4Regression_Housing_ANN(MLP).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Regression_Housing_ANN(MLP)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import fetch_california_housing\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "housing = fetch_california_housing()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train_full, X_test, y_train_full, y_test = train_test_split(housing.data, housing.target)\n",
    "X_train, X_valid, y_train, y_valid = train_test_split(X_train_full, y_train_full)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((15480, 8), (11610, 8), (3870, 8), (5160, 8))"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_full.shape    ,  X_train.shape,   X_valid.shape,      X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = StandardScaler()\n",
    "X_train = scaler.fit_transform(X_train)\n",
    "X_valid = scaler.transform(X_valid)\n",
    "X_test = scaler.transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow import keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.Sequential(\n",
    "        [keras.layers.Dense(30, activation=\"relu\", input_shape=X_train.shape[1:]),\n",
    "         keras.layers.Dense(1)]\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 30)                270       \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 1)                 31        \n",
      "=================================================================\n",
      "Total params: 301\n",
      "Trainable params: 301\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"mean_squared_error\", optimizer=\"sgd\")\n",
    "# model.compile(loss=\"mean_squared_error\", optimizer=keras.optimizers.SGD(lr=1e-3))  #if we want to change the learning rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "early_stopping_cb = keras.callbacks.EarlyStopping(patience=10,restore_best_weights=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/200\n",
      "363/363 [==============================] - 0s 887us/step - loss: 0.9724 - val_loss: 1.0628\n",
      "Epoch 2/200\n",
      "363/363 [==============================] - 0s 631us/step - loss: 0.5638 - val_loss: 0.4804\n",
      "Epoch 3/200\n",
      "363/363 [==============================] - 0s 605us/step - loss: 0.4821 - val_loss: 0.4329\n",
      "Epoch 4/200\n",
      "363/363 [==============================] - 0s 608us/step - loss: 0.4583 - val_loss: 0.4251\n",
      "Epoch 5/200\n",
      "363/363 [==============================] - 0s 607us/step - loss: 0.4478 - val_loss: 0.4077\n",
      "Epoch 6/200\n",
      "363/363 [==============================] - 0s 599us/step - loss: 0.4260 - val_loss: 0.3925\n",
      "Epoch 7/200\n",
      "363/363 [==============================] - 0s 612us/step - loss: 0.4154 - val_loss: 0.3860\n",
      "Epoch 8/200\n",
      "363/363 [==============================] - 0s 589us/step - loss: 0.4065 - val_loss: 0.4062\n",
      "Epoch 9/200\n",
      "363/363 [==============================] - 0s 602us/step - loss: 0.3999 - val_loss: 0.3728\n",
      "Epoch 10/200\n",
      "363/363 [==============================] - 0s 620us/step - loss: 0.3945 - val_loss: 0.3934\n",
      "Epoch 11/200\n",
      "363/363 [==============================] - 0s 624us/step - loss: 0.3927 - val_loss: 0.3788\n",
      "Epoch 12/200\n",
      "363/363 [==============================] - 0s 626us/step - loss: 0.3854 - val_loss: 0.3623\n",
      "Epoch 13/200\n",
      "363/363 [==============================] - 0s 605us/step - loss: 0.3860 - val_loss: 0.3643\n",
      "Epoch 14/200\n",
      "363/363 [==============================] - 0s 758us/step - loss: 0.3768 - val_loss: 0.3603\n",
      "Epoch 15/200\n",
      "363/363 [==============================] - 0s 873us/step - loss: 0.3764 - val_loss: 0.3590\n",
      "Epoch 16/200\n",
      "363/363 [==============================] - 0s 587us/step - loss: 0.3773 - val_loss: 0.3607\n",
      "Epoch 17/200\n",
      "363/363 [==============================] - 0s 614us/step - loss: 0.3716 - val_loss: 0.3566\n",
      "Epoch 18/200\n",
      "363/363 [==============================] - 0s 605us/step - loss: 0.3746 - val_loss: 0.3742\n",
      "Epoch 19/200\n",
      "363/363 [==============================] - 0s 618us/step - loss: 0.3845 - val_loss: 0.3551\n",
      "Epoch 20/200\n",
      "363/363 [==============================] - 0s 603us/step - loss: 0.3677 - val_loss: 0.3884\n",
      "Epoch 21/200\n",
      "363/363 [==============================] - 0s 613us/step - loss: 0.3664 - val_loss: 0.3533\n",
      "Epoch 22/200\n",
      "363/363 [==============================] - 0s 1ms/step - loss: 0.3616 - val_loss: 0.3499\n",
      "Epoch 23/200\n",
      "363/363 [==============================] - 0s 944us/step - loss: 0.3633 - val_loss: 0.4913\n",
      "Epoch 24/200\n",
      "363/363 [==============================] - 0s 699us/step - loss: 0.3709 - val_loss: 0.3478\n",
      "Epoch 25/200\n",
      "363/363 [==============================] - 0s 602us/step - loss: 0.3604 - val_loss: 0.3834\n",
      "Epoch 26/200\n",
      "363/363 [==============================] - 0s 632us/step - loss: 0.3600 - val_loss: 0.3483\n",
      "Epoch 27/200\n",
      "363/363 [==============================] - 0s 669us/step - loss: 0.3514 - val_loss: 0.3454\n",
      "Epoch 28/200\n",
      "363/363 [==============================] - 0s 610us/step - loss: 0.3502 - val_loss: 0.3636\n",
      "Epoch 29/200\n",
      "363/363 [==============================] - 0s 610us/step - loss: 0.3704 - val_loss: 0.5188\n",
      "Epoch 30/200\n",
      "363/363 [==============================] - 0s 624us/step - loss: 0.3772 - val_loss: 0.3472\n",
      "Epoch 31/200\n",
