Notebooks/3. Dense_Forecast.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Dense Forecast.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dPVvzVmcuG44",
        "colab_type": "text"
      },
      "source": [
        "# Dense Forecasting\n",
        "\n",
        "A dense layer is just a regular layer of neurons in a neural network. Each neuron recieves input from all the neurons in the previous layer, thus densely connected. The layer has a weight matrix W, a bias vector b, and the activations of previous layer a. The following is te docstring of class Dense from the keras documentation:\n",
        "\n",
        "output = activation(dot(input, kernel) + bias)where activation is the element-wise activation function passed as the activation argument, kernel is a weights matrix created by the layer, and bias is a bias vector created by the layer."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "vidayERjaO5q"
      },
      "source": [
        "## Setup"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "gqWabzlJ63nL",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        },
        "outputId": "d7289e37-c4d6-4688-9bd8-bb696973e6ed"
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf\n",
        "import seaborn as sns\n",
        "\n",
        "from sklearn.preprocessing import MinMaxScaler\n",
        "\n",
        "keras = tf.keras\n",
        "\n",
        "# set style of charts\n",
        "sns.set(style=\"darkgrid\")\n",
        "plt.rcParams['figure.figsize'] = [10, 10]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/statsmodels/tools/_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n",
            "  import pandas.util.testing as tm\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "cg1hfKCPldZG",
        "colab": {}
      },
      "source": [
        "def plot_series(time, series, format=\"-\", start=0, end=None, label=None):\n",
        "    plt.plot(time[start:end], series[start:end], format, label=label)\n",
        "    plt.xlabel(\"Time\")\n",
        "    plt.ylabel(\"Value\")\n",
        "    if label:\n",
        "        plt.legend(fontsize=14)\n",
        "    plt.grid(True)\n",
        "\n",
        "def model_forecast(model, series, window_size):\n",
        "    ds = tf.data.Dataset.from_tensor_slices(series)\n",
        "    ds = ds.window(window_size, shift=1, drop_remainder=True)\n",
        "    ds = ds.flat_map(lambda w: w.batch(window_size))\n",
        "    ds = ds.batch(32).prefetch(1)\n",
        "    forecast = model.predict(ds)\n",
        "    return forecast\n",
        "\n",
        "def window_dataset(series, window_size, batch_size=128,\n",
        "                   shuffle_buffer=1000):\n",
        "    dataset = tf.data.Dataset.from_tensor_slices(series)\n",
        "    dataset = dataset.window(window_size + 1, shift=1, drop_remainder=True)\n",
        "    dataset = dataset.flat_map(lambda window: window.batch(window_size + 1))\n",
        "    dataset = dataset.shuffle(shuffle_buffer)\n",
        "    dataset = dataset.map(lambda window: (window[:-1], window[-1]))\n",
        "    dataset = dataset.batch(batch_size).prefetch(1)\n",
        "    return dataset"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "ViWVB9qd8OIR"
      },
      "source": [
        "## Forecasting with Machine Learning\n",
        "\n",
        "First, we will train a model to forecast the next step given the previous 30 steps, therefore, we need to create a dataset of 20-step windows for training. Every 20 steps is 1 trading month since the markets are only open Monday - Friday."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Zmp1JXKxk9Vb",
        "colab": {}
      },
      "source": [
        "# Read in data\n",
        "spy = pd.read_csv('SPY.csv')\n",
        "\n",
        "# Convert series into datetime type\n",
        "spy['Date'] = pd.to_datetime(spy['Date'])\n",
        "\n",
        "# Save target series\n",
        "series = spy['Close']\n",
        "\n",
        "# Create train data set\n",
        "train_split_date = '2014-12-31'\n",
        "train_split_index = np.where(spy.Date == train_split_date)[0][0]\n",
        "x_train = spy.loc[spy['Date'] <= train_split_date]['Close']\n",
        "\n",
        "# Create test data set\n",
        "test_split_date = '2019-01-02'\n",
        "test_split_index = np.where(spy.Date == test_split_date)[0][0]\n",
        "x_test = spy.loc[spy['Date'] >= test_split_date]['Close']\n",
        "\n",
        "# Create valid data set\n",
        "valid_split_index = (train_split_index.max(),test_split_index.min())\n",
        "x_valid = spy.loc[(spy['Date'] < test_split_date) & (spy['Date'] > train_split_date)]['Close']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xxpJz3EkErei",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "de75b690-3577-4529-ef77-8abefc82932f"
      },
      "source": [
        "# Plot all lines on one chart to see where one segment starts and another ends\n",
        "plt.plot(x_train, label = 'Train')\n",
        "plt.plot(x_valid, label = 'Validate')\n",
        "plt.plot(x_test, label = 'Test')\n",
        "plt.legend()\n",
        "print(x_train.index.max(),x_valid.index.min(),x_valid.index.max(),x_test.index.min(),x_test.index.max())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "5521 5522 6527 6528 6949\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xTOs5lgVEtuG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Reshape values\n",
