3-housing-prices/simple_regression.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "regression.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/unclepeddy/deeplearning/blob/master/3-housing-prices/simple_regression.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "metadata": {
        "id": "6-Ps0NBU7C7R",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "! pip install tf-nightly-2.0-preview"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "Ry227hIG7yjd",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "6292994e-9d33-442d-98e5-d28486485b5f"
      },
      "cell_type": "code",
      "source": [
        "import tensorflow as tf\n",
        "\n",
        "from tensorflow.keras import models, layers\n",
        "from tensorflow.keras.datasets import boston_housing\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "\n",
        "print(tf.__version__)"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2.0.0-dev20190330\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "WSBJvqa97xYg",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Import Boston housing house price data set\n",
        "(train_data, train_targets), (test_data, test_targets) = boston_housing.load_data()\n",
        "\n",
        "# Training hyperparameters\n",
        "num_validation_folds = 4\n",
        "num_epochs = 100\n",
        "is_verbose = 0\n",
        "\n",
        "# Do feature-wise normalization\n",
        "mean = train_data.mean(axis=0)\n",
        "std = train_data.std(axis=0)\n",
        "train_data -= mean\n",
        "train_data /= std\n",
        "test_data -= mean\n",
        "test_data /= std"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "I2UedRzo8VeL",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Function to build a simple 2 layer model \n",
        "def build_model():\n",
        "  model = models.Sequential()\n",
        "  model.add(layers.Dense(64, activation='relu', input_shape=(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"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "NlTwKvdT8Vxt",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 89
        },
        "outputId": "f0c23c06-e3ba-4e5c-dd90-6c2d0f413076"
      },
      "cell_type": "code",
      "source": [
        "# Calculate number of examples per validation fold \n",
        "fold_size = len(train_data) // num_validation_folds\n",
        "all_scores = []\n",
        "all_mae_histories = []\n",
        "\n",
        "# Train using k-fold validation technique\n",
        "for i in range(num_validation_folds):\n",
        "  print(\"processing fold #\", i)\n",
        "  \n",
        "  # Select validation set\n",
        "  val_data = train_data[i * fold_size: (i+1) * fold_size]\n",
        "  val_targets = train_targets[i * fold_size: (i+1) * fold_size]\n",
        "  \n",
        "  # Select training set using the rest of the data\n",
        "  partial_training_data = np.concatenate(\n",
        "      [train_data[:i * fold_size], \n",
        "      train_data[(i+1) * fold_size:]], axis=0)\n",
        "  partial_training_targets = np.concatenate(\n",
        "      [train_targets[:i * fold_size], \n",
        "      train_targets[(i+1) * fold_size:]], axis=0)\n",
        "  \n",
        "  # Re-compile model and train\n",
        "  model = build_model()\n",
        "  history = model.fit(partial_training_data, \n",
        "                      partial_training_targets, \n",
        "                      validation_data=(val_data, val_targets),\n",
        "                      epochs=num_epochs, \n",
        "                      batch_size=1, \n",
        "                      verbose=is_verbose)\n",
        "  \n",
        "  # Capture absolute error metric across this run\n",
        "  mae_history = history.history['val_mae']\n",
        "  all_mae_histories.append(mae_history)"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "processing fold # 0\n",
            "processing fold # 1\n",
            "processing fold # 2\n",
            "processing fold # 3\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "0lDn-bwe9O9R",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 361
        },
        "outputId": "2868879a-b565-490f-a5ee-d34f2cf1bcab"
      },
      "cell_type": "code",
      "source": [
        "# Calculate average MAE for each epoch\n",
        "average_mae_history = [np.mean([x[i] for x in all_mae_histories]) for i in range(num_epochs)]\n",
        "\n",
        "# Cut out first 10 epochs and implement explonential moving average\n",
        "def smooth_curve(points, factor=0.9):\n",
        "  smoothed_points = []\n",
        "  for point in points:\n",
        "    if smoothed_points:\n",
        "      previous = smoothed_points[-1] \n",
        "      smoothed_points.append(previous * factor + point * (1-factor))\n",
        "    else:\n",
        "      smoothed_points.append(point)\n",
        "  return smoothed_points\n",
        "\n",
        "smoothed_mae_history = smooth_curve(average_mae_history[10:])\n",
        "\n",
        "# Plot the MAE over epochs after 10\n",
        "plt.plot(range(10, len(smoothed_mae_history) + 10), smoothed_mae_history)\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Validation MAE')\n",
        "plt.show()"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFYCAYAAABKymUhAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nP6GgLiIi4icU1EVERPzE/wOHXIYwTDPNpQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "ifOg2Ikj_aFa",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        },
        "outputId": "51035dd2-393f-4c7a-b05c-78c4df46e276"
      },
      "cell_type": "code",
      "source": [
        "loss, mae = model.evaluate(test_data, test_targets)\n",
        "\n",
        "print(\"Mean Absolute Error: $\", mae * 1000)"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "102/102 [==============================] - 0s 962us/sample - loss: 20.5090 - mae: 2.7885\n",
            "Mean Absolute Error: $ 2788.5377407073975\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}