      "363/363 [==============================] - 0s 790us/step - loss: 0.3477 - val_loss: 0.3595\n",
      "Epoch 32/200\n",
      "363/363 [==============================] - 0s 602us/step - loss: 0.3613 - val_loss: 0.3483\n",
      "Epoch 33/200\n",
      "363/363 [==============================] - 0s 623us/step - loss: 0.3486 - val_loss: 0.3435\n",
      "Epoch 34/200\n",
      "363/363 [==============================] - 0s 894us/step - loss: 0.3416 - val_loss: 0.3403\n",
      "Epoch 35/200\n",
      "363/363 [==============================] - 0s 1ms/step - loss: 0.3410 - val_loss: 0.3427\n",
      "Epoch 36/200\n",
      "363/363 [==============================] - 0s 672us/step - loss: 0.3439 - val_loss: 0.3397\n",
      "Epoch 37/200\n",
      "363/363 [==============================] - 0s 715us/step - loss: 0.3382 - val_loss: 0.3386\n",
      "Epoch 38/200\n",
      "363/363 [==============================] - 0s 959us/step - loss: 0.3629 - val_loss: 0.3344\n",
      "Epoch 39/200\n",
      "363/363 [==============================] - 0s 757us/step - loss: 0.3372 - val_loss: 0.3363\n",
      "Epoch 40/200\n",
      "363/363 [==============================] - 0s 607us/step - loss: 0.3381 - val_loss: 0.3385\n",
      "Epoch 41/200\n",
      "363/363 [==============================] - 0s 628us/step - loss: 0.3332 - val_loss: 0.3431\n",
      "Epoch 42/200\n",
      "363/363 [==============================] - 0s 628us/step - loss: 0.3330 - val_loss: 0.3410\n",
      "Epoch 43/200\n",
      "363/363 [==============================] - 0s 734us/step - loss: 0.3328 - val_loss: 0.3350\n",
      "Epoch 44/200\n",
      "363/363 [==============================] - 0s 745us/step - loss: 0.3309 - val_loss: 0.3602\n",
      "Epoch 45/200\n",
      "363/363 [==============================] - 0s 856us/step - loss: 0.3296 - val_loss: 0.3432\n",
      "Epoch 46/200\n",
      "363/363 [==============================] - 0s 850us/step - loss: 0.3295 - val_loss: 0.3569\n",
      "Epoch 47/200\n",
      "363/363 [==============================] - 0s 616us/step - loss: 0.3281 - val_loss: 0.3377\n",
      "Epoch 48/200\n",
      "363/363 [==============================] - 0s 604us/step - loss: 0.3271 - val_loss: 0.3363\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X_train, y_train,     epochs=200,      validation_data=(X_valid, y_valid), callbacks=[early_stopping_cb])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd"
   ]
  },
  {
   "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>loss</th>\n",
       "      <th>val_loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.972359</td>\n",
       "      <td>1.062751</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.563761</td>\n",
       "      <td>0.480439</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.482126</td>\n",
       "      <td>0.432931</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.458319</td>\n",
       "      <td>0.425117</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.447764</td>\n",
       "      <td>0.407670</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.426000</td>\n",
       "      <td>0.392452</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.415378</td>\n",
       "      <td>0.386023</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.406504</td>\n",
       "      <td>0.406157</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0.399916</td>\n",
       "      <td>0.372791</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0.394467</td>\n",
       "      <td>0.393397</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>0.392668</td>\n",
       "      <td>0.378789</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>0.385409</td>\n",
       "      <td>0.362268</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>0.385963</td>\n",
       "      <td>0.364314</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>0.376813</td>\n",
       "      <td>0.360275</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>0.376429</td>\n",
       "      <td>0.359045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>0.377344</td>\n",
       "      <td>0.360698</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>0.371642</td>\n",
       "      <td>0.356631</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>0.374599</td>\n",
       "      <td>0.374159</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>0.384506</td>\n",
       "      <td>0.355070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>0.367689</td>\n",
       "      <td>0.388356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>0.366371</td>\n",
       "      <td>0.353317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>0.361647</td>\n",
       "      <td>0.349939</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>0.363305</td>\n",
       "      <td>0.491346</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>0.370926</td>\n",
       "      <td>0.347807</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>0.360350</td>\n",
       "      <td>0.383398</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>0.359994</td>\n",
       "      <td>0.348262</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>0.351441</td>\n",
       "      <td>0.345435</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>0.350236</td>\n",
       "      <td>0.363634</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>0.370398</td>\n",
       "      <td>0.518787</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>0.377151</td>\n",
       "      <td>0.347182</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>0.347683</td>\n",
       "      <td>0.359527</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>0.361329</td>\n",