        "x_train_values = x_train.values.reshape(-1, 1)\n",
        "x_valid_values = x_valid.values.reshape(-1, 1)\n",
        "x_test_values = x_test.values.reshape(-1, 1)\n",
        "\n",
        "#  Create Scaler Object\n",
        "x_train_scaler = MinMaxScaler(feature_range=(0, 1))\n",
        "\n",
        "# Fit x_train values\n",
        "normalized_x_train = x_train_scaler.fit_transform(x_train_values)\n",
        "\n",
        "# Fit x_valid values\n",
        "normalized_x_valid = x_train_scaler.transform(x_valid_values)\n",
        "\n",
        "# Fit x_test values\n",
        "normalized_x_test = x_train_scaler.transform(x_test_values)\n",
        "\n",
        "# All values normalized to training data\n",
        "spy_normalized_to_traindata = x_train_scaler.transform(series.values.reshape(-1, 1))\n",
        "\n",
        "# Example of how to iverse\n",
        "# inversed = scaler.inverse_transform(normalized_x_train).flatten()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "9nEM33dZ8OIp"
      },
      "source": [
        "### Dense Model Forecasting"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZZiFRsKOW65u",
        "colab_type": "text"
      },
      "source": [
        "#### Find Learning Rate"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "RhGTv4G_8OIp",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "ef71f8d0-d990-4f63-c76e-f67a5d7de524"
      },
      "source": [
        "keras.backend.clear_session()\n",
        "tf.random.set_seed(42)\n",
        "np.random.seed(42)\n",
        "\n",
        "window_size = 20\n",
        "train_set = window_dataset(normalized_x_train.flatten(), window_size)\n",
        "\n",
        "model = keras.models.Sequential([\n",
        "  keras.layers.Dense(10, activation=\"relu\", input_shape=[window_size]),\n",
        "  keras.layers.Dense(10, activation=\"relu\"),\n",
        "  keras.layers.Dense(1)\n",
        "])\n",
        "\n",
        "lr_schedule = keras.callbacks.LearningRateScheduler(\n",
        "    lambda epoch: 1e-7 * 10**(epoch / 20))\n",
        "optimizer = keras.optimizers.Nadam(lr=1e-7)\n",
        "model.compile(loss=keras.losses.Huber(),\n",
        "              optimizer=optimizer,\n",
        "              metrics=[\"mae\"])\n",
        "history = model.fit(train_set, epochs=100, callbacks=[lr_schedule])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.2011 - mae: 0.5516\n",
            "Epoch 2/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.2011 - mae: 0.5515\n",
            "Epoch 3/100\n",
            "43/43 [==============================] - 1s 15ms/step - loss: 0.2010 - mae: 0.5514\n",
            "Epoch 4/100\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 0.2009 - mae: 0.5512\n",
            "Epoch 5/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.2008 - mae: 0.5511\n",
            "Epoch 6/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.2007 - mae: 0.5510\n",
            "Epoch 7/100\n",
            "43/43 [==============================] - 1s 15ms/step - loss: 0.2006 - mae: 0.5508\n",
            "Epoch 8/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.2005 - mae: 0.5506\n",
            "Epoch 9/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.2003 - mae: 0.5504\n",
            "Epoch 10/100\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 0.2002 - mae: 0.5502\n",
            "Epoch 11/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.2000 - mae: 0.5499\n",
            "Epoch 12/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1998 - mae: 0.5496\n",
            "Epoch 13/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1996 - mae: 0.5493\n",
            "Epoch 14/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1994 - mae: 0.5489\n",
            "Epoch 15/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1991 - mae: 0.5485\n",
            "Epoch 16/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.1988 - mae: 0.5481\n",
            "Epoch 17/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.1984 - mae: 0.5475\n",
            "Epoch 18/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1981 - mae: 0.5470\n",
            "Epoch 19/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1976 - mae: 0.5463\n",
            "Epoch 20/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1972 - mae: 0.5456\n",
            "Epoch 21/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.1966 - mae: 0.5448\n",
            "Epoch 22/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1960 - mae: 0.5439\n",
            "Epoch 23/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1953 - mae: 0.5428\n",
            "Epoch 24/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1946 - mae: 0.5417\n",
            "Epoch 25/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.1937 - mae: 0.5404\n",
            "Epoch 26/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1928 - mae: 0.5390\n",
            "Epoch 27/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1917 - mae: 0.5374\n",
            "Epoch 28/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1906 - mae: 0.5356\n",
            "Epoch 29/100\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 0.1893 - mae: 0.5335\n",
            "Epoch 30/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.1878 - mae: 0.5313\n",
