  0.99560  3.19   \n",
       "4         0.058                 47.0                 186.0  0.99560  3.19   \n",
       "...         ...                  ...                   ...      ...   ...   \n",
       "6491      0.068                 28.0                  38.0  0.99651  3.42   \n",
       "6492      0.090                 32.0                  44.0  0.99490  3.45   \n",
       "6494      0.076                 29.0                  40.0  0.99574  3.42   \n",
       "6495      0.075                 32.0                  44.0  0.99547  3.57   \n",
       "6496      0.067                 18.0                  42.0  0.99549  3.39   \n",
       "\n",
       "      sulphates  alcohol  quality  \n",
       "0          0.45      8.8        6  \n",
       "1          0.49      9.5        6  \n",
       "2          0.44     10.1        6  \n",
       "3          0.40      9.9        6  \n",
       "4          0.40      9.9        6  \n",
       "...         ...      ...      ...  \n",
       "6491       0.82      9.5        6  \n",
       "6492       0.58     10.5        5  \n",
       "6494       0.75     11.0        6  \n",
       "6495       0.71     10.2        5  \n",
       "6496       0.66     11.0        6  \n",
       "\n",
       "[6463 rows x 13 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "wine_data.loc[wine_data[\"white_wine\"] == \"white\", \"white_wine\"] = 1\n",
    "wine_data.loc[wine_data[\"white_wine\"] == \"red\", \"white_wine\"] = 0\n",
    "wine_data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Change Quality Based on Condition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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>white_wine</th>\n",
       "      <th>fixed acidity</th>\n",
       "      <th>volatile acidity</th>\n",
       "      <th>citric acid</th>\n",
       "      <th>residual sugar</th>\n",
       "      <th>chlorides</th>\n",
       "      <th>free sulfur dioxide</th>\n",
       "      <th>total sulfur dioxide</th>\n",
       "      <th>density</th>\n",
       "      <th>pH</th>\n",
       "      <th>sulphates</th>\n",
       "      <th>alcohol</th>\n",
       "      <th>quality</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.270</td>\n",
       "      <td>0.36</td>\n",
       "      <td>20.7</td>\n",
       "      <td>0.045</td>\n",
       "      <td>45.0</td>\n",
       "      <td>170.0</td>\n",
       "      <td>1.00100</td>\n",
       "      <td>3.00</td>\n",
       "      <td>0.45</td>\n",
       "      <td>8.8</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>6.3</td>\n",
       "      <td>0.300</td>\n",
       "      <td>0.34</td>\n",
       "      <td>1.6</td>\n",
       "      <td>0.049</td>\n",
       "      <td>14.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>0.99400</td>\n",
       "      <td>3.30</td>\n",
       "      <td>0.49</td>\n",
       "      <td>9.5</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>8.1</td>\n",
       "      <td>0.280</td>\n",
       "      <td>0.40</td>\n",
       "      <td>6.9</td>\n",
       "      <td>0.050</td>\n",
       "      <td>30.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>0.99510</td>\n",
       "      <td>3.26</td>\n",
       "      <td>0.44</td>\n",
       "      <td>10.1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.230</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.99560</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.230</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.99560</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6491</th>\n",
       "      <td>0</td>\n",
       "      <td>6.8</td>\n",
       "      <td>0.620</td>\n",
       "      <td>0.08</td>\n",
       "      <td>1.9</td>\n",
       "      <td>0.068</td>\n",
       "      <td>28.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.99651</td>\n",
       "      <td>3.42</td>\n",
       "      <td>0.82</td>\n",
       "      <td>9.5</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6492</th>\n",
       "      <td>0</td>\n",
       "      <td>6.2</td>\n",
       "      <td>0.600</td>\n",
       "      <td>0.08</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.090</td>\n",
       "      <td>32.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0.99490</td>\n",
       "      <td>3.45</td>\n",
       "      <td>0.58</td>\n",
       "      <td>10.5</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6494</th>\n",
       "      <td>0</td>\n",
       "      <td>6.3</td>\n",
       "      <td>0.510</td>\n",
       "      <td>0.13</td>\n",
       "      <td>2.3</td>\n",
       "      <td>0.076</td>\n",
       "      <td>29.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>0.99574</td>\n",
       "      <td>3.42</td>\n",
       "      <td>0.75</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6495</th>\n",
       "      <td>0</td>\n",
       "      <td>5.9</td>\n",
       "      <td>0.645</td>\n",
       "      <td>0.12</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.075</td>\n",
       "      <td>32.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0.99547</td>\n",
       "      <td>3.57</td>\n",
       "      <td>0.71</td>\n",
       "      <td>10.2</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6496</th>\n",
       "      <td>0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.310</td>\n",
       "      <td>0.47</td>\n",
       "      <td>3.6</td>\n",
       "      <td>0.067</td>\n",
       "      <td>18.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>0.99549</td>\n",
       "      <td>3.39</td>\n",
       "      <td>0.66</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6463 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     white_wine  fixed acidity  volatile acidity  citric acid  residual sugar  \\\n",
       "0             1            7.0             0.270         0.36            20.7   \n",
       "1             1            6.3             0.300         0.34             1.6   \n",
       "2             1            8.1             0.280         0.40             6.9   \n",
       "3             1            7.2             0.230         0.32             8.5   \n",
       "4             1            7.2             0.230         0.32             8.5   \n",
       "...         ...            ...               ...          ...             ...   \n",
       "6491          0            6.8             0.620         0.08             1.9   \n",
       "6492          0            6.2             0.600         0.08             2.0   \n",
       "6494          0            6.3             0.510         0.13             2.3   \n",
       "6495          0            5.9             0.645         0.12             2.0   \n",
       "6496          0            6.0             0.310         0.47             3.6   \n",
       "\n",
       "      chlorides  free sulfur dioxide  total sulfur dioxide  density    pH  \\\n",
       "0         0.045                 45.0                 170.0  1.00100  3.00   \n",
       "1         0.049                 14.0                 132.0  0.99400  3.30   \n",
       "2         0.050                 30.0                  97.0  0.99510  3.26   \n",
       "3         0.058                 47.0                 186.0  0.99560  3.19   \n",
       "4         0.058                 47.0                 186.0  0.99560  3.19   \n",
       "...         ...                  ...                   ...      ...   ...   \n",
       "6491      0.068                 28.0                  38.0  0.99651  3.42   \n",
       "6492      0.090                 32.0                  44.0  0.99490  3.45   \n",
       "6494      0.076                 29.0                  40.0  0.99574  3.42   \n",
       "6495      0.075                 32.0                  44.0  0.99547  3.57   \n",
       "6496      0.067                 18.0                  42.0  0.99549  3.39   \n",
       "\n",
       "      sulphates  alcohol  quality  \n",
       "0          0.45      8.8        0  \n",
       "1          0.49      9.5        0  \n",
       "2          0.44     10.1        0  \n",
       "3          0.40      9.9        0  \n",
       "4          0.40      9.9        0  \n",
       "...         ...      ...      ...  \n",
       "6491       0.82      9.5        0  \n",
       "6492       0.58     10.5        0  \n",
       "6494       0.75     11.0        0  \n",
       "6495       0.71     10.2        0  \n",
       "6496       0.66     11.0        0  \n",
       "\n",
       "[6463 rows x 13 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "wine_data.loc[wine_data[\"quality\"] <= 6, \"quality\"] = 0\n",
