Hold-out Validation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "0c1be704",
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.datasets import boston_housing\n",
    "import numpy as np\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from keras import layers, models\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "17ba82d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "(train_data, train_targets), (test_data, test_targets) = boston_housing.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "fabdff3e",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(404, 13)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bbb93fb3",
   "metadata": {},
   "source": [
    "### Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "2f1d2774",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a MinMaxScaler object\n",
    "scaler = MinMaxScaler()\n",
    "\n",
    "# Fit the scaler on the training data\n",
    "scaler.fit(train_data)\n",
    "\n",
    "# Normalize the training data\n",
    "train_data_normalized = scaler.transform(train_data)\n",
    "\n",
    "# Normalize the test data using the same scaler\n",
    "test_data_normalized = scaler.transform(test_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eeab99d4",
   "metadata": {},
   "source": [
    "# Hold-out Validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "a70e1fae",
   "metadata": {},
   "outputs": [],
   "source": [
    "num_validation_samples = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "5f83489a",
   "metadata": {},
   "outputs": [],
   "source": [
    "np.random.shuffle(train_data_normalized)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "cef853be",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((354, 13), (50, 13))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "validation_data = train_data_normalized[:num_validation_samples]\n",
    "final_train_data = train_data_normalized[num_validation_samples:]\n",
    "final_train_data.shape,  validation_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "4d9dec5e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(354,)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "final_train_targets = train_targets[num_validation_samples:]\n",
    "final_train_targets.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "e6f5cc66",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50,)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target_validation_data = train_targets[:num_validation_samples]\n",
    "target_validation_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "5951b56c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_model():\n",
    "    model = models.Sequential()\n",
    "    model.add(layers.Dense(64, activation='relu', input_shape = (final_train_data.shape[1],)))\n",
    "    model.add(layers.Dense(64, activation='relu'))\n",
    "    model.add(layers.Dense(1))\n",
    "    model.compile(optimizer='rmsprop', loss='mse', metrics = ['mae'])\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d6097ec5",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "23/23 [==============================] - 1s 19ms/step - loss: 550.0576 - mae: 21.5396 - val_loss: 429.0368 - val_mae: 18.7199\n",
      "Epoch 2/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 411.4131 - mae: 18.0529 - val_loss: 268.8276 - val_mae: 13.9162\n",
      "Epoch 3/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 247.8347 - mae: 12.7449 - val_loss: 140.2421 - val_mae: 8.8443\n",
      "Epoch 4/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 137.1097 - mae: 8.6304 - val_loss: 77.7565 - val_mae: 6.5837\n",
      "Epoch 5/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 107.3173 - mae: 7.5730 - val_loss: 74.3744 - val_mae: 6.8231\n",
      "Epoch 6/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 100.5567 - mae: 7.3551 - val_loss: 75.2602 - val_mae: 6.9190\n",
      "Epoch 7/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 96.8682 - mae: 7.2480 - val_loss: 72.9685 - val_mae: 6.2043\n",
      "Epoch 8/40\n",
      "23/23 [==============================] - 0s 9ms/step - loss: 93.5362 - mae: 6.9749 - val_loss: 73.7743 - val_mae: 6.1626\n",
      "Epoch 9/40\n",
      "23/23 [==============================] - 0s 7ms/step - loss: 91.8839 - mae: 6.9827 - val_loss: 75.8597 - val_mae: 6.0539\n",
      "Epoch 10/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 89.4232 - mae: 6.7083 - val_loss: 79.5443 - val_mae: 7.0609\n",
      "Epoch 11/40\n",
      "23/23 [==============================] - 0s 7ms/step - loss: 88.7814 - mae: 6.8717 - val_loss: 78.7221 - val_mae: 6.8462\n",
      "Epoch 12/40\n",
      "23/23 [==============================] - 0s 7ms/step - loss: 88.4118 - mae: 6.8883 - val_loss: 78.0414 - val_mae: 6.5861\n",
      "Epoch 13/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 87.3332 - mae: 6.8349 - val_loss: 85.1632 - val_mae: 7.4597\n",
      "Epoch 14/40\n",
      "23/23 [==============================] - 0s 7ms/step - loss: 86.8152 - mae: 6.9151 - val_loss: 78.9240 - val_mae: 6.4809\n",
