時系列株価予測git用.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "view-in-github"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Ken-huea/Time-series-analysis/blob/main/%E6%99%82%E7%B3%BB%E5%88%97%E6%A0%AA%E4%BE%A1%E4%BA%88%E6%B8%ACgit%E7%94%A8.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "50X6tDUGNmkT"
      },
      "source": [
        "## 問題\n",
        "- LSTMを用いて<font color=\"red\">終値</font>の予測を行えと。\n",
        "- 予測精度をRMSEを用いて計測。小さくなるようにするべし。\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FnNfK8-balxc"
      },
      "source": [
        "<font color=\"red\">ポイント</font>\n",
        "\n",
        "・時系列データのLSTM解析。LSTMの都合上、大部分のデータをpandasではなく<font color=\"red\">numpy</font>で行う。⇒面喰った。pandasに慣れているので。\n",
        "\n",
        "・データ分析の定石、時系列カラムをdatetime形式にしてindexにぶち込む。⇒今回は不要。\n",
        "\n",
        "pandas/dfで扱う際は有効だけど、今回はnumpuyなので。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FSdbAtLkxImo"
      },
      "source": [
        "### データ読み込み\n",
        "このセルを実行すればデータを取得できます。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zfgnBFE5wcAT"
      },
      "source": [
        "取り出したデータの値はLSTMでの分析に適合させるため、float型（浮動小数点型）に変換して読み込みます。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AbKH7GVgXqwy"
      },
      "source": [
        "何はともあれまずはデータの確認。確認してみると末尾に不要なデータがありそうですね。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "c8GMDE50Zvtu"
      },
      "outputs": [],
      "source": [
        "# 初めの一回だけこのセルを実行。データセットをダウンロードして展開。\n",
        "# 一回実行すれば、データセットはダウンロードされたまま。再起動後等再び実行する必要はなし。\n",
        "import urllib.request\n",
        "import zipfile\n",
        "%matplotlib inline\n",
        "\n",
        "\n",
        "# URLを指定\n",
        "url = \"https://storage.googleapis.com/tutor-contents-dataset/6160_sales_forecast_data.zip\"\n",
        "save_name = url.split('/')[-1]\n",
        "\n",
        "# ダウンロードする\n",
        "mem = urllib.request.urlopen(url).read()\n",
        "\n",
        "# ファイルへ保存\n",
        "with open(save_name, mode='wb') as f:\n",
        "    f.write(mem)\n",
        "\n",
        "# zipファイルをカレントディレクトリに展開する\n",
        "zfile = zipfile.ZipFile(save_name)\n",
        "zfile.extractall('.')\n",
        "\n",
        "!mv ./6160_sales_forecast_data/nikkei225.csv nikkei225.csv \n",
        "\n",
        "# 適当なモジュールのインポート\n",
        "import numpy\n",
        "import matplotlib.pyplot as plt\n",
        "from pandas import read_csv\n",
        "import math\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "from sklearn.preprocessing import MinMaxScaler\n",
        "from sklearn.metrics import mean_squared_error\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 143
        },
        "id": "g6LPoQEWXT5r",
        "outputId": "6f135cdd-9b9d-47e0-aa11-f4d64b1b6407"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-24949b4c-47e9-464e-9710-4b70a7eff530\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>日付け</th>\n",
              "      <th>終値</th>\n",
              "      <th>始値</th>\n",
              "      <th>高値</th>\n",
              "      <th>安値</th>\n",
              "      <th>出来高</th>\n",
              "      <th>前日比%</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>2063</th>\n",
              "      <td>2009年03月04日</td>\n",
              "      <td>7270</td>\n",
              "      <td>7200</td>\n",
              "      <td>7320</td>\n",
              "      <td>7090</td>\n",
              "      <td>79.49K</td>\n",
              "      <td>0.83</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2064</th>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2065</th>\n",
              "      <td>NaN</td>\n",
              "      <td>高値:20,950.0</td>\n",
              "      <td>安値:6,960.0</td>\n",
              "      <td>差:13,990.0</td>\n",
              "      <td>平均:13,086.6</td>\n",