       "      <td>0.348265</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>0.348569</td>\n",
       "      <td>0.343530</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>0.341575</td>\n",
       "      <td>0.340346</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>0.340989</td>\n",
       "      <td>0.342660</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>0.343920</td>\n",
       "      <td>0.339721</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>0.338176</td>\n",
       "      <td>0.338565</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>0.362874</td>\n",
       "      <td>0.334376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>0.337169</td>\n",
       "      <td>0.336300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>0.338127</td>\n",
       "      <td>0.338538</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>0.333221</td>\n",
       "      <td>0.343107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>0.332960</td>\n",
       "      <td>0.340981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>0.332817</td>\n",
       "      <td>0.334981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>0.330936</td>\n",
       "      <td>0.360153</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>0.329583</td>\n",
       "      <td>0.343180</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>0.329544</td>\n",
       "      <td>0.356876</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>0.328051</td>\n",
       "      <td>0.337719</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>0.327095</td>\n",
       "      <td>0.336283</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        loss  val_loss\n",
       "0   0.972359  1.062751\n",
       "1   0.563761  0.480439\n",
       "2   0.482126  0.432931\n",
       "3   0.458319  0.425117\n",
       "4   0.447764  0.407670\n",
       "5   0.426000  0.392452\n",
       "6   0.415378  0.386023\n",
       "7   0.406504  0.406157\n",
       "8   0.399916  0.372791\n",
       "9   0.394467  0.393397\n",
       "10  0.392668  0.378789\n",
       "11  0.385409  0.362268\n",
       "12  0.385963  0.364314\n",
       "13  0.376813  0.360275\n",
       "14  0.376429  0.359045\n",
       "15  0.377344  0.360698\n",
       "16  0.371642  0.356631\n",
       "17  0.374599  0.374159\n",
       "18  0.384506  0.355070\n",
       "19  0.367689  0.388356\n",
       "20  0.366371  0.353317\n",
       "21  0.361647  0.349939\n",
       "22  0.363305  0.491346\n",
       "23  0.370926  0.347807\n",
       "24  0.360350  0.383398\n",
       "25  0.359994  0.348262\n",
       "26  0.351441  0.345435\n",
       "27  0.350236  0.363634\n",
       "28  0.370398  0.518787\n",
       "29  0.377151  0.347182\n",
       "30  0.347683  0.359527\n",
       "31  0.361329  0.348265\n",
       "32  0.348569  0.343530\n",
       "33  0.341575  0.340346\n",
       "34  0.340989  0.342660\n",
       "35  0.343920  0.339721\n",
       "36  0.338176  0.338565\n",
       "37  0.362874  0.334376\n",
       "38  0.337169  0.336300\n",
       "39  0.338127  0.338538\n",
       "40  0.333221  0.343107\n",
       "41  0.332960  0.340981\n",
       "42  0.332817  0.334981\n",
       "43  0.330936  0.360153\n",
       "44  0.329583  0.343180\n",
       "45  0.329544  0.356876\n",
       "46  0.328051  0.337719\n",
       "47  0.327095  0.336283"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(history.history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(history.history).plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Or"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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(pd.DataFrame(history.history))\n",
    "plt.grid(True)\n",
    "plt.gca().set_ylim(0, 1)\n",
    "plt.legend(['Train Loss','Validation Loss'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "162/162 [==============================] - 0s 484us/step - loss: 0.3323\n"
     ]
    }
   ],
   "source": [
    "mse_test = model.evaluate(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_new = X_test[:3] # pretend these are new instances\n",
    "y_pred = model.predict(X_new)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1.3877509],\n",
       "       [3.1859558],\n",
       "       [4.624016 ]], dtype=float32)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.925  , 3.821  , 5.00001])"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test[:3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "***\n",
    "### EXTRA:\n",
    "### SAVING MODEL:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save(\"my_keras_model.h5\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### LOADING MODEL:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "model2 = keras.models.load_model(\"my_keras_model.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_new = X_test[:3] # pretend these are new instances\n",
    "y_pred = model2.predict(X_new)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1.3877509],\n",
       "       [3.1859558],\n",
       "       [4.624016 ]], dtype=float32)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.925  , 3.821  , 5.00001])"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test[:3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### SAVING WEIGHTS:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save_weights(\"my_keras_weights.ckpt\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### LOADING WEIGHTS:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x1883e5da9c8>"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.load_weights(\"my_keras_weights.ckpt\")"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}