    "wine_data.loc[wine_data[\"quality\"] >= 7, \"quality\"] = 1\n",
    "wine_data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How many Good Wines?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1271"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(wine_data['quality'] == 1).sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Normalize Our Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will convert our non-numeric data into numbers but all of our data is on different scales. We normalize our data so that it is using a common scale. We can do this while also not distorting the differences in the wide range of values. We will convert all values to between 0 and 1 while preserving the original distribution.\n",
    "\n",
    "We do this because machine learning algorithms work better when features have a similar scale."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "ct = make_column_transformer(\n",
    "    (MinMaxScaler(), [\"fixed acidity\", \"volatile acidity\", \"citric acid\",\n",
    "     \"residual sugar\", \"chlorides\", \"free sulfur dioxide\",\n",
    "     \"total sulfur dioxide\", \"density\", \"pH\", \"sulphates\",\n",
    "     \"alcohol\"]), # Normalize these columns\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Separate Features & Labels"
   ]
  },
  {
   "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>white_wine</th>\n",
       "      <th>fixed acidity</th>\n",
       "      <th>volatile acidity</th>\n",
       "      <th>citric acid</th>\n",
       "      <th>residual sugar</th>\n",
       "      <th>chlorides</th>\n",
       "      <th>free sulfur dioxide</th>\n",
       "      <th>total sulfur dioxide</th>\n",
       "      <th>density</th>\n",
       "      <th>pH</th>\n",
       "      <th>sulphates</th>\n",
       "      <th>alcohol</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>7.0</td>\n",
       "      <td>0.270</td>\n",
       "      <td>0.36</td>\n",
       "      <td>20.7</td>\n",
       "      <td>0.045</td>\n",
       "      <td>45.0</td>\n",
       "      <td>170.0</td>\n",
       "      <td>1.00100</td>\n",
       "      <td>3.00</td>\n",
       "      <td>0.45</td>\n",
       "      <td>8.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>6.3</td>\n",
       "      <td>0.300</td>\n",
       "      <td>0.34</td>\n",
       "      <td>1.6</td>\n",
       "      <td>0.049</td>\n",
       "      <td>14.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>0.99400</td>\n",
       "      <td>3.30</td>\n",
       "      <td>0.49</td>\n",
       "      <td>9.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>8.1</td>\n",
       "      <td>0.280</td>\n",
       "      <td>0.40</td>\n",
       "      <td>6.9</td>\n",
       "      <td>0.050</td>\n",
       "      <td>30.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>0.99510</td>\n",
       "      <td>3.26</td>\n",
       "      <td>0.44</td>\n",
       "      <td>10.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.230</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.99560</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>7.2</td>\n",
       "      <td>0.230</td>\n",
       "      <td>0.32</td>\n",
       "      <td>8.5</td>\n",
       "      <td>0.058</td>\n",
       "      <td>47.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.99560</td>\n",
       "      <td>3.19</td>\n",
       "      <td>0.40</td>\n",
       "      <td>9.9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6491</th>\n",
       "      <td>0</td>\n",
       "      <td>6.8</td>\n",
       "      <td>0.620</td>\n",
       "      <td>0.08</td>\n",
       "      <td>1.9</td>\n",
       "      <td>0.068</td>\n",
       "      <td>28.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.99651</td>\n",
       "      <td>3.42</td>\n",
       "      <td>0.82</td>\n",
       "      <td>9.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6492</th>\n",
       "      <td>0</td>\n",
       "      <td>6.2</td>\n",
       "      <td>0.600</td>\n",
       "      <td>0.08</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.090</td>\n",
       "      <td>32.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0.99490</td>\n",
       "      <td>3.45</td>\n",
       "      <td>0.58</td>\n",
       "      <td>10.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6494</th>\n",
       "      <td>0</td>\n",
       "      <td>6.3</td>\n",
       "      <td>0.510</td>\n",
       "      <td>0.13</td>\n",
       "      <td>2.3</td>\n",
       "      <td>0.076</td>\n",
       "      <td>29.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>0.99574</td>\n",
       "      <td>3.42</td>\n",
       "      <td>0.75</td>\n",
       "      <td>11.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6495</th>\n",
       "      <td>0</td>\n",
       "      <td>5.9</td>\n",
       "      <td>0.645</td>\n",
       "      <td>0.12</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.075</td>\n",
       "      <td>32.0</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0.99547</td>\n",
       "      <td>3.57</td>\n",
       "      <td>0.71</td>\n",
       "      <td>10.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6496</th>\n",
       "      <td>0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.310</td>\n",
       "      <td>0.47</td>\n",
       "      <td>3.6</td>\n",
       "      <td>0.067</td>\n",
       "      <td>18.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>0.99549</td>\n",
       "      <td>3.39</td>\n",
       "      <td>0.66</td>\n",
       "      <td>11.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6463 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     white_wine  fixed acidity  volatile acidity  citric acid  residual sugar  \\\n",
       "0             1            7.0             0.270         0.36            20.7   \n",
       "1             1            6.3             0.300         0.34             1.6   \n",
       "2             1            8.1             0.280         0.40             6.9   \n",
       "3             1            7.2             0.230         0.32             8.5   \n",
       "4             1            7.2             0.230         0.32             8.5   \n",
       "...         ...            ...               ...          ...             ...   \n",
       "6491          0            6.8             0.620         0.08             1.9   \n",
       "6492          0            6.2             0.600         0.08             2.0   \n",
       "6494          0            6.3             0.510         0.13             2.3   \n",
       "6495          0            5.9             0.645         0.12             2.0   \n",
       "6496          0            6.0             0.310         0.47             3.6   \n",
       "\n",
       "      chlorides  free sulfur dioxide  total sulfur dioxide  density    pH  \\\n",
       "0         0.045                 45.0                 170.0  1.00100  3.00   \n",
       "1         0.049                 14.0                 132.0  0.99400  3.30   \n",
       "2         0.050                 30.0                  97.0  0.99510  3.26   \n",
       "3         0.058                 47.0                 186.0  0.99560  3.19   \n",
       "4         0.058                 47.0                 186.0  0.99560  3.19   \n",
       "...         ...                  ...                   ...      ...   ...   \n",
       "6491      0.068                 28.0                  38.0  0.99651  3.42   \n",
       "6492      0.090                 32.0                  44.0  0.99490  3.45   \n",
       "6494      0.076                 29.0                  40.0  0.99574  3.42   \n",
       "6495      0.075                 32.0                  44.0  0.99547  3.57   \n",
       "6496      0.067                 18.0                  42.0  0.99549  3.39   \n",
       "\n",
       "      sulphates  alcohol  \n",
       "0          0.45      8.8  \n",
       "1          0.49      9.5  \n",
       "2          0.44     10.1  \n",
       "3          0.40      9.9  \n",
       "4          0.40      9.9  \n",
       "...         ...      ...  \n",
       "6491       0.82      9.5  \n",
       "6492       0.58     10.5  \n",
       "6494       0.75     11.0  \n",
       "6495       0.71     10.2  \n",
       "6496       0.66     11.0  \n",
       "\n",
       "[6463 rows x 12 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = wine_data.drop(\"quality\", axis=1) # Features\n",
    "y = wine_data[\"quality\"] # Label\n",