            "Epoch 31/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.1861 - mae: 0.5287\n",
            "Epoch 32/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1843 - mae: 0.5259\n",
            "Epoch 33/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1823 - mae: 0.5227\n",
            "Epoch 34/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1800 - mae: 0.5192\n",
            "Epoch 35/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1775 - mae: 0.5152\n",
            "Epoch 36/100\n",
            "43/43 [==============================] - 0s 11ms/step - loss: 0.1747 - mae: 0.5108\n",
            "Epoch 37/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1716 - mae: 0.5058\n",
            "Epoch 38/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1682 - mae: 0.5003\n",
            "Epoch 39/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1644 - mae: 0.4942\n",
            "Epoch 40/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1602 - mae: 0.4874\n",
            "Epoch 41/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1555 - mae: 0.4798\n",
            "Epoch 42/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1505 - mae: 0.4714\n",
            "Epoch 43/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.1450 - mae: 0.4621\n",
            "Epoch 44/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1389 - mae: 0.4518\n",
            "Epoch 45/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.1324 - mae: 0.4405\n",
            "Epoch 46/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1254 - mae: 0.4280\n",
            "Epoch 47/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.1178 - mae: 0.4144\n",
            "Epoch 48/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1097 - mae: 0.3995\n",
            "Epoch 49/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.1012 - mae: 0.3831\n",
            "Epoch 50/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.0922 - mae: 0.3653\n",
            "Epoch 51/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.0829 - mae: 0.3459\n",
            "Epoch 52/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.0734 - mae: 0.3250\n",
            "Epoch 53/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.0639 - mae: 0.3027\n",
            "Epoch 54/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.0545 - mae: 0.2793\n",
            "Epoch 55/100\n",
            "43/43 [==============================] - 1s 15ms/step - loss: 0.0463 - mae: 0.2571\n",
            "Epoch 56/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.0394 - mae: 0.2367\n",
            "Epoch 57/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.0330 - mae: 0.2162\n",
            "Epoch 58/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.0269 - mae: 0.1947\n",
            "Epoch 59/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.0213 - mae: 0.1725\n",
            "Epoch 60/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.0162 - mae: 0.1500\n",
            "Epoch 61/100\n",
            "43/43 [==============================] - 1s 15ms/step - loss: 0.0119 - mae: 0.1277\n",
            "Epoch 62/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.0083 - mae: 0.1062\n",
            "Epoch 63/100\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 0.0056 - mae: 0.0864\n",
            "Epoch 64/100\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 0.0037 - mae: 0.0691\n",
            "Epoch 65/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.0024 - mae: 0.0553\n",
            "Epoch 66/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.0017 - mae: 0.0457\n",
            "Epoch 67/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.0014 - mae: 0.0404\n",
            "Epoch 68/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 0.0012 - mae: 0.0381\n",
            "Epoch 69/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 0.0011 - mae: 0.0373\n",
            "Epoch 70/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.0011 - mae: 0.0369\n",
            "Epoch 71/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 0.0010 - mae: 0.0364\n",
            "Epoch 72/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 9.7454e-04 - mae: 0.0357\n",
            "Epoch 73/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 9.0331e-04 - mae: 0.0345\n",
            "Epoch 74/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 7.6208e-04 - mae: 0.0316\n",
            "Epoch 75/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 6.0085e-04 - mae: 0.0282\n",
            "Epoch 76/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 4.2562e-04 - mae: 0.0238\n",
            "Epoch 77/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 2.9549e-04 - mae: 0.0197\n",
            "Epoch 78/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 2.2111e-04 - mae: 0.0165\n",
            "Epoch 79/100\n",
            "43/43 [==============================] - 1s 15ms/step - loss: 1.9089e-04 - mae: 0.0145\n",
            "Epoch 80/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 1.8168e-04 - mae: 0.0137\n",
            "Epoch 81/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 1.7846e-04 - mae: 0.0135\n",
            "Epoch 82/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 1.7742e-04 - mae: 0.0135\n",
            "Epoch 83/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 1.7492e-04 - mae: 0.0134\n",
