house_price_prediction_changed.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('housepricedata.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>LotArea</th>\n",
       "      <th>OverallQual</th>\n",
       "      <th>OverallCond</th>\n",
       "      <th>TotalBsmtSF</th>\n",
       "      <th>FullBath</th>\n",
       "      <th>HalfBath</th>\n",
       "      <th>BedroomAbvGr</th>\n",
       "      <th>TotRmsAbvGrd</th>\n",
       "      <th>Fireplaces</th>\n",
       "      <th>GarageArea</th>\n",
       "      <th>AboveMedianPrice</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>8450</td>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>856</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>548</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9600</td>\n",
       "      <td>6</td>\n",
       "      <td>8</td>\n",
       "      <td>1262</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>460</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>11250</td>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>920</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>608</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>9550</td>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>756</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>642</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>14260</td>\n",
       "      <td>8</td>\n",
       "      <td>5</td>\n",
       "      <td>1145</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "      <td>1</td>\n",
       "      <td>836</td>\n",
       "      <td>1</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1455</th>\n",
       "      <td>7917</td>\n",
       "      <td>6</td>\n",
       "      <td>5</td>\n",
       "      <td>953</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>460</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1456</th>\n",
       "      <td>13175</td>\n",
       "      <td>6</td>\n",
       "      <td>6</td>\n",
       "      <td>1542</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>500</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1457</th>\n",
       "      <td>9042</td>\n",
       "      <td>7</td>\n",
       "      <td>9</td>\n",
       "      <td>1152</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "      <td>2</td>\n",
       "      <td>252</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1458</th>\n",
       "      <td>9717</td>\n",
       "      <td>5</td>\n",
       "      <td>6</td>\n",
       "      <td>1078</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>240</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1459</th>\n",
       "      <td>9937</td>\n",
       "      <td>5</td>\n",
       "      <td>6</td>\n",
       "      <td>1256</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>276</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1460 rows × 11 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      LotArea  OverallQual  OverallCond  TotalBsmtSF  FullBath  HalfBath  \\\n",
       "0        8450            7            5          856         2         1   \n",
       "1        9600            6            8         1262         2         0   \n",
       "2       11250            7            5          920         2         1   \n",
       "3        9550            7            5          756         1         0   \n",
       "4       14260            8            5         1145         2         1   \n",
       "...       ...          ...          ...          ...       ...       ...   \n",
       "1455     7917            6            5          953         2         1   \n",
       "1456    13175            6            6         1542         2         0   \n",
       "1457     9042            7            9         1152         2         0   \n",
       "1458     9717            5            6         1078         1         0   \n",
       "1459     9937            5            6         1256         1         1   \n",
       "\n",
       "      BedroomAbvGr  TotRmsAbvGrd  Fireplaces  GarageArea  AboveMedianPrice  \n",
       "0                3             8           0         548                 1  \n",
       "1                3             6           1         460                 1  \n",
       "2                3             6           1         608                 1  \n",
       "3                3             7           1         642                 0  \n",
       "4                4             9           1         836                 1  \n",
       "...            ...           ...         ...         ...               ...  \n",
       "1455             3             7           1         460                 1  \n",
       "1456             3             7           2         500                 1  \n",
       "1457             4             9           2         252                 1  \n",