    "X"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Split into Training & Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5170.400000000001"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get first 80%\n",
    "len(wine_data) * .80"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Build train (80%) & test (20%) datasets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, \n",
    "random_state=66)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Transform Training & Test Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Transform training data \n",
    "ct.fit(X_train)\n",
    "\n",
    "# Transform training and test data with normalization \n",
    "# (MinMaxScalar) \n",
    "X_train_normal = ct.transform(X_train)\n",
    "X_test_normal = ct.transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.21487603, 0.11333333, 0.22289157, ..., 0.34645669, 0.1011236 ,\n",
       "        0.44927536],\n",
       "       [0.23966942, 0.12666667, 0.18674699, ..., 0.40944882, 0.19101124,\n",
       "        0.23188406],\n",
       "       [0.63636364, 0.15333333, 0.30722892, ..., 0.22834646, 0.3988764 ,\n",
       "        0.26086957],\n",
       "       ...,\n",
       "       [0.27272727, 0.12      , 0.20481928, ..., 0.25984252, 0.1741573 ,\n",
       "        0.15942029],\n",
       "       [0.18181818, 0.08      , 0.14457831, ..., 0.4488189 , 0.14044944,\n",
       "        0.50724638],\n",
       "       [0.30578512, 0.07333333, 0.20481928, ..., 0.28346457, 0.15168539,\n",
       "        0.44927536]])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_normal"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## TensorFlow Playground"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Reproduce results from https://playground.tensorflow.org/ using ReLU with code.\n",
    "\n",
    "- Ratio of Training to Test : 80%\n",
    "- Learning Rate : 0.001\n",
    "- Activiation : ReLU\n",
    "- Regularization : None\n",
    "- Regularization Rate : 0\n",
    "- Problem Type : Classification\n",
    "\n",
    "ReLU, Tanh and Sigmoid are for non-linear data\n",
    "When analyzing pictures a neural network draws many straight and non-straight lines around objects in images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x13720c03760>]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create a example tensor with values from -15 to 15\n",
    "samp_ten = tf.cast(tf.range(-15, 15), tf.float32)\n",
    "plt.plot(samp_ten)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x1368108a4d0>]"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Turn negative values to zero and other wise leave them as they are\n",
    "def relu(x):\n",
    "    return tf.maximum(0, x)\n",
    "\n",
    "# Pass toy tensor to our relu function\n",
    "plt.plot(relu(samp_ten))\n",
    "\n",
    "# When we give our model non-linear functions it's able to find patterns\n",
    "# in non-linear data (Neural networks use linear and non-linear activations\n",
    "# to find patterns in data)\n",
    "# Now we can draw patterns that are curved and bent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x13720c73280>]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Replicate sigmoid : sigmoid(x) = 1 / (1 + exp(-x))\n",
    "def sigmoid(x):\n",
    "    return 1 / (1 + tf.exp(-x))\n",
    "\n",
    "# Use the sigmoid function on our toy tensor\n",
    "# This allows use to draw non-straight lines\n",
    "plt.plot(sigmoid(samp_ten))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x13720cde710>]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def tanh(x):\n",
    "    return tf.tanh(x)\n",
    "\n",
    "# Pass toy tensor to our tanh function\n",
    "plt.plot(tanh(samp_ten))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.6386 - accuracy: 0.7749\n",
      "Epoch 2/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5157 - accuracy: 0.8056\n",
      "Epoch 3/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.4901 - accuracy: 0.8056\n",
      "Epoch 4/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4839 - accuracy: 0.8056\n",
      "Epoch 5/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4676 - accuracy: 0.8056\n",
      "Epoch 6/100\n",
      "162/162 [==============================] - 1s 4ms/step - loss: 0.4435 - accuracy: 0.8056\n",
      "Epoch 7/100\n",
      "162/162 [==============================] - 1s 4ms/step - loss: 0.4242 - accuracy: 0.8056\n",
      "Epoch 8/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.4121 - accuracy: 0.8101\n",
      "Epoch 9/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.4062 - accuracy: 0.8151\n",
      "Epoch 10/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.4028 - accuracy: 0.8143\n",
      "Epoch 11/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.4007 - accuracy: 0.8155\n",
      "Epoch 12/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3989 - accuracy: 0.8133\n",
      "Epoch 13/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3978 - accuracy: 0.8120\n",
      "Epoch 14/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3965 - accuracy: 0.8126\n",
      "Epoch 15/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3958 - accuracy: 0.8145\n",
      "Epoch 16/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3947 - accuracy: 0.8172\n",
      "Epoch 17/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3939 - accuracy: 0.8141\n",
      "Epoch 18/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3942 - accuracy: 0.8170\n",
      "Epoch 19/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3925 - accuracy: 0.8166\n",
      "Epoch 20/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3921 - accuracy: 0.8155\n",
      "Epoch 21/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3914 - accuracy: 0.8180\n",
      "Epoch 22/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3910 - accuracy: 0.8191\n",
      "Epoch 23/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3907 - accuracy: 0.8161\n",
      "Epoch 24/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3898 - accuracy: 0.8195\n",
      "Epoch 25/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3894 - accuracy: 0.8191\n",
      "Epoch 26/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3892 - accuracy: 0.8184\n",
      "Epoch 27/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3885 - accuracy: 0.8215\n",
      "Epoch 28/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3882 - accuracy: 0.8209\n",
      "Epoch 29/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3875 - accuracy: 0.8186\n",
      "Epoch 30/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3872 - accuracy: 0.8215\n",
      "Epoch 31/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3872 - accuracy: 0.8191\n",
      "Epoch 32/100\n",
      "162/162 [==============================] - 1s 7ms/step - loss: 0.3863 - accuracy: 0.8213\n",
      "Epoch 33/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3862 - accuracy: 0.8213\n",
      "Epoch 34/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3862 - accuracy: 0.8219\n",
      "Epoch 35/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3861 - accuracy: 0.8217\n",
      "Epoch 36/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3858 - accuracy: 0.8199\n",
      "Epoch 37/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3853 - accuracy: 0.8221\n",
      "Epoch 38/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3852 - accuracy: 0.8217\n",
      "Epoch 39/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3847 - accuracy: 0.8222\n",
      "Epoch 40/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3846 - accuracy: 0.8226\n",
      "Epoch 41/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3845 - accuracy: 0.8201\n",
      "Epoch 42/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3840 - accuracy: 0.8224\n",
      "Epoch 43/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3839 - accuracy: 0.8205\n",
      "Epoch 44/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3834 - accuracy: 0.8219\n",
      "Epoch 45/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3829 - accuracy: 0.8215\n",
      "Epoch 46/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3830 - accuracy: 0.8232\n",
      "Epoch 47/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3829 - accuracy: 0.8217\n",
      "Epoch 48/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3827 - accuracy: 0.8234\n",
      "Epoch 49/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3829 - accuracy: 0.8236\n",