            "Epoch 84/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.7351e-04 - mae: 0.0134\n",
            "Epoch 85/100\n",
            "43/43 [==============================] - 1s 14ms/step - loss: 1.7189e-04 - mae: 0.0133\n",
            "Epoch 86/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.7037e-04 - mae: 0.0132\n",
            "Epoch 87/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.6769e-04 - mae: 0.0131\n",
            "Epoch 88/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.6421e-04 - mae: 0.0130\n",
            "Epoch 89/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 1.6325e-04 - mae: 0.0129\n",
            "Epoch 90/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.6127e-04 - mae: 0.0128\n",
            "Epoch 91/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.5731e-04 - mae: 0.0126\n",
            "Epoch 92/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.5698e-04 - mae: 0.0126\n",
            "Epoch 93/100\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 1.5572e-04 - mae: 0.0127\n",
            "Epoch 94/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 1.5131e-04 - mae: 0.0125\n",
            "Epoch 95/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.4620e-04 - mae: 0.0122\n",
            "Epoch 96/100\n",
            "43/43 [==============================] - 1s 12ms/step - loss: 1.5027e-04 - mae: 0.0124\n",
            "Epoch 97/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.4509e-04 - mae: 0.0122\n",
            "Epoch 98/100\n",
            "43/43 [==============================] - 1s 15ms/step - loss: 1.5175e-04 - mae: 0.0124\n",
            "Epoch 99/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.4290e-04 - mae: 0.0126\n",
            "Epoch 100/100\n",
            "43/43 [==============================] - 1s 13ms/step - loss: 1.7159e-04 - mae: 0.0131\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "5g-nC_em8OIq",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 290
        },
        "outputId": "b3b4c449-4c8d-4c26-deb1-ac9174cbb349"
      },
      "source": [
        "plt.semilogx(history.history[\"lr\"], history.history[\"loss\"])\n",
        "plt.axis([1e-7, 1, 0, .1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(1e-07, 1.0, 0.0, 0.1)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0uzfjgZLXDVF",
        "colab_type": "text"
      },
      "source": [
        "#### Create Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "B7t0VrCH8OIr",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "837c3482-516c-4719-d513-42ef36ef99ec"
      },
      "source": [
        "# Clear back end\n",
        "keras.backend.clear_session()\n",
        "\n",
        "# Ensure reproducibility\n",
        "tf.random.set_seed(42)\n",
        "np.random.seed(42)\n",
        "\n",
        "# Set Window Size\n",
        "window_size = 30\n",
        "train_set = window_dataset(normalized_x_train.flatten(), window_size)\n",
        "valid_set = window_dataset(normalized_x_valid.flatten(), window_size)\n",
        "\n",
        "# Build 2 layer model with 10 neurons each and 1 output layer\n",
        "model = keras.models.Sequential([\n",
        "  keras.layers.Dense(10, activation=\"relu\", input_shape=[window_size]),\n",
        "  keras.layers.Dense(10, activation=\"relu\"),\n",
        "  keras.layers.Dense(1)\n",
        "])\n",
        "\n",
        "# Set optimizer\n",
        "optimizer = keras.optimizers.Nadam(lr=1e-2)\n",
        "model.compile(loss=keras.losses.Huber(),\n",
        "              optimizer=optimizer,\n",
        "              metrics=[\"mae\"])\n",
        "\n",
        "# Set early Stopping\n",
        "early_stopping = keras.callbacks.EarlyStopping(patience=20)\n",
        "\n",
        "# create save points for best model\n",
        "model_checkpoint = keras.callbacks.ModelCheckpoint(\n",
        "    \"my_checkpoint\", save_best_only=True)\n",
        "\n",
        "# Fit model\n",
        "history = model.fit(train_set, epochs=500,\n",
        "                    validation_data=valid_set,\n",
        "                    callbacks=[early_stopping, model_checkpoint])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/500\n",
            "     36/Unknown - 1s 14ms/step - loss: 0.0015 - mae: 0.0411WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/tracking/tracking.py:111: Model.state_updates (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "This property should not be used in TensorFlow 2.0, as updates are applied automatically.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/tracking/tracking.py:111: Layer.updates (from tensorflow.python.keras.engine.base_layer) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "This property should not be used in TensorFlow 2.0, as updates are applied automatically.\n",
            "INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 34ms/step - loss: 0.0029 - mae: 0.0486 - val_loss: 6.9175e-04 - val_mae: 0.0291\n",
            "Epoch 2/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 2.4322e-04 - mae: 0.0164 - val_loss: 0.0030 - val_mae: 0.0714\n",
            "Epoch 3/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 0.0020 - mae: 0.0329 - val_loss: 0.0034 - val_mae: 0.0752\n",
            "Epoch 4/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 2.1677e-04 - mae: 0.0156 - val_loss: 7.0642e-04 - val_mae: 0.0256\n",