       "1458             2             5           0         240                 0  \n",
       "1459             3             6           0         276                 0  \n",
       "\n",
       "[1460 rows x 11 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The dataset that we have now is in what we call a pandas dataframe. To convert it to an array, simply access its values:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = df.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 8450,     7,     5, ...,     0,   548,     1],\n",
       "       [ 9600,     6,     8, ...,     1,   460,     1],\n",
       "       [11250,     7,     5, ...,     1,   608,     1],\n",
       "       ...,\n",
       "       [ 9042,     7,     9, ...,     2,   252,     1],\n",
       "       [ 9717,     5,     6, ...,     0,   240,     0],\n",
       "       [ 9937,     5,     6, ...,     0,   276,     0]])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, we split the dataset into our input features and the label we wish to predict."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = dataset[:,0:10]\n",
    "Y = dataset[:,10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Normalizing our data is very important, as we want the input features to be on the same order of magnitude to make our training easier. We'll use a min-max scaler from scikit-learn which scales our data to be between 0 and 1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "min_max_scaler = preprocessing.MinMaxScaler()\n",
    "X_scale = min_max_scaler.fit_transform(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.0334198 , 0.66666667, 0.5       , ..., 0.5       , 0.        ,\n",
       "        0.3864598 ],\n",
       "       [0.03879502, 0.55555556, 0.875     , ..., 0.33333333, 0.33333333,\n",
       "        0.32440056],\n",
       "       [0.04650728, 0.66666667, 0.5       , ..., 0.33333333, 0.33333333,\n",
       "        0.42877292],\n",
       "       ...,\n",
       "       [0.03618687, 0.66666667, 1.        , ..., 0.58333333, 0.66666667,\n",
       "        0.17771509],\n",
       "       [0.03934189, 0.44444444, 0.625     , ..., 0.25      , 0.        ,\n",
       "        0.16925247],\n",
       "       [0.04037019, 0.44444444, 0.625     , ..., 0.33333333, 0.        ,\n",
       "        0.19464034]])"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_scale"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lastly, we wish to set aside some parts of our dataset for a validation set and a test set. We use the function train_test_split from scikit-learn to do that."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_val_and_test, Y_train, Y_val_and_test = train_test_split(X_scale, Y, test_size=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_val, X_test, Y_val, Y_test = train_test_split(X_val_and_test, Y_val_and_test, test_size=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1022, 10) (219, 10) (219, 10) (1022,) (219,) (219,)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape, X_val.shape, X_test.shape, Y_train.shape, Y_val.shape, Y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Building and Training Our First Neural Network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will be using Keras to build our architecture. Let's import the code from Keras that we will need to use:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow.keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will be using the Sequential model, which means that we merely need to describe the layers above in sequence. Our neural network has three layers:\n",
    "\n",
    "- Hidden layer 1: 30 neurons, ReLU activation\n",
    "- Hidden layer 2: 30 neurons, ReLU activation\n",
    "- Output Layer: 1 neuron, Sigmoid activation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential([\n",
    "    Dense(32, activation='relu', input_shape=(10,)),\n",
    "    Dense(32, activation='relu'),\n",
    "    Dense(1, activation='sigmoid'),\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we've got our architecture specified, we need to find the best numbers for it. Before we start our training, we have to configure the model by\n",
    "- Telling it what algorithm you want to use to do the optimization (we'll use stochastic gradient descent)\n",
    "- Telling it what loss function to use (for binary classification, we will use binary cross entropy)\n",
    "- Telling it what other metrics you want to track apart from the loss function (we want to track accuracy as well)\n",
    "\n",
    "We do so below:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# changing loss from binary_crossentropy to mean_squared_error\n",
    "model.compile(optimizer='sgd',\n",
    "              loss='mean_squared_error',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Training on the data is pretty straightforward and requires us to write one line of code. The function is called 'fit' as we are fitting the parameters to the data. We specify:\n",