      "Epoch 50/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3824 - accuracy: 0.8242\n",
      "Epoch 51/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3817 - accuracy: 0.8248\n",
      "Epoch 52/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3820 - accuracy: 0.8232\n",
      "Epoch 53/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3816 - accuracy: 0.8228\n",
      "Epoch 54/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3819 - accuracy: 0.8221\n",
      "Epoch 55/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3815 - accuracy: 0.8207\n",
      "Epoch 56/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3815 - accuracy: 0.8234\n",
      "Epoch 57/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3814 - accuracy: 0.8238\n",
      "Epoch 58/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3810 - accuracy: 0.8230\n",
      "Epoch 59/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3808 - accuracy: 0.8228\n",
      "Epoch 60/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3807 - accuracy: 0.8240\n",
      "Epoch 61/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3808 - accuracy: 0.8232\n",
      "Epoch 62/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3804 - accuracy: 0.8207\n",
      "Epoch 63/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3800 - accuracy: 0.8246\n",
      "Epoch 64/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3797 - accuracy: 0.8238\n",
      "Epoch 65/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3798 - accuracy: 0.8240\n",
      "Epoch 66/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3797 - accuracy: 0.8236\n",
      "Epoch 67/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3792 - accuracy: 0.8238\n",
      "Epoch 68/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3795 - accuracy: 0.8236\n",
      "Epoch 69/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3789 - accuracy: 0.8236\n",
      "Epoch 70/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3791 - accuracy: 0.8255\n",
      "Epoch 71/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3791 - accuracy: 0.8240\n",
      "Epoch 72/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3787 - accuracy: 0.8248\n",
      "Epoch 73/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3786 - accuracy: 0.8248\n",
      "Epoch 74/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3787 - accuracy: 0.8246\n",
      "Epoch 75/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3782 - accuracy: 0.8240\n",
      "Epoch 76/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3780 - accuracy: 0.8224\n",
      "Epoch 77/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3776 - accuracy: 0.8234\n",
      "Epoch 78/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3778 - accuracy: 0.8244\n",
      "Epoch 79/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3776 - accuracy: 0.8221\n",
      "Epoch 80/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3775 - accuracy: 0.8236\n",
      "Epoch 81/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3778 - accuracy: 0.8265\n",
      "Epoch 82/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3778 - accuracy: 0.8248\n",
      "Epoch 83/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3768 - accuracy: 0.8246\n",
      "Epoch 84/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3771 - accuracy: 0.8244\n",
      "Epoch 85/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3770 - accuracy: 0.8255\n",
      "Epoch 86/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3769 - accuracy: 0.8236\n",
      "Epoch 87/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3763 - accuracy: 0.8228\n",
      "Epoch 88/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3766 - accuracy: 0.8244\n",
      "Epoch 89/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3768 - accuracy: 0.8222\n",
      "Epoch 90/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3765 - accuracy: 0.8251\n",
      "Epoch 91/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3761 - accuracy: 0.8250\n",
      "Epoch 92/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3757 - accuracy: 0.8240\n",
      "Epoch 93/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3758 - accuracy: 0.8250\n",
      "Epoch 94/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3758 - accuracy: 0.8251\n",
      "Epoch 95/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3758 - accuracy: 0.8240\n",
      "Epoch 96/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3756 - accuracy: 0.8232\n",
      "Epoch 97/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3759 - accuracy: 0.8230\n",
      "Epoch 98/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3756 - accuracy: 0.8250\n",
      "Epoch 99/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3760 - accuracy: 0.8236\n",
      "Epoch 100/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3754 - accuracy: 0.8234\n"
     ]
    }
   ],
   "source": [
    "# Set random seed\n",
    "tf.random.set_seed(66)\n",
    "\n",
    "# Create a model\n",
    "# Activation functions help our network decide what is important data.\n",
    "# They receive data and add weightings to it and pass the results to other hidden layers\n",
    "# Depending on the error value the weights are updated\n",
    "model_1 = tf.keras.Sequential([\n",
    "  tf.keras.layers.Dense(4, activation=tf.keras.activations.relu), # hidden layer 1, ReLU activation\n",
    "  tf.keras.layers.Dense(4, activation=tf.keras.activations.relu), # hidden layer 2, ReLU activation\n",
    "  tf.keras.layers.Dense(1, activation=tf.keras.activations.sigmoid) # ouput layer, sigmoid activation\n",
    "])\n",
    "\n",
    "# Compile the model\n",
    "# Compile the model (Calc errors, optimize and evaluate)\n",
    "# Loss : Finds the average of the errors between predictions and \n",
    "# observed data (How incorrect are our predictions)\n",
    "# Adam : Optimizes our guesses by smoothing the predictions \n",
    "# to the observed data \n",
    "# Metrics test the accuracy of the model\n",
    "model_1.compile(loss=tf.keras.losses.binary_crossentropy,\n",
    "                optimizer=tf.keras.optimizers.Adam(),\n",
    "                metrics=['accuracy'])\n",
    "\n",
    "# Fit the model\n",
    "# Take the data and try to fit the data with 100 tries creating a trained model\n",
    "# An epoch is a full iteration over the training data\n",
    "history = model_1.fit(X_train_normal, y_train, epochs=100)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- The features passed are converted into a 4 value vector\n",
    "- That vector is then turned into an additional optimized vector\n",
    "- Those are converted into a prediction of either good or bad wine"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluate Accuracy of Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "162/162 [==============================] - 1s 3ms/step - loss: 0.3764 - accuracy: 0.8248\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.3763878643512726, 0.8247582316398621]"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_1.evaluate(X_train_normal, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How to Improve Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Fit on more data\n",
    "- Increase hidden units : Add weightings to get desired output\n",
    "- Try different activation functions : Helps decide what data is most important\n",
    "- Try different optimization functions : Optimize by smoothing data to expected results\n",
    "- Increase learning rate : How much to change the model based on estimated error\n",
    "- Fit longer : Increase the number of times we try to fit the data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Correlation Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2160x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# When we say data is correlated that means the data is related to each other. When a feature moves in a\n",
    "# certain direction a label also tends to move in that same direction\n",
    "plt.figure(figsize=(30, 20))\n",
    "sns.heatmap(wine_data.corr().abs(),  annot=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Correlation matrix shows some correlation between quality and density and alcohol percentage."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot Loss Curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Model 1 Loss Curve')"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(history.history).plot()\n",