            "Epoch 5/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 5.0527e-04 - mae: 0.0202 - val_loss: 0.0128 - val_mae: 0.1549\n",
            "Epoch 6/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.9926e-04 - mae: 0.0151 - val_loss: 8.0172e-04 - val_mae: 0.0340\n",
            "Epoch 7/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 6.2152e-04 - mae: 0.0194 - val_loss: 0.0643 - val_mae: 0.3531\n",
            "Epoch 8/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 0.0014 - mae: 0.0300 - val_loss: 0.0434 - val_mae: 0.2896\n",
            "Epoch 9/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 0.0010 - mae: 0.0285 - val_loss: 0.0431 - val_mae: 0.2887\n",
            "Epoch 10/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 2.6672e-04 - mae: 0.0170 - val_loss: 0.0011 - val_mae: 0.0353\n",
            "Epoch 11/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.7195e-04 - mae: 0.0138 - val_loss: 0.0037 - val_mae: 0.0793\n",
            "Epoch 12/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.7421e-04 - mae: 0.0139 - val_loss: 0.0047 - val_mae: 0.0920\n",
            "Epoch 13/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.8074e-04 - mae: 0.0140 - val_loss: 0.0046 - val_mae: 0.0907\n",
            "Epoch 14/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 2.5683e-04 - mae: 0.0154 - val_loss: 0.0174 - val_mae: 0.1819\n",
            "Epoch 15/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 3.9480e-04 - mae: 0.0186 - val_loss: 0.0474 - val_mae: 0.3034\n",
            "Epoch 16/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 3.1231e-04 - mae: 0.0189 - val_loss: 0.0044 - val_mae: 0.0886\n",
            "Epoch 17/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 3.9205e-04 - mae: 0.0178 - val_loss: 0.0355 - val_mae: 0.2618\n",
            "Epoch 18/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 7.8858e-04 - mae: 0.0240 - val_loss: 0.0717 - val_mae: 0.3736\n",
            "Epoch 19/500\n",
            "37/43 [========================>.....] - ETA: 0s - loss: 4.0185e-04 - mae: 0.0210INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 28ms/step - loss: 3.8366e-04 - mae: 0.0205 - val_loss: 6.8575e-04 - val_mae: 0.0312\n",
            "Epoch 20/500\n",
            "43/43 [==============================] - ETA: 0s - loss: 1.2275e-04 - mae: 0.0115INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 27ms/step - loss: 1.2275e-04 - mae: 0.0115 - val_loss: 4.0509e-04 - val_mae: 0.0215\n",
            "Epoch 21/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.9564e-04 - mae: 0.0137 - val_loss: 0.0114 - val_mae: 0.1468\n",
            "Epoch 22/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 1.4554e-04 - mae: 0.0128 - val_loss: 0.0013 - val_mae: 0.0420\n",
            "Epoch 23/500\n",
            "43/43 [==============================] - 1s 19ms/step - loss: 6.0419e-04 - mae: 0.0195 - val_loss: 0.0635 - val_mae: 0.3510\n",
            "Epoch 24/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 5.5263e-04 - mae: 0.0252 - val_loss: 0.0127 - val_mae: 0.1541\n",
            "Epoch 25/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 2.3583e-04 - mae: 0.0151 - val_loss: 0.0110 - val_mae: 0.1447\n",
            "Epoch 26/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 6.2320e-04 - mae: 0.0209 - val_loss: 0.0285 - val_mae: 0.2348\n",
            "Epoch 27/500\n",
            "36/43 [========================>.....] - ETA: 0s - loss: 1.9247e-04 - mae: 0.0148INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 28ms/step - loss: 1.8275e-04 - mae: 0.0144 - val_loss: 3.5815e-04 - val_mae: 0.0196\n",
            "Epoch 28/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 1.5982e-04 - mae: 0.0127 - val_loss: 0.0031 - val_mae: 0.0741\n",
            "Epoch 29/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.2635e-04 - mae: 0.0117 - val_loss: 0.0023 - val_mae: 0.0632\n",
            "Epoch 30/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 6.8207e-04 - mae: 0.0198 - val_loss: 0.0235 - val_mae: 0.2131\n",
            "Epoch 31/500\n",
            "43/43 [==============================] - 1s 19ms/step - loss: 2.2992e-04 - mae: 0.0157 - val_loss: 0.0076 - val_mae: 0.1190\n",
            "Epoch 32/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.4958e-04 - mae: 0.0127 - val_loss: 0.0024 - val_mae: 0.0640\n",
            "Epoch 33/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.3009e-04 - mae: 0.0119 - val_loss: 0.0024 - val_mae: 0.0647\n",
            "Epoch 34/500\n",
            "36/43 [========================>.....] - ETA: 0s - loss: 9.8502e-05 - mae: 0.0101INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 27ms/step - loss: 1.0103e-04 - mae: 0.0104 - val_loss: 3.2230e-04 - val_mae: 0.0184\n",
            "Epoch 35/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 9.5329e-05 - mae: 0.0098 - val_loss: 9.3711e-04 - val_mae: 0.0353\n",
            "Epoch 36/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 9.5531e-05 - mae: 0.0099 - val_loss: 0.0011 - val_mae: 0.0423\n",
            "Epoch 37/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 4.0171e-04 - mae: 0.0156 - val_loss: 0.0621 - val_mae: 0.3473\n",
            "Epoch 38/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 4.4138e-04 - mae: 0.0216 - val_loss: 0.0073 - val_mae: 0.1178\n",
            "Epoch 39/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.8919e-04 - mae: 0.0141 - val_loss: 0.0103 - val_mae: 0.1402\n",