    "- what data we are training on, which is X_train and Y_train\n",
    "- the size of our mini-batch \n",
    "- how long we want to train it for (epochs)\n",
    "- what our validation data is so that the model will tell us how we are doing on the validation data at each point.\n",
    "\n",
    "This function will output a history, which we save under the variable hist. We'll use this variable a little later."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "32/32 [==============================] - 2s 35ms/step - loss: 0.2388 - accuracy: 0.5065 - val_loss: 0.2349 - val_accuracy: 0.5160\n",
      "Epoch 2/5\n",
      "32/32 [==============================] - 0s 2ms/step - loss: 0.2362 - accuracy: 0.5057 - val_loss: 0.2330 - val_accuracy: 0.5251\n",
      "Epoch 3/5\n",
      "32/32 [==============================] - 0s 3ms/step - loss: 0.2355 - accuracy: 0.5110 - val_loss: 0.2310 - val_accuracy: 0.5571\n",
      "Epoch 4/5\n",
      "32/32 [==============================] - 0s 3ms/step - loss: 0.2321 - accuracy: 0.5719 - val_loss: 0.2287 - val_accuracy: 0.6164\n",
      "Epoch 5/5\n",
      "32/32 [==============================] - 0s 3ms/step - loss: 0.2298 - accuracy: 0.6345 - val_loss: 0.2261 - val_accuracy: 0.7032\n"
     ]
    }
   ],
   "source": [
    "hist = model.fit(X_train, Y_train,\n",
    "          batch_size=32, epochs=5,\n",
    "          validation_data=(X_val, Y_val))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Evaluating our data on the test set:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7/7 [==============================] - 0s 990us/step - loss: 0.2283 - accuracy: 0.6530\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.6529680490493774"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(X_test, Y_test)[1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Visualizing Loss and Accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import the relevant package we need to do the visualization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We want to visualize the training loss and the validation loss like this:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(hist.history['loss'])\n",
    "plt.plot(hist.history['val_loss'])\n",
    "plt.title('Model loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Val'], loc='upper right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can also visualize the training accuracy and the validation accuracy like this:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(hist.history['accuracy'])\n",
    "plt.plot(hist.history['val_accuracy'])\n",
    "plt.title('Model accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Val'], loc='lower right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Adding Regularization to our Neural Network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll train a model which will overfit, which we call Model 2. This might take a few minutes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "32/32 [==============================] - 1s 26ms/step - loss: 0.1810 - accuracy: 0.7599 - val_loss: 0.1081 - val_accuracy: 0.8493\n",
      "Epoch 2/10\n",
      "32/32 [==============================] - 1s 21ms/step - loss: 0.1126 - accuracy: 0.8520 - val_loss: 0.1008 - val_accuracy: 0.8447\n",
      "Epoch 3/10\n",
      "32/32 [==============================] - 1s 21ms/step - loss: 0.1012 - accuracy: 0.8458 - val_loss: 0.0836 - val_accuracy: 0.8904\n",
      "Epoch 4/10\n",
      "32/32 [==============================] - 1s 21ms/step - loss: 0.0918 - accuracy: 0.8704 - val_loss: 0.0842 - val_accuracy: 0.8676\n",
      "Epoch 5/10\n",
      "32/32 [==============================] - 1s 21ms/step - loss: 0.0876 - accuracy: 0.8701 - val_loss: 0.0724 - val_accuracy: 0.9041\n",
      "Epoch 6/10\n",
      "32/32 [==============================] - 1s 21ms/step - loss: 0.0773 - accuracy: 0.8984 - val_loss: 0.0931 - val_accuracy: 0.8676\n",
      "Epoch 7/10\n",
      "32/32 [==============================] - 1s 22ms/step - loss: 0.0796 - accuracy: 0.8820 - val_loss: 0.0792 - val_accuracy: 0.9178\n",
      "Epoch 8/10\n",
      "32/32 [==============================] - 1s 22ms/step - loss: 0.0812 - accuracy: 0.8824 - val_loss: 0.0721 - val_accuracy: 0.9041\n",
      "Epoch 9/10\n",
      "32/32 [==============================] - 1s 22ms/step - loss: 0.0632 - accuracy: 0.9189 - val_loss: 0.0825 - val_accuracy: 0.8858\n",
      "Epoch 10/10\n",
      "32/32 [==============================] - 1s 21ms/step - loss: 0.0740 - accuracy: 0.8978 - val_loss: 0.0733 - val_accuracy: 0.9132\n"
     ]
    }
   ],
   "source": [
    "model_2 = Sequential([\n",
    "    Dense(1000, activation='relu', input_shape=(10,)),\n",
    "    Dense(1000, activation='relu'),\n",
    "    Dense(1000, activation='relu'),\n",