      "Epoch 15/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 87.3371 - mae: 6.8257 - val_loss: 79.3330 - val_mae: 6.5067\n",
      "Epoch 16/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 86.8143 - mae: 6.8331 - val_loss: 79.6910 - val_mae: 6.2694\n",
      "Epoch 17/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 86.6936 - mae: 6.7612 - val_loss: 82.4329 - val_mae: 6.9960\n",
      "Epoch 18/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 86.1566 - mae: 6.8625 - val_loss: 82.0285 - val_mae: 6.9110\n",
      "Epoch 19/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 85.9223 - mae: 6.8506 - val_loss: 80.2655 - val_mae: 6.5075\n",
      "Epoch 20/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 86.2826 - mae: 6.8128 - val_loss: 80.7940 - val_mae: 6.5865\n",
      "Epoch 21/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 85.5381 - mae: 6.7767 - val_loss: 82.1403 - val_mae: 6.8836\n",
      "Epoch 22/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 86.0548 - mae: 6.8156 - val_loss: 82.0952 - val_mae: 6.8516\n",
      "Epoch 23/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 85.4156 - mae: 6.7927 - val_loss: 81.9487 - val_mae: 6.7973\n",
      "Epoch 24/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 84.5672 - mae: 6.7390 - val_loss: 84.3231 - val_mae: 7.1462\n",
      "Epoch 25/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 83.8482 - mae: 6.8840 - val_loss: 80.4381 - val_mae: 6.1945\n",
      "Epoch 26/40\n",
      "23/23 [==============================] - 0s 4ms/step - loss: 85.3883 - mae: 6.7596 - val_loss: 79.7891 - val_mae: 6.3807\n",
      "Epoch 27/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 85.1348 - mae: 6.7253 - val_loss: 85.0786 - val_mae: 7.1811\n",
      "Epoch 28/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 84.1608 - mae: 6.8341 - val_loss: 80.6436 - val_mae: 6.3626\n",
      "Epoch 29/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 84.4834 - mae: 6.6746 - val_loss: 84.5685 - val_mae: 7.0677\n",
      "Epoch 30/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 84.2301 - mae: 6.7784 - val_loss: 84.3617 - val_mae: 6.9789\n",
      "Epoch 31/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 84.8962 - mae: 6.8053 - val_loss: 82.1500 - val_mae: 6.4862\n",
      "Epoch 32/40\n",
      "23/23 [==============================] - 0s 7ms/step - loss: 84.3043 - mae: 6.7406 - val_loss: 81.8790 - val_mae: 6.4227\n",
      "Epoch 33/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 84.5896 - mae: 6.7112 - val_loss: 84.8242 - val_mae: 7.0693\n",
      "Epoch 34/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 83.9882 - mae: 6.7838 - val_loss: 81.7749 - val_mae: 6.5497\n",
      "Epoch 35/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 83.8149 - mae: 6.7012 - val_loss: 88.5065 - val_mae: 7.4945\n",
      "Epoch 36/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 83.5203 - mae: 6.8717 - val_loss: 81.0184 - val_mae: 6.5325\n",
      "Epoch 37/40\n",
      "23/23 [==============================] - 0s 6ms/step - loss: 83.9868 - mae: 6.7196 - val_loss: 80.5539 - val_mae: 6.4174\n",
      "Epoch 38/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 84.1863 - mae: 6.7013 - val_loss: 85.9013 - val_mae: 7.2286\n",
      "Epoch 39/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 83.7473 - mae: 6.7760 - val_loss: 83.3650 - val_mae: 6.8317\n",
      "Epoch 40/40\n",
      "23/23 [==============================] - 0s 5ms/step - loss: 83.0318 - mae: 6.6331 - val_loss: 90.1919 - val_mae: 7.6638\n"
     ]
    }
   ],
   "source": [
    "model= get_model()\n",
    "\n",
    "# Train the model\n",
    "epochs = 40\n",
    "batch_size = 16\n",
    "\n",
    "history = model.fit(final_train_data, final_train_targets, validation_data=(validation_data, target_validation_data), epochs=epochs, batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "5496e85a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 [==============================] - 0s 5ms/step - loss: 86.7708 - mae: 7.1395\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[86.7707748413086, 7.139461517333984]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_score=model.evaluate(test_data_normalized, test_targets)\n",
    "test_score"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "950b68da",
   "metadata": {},
   "source": [
    "## Ploting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "26f08198",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['loss', 'mae', 'val_loss', 'val_mae'])"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history_dict = history.history\n",
    "history_dict.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "feaf2069",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_losses = history_dict['loss']\n",
    "train_mae = history_dict['mae']\n",
    "val_losses = history_dict['val_loss']\n",
    "val_mae = history_dict['val_mae']\n",