            "Epoch 40/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 2.5158e-04 - mae: 0.0150 - val_loss: 0.0244 - val_mae: 0.2173\n",
            "Epoch 41/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.7997e-04 - mae: 0.0140 - val_loss: 0.0020 - val_mae: 0.0579\n",
            "Epoch 42/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.9304e-04 - mae: 0.0127 - val_loss: 0.0180 - val_mae: 0.1857\n",
            "Epoch 43/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.3784e-04 - mae: 0.0124 - val_loss: 3.3682e-04 - val_mae: 0.0177\n",
            "Epoch 44/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 1.7533e-04 - mae: 0.0122 - val_loss: 0.0114 - val_mae: 0.1476\n",
            "Epoch 45/500\n",
            "43/43 [==============================] - ETA: 0s - loss: 1.1406e-04 - mae: 0.0114INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 28ms/step - loss: 1.1406e-04 - mae: 0.0114 - val_loss: 3.1410e-04 - val_mae: 0.0196\n",
            "Epoch 46/500\n",
            "43/43 [==============================] - 1s 15ms/step - loss: 9.4102e-04 - mae: 0.0208 - val_loss: 0.0644 - val_mae: 0.3533\n",
            "Epoch 47/500\n",
            "43/43 [==============================] - 1s 19ms/step - loss: 3.7562e-04 - mae: 0.0205 - val_loss: 0.0034 - val_mae: 0.0766\n",
            "Epoch 48/500\n",
            "37/43 [========================>.....] - ETA: 0s - loss: 1.1360e-04 - mae: 0.0108INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 30ms/step - loss: 1.1491e-04 - mae: 0.0110 - val_loss: 2.8048e-04 - val_mae: 0.0169\n",
            "Epoch 49/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 4.9222e-04 - mae: 0.0200 - val_loss: 0.0072 - val_mae: 0.1169\n",
            "Epoch 50/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.5612e-04 - mae: 0.0130 - val_loss: 0.0019 - val_mae: 0.0572\n",
            "Epoch 51/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 8.8285e-05 - mae: 0.0097 - val_loss: 0.0014 - val_mae: 0.0483\n",
            "Epoch 52/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 8.2319e-05 - mae: 0.0092 - val_loss: 6.1766e-04 - val_mae: 0.0308\n",
            "Epoch 53/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.6749e-04 - mae: 0.0118 - val_loss: 0.0066 - val_mae: 0.1113\n",
            "Epoch 54/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.1897e-04 - mae: 0.0114 - val_loss: 0.0031 - val_mae: 0.0750\n",
            "Epoch 55/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.0451e-04 - mae: 0.0105 - val_loss: 0.0034 - val_mae: 0.0791\n",
            "Epoch 56/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.2336e-04 - mae: 0.0110 - val_loss: 0.0055 - val_mae: 0.1014\n",
            "Epoch 57/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 8.7549e-05 - mae: 0.0099 - val_loss: 5.4237e-04 - val_mae: 0.0286\n",
            "Epoch 58/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 4.4162e-04 - mae: 0.0154 - val_loss: 0.0584 - val_mae: 0.3370\n",
            "Epoch 59/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 3.2605e-04 - mae: 0.0188 - val_loss: 3.7197e-04 - val_mae: 0.0220\n",
            "Epoch 60/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 8.3775e-05 - mae: 0.0096 - val_loss: 6.8746e-04 - val_mae: 0.0330\n",
            "Epoch 61/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 4.3107e-04 - mae: 0.0158 - val_loss: 0.0275 - val_mae: 0.2313\n",
            "Epoch 62/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.6381e-04 - mae: 0.0132 - val_loss: 3.6612e-04 - val_mae: 0.0188\n",
            "Epoch 63/500\n",
            "42/43 [============================>.] - ETA: 0s - loss: 6.9676e-05 - mae: 0.0086INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 30ms/step - loss: 7.0046e-05 - mae: 0.0086 - val_loss: 2.4549e-04 - val_mae: 0.0161\n",
            "Epoch 64/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 2.2372e-04 - mae: 0.0128 - val_loss: 0.0175 - val_mae: 0.1836\n",
            "Epoch 65/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 2.7910e-04 - mae: 0.0167 - val_loss: 0.0118 - val_mae: 0.1499\n",
            "Epoch 66/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.3506e-04 - mae: 0.0123 - val_loss: 0.0015 - val_mae: 0.0495\n",
            "Epoch 67/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.0164e-04 - mae: 0.0101 - val_loss: 0.0041 - val_mae: 0.0872\n",
            "Epoch 68/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 8.0919e-05 - mae: 0.0093 - val_loss: 3.3062e-04 - val_mae: 0.0210\n",
            "Epoch 69/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.3207e-04 - mae: 0.0106 - val_loss: 0.0091 - val_mae: 0.1316\n",
            "Epoch 70/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.2811e-04 - mae: 0.0115 - val_loss: 0.0064 - val_mae: 0.1096\n",
            "Epoch 71/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 3.4479e-04 - mae: 0.0157 - val_loss: 0.0346 - val_mae: 0.2588\n",
            "Epoch 72/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 2.9486e-04 - mae: 0.0179 - val_loss: 0.0040 - val_mae: 0.0868\n",
            "Epoch 73/500\n",
            "39/43 [==========================>...] - ETA: 0s - loss: 9.0584e-05 - mae: 0.0099INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 29ms/step - loss: 9.0363e-05 - mae: 0.0099 - val_loss: 2.4221e-04 - val_mae: 0.0167\n",