    "    Dense(1000, activation='relu'),\n",
    "    Dense(1, activation='sigmoid'),\n",
    "])\n",
    "model_2.compile(optimizer='adam',\n",
    "              loss='mean_squared_error',\n",
    "              metrics=['accuracy'])\n",
    "hist_2 = model_2.fit(X_train, Y_train,\n",
    "          batch_size=32, epochs=10,\n",
    "          validation_data=(X_val, Y_val))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's do the same visualization to see what overfitting looks like in terms of the loss and accuracy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(hist_2.history['loss'])\n",
    "plt.plot(hist_2.history['val_loss'])\n",
    "plt.title('Model loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Val'], loc='upper right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(hist_2.history['accuracy'])\n",
    "plt.plot(hist_2.history['val_accuracy'])\n",
    "plt.title('Model accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Val'], loc='lower right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To address the overfitting we see in Model 2, we'll incorporate L2 regularization and dropout in our third model here (Model 3)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.layers import Dropout\n",
    "from keras import regularizers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_3 = Sequential([\n",
    "    Dense(1000, activation='relu', kernel_regularizer=regularizers.l2(0.01), input_shape=(10,)),\n",
    "    Dropout(0.3),\n",
    "    Dense(1000, activation='relu', kernel_regularizer=regularizers.l2(0.01)),\n",
    "    Dropout(0.3),\n",
    "    Dense(1000, activation='relu', kernel_regularizer=regularizers.l2(0.01)),\n",
    "    Dropout(0.3),\n",
    "    Dense(1000, activation='relu', kernel_regularizer=regularizers.l2(0.01)),\n",
    "    Dropout(0.3),\n",
    "    Dense(1, activation='sigmoid', kernel_regularizer=regularizers.l2(0.01)),\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "32/32 [==============================] - 2s 34ms/step - loss: 20.5972 - accuracy: 0.5290 - val_loss: 3.4438 - val_accuracy: 0.4840\n",
      "Epoch 2/10\n",
      "32/32 [==============================] - 1s 29ms/step - loss: 2.0820 - accuracy: 0.5039 - val_loss: 0.3552 - val_accuracy: 0.4840\n",
      "Epoch 3/10\n",
      "32/32 [==============================] - 1s 33ms/step - loss: 0.3021 - accuracy: 0.4935 - val_loss: 0.2529 - val_accuracy: 0.4840\n",
      "Epoch 4/10\n",
      "32/32 [==============================] - 1s 31ms/step - loss: 0.2519 - accuracy: 0.4816 - val_loss: 0.2504 - val_accuracy: 0.4840\n",
      "Epoch 5/10\n",
      "32/32 [==============================] - 1s 32ms/step - loss: 0.2499 - accuracy: 0.5319 - val_loss: 0.2502 - val_accuracy: 0.4840\n",
      "Epoch 6/10\n",
      "32/32 [==============================] - 1s 31ms/step - loss: 0.2501 - accuracy: 0.4942 - val_loss: 0.2501 - val_accuracy: 0.4840\n",
      "Epoch 7/10\n",
      "32/32 [==============================] - 1s 32ms/step - loss: 0.2500 - accuracy: 0.5181 - val_loss: 0.2501 - val_accuracy: 0.4840\n",
      "Epoch 8/10\n",
      "32/32 [==============================] - 1s 30ms/step - loss: 0.2499 - accuracy: 0.5234 - val_loss: 0.2501 - val_accuracy: 0.4840\n",
      "Epoch 9/10\n",
      "32/32 [==============================] - 1s 31ms/step - loss: 0.2501 - accuracy: 0.4904 - val_loss: 0.2501 - val_accuracy: 0.4840\n",
      "Epoch 10/10\n",
      "32/32 [==============================] - 1s 30ms/step - loss: 0.2499 - accuracy: 0.5187 - val_loss: 0.2501 - val_accuracy: 0.4840\n"
     ]
    }
   ],
   "source": [
    "model_3.compile(optimizer='adam',\n",
    "              loss='mean_squared_error',\n",
    "              metrics=['accuracy'])\n",
    "hist_3 = model_3.fit(X_train, Y_train,\n",
    "          batch_size=32, epochs=10,\n",
    "          validation_data=(X_val, Y_val))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll now plot the loss and accuracy graphs for Model 3. You'll notice that the loss is a lot higher at the start, and that's because we've changed our loss function. To plot such that the window is zoomed in between 0 and 1.2 for the loss, we add an additional line of code (plt.ylim) when plotting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(hist_3.history['loss'])\n",
    "plt.plot(hist_3.history['val_loss'])\n",
    "plt.title('Model loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Val'], loc='upper right')\n",
    "plt.ylim(top=1.2, bottom=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(hist_3.history['accuracy'])\n",
    "plt.plot(hist_3.history['val_accuracy'])\n",
    "plt.title('Model accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Val'], loc='lower right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As compared to Model 2, you should see that there's less overfitting!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}