    "plt.title(\"Model 1 Loss Curve\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Use Callback to Find Ideal Learning Rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.6945 - accuracy: 0.4956 - lr: 1.0000e-04\n",
      "Epoch 2/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.6780 - accuracy: 0.7878 - lr: 1.1220e-04\n",
      "Epoch 3/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.6640 - accuracy: 0.8035 - lr: 1.2589e-04\n",
      "Epoch 4/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.6492 - accuracy: 0.8054 - lr: 1.4125e-04\n",
      "Epoch 5/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.6304 - accuracy: 0.8056 - lr: 1.5849e-04\n",
      "Epoch 6/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.6029 - accuracy: 0.8056 - lr: 1.7783e-04\n",
      "Epoch 7/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5694 - accuracy: 0.8056 - lr: 1.9953e-04\n",
      "Epoch 8/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5370 - accuracy: 0.8056 - lr: 2.2387e-04\n",
      "Epoch 9/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5129 - accuracy: 0.8056 - lr: 2.5119e-04\n",
      "Epoch 10/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4993 - accuracy: 0.8056 - lr: 2.8184e-04\n",
      "Epoch 11/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4934 - accuracy: 0.8056 - lr: 3.1623e-04\n",
      "Epoch 12/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4903 - accuracy: 0.8056 - lr: 3.5481e-04\n",
      "Epoch 13/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4879 - accuracy: 0.8056 - lr: 3.9811e-04\n",
      "Epoch 14/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4851 - accuracy: 0.8056 - lr: 4.4668e-04\n",
      "Epoch 15/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4807 - accuracy: 0.8056 - lr: 5.0119e-04\n",
      "Epoch 16/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4721 - accuracy: 0.8056 - lr: 5.6234e-04\n",
      "Epoch 17/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4533 - accuracy: 0.8056 - lr: 6.3096e-04\n",
      "Epoch 18/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4365 - accuracy: 0.8054 - lr: 7.0795e-04\n",
      "Epoch 19/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4215 - accuracy: 0.8075 - lr: 7.9433e-04\n",
      "Epoch 20/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4117 - accuracy: 0.8085 - lr: 8.9125e-04\n",
      "Epoch 21/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4058 - accuracy: 0.8143 - lr: 0.0010\n",
      "Epoch 22/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4023 - accuracy: 0.8133 - lr: 0.0011\n",
      "Epoch 23/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4002 - accuracy: 0.8149 - lr: 0.0013\n",
      "Epoch 24/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3981 - accuracy: 0.8155 - lr: 0.0014\n",
      "Epoch 25/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3965 - accuracy: 0.8147 - lr: 0.0016\n",
      "Epoch 26/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3951 - accuracy: 0.8149 - lr: 0.0018\n",
      "Epoch 27/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3938 - accuracy: 0.8164 - lr: 0.0020\n",
      "Epoch 28/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3925 - accuracy: 0.8174 - lr: 0.0022\n",
      "Epoch 29/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3913 - accuracy: 0.8153 - lr: 0.0025\n",
      "Epoch 30/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3905 - accuracy: 0.8186 - lr: 0.0028\n",
      "Epoch 31/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3904 - accuracy: 0.8191 - lr: 0.0032\n",
      "Epoch 32/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3884 - accuracy: 0.8172 - lr: 0.0035\n",
      "Epoch 33/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3876 - accuracy: 0.8182 - lr: 0.0040\n",
      "Epoch 34/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3874 - accuracy: 0.8207 - lr: 0.0045\n",
      "Epoch 35/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3880 - accuracy: 0.8182 - lr: 0.0050\n",
      "Epoch 36/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3847 - accuracy: 0.8209 - lr: 0.0056\n",
      "Epoch 37/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3885 - accuracy: 0.8182 - lr: 0.0063\n",
      "Epoch 38/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3851 - accuracy: 0.8161 - lr: 0.0071\n",
      "Epoch 39/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3829 - accuracy: 0.8191 - lr: 0.0079\n",
      "Epoch 40/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3832 - accuracy: 0.8191 - lr: 0.0089\n",
      "Epoch 41/100\n",
      "162/162 [==============================] - 1s 4ms/step - loss: 0.3855 - accuracy: 0.8186 - lr: 0.0100\n",
      "Epoch 42/100\n",
      "162/162 [==============================] - 1s 4ms/step - loss: 0.3831 - accuracy: 0.8205 - lr: 0.0112\n",
      "Epoch 43/100\n",
      "162/162 [==============================] - 1s 4ms/step - loss: 0.3858 - accuracy: 0.8176 - lr: 0.0126\n",
      "Epoch 44/100\n",
      "162/162 [==============================] - 1s 4ms/step - loss: 0.3805 - accuracy: 0.8232 - lr: 0.0141\n",
      "Epoch 45/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3792 - accuracy: 0.8240 - lr: 0.0158\n",
      "Epoch 46/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3801 - accuracy: 0.8219 - lr: 0.0178\n",
      "Epoch 47/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3855 - accuracy: 0.8170 - lr: 0.0200\n",
      "Epoch 48/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.3808 - accuracy: 0.8246 - lr: 0.0224\n",
      "Epoch 49/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3840 - accuracy: 0.8162 - lr: 0.0251\n",
      "Epoch 50/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3811 - accuracy: 0.8203 - lr: 0.0282\n",
      "Epoch 51/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3842 - accuracy: 0.8162 - lr: 0.0316\n",
      "Epoch 52/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3861 - accuracy: 0.8166 - lr: 0.0355\n",
      "Epoch 53/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3777 - accuracy: 0.8230 - lr: 0.0398\n",
      "Epoch 54/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3865 - accuracy: 0.8203 - lr: 0.0447\n",
      "Epoch 55/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3868 - accuracy: 0.8159 - lr: 0.0501\n",
      "Epoch 56/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3866 - accuracy: 0.8201 - lr: 0.0562\n",
      "Epoch 57/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3871 - accuracy: 0.8201 - lr: 0.0631\n",
      "Epoch 58/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3883 - accuracy: 0.8176 - lr: 0.0708\n",
      "Epoch 59/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3850 - accuracy: 0.8248 - lr: 0.0794\n",
      "Epoch 60/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3935 - accuracy: 0.8149 - lr: 0.0891\n",
      "Epoch 61/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3963 - accuracy: 0.8178 - lr: 0.1000\n",
      "Epoch 62/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3936 - accuracy: 0.8190 - lr: 0.1122\n",
      "Epoch 63/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3967 - accuracy: 0.8191 - lr: 0.1259\n",
      "Epoch 64/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.3914 - accuracy: 0.8201 - lr: 0.1413\n",
      "Epoch 65/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4040 - accuracy: 0.8103 - lr: 0.1585\n",
      "Epoch 66/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4009 - accuracy: 0.8164 - lr: 0.1778\n",
      "Epoch 67/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4256 - accuracy: 0.8058 - lr: 0.1995\n",
      "Epoch 68/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4356 - accuracy: 0.8041 - lr: 0.2239\n",
      "Epoch 69/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4123 - accuracy: 0.8031 - lr: 0.2512\n",
      "Epoch 70/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4100 - accuracy: 0.8093 - lr: 0.2818\n",
      "Epoch 71/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4076 - accuracy: 0.8035 - lr: 0.3162\n",
      "Epoch 72/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4223 - accuracy: 0.8050 - lr: 0.3548\n",
      "Epoch 73/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4447 - accuracy: 0.8075 - lr: 0.3981\n",
      "Epoch 74/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4311 - accuracy: 0.7990 - lr: 0.4467\n",
      "Epoch 75/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4336 - accuracy: 0.8027 - lr: 0.5012\n",
      "Epoch 76/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.4324 - accuracy: 0.8004 - lr: 0.5623\n",