            "Epoch 74/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.2586e-04 - mae: 0.0103 - val_loss: 0.0062 - val_mae: 0.1084\n",
            "Epoch 75/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 8.2119e-05 - mae: 0.0096 - val_loss: 2.4618e-04 - val_mae: 0.0171\n",
            "Epoch 76/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 7.3297e-05 - mae: 0.0086 - val_loss: 0.0027 - val_mae: 0.0698\n",
            "Epoch 77/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 4.9082e-04 - mae: 0.0163 - val_loss: 0.0272 - val_mae: 0.2303\n",
            "Epoch 78/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 0.0015 - mae: 0.0317 - val_loss: 2.6856e-04 - val_mae: 0.0174\n",
            "Epoch 79/500\n",
            "43/43 [==============================] - ETA: 0s - loss: 3.6879e-04 - mae: 0.0172INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 33ms/step - loss: 3.6879e-04 - mae: 0.0172 - val_loss: 2.3838e-04 - val_mae: 0.0146\n",
            "Epoch 80/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 9.8251e-05 - mae: 0.0111 - val_loss: 2.6916e-04 - val_mae: 0.0152\n",
            "Epoch 81/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 7.3113e-05 - mae: 0.0092 - val_loss: 3.0673e-04 - val_mae: 0.0167\n",
            "Epoch 82/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 6.9199e-05 - mae: 0.0087 - val_loss: 3.8825e-04 - val_mae: 0.0237\n",
            "Epoch 83/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 6.6651e-05 - mae: 0.0085 - val_loss: 5.3513e-04 - val_mae: 0.0255\n",
            "Epoch 84/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 6.7465e-05 - mae: 0.0085 - val_loss: 5.2859e-04 - val_mae: 0.0253\n",
            "Epoch 85/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 6.2883e-05 - mae: 0.0082 - val_loss: 3.5918e-04 - val_mae: 0.0190\n",
            "Epoch 86/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 6.9616e-05 - mae: 0.0085 - val_loss: 0.0013 - val_mae: 0.0468\n",
            "Epoch 87/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 6.4492e-05 - mae: 0.0084 - val_loss: 2.8320e-04 - val_mae: 0.0193\n",
            "Epoch 88/500\n",
            "43/43 [==============================] - 1s 21ms/step - loss: 8.9155e-05 - mae: 0.0093 - val_loss: 0.0035 - val_mae: 0.0802\n",
            "Epoch 89/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.8532e-04 - mae: 0.0126 - val_loss: 0.0169 - val_mae: 0.1806\n",
            "Epoch 90/500\n",
            "40/43 [==========================>...] - ETA: 0s - loss: 1.2953e-04 - mae: 0.0121INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 27ms/step - loss: 1.2638e-04 - mae: 0.0119 - val_loss: 2.1172e-04 - val_mae: 0.0148\n",
            "Epoch 91/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.5331e-04 - mae: 0.0111 - val_loss: 0.0121 - val_mae: 0.1526\n",
            "Epoch 92/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.1104e-04 - mae: 0.0114 - val_loss: 9.9723e-04 - val_mae: 0.0395\n",
            "Epoch 93/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.4646e-04 - mae: 0.0112 - val_loss: 0.0102 - val_mae: 0.1399\n",
            "Epoch 94/500\n",
            "43/43 [==============================] - 1s 19ms/step - loss: 2.6999e-04 - mae: 0.0152 - val_loss: 0.0213 - val_mae: 0.2029\n",
            "Epoch 95/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.9563e-04 - mae: 0.0148 - val_loss: 0.0056 - val_mae: 0.1034\n",
            "Epoch 96/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.2895e-04 - mae: 0.0116 - val_loss: 0.0074 - val_mae: 0.1190\n",
            "Epoch 97/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.1521e-04 - mae: 0.0115 - val_loss: 0.0036 - val_mae: 0.0812\n",
            "Epoch 98/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 7.7786e-05 - mae: 0.0094 - val_loss: 0.0020 - val_mae: 0.0589\n",
            "Epoch 99/500\n",
            "43/43 [==============================] - 1s 20ms/step - loss: 2.6625e-04 - mae: 0.0137 - val_loss: 0.0169 - val_mae: 0.1810\n",
            "Epoch 100/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.1992e-04 - mae: 0.0117 - val_loss: 0.0013 - val_mae: 0.0457\n",
            "Epoch 101/500\n",
            "43/43 [==============================] - ETA: 0s - loss: 5.9928e-05 - mae: 0.0082INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 29ms/step - loss: 5.9928e-05 - mae: 0.0082 - val_loss: 2.0259e-04 - val_mae: 0.0148\n",
            "Epoch 102/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 5.7421e-05 - mae: 0.0079 - val_loss: 0.0011 - val_mae: 0.0416\n",
            "Epoch 103/500\n",
            "43/43 [==============================] - 1s 19ms/step - loss: 7.0938e-05 - mae: 0.0086 - val_loss: 0.0021 - val_mae: 0.0616\n",
            "Epoch 104/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 2.9503e-04 - mae: 0.0136 - val_loss: 0.0373 - val_mae: 0.2693\n",
            "Epoch 105/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 2.3472e-04 - mae: 0.0158 - val_loss: 9.7447e-04 - val_mae: 0.0394\n",
            "Epoch 106/500\n",
            "37/43 [========================>.....] - ETA: 0s - loss: 6.3148e-05 - mae: 0.0083INFO:tensorflow:Assets written to: my_checkpoint/assets\n",
            "43/43 [==============================] - 1s 30ms/step - loss: 6.4445e-05 - mae: 0.0085 - val_loss: 1.9007e-04 - val_mae: 0.0137\n",