    "epochs = range(1, len(val_mae)+1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "1f9307ba",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x25eb1dbba30>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(epochs, train_losses, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_losses, 'b', label= 'Validation loss')\n",
    "plt.title('Training and Validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "9e6aa329",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x25eb2e9eec0>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.clf()\n",
    "plt.plot(epochs, train_mae, 'bo', label='Training MAE')\n",
    "plt.plot(epochs, val_mae, 'b', label= 'Validation MAE')\n",
    "plt.title('Training and Validation MAE')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('MAE')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e81a12f4",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "206c21d7",
   "metadata": {},
   "source": [
    "## Final Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "3a7168bb",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/35\n",
      "23/23 [==============================] - 1s 4ms/step - loss: 526.5774 - mae: 20.9504\n",
      "Epoch 2/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 368.4770 - mae: 16.6865\n",
      "Epoch 3/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 199.0014 - mae: 11.0171\n",
      "Epoch 4/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 121.6050 - mae: 8.0685\n",
      "Epoch 5/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 104.8343 - mae: 7.5131\n",
      "Epoch 6/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 98.6687 - mae: 7.1773\n",
      "Epoch 7/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 94.4780 - mae: 7.1493\n",
      "Epoch 8/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 91.1685 - mae: 6.9691\n",
      "Epoch 9/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 89.8299 - mae: 6.9092\n",
      "Epoch 10/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 87.9038 - mae: 6.8509\n",
      "Epoch 11/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 87.5593 - mae: 6.8073\n",
      "Epoch 12/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 86.6902 - mae: 6.7745\n",
      "Epoch 13/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 87.2202 - mae: 6.7627\n",
      "Epoch 14/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 86.4436 - mae: 6.8323\n",
      "Epoch 15/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 86.6555 - mae: 6.7157\n",
      "Epoch 16/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 85.9861 - mae: 6.7892\n",
      "Epoch 17/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 86.4550 - mae: 6.7206\n",
      "Epoch 18/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 86.0754 - mae: 6.8182\n",
      "Epoch 19/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 85.4710 - mae: 6.7791\n",
      "Epoch 20/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 85.2859 - mae: 6.8013\n",
      "Epoch 21/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 85.1582 - mae: 6.8273\n",
      "Epoch 22/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 85.2838 - mae: 6.7897\n",
      "Epoch 23/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 84.5223 - mae: 6.7913\n",
      "Epoch 24/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 84.5365 - mae: 6.9192\n",
      "Epoch 25/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 85.4073 - mae: 6.7184\n",
      "Epoch 26/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 84.8701 - mae: 6.8563\n",
      "Epoch 27/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 84.8195 - mae: 6.7558\n",
      "Epoch 28/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 84.7150 - mae: 6.7556\n",
      "Epoch 29/35\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 82.1142 - mae: 6.5775\n",
      "Epoch 30/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 84.3074 - mae: 6.9904\n",
      "Epoch 31/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 84.4317 - mae: 6.7241\n",
      "Epoch 32/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 84.7134 - mae: 6.7011\n",
      "Epoch 33/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 84.5550 - mae: 6.7007\n",
      "Epoch 34/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 84.1012 - mae: 6.7771\n",
      "Epoch 35/35\n",
      "23/23 [==============================] - 0s 3ms/step - loss: 83.7133 - mae: 6.6862\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x25ee64194b0>"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "final_model = get_model() \n",
    "final_model.fit(final_train_data, final_train_targets, epochs=35, batch_size=16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "f65ace4e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 [==============================] - 0s 4ms/step - loss: 81.7284 - mae: 6.7466\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[81.7283935546875, 6.7465500831604]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_score=final_model.evaluate(test_data_normalized, test_targets)\n",
    "test_score"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}