      "Epoch 77/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5043 - accuracy: 0.8033 - lr: 0.6310\n",
      "Epoch 78/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5011 - accuracy: 0.8056 - lr: 0.7079\n",
      "Epoch 79/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5038 - accuracy: 0.8056 - lr: 0.7943\n",
      "Epoch 80/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5052 - accuracy: 0.8056 - lr: 0.8913\n",
      "Epoch 81/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.5182 - accuracy: 0.8056 - lr: 1.0000\n",
      "Epoch 82/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5081 - accuracy: 0.8056 - lr: 1.1220\n",
      "Epoch 83/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5082 - accuracy: 0.8056 - lr: 1.2589\n",
      "Epoch 84/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5190 - accuracy: 0.8056 - lr: 1.4125\n",
      "Epoch 85/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5099 - accuracy: 0.8056 - lr: 1.5849\n",
      "Epoch 86/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5186 - accuracy: 0.8056 - lr: 1.7783\n",
      "Epoch 87/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5306 - accuracy: 0.8056 - lr: 1.9953\n",
      "Epoch 88/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5297 - accuracy: 0.8056 - lr: 2.2387\n",
      "Epoch 89/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5255 - accuracy: 0.7986 - lr: 2.5119\n",
      "Epoch 90/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5443 - accuracy: 0.7928 - lr: 2.8184\n",
      "Epoch 91/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5487 - accuracy: 0.7809 - lr: 3.1623\n",
      "Epoch 92/100\n",
      "162/162 [==============================] - 1s 5ms/step - loss: 0.5510 - accuracy: 0.7975 - lr: 3.5481\n",
      "Epoch 93/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5562 - accuracy: 0.7979 - lr: 3.9811\n",
      "Epoch 94/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5298 - accuracy: 0.8014 - lr: 4.4668\n",
      "Epoch 95/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5419 - accuracy: 0.7897 - lr: 5.0119\n",
      "Epoch 96/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5730 - accuracy: 0.7634 - lr: 5.6234\n",
      "Epoch 97/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5317 - accuracy: 0.7959 - lr: 6.3096\n",
      "Epoch 98/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.5461 - accuracy: 0.7979 - lr: 7.0795\n",
      "Epoch 99/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.6076 - accuracy: 0.7677 - lr: 7.9433\n",
      "Epoch 100/100\n",
      "162/162 [==============================] - 1s 6ms/step - loss: 0.6747 - accuracy: 0.7511 - lr: 8.9125\n"
     ]
    }
   ],
   "source": [
    "# Create model to fit on the training data and test on testing data\n",
    "tf.random.set_seed(66)\n",
    "\n",
    "model_2 = tf.keras.Sequential([\n",
    "    tf.keras.layers.Dense(4, activation=\"relu\"),\n",
    "    tf.keras.layers.Dense(4, activation=\"relu\"),\n",
    "    tf.keras.layers.Dense(1, activation=\"sigmoid\")\n",
    "])\n",
    "\n",
    "model_2.compile(loss=\"binary_crossentropy\",\n",
    "optimizer=\"Adam\",\n",
    "metrics=[\"accuracy\"])\n",
    "\n",
    "# The callback works during model training\n",
    "# At each epoch if we change the learning rate and the \n",
    "# schedular is going to give Adam the updated learning rate\n",
    "\n",
    "# The learning rate starts at 1e-4 (.0001) and after each epoch \n",
    "# increase the learning rate 10^(epoch/20)\n",
    "scheduler = tf.keras.callbacks.LearningRateScheduler(lambda epoch: 1e-4 * 10**(epoch/20))\n",
    "\n",
    "history_lr = model_2.fit(X_train_normal, y_train, \n",
    "epochs=100,\n",
    "callbacks=[scheduler]) # You can have multiple callbacks"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot Learning Rate Versus the Loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Learning Rate vs. Loss')"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lrs = 1e-4 * (10 ** (tf.range(100)/20))\n",
    "plt.figure(figsize=(20, 10))\n",
    "plt.semilogx(lrs, history_lr.history['loss'])\n",
    "plt.xlabel(\"Learning Rate\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.title(\"Learning Rate vs. Loss\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$ 10^{-3} $ or 0.001 is around the best learning rate\n",
    "\n",
    "To find the learning rate to begin training take the learning rate where loss is decreasing but hasn't flattened out. \n",
    "\n",
    "This is a log scale so the next dot is .02, .03, ..."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluate with Test Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "41/41 [==============================] - 0s 7ms/step - loss: 0.4086 - accuracy: 0.8097\n",
      "Model Loss (Test Set) : 0.40856754779815674\n",
      "Model Accuracy (Test Set) : 0.8097447752952576\n"
     ]
    }
   ],
   "source": [
    "loss, acc = model_1.evaluate(X_test_normal, y_test)\n",
    "print(f\"Model Loss (Test Set) : {loss}\")\n",
    "print(f\"Model Accuracy (Test Set) : {acc}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Understanding the Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Each hidden layer tries to find patterns and then feeds \n",
    "# the results to the next layer\n",
    "from tensorflow.keras.utils import plot_model\n",
    "plot_model(model_2, show_shapes=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([[-0.92357695, -1.8599589 , -0.5405463 , -0.33853838],\n",
       "        [-2.5003278 ,  0.05589265, -2.1297085 , -2.0484781 ],\n",
       "        [-3.1383114 , -3.2839518 , -0.64899147, -1.2386305 ],\n",
       "        [-0.06084743, -0.6045449 ,  0.48507988,  0.82604545]],\n",
       "       dtype=float32),\n",
       " (4, 4))"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get Weight values & size\n",
    "weights, biases = model_2.layers[1].get_weights()\n",
    "weights, weights.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For each of the 11 data points a 4 value tensor is created. This is referred to as the weight. It adjusts to find patterns in those 11 data points.\n",
    "\n",
    "These 4 values start off as random weightings and adjust to decide how to best use these values to influence the neurla networks decisions (Guesses). Our neural network updates weights and biases as it improves its predictions."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Each of the 11 features is assigned a 4 value tensor."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([-2.596155 , -0.623707 , -1.7804542, -2.7812173], dtype=float32), (4,))"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Bias values & size\n",
    "biases, biases.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There is one bias vector for each hidden unit in our neural network. It is paired with the weight tensor. The bias tensor is responsible for defining how much the patterns in the weights tensor should influence the next layer."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Multi-class Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When you have more than 2 classes it is called a multi-class classification. Build neural network to classify written numbers from 0 to 9. We'll use the Tensorflow MNIST handwritten digit database. 60,000 training examples and a test set of 10,000. 28x28 grayscale image associated with 10 classes. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Download Data and Split into Training & Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n",
      "11493376/11490434 [==============================] - 0s 0us/step\n",
      "11501568/11490434 [==============================] - 0s 0us/step\n"
     ]
    }
   ],
   "source": [
    "mnist = tf.keras.datasets.mnist \n",