            "Epoch 107/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 2.9364e-04 - mae: 0.0130 - val_loss: 0.0414 - val_mae: 0.2836\n",
            "Epoch 108/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 2.6052e-04 - mae: 0.0166 - val_loss: 6.3941e-04 - val_mae: 0.0301\n",
            "Epoch 109/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.0048e-04 - mae: 0.0102 - val_loss: 0.0044 - val_mae: 0.0907\n",
            "Epoch 110/500\n",
            "43/43 [==============================] - 1s 20ms/step - loss: 7.5640e-05 - mae: 0.0094 - val_loss: 6.5307e-04 - val_mae: 0.0304\n",
            "Epoch 111/500\n",
            "43/43 [==============================] - 1s 19ms/step - loss: 1.0837e-04 - mae: 0.0099 - val_loss: 0.0049 - val_mae: 0.0962\n",
            "Epoch 112/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 8.0640e-05 - mae: 0.0096 - val_loss: 0.0016 - val_mae: 0.0535\n",
            "Epoch 113/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.4767e-04 - mae: 0.0111 - val_loss: 0.0172 - val_mae: 0.1824\n",
            "Epoch 114/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 1.5854e-04 - mae: 0.0135 - val_loss: 0.0055 - val_mae: 0.1021\n",
            "Epoch 115/500\n",
            "43/43 [==============================] - 1s 19ms/step - loss: 9.7914e-05 - mae: 0.0106 - val_loss: 0.0028 - val_mae: 0.0715\n",
            "Epoch 116/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 7.2016e-05 - mae: 0.0090 - val_loss: 0.0019 - val_mae: 0.0592\n",
            "Epoch 117/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 5.9175e-05 - mae: 0.0082 - val_loss: 5.7452e-04 - val_mae: 0.0281\n",
            "Epoch 118/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 3.7664e-04 - mae: 0.0140 - val_loss: 0.0327 - val_mae: 0.2525\n",
            "Epoch 119/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 3.5273e-04 - mae: 0.0187 - val_loss: 0.0069 - val_mae: 0.1149\n",
            "Epoch 120/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.1029e-04 - mae: 0.0114 - val_loss: 0.0011 - val_mae: 0.0424\n",
            "Epoch 121/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 6.5445e-05 - mae: 0.0085 - val_loss: 0.0012 - val_mae: 0.0437\n",
            "Epoch 122/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 5.4603e-05 - mae: 0.0078 - val_loss: 2.3933e-04 - val_mae: 0.0177\n",
            "Epoch 123/500\n",
            "43/43 [==============================] - 1s 17ms/step - loss: 8.5345e-05 - mae: 0.0091 - val_loss: 0.0031 - val_mae: 0.0761\n",
            "Epoch 124/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 1.0660e-04 - mae: 0.0104 - val_loss: 0.0044 - val_mae: 0.0912\n",
            "Epoch 125/500\n",
            "43/43 [==============================] - 1s 16ms/step - loss: 8.8041e-05 - mae: 0.0099 - val_loss: 0.0026 - val_mae: 0.0693\n",
            "Epoch 126/500\n",
            "43/43 [==============================] - 1s 18ms/step - loss: 2.6590e-04 - mae: 0.0141 - val_loss: 0.0093 - val_mae: 0.1337\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NfVoTx5sjKql",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = keras.models.load_model(\"my_checkpoint\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "RqQbX6DZ8OIu",
        "colab": {}
      },
      "source": [
        "dense_forecast = model_forecast(model, spy_normalized_to_traindata.flatten()[x_test.index.min() - window_size:-1], window_size)[:, 0]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9EZLjPLgyqci",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "b54335be-54b1-46e4-ee80-90a080494c4f"
      },
      "source": [
        "spy_normalized_to_traindata.flatten().shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(6950,)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RUhD-RDML_WY",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "b17fbc08-6c62-432f-f7cf-7dba0a0ae548"
      },
      "source": [
        "# Undo the scaling\n",
        "dense_forecast = x_train_scaler.inverse_transform(dense_forecast.reshape(-1,1)).flatten()\n",
        "dense_forecast.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(422,)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "98zwAuIo8OIv",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 410
        },
        "outputId": "cbe3d2d6-7e7c-4074-9a2b-2d7bae6adbf9"
      },
      "source": [
        "# set style of charts\n",
        "sns.set(style=\"darkgrid\")\n",
        "plt.rcParams['figure.figsize'] = [10, 10]\n",
        "\n",
        "plt.figure(figsize=(10, 6))\n",
        "plt.title('Fully Dense Forecast')\n",
        "plt.ylabel('Dollars $')\n",
        "plt.xlabel('Timestep in Days')\n",
        "plot_series(x_test.index, x_test)\n",
        "plot_series(x_test.index, dense_forecast)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NO6jYts6g6wV",
        "colab_type": "text"
      },
      "source": [
        "#### Dense Model Result"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "EgkELN-58OIw",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "8289bbbc-f700-4f27-bab3-2f81814be099"
      },
      "source": [
        "keras.metrics.mean_absolute_error(x_test, dense_forecast).numpy()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "5.4198823"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 42
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "S9fM69__PvMi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}