    "(X_train, y_train),(X_test, y_test) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## View Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAAAN8klEQVR4nO3df6jVdZ7H8ddrbfojxzI39iZOrWOEUdE6i9nSyjYRTj8o7FYMIzQ0JDl/JDSwyIb7xxSLIVu6rBSDDtXYMus0UJHFMNVm5S6BdDMrs21qoxjlphtmmv1a9b1/3K9xp+75nOs53/PD+34+4HDO+b7P93zffPHl99f53o8jQgAmvj/rdQMAuoOwA0kQdiAJwg4kQdiBJE7o5sJsc+of6LCI8FjT29qy277C9lu237F9ezvfBaCz3Op1dtuTJP1B0gJJOyW9JGlRROwozMOWHeiwTmzZ50l6JyLejYgvJf1G0sI2vg9AB7UT9hmS/jjq/c5q2p+wvcT2kO2hNpYFoE0dP0EXEeskrZPYjQd6qZ0t+y5JZ4x6/51qGoA+1E7YX5J0tu3v2j5R0o8kbaynLQB1a3k3PiIO2V4q6SlJkyQ9EBFv1NYZgFq1fOmtpYVxzA50XEd+VAPg+EHYgSQIO5AEYQeSIOxAEoQdSIKwA0kQdiAJwg4kQdiBJAg7kARhB5Ig7EAShB1IgrADSRB2IAnCDiRB2IEkCDuQBGEHkiDsQBKEHUiCsANJEHYgCcIOJEHYgSQIO5AEYQeSIOxAEi0P2Yzjw6RJk4r1U045paPLX7p0acPaSSedVJx39uzZxfqtt95arN9zzz0Na4sWLSrO+/nnnxfrK1euLNbvvPPOYr0X2gq77fckHZB0WNKhiJhbR1MA6lfHlv3SiPiwhu8B0EEcswNJtBv2kPS07ZdtLxnrA7aX2B6yPdTmsgC0od3d+PkRscv2X0h6xvZ/R8Tm0R+IiHWS1kmS7WhzeQBa1NaWPSJ2Vc97JD0maV4dTQGoX8thtz3Z9pSjryX9QNL2uhoDUK92duMHJD1m++j3/HtE/L6WriaYM888s1g/8cQTi/WLL764WJ8/f37D2tSpU4vzXn/99cV6L+3cubNYX7NmTbE+ODjYsHbgwIHivK+++mqx/sILLxTr/ajlsEfEu5L+qsZeAHQQl96AJAg7kARhB5Ig7EAShB1IwhHd+1HbRP0F3Zw5c4r1TZs2Feudvs20Xx05cqRYv/nmm4v1Tz75pOVlDw8PF+sfffRRsf7WW2+1vOxOiwiPNZ0tO5AEYQeSIOxAEoQdSIKwA0kQdiAJwg4kwXX2GkybNq1Y37JlS7E+a9asOtupVbPe9+3bV6xfeumlDWtffvllcd6svz9oF9fZgeQIO5AEYQeSIOxAEoQdSIKwA0kQdiAJhmyuwd69e4v1ZcuWFetXX311sf7KK68U683+pHLJtm3bivUFCxYU6wcPHizWzzvvvIa12267rTgv6sWWHUiCsANJEHYgCcIOJEHYgSQIO5AEYQeS4H72PnDyyScX682GF167dm3D2uLFi4vz3njjjcX6hg0binX0n5bvZ7f9gO09trePmjbN9jO2366eT62zWQD1G89u/K8kXfG1abdLejYizpb0bPUeQB9rGvaI2Czp678HXShpffV6vaRr620LQN1a/W38QEQcHSzrA0kDjT5oe4mkJS0uB0BN2r4RJiKidOItItZJWidxgg7opVYvve22PV2Squc99bUEoBNaDftGSTdVr2+S9Hg97QDolKa78bY3SPq+pNNs75T0c0krJf3W9mJJ70v6YSebnOj279/f1vwff/xxy/PecsstxfrDDz9crDcbYx39o2nYI2JRg9JlNfcCoIP4uSyQBGEHkiDsQBKEHUiCsANJcIvrBDB58uSGtSeeeKI47yWXXFKsX3nllcX6008/Xayj+xiyGUiOsANJEHYgCcIOJEHYgSQIO5AEYQeS4Dr7BHfWWWcV61u3bi3W9+3bV6w/99xzxfrQ0FDD2n333Vect5v/NicSrrMDyRF2IAnCDiRB2IEkCDuQBGEHkiDsQBJcZ09ucHCwWH/wwQeL9SlTprS87OXLlxfrDz30ULE+PDxcrGfFdXYgOcIOJEHYgSQIO5AEYQeSIOxAEoQdSILr7Cg6//zzi/XVq1cX65dd1vpgv2vXri3WV6xYUazv2rWr5WUfz1q+zm77Adt7bG8fNe0O27tsb6seV9XZLID6jWc3/leSrhhj+r9ExJzq8bt62wJQt6Zhj4jNkvZ2oRcAHdTOCbqltl+rdvNPbfQh20tsD9lu/MfIAHRcq2H/haSzJM2RNCxpVaMPRsS6iJgbEXNbXBaAGrQU9ojYHRGHI+KIpF9KmldvWwDq1lLYbU8f9XZQ0vZGnwXQH5peZ7e9QdL3JZ0mabekn1fv50gKSe9J+mlENL25mOvsE8/UqVOL9WuuuaZhrdm98vaYl4u/smnTpmJ9wYIFxfpE1eg6+wnjmHHRGJPvb7sjAF3Fz2WBJAg7kARhB5Ig7EAShB1Igltc0TNffPFFsX7CCeWLRYcOHSrWL7/88oa1559/vjjv8Yw/JQ0kR9iBJAg7kARhB5Ig7EAShB1IgrADSTS96w25XXDBBcX6DTfcUKxfeOGFDWvNrqM3s2PHjmJ98+bNbX3/RMOWHUiCsANJEHYgCcIOJEHYgSQIO5AEYQeS4Dr7BDd79uxifenSpcX6ddddV6yffvrpx9zTeB0+fLhYHx4u//XyI0eO1NnOcY8tO5AEYQeSIOxAEoQdSIKwA0kQdiAJwg4kwXX240Cza9mLFo010O6IZtfRZ86c2UpLtRgaGirWV6xYUaxv3LixznYmvKZbdttn2H7O9g7bb9i+rZo+zfYztt+unk/tfLsAWjWe3fhDkv4+Is6V9DeSbrV9rqTbJT0bEWdLerZ6D6BPNQ17RAxHxNbq9QFJb0qaIWmhpPXVx9ZLurZDPQKowTEds9ueKel7krZIGoiIoz9O/kDSQIN5lkha0kaPAGow7rPxtr8t6RFJP4uI/aNrMTI65JiDNkbEuoiYGxFz2+oUQFvGFXbb39JI0H8dEY9Wk3fbnl7Vp0va05kWAdSh6W68bUu6X9KbEbF6VGmjpJskrayeH+9IhxPAwMCYRzhfOffcc4v1e++9t1g/55xzjrmnumzZsqVYv/vuuxvWHn+8/E+GW1TrNZ5j9r+V9GNJr9veVk1brpGQ/9b2YknvS/phRzoEUIumYY+I/5I05uDuki6rtx0AncLPZYEkCDuQBGEHkiDsQBKEHUiCW1zHadq0aQ1ra9euLc47Z86cYn3WrFmttFSLF198sVhftWpVsf7UU08V65999tkx94TOYMsOJEHYgSQIO5AEYQeSIOxAEoQdSIKwA0mkuc5+0UUXFevLli0r1ufNm9ewNmPGjJZ6qsunn37asLZmzZrivHfddVexfvDgwZZ6Qv9hyw4kQdiBJAg7kARhB5Ig7EAShB1IgrADSaS5zj44ONhWvR07duwo1p988sli/dChQ8V66Z7zffv2FedFHmzZgSQIO5AEYQeSIOxAEoQdSIKwA0kQdiAJR0T5A/YZkh6SNCApJK2LiH+1fYekWyT9b/XR5RHxuybfVV4YgLZFxJijLo8n7NMlTY+IrbanSHpZ0rUaGY/9k4i4Z7xNEHag8xqFfTzjsw9LGq5eH7D9pqTe/mkWAMfsmI7Zbc+U9D1JW6pJS22/ZvsB26c2mGeJ7SHbQ+21CqAdTXfjv/qg/W1JL0haERGP2h6Q9KFGjuP/SSO7+jc3+Q5244EOa/mYXZJsf0vSk5KeiojVY9RnSnoyIs5v8j2EHeiwRmFvuhtv25Lul/Tm6KBXJ+6OGpS0vd0mAXTOeM7Gz5f0n5Jel3Skmrxc0iJJczSyG/+epJ9WJ/NK38WWHeiwtnbj60LYgc5reTcewMRA2IEkCDuQBGEHkiDsQBKEHUiCsANJEHYgCcIOJEHYgSQIO5AEYQeSIOxAEoQdSKLbQzZ/KOn9Ue9Pq6b1o37trV/7kuitVXX29peNCl29n/0bC7eHImJuzxoo6Nfe+rUvid5a1a3e2I0HkiDsQBK9Dvu6Hi+/pF9769e+JHprVVd66+kxO4Du6fWWHUCXEHYgiZ6E3fYVtt+y/Y7t23vRQyO237P9uu1tvR6frhpDb4/t7aOmTbP9jO23q+cxx9jrUW932N5Vrbtttq/qUW9n2H7O9g7bb9i+rZre03VX6Ksr663rx+y2J0n6g6QFknZKeknSoojY0dVGGrD9nqS5EdHzH2DY/jtJn0h66OjQWrb/WdLeiFhZ/Ud5akT8Q5/0doeOcRjvDvXWaJjxn6iH667O4c9b0Yst+zxJ70TEuxHxpaTfSFrYgz76XkRslrT3a5MXSlpfvV6vkX8sXdegt74QEcMRsbV6fUDS0WHGe7ruCn11RS/CPkPSH0e936n+Gu89JD1t+2XbS3rdzBgGRg2z9YGkgV42M4amw3h309eGGe+bddfK8Oft4gTdN82PiL+WdKWkW6vd1b4UI8dg/XTt9BeSztLIGIDDklb1splqmPFHJP0sIvaPrvVy3Y3RV1fWWy/CvkvSGaPef6ea1hciYlf1vEfSYxo57Ognu4+OoFs97+lxP1+JiN0RcTgijkj6pXq47qphxh+R9OuIeLSa3PN1N1Zf3VpvvQj7S5LOtv1d2ydK+pGkjT3o4xtsT65OnMj2ZEk/UP8NRb1R0k3V65skPd7DXv5Evwzj3WiYcfV43fV8+POI6PpD0lUaOSP/P5L+sRc9NOhrlqRXq8cbve5N0gaN7Nb9n0bObSyW9OeSnpX0tqT/kDStj3r7N40M7f2aRoI1vUe9zdfILvprkrZVj6t6ve4KfXVlvfFzWSAJTtABSRB2IAnCDiRB2IEkCDuQBGEHkiDsQBL/DyJ7caZa7LphAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(X_train[0], cmap=\"gray\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Normalize the Features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = tf.keras.utils.normalize(X_train, axis=1)\n",
    "X_test = tf.keras.utils.normalize(X_test, axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Show Unique values in Label Numpy Array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9], dtype=uint8)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.unique(y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Store Label Names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "label_names = np.unique(y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "| Hyperparameter | Multiclass Classification |\n",
    "| :---: | :---: |\n",
    "| Input Layer | Shape == number of features |\n",
    "| Hidden Layers | Minimum of 1 to Unlimited |\n",
    "| Neurons in Hidden Layer | Normally 10 to 100 |\n",
    "| Output layer | 1 per Class |\n",
    "| Hidden activation | Usually Rectified Linear Unit (ReLU) |\n",
    "| Output activation | Softmax |\n",
    "| Loss function | CategoricalCrossentropy |\n",
    "| Optimizer | Stochastic Gradient Descent, Adam |"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "1875/1875 [==============================] - 10s 4ms/step - loss: 1.2819 - accuracy: 0.5605 - val_loss: 0.8054 - val_accuracy: 0.7323\n",
      "Epoch 2/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.7314 - accuracy: 0.7563 - val_loss: 0.6761 - val_accuracy: 0.7763\n",
      "Epoch 3/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.6551 - accuracy: 0.7826 - val_loss: 0.6293 - val_accuracy: 0.7924\n",
      "Epoch 4/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.6188 - accuracy: 0.7971 - val_loss: 0.6011 - val_accuracy: 0.8064\n",
      "Epoch 5/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5958 - accuracy: 0.8137 - val_loss: 0.5843 - val_accuracy: 0.8229\n",
      "Epoch 6/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5712 - accuracy: 0.8283 - val_loss: 0.5629 - val_accuracy: 0.8380\n",
      "Epoch 7/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.5501 - accuracy: 0.8384 - val_loss: 0.5454 - val_accuracy: 0.8409\n",
      "Epoch 8/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5367 - accuracy: 0.8421 - val_loss: 0.5329 - val_accuracy: 0.8465\n",
      "Epoch 9/10\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5270 - accuracy: 0.8441 - val_loss: 0.5253 - val_accuracy: 0.8478\n",
      "Epoch 10/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.5201 - accuracy: 0.8464 - val_loss: 0.5196 - val_accuracy: 0.8487\n"
     ]
    }
   ],
   "source": [
    "# Set random seed\n",
    "tf.random.set_seed(66)\n",
    "\n",
    "# Create the model\n",
    "# We have to flatten the 28x28 image\n",
    "# Create a 4 value tensor for each of the 784 cells\n",
    "# Optimize and create another 4 value tensor\n",
    "# Output is 1 of the 10 possible labels\n",
    "model_3 = tf.keras.Sequential([\n",
    "  tf.keras.layers.Flatten(input_shape=(28, 28)),  \n",
    "  tf.keras.layers.Dense(4, activation=\"relu\"),\n",
    "  tf.keras.layers.Dense(4, activation=\"relu\"),\n",
    "  tf.keras.layers.Dense(10, activation=\"softmax\")\n",
    "])\n",
    "\n",
    "# Use SparseCategoricalCrossentropy if data isn't \n",
    "# normalized or one-hot encoded\n",
    "model_3.compile(loss=tf.keras.losses.SparseCategoricalCrossentropy(),\n",
    "                 optimizer=tf.keras.optimizers.Adam(),\n",
    "                 metrics=[\"accuracy\"])\n",
    "\n",
    "# Fit the model\n",
    "history = model_3.fit(X_train, \n",
    "                      y_train,\n",
    "                      epochs=10,\n",
    "                      validation_data=(X_test, y_test)) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Find Ideal learning Rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "1875/1875 [==============================] - 9s 5ms/step - loss: 1.2819 - accuracy: 0.5605 - val_loss: 0.8054 - val_accuracy: 0.7323 - lr: 0.0010\n",
      "Epoch 2/40\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.7270 - accuracy: 0.7580 - val_loss: 0.6705 - val_accuracy: 0.7764 - lr: 0.0011\n",
      "Epoch 3/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.6473 - accuracy: 0.7851 - val_loss: 0.6213 - val_accuracy: 0.7949 - lr: 0.0013\n",
      "Epoch 4/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.6101 - accuracy: 0.8041 - val_loss: 0.5868 - val_accuracy: 0.8230 - lr: 0.0014\n",
      "Epoch 5/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5771 - accuracy: 0.8267 - val_loss: 0.5558 - val_accuracy: 0.8356 - lr: 0.0016\n",
      "Epoch 6/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5490 - accuracy: 0.8367 - val_loss: 0.5446 - val_accuracy: 0.8402 - lr: 0.0018\n",
      "Epoch 7/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5344 - accuracy: 0.8425 - val_loss: 0.5242 - val_accuracy: 0.8478 - lr: 0.0020\n",
      "Epoch 8/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5274 - accuracy: 0.8433 - val_loss: 0.5168 - val_accuracy: 0.8507 - lr: 0.0022\n",
      "Epoch 9/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5209 - accuracy: 0.8457 - val_loss: 0.5129 - val_accuracy: 0.8481 - lr: 0.0025\n",
      "Epoch 10/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5153 - accuracy: 0.8471 - val_loss: 0.5107 - val_accuracy: 0.8490 - lr: 0.0028\n",
      "Epoch 11/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5098 - accuracy: 0.8483 - val_loss: 0.5049 - val_accuracy: 0.8509 - lr: 0.0032\n",
      "Epoch 12/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5048 - accuracy: 0.8494 - val_loss: 0.4962 - val_accuracy: 0.8525 - lr: 0.0035\n",
      "Epoch 13/40\n",
      "1875/1875 [==============================] - 9s 5ms/step - loss: 0.4993 - accuracy: 0.8515 - val_loss: 0.5099 - val_accuracy: 0.8473 - lr: 0.0040\n",
      "Epoch 14/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4970 - accuracy: 0.8528 - val_loss: 0.5005 - val_accuracy: 0.8481 - lr: 0.0045\n",
      "Epoch 15/40\n",
      "1875/1875 [==============================] - 8s 5ms/step - loss: 0.4916 - accuracy: 0.8541 - val_loss: 0.4909 - val_accuracy: 0.8541 - lr: 0.0050\n",
      "Epoch 16/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4838 - accuracy: 0.8552 - val_loss: 0.4688 - val_accuracy: 0.8659 - lr: 0.0056\n",
      "Epoch 17/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4780 - accuracy: 0.8577 - val_loss: 0.5298 - val_accuracy: 0.8443 - lr: 0.0063\n",
      "Epoch 18/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4723 - accuracy: 0.8591 - val_loss: 0.4753 - val_accuracy: 0.8594 - lr: 0.0071\n",
      "Epoch 19/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4709 - accuracy: 0.8597 - val_loss: 0.4681 - val_accuracy: 0.8666 - lr: 0.0079\n",
      "Epoch 20/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4712 - accuracy: 0.8578 - val_loss: 0.4905 - val_accuracy: 0.8585 - lr: 0.0089\n",
      "Epoch 21/40\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.4734 - accuracy: 0.8582 - val_loss: 0.4775 - val_accuracy: 0.8675 - lr: 0.0100\n",
      "Epoch 22/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4747 - accuracy: 0.8566 - val_loss: 0.4765 - val_accuracy: 0.8652 - lr: 0.0112\n",
      "Epoch 23/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4769 - accuracy: 0.8566 - val_loss: 0.5056 - val_accuracy: 0.8513 - lr: 0.0126\n",
      "Epoch 24/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4800 - accuracy: 0.8557 - val_loss: 0.4691 - val_accuracy: 0.8636 - lr: 0.0141\n",
      "Epoch 25/40\n",
      "1875/1875 [==============================] - 9s 5ms/step - loss: 0.4851 - accuracy: 0.8550 - val_loss: 0.4801 - val_accuracy: 0.8640 - lr: 0.0158\n",
      "Epoch 26/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4935 - accuracy: 0.8513 - val_loss: 0.5174 - val_accuracy: 0.8497 - lr: 0.0178\n",
      "Epoch 27/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.4989 - accuracy: 0.8493 - val_loss: 0.5353 - val_accuracy: 0.8419 - lr: 0.0200\n",
      "Epoch 28/40\n",
      "1875/1875 [==============================] - 9s 5ms/step - loss: 0.5108 - accuracy: 0.8456 - val_loss: 0.5136 - val_accuracy: 0.8520 - lr: 0.0224\n",
      "Epoch 29/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5175 - accuracy: 0.8445 - val_loss: 0.5750 - val_accuracy: 0.8386 - lr: 0.0251\n",
      "Epoch 30/40\n",
      "1875/1875 [==============================] - 9s 5ms/step - loss: 0.5278 - accuracy: 0.8417 - val_loss: 0.6092 - val_accuracy: 0.8170 - lr: 0.0282\n",
      "Epoch 31/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5442 - accuracy: 0.8377 - val_loss: 0.5221 - val_accuracy: 0.8518 - lr: 0.0316\n",
      "Epoch 32/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5566 - accuracy: 0.8362 - val_loss: 0.6408 - val_accuracy: 0.8162 - lr: 0.0355\n",
      "Epoch 33/40\n",
      "1875/1875 [==============================] - 8s 5ms/step - loss: 0.5744 - accuracy: 0.8334 - val_loss: 0.5699 - val_accuracy: 0.8429 - lr: 0.0398\n",
      "Epoch 34/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.5958 - accuracy: 0.8277 - val_loss: 0.6559 - val_accuracy: 0.8183 - lr: 0.0447\n",
      "Epoch 35/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.6169 - accuracy: 0.8202 - val_loss: 0.6972 - val_accuracy: 0.7924 - lr: 0.0501\n",
      "Epoch 36/40\n",
      "1875/1875 [==============================] - 9s 5ms/step - loss: 0.6664 - accuracy: 0.8067 - val_loss: 0.6349 - val_accuracy: 0.8127 - lr: 0.0562\n",
      "Epoch 37/40\n",
      "1875/1875 [==============================] - 8s 5ms/step - loss: 0.7218 - accuracy: 0.7890 - val_loss: 0.6509 - val_accuracy: 0.8213 - lr: 0.0631\n",
      "Epoch 38/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.7518 - accuracy: 0.7794 - val_loss: 0.6304 - val_accuracy: 0.8206 - lr: 0.0708\n",
      "Epoch 39/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.7720 - accuracy: 0.7713 - val_loss: 0.8606 - val_accuracy: 0.7741 - lr: 0.0794\n",
      "Epoch 40/40\n",
      "1875/1875 [==============================] - 8s 4ms/step - loss: 0.9161 - accuracy: 0.7153 - val_loss: 0.9719 - val_accuracy: 0.6422 - lr: 0.0891\n"
     ]
    }
   ],
   "source": [
    "# The ideal learning rate is where the loss decreases the most\n",
    "# Set random seed\n",
    "tf.random.set_seed(66)\n",
    "\n",
    "# Create the model\n",
    "model_4 = tf.keras.Sequential([\n",
    "  tf.keras.layers.Flatten(input_shape=(28, 28)), \n",
    "  tf.keras.layers.Dense(4, activation=\"relu\"),\n",
    "  tf.keras.layers.Dense(4, activation=\"relu\"),\n",
    "  tf.keras.layers.Dense(10, activation=\"softmax\") \n",
    "])\n",
    "\n",
    "# Compile the model\n",
    "model_4.compile(loss=tf.keras.losses.SparseCategoricalCrossentropy(), \n",
    "                 optimizer=tf.keras.optimizers.Adam(),\n",
    "                 metrics=[\"accuracy\"])\n",
    "\n",
    "# Learning rate callback\n",
    "scheduler = tf.keras.callbacks.LearningRateScheduler(lambda epoch: 1e-3 * 10**(epoch/20))\n",
    "\n",
    "# Fit the model\n",
    "history_lr = model_4.fit(X_train, \n",
    "                              y_train,\n",
    "                              epochs=40,\n",
    "                              validation_data=(X_test, y_test),\n",
    "                              callbacks=[scheduler])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot to Find Ideal Learning rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Learning Rate vs. Loss')"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lrs = 1e-4 * (10 ** (tf.range(40)/20))\n",
    "plt.figure(figsize=(20, 10))\n",
    "plt.semilogx(lrs, history_lr.history['loss'])\n",
    "plt.xlabel(\"Learning Rate\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.title(\"Learning Rate vs. Loss\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$ 10^{-3} $ or 0.001 is around the best learning rate"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Make Predictions with Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.6335019e-19, 1.5874131e-09, 9.3062638e-08, 3.1929246e-09,\n",
       "       7.0527524e-02, 1.8261727e-11, 1.3014530e-18, 9.2456114e-01,\n",
       "       6.1137167e-12, 4.9111922e-03], dtype=float32)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Prediction Probabilities\n",
    "y_prob = model_4.predict(X_test) \n",
    "y_prob[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get the actual label name our model predicted\n",
    "label_names[tf.argmax(y_prob[0])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get the actual number from test label\n",
    "y_test[0]"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "52eb59783876bcfbd73674135f70218772b928a518e0b6bdca31ef39a84411bc"
  },
  "kernelspec": {
   "display_name": "Python 3.10.4 64-bit",
   "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": "3.10.4"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}