              "      <td>前日比%:163.2</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-24949b4c-47e9-464e-9710-4b70a7eff530')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-24949b4c-47e9-464e-9710-4b70a7eff530 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-24949b4c-47e9-464e-9710-4b70a7eff530');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "              日付け           終値          始値          高値           安値  \\\n",
              "2063  2009年03月04日         7270        7200        7320         7090   \n",
              "2064          NaN          NaN         NaN         NaN          NaN   \n",
              "2065          NaN  高値:20,950.0  安値:6,960.0  差:13,990.0  平均:13,086.6   \n",
              "\n",
              "             出来高  前日比%  \n",
              "2063      79.49K  0.83  \n",
              "2064         NaN   NaN  \n",
              "2065  前日比%:163.2   NaN  "
            ]
          },
          "execution_count": 2,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "import pandas as pd\n",
        "# データの読込\n",
        "df = pd.read_csv(\"/content/nikkei225.csv\")\n",
        "df.tail(3)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fg3l-ccjZCNr"
      },
      "source": [
        "だからデータ読み込みの際に末尾をカットしましょうと。\n",
        "\n",
        "skipfooter：下から数行を読み飛ばす。　[詳細はこちら](https://www.yutaka-note.com/entry/pandas_read_csv#skipfooter%E4%B8%8B%E3%81%8B%E3%82%89%E6%95%B0%E8%A1%8C%E3%82%92%E8%AA%AD%E3%81%BF%E9%A3%9B%E3%81%B0%E3%81%99)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "VUAj3TBCYkfh",
        "outputId": "a665cf79-824e-4e7e-ccf8-09093d55d2b6"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-ad87216b-0864-4911-9136-e68224b66494\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>日付け</th>\n",
              "      <th>終値</th>\n",
              "      <th>始値</th>\n",
              "      <th>高値</th>\n",
              "      <th>安値</th>\n",
              "      <th>出来高</th>\n",
              "      <th>前日比%</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2017年01月31日</td>\n",
              "      <td>18980.0</td>\n",
              "      <td>19350.0</td>\n",
              "      <td>19390.0</td>\n",
              "      <td>18970.0</td>\n",
              "      <td>82.98K</td>\n",
              "      <td>-1.81</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2017年01月30日</td>\n",
              "      <td>19330.0</td>\n",
              "      <td>19500.0</td>\n",
              "      <td>19560.0</td>\n",
              "      <td>19280.0</td>\n",
              "      <td>48.95K</td>\n",
              "      <td>-0.67</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2017年01月27日</td>\n",
              "      <td>19460.0</td>\n",
              "      <td>19450.0</td>\n",
              "      <td>19560.0</td>\n",
              "      <td>19390.0</td>\n",
              "      <td>63.14K</td>\n",
              "      <td>0.00</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2017年01月26日</td>\n",
              "      <td>19460.0</td>\n",
              "      <td>19070.0</td>\n",
              "      <td>19460.0</td>\n",
              "      <td>19050.0</td>\n",
              "      <td>74.06K</td>\n",
              "      <td>2.05</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2017年01月25日</td>\n",
              "      <td>19070.0</td>\n",
              "      <td>18850.0</td>\n",
              "      <td>19130.0</td>\n",
              "      <td>18830.0</td>\n",
              "      <td>61.89K</td>\n",
              "      <td>1.22</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2059</th>\n",
              "      <td>2009年03月10日</td>\n",
              "      <td>7040.0</td>\n",
              "      <td>7050.0</td>\n",
              "      <td>7110.0</td>\n",
              "      <td>6960.0</td>\n",
              "      <td>85.37K</td>\n",
              "      <td>-0.14</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2060</th>\n",
              "      <td>2009年03月09日</td>\n",
              "      <td>7050.0</td>\n",
              "      <td>7180.0</td>\n",
              "      <td>7230.0</td>\n",
              "      <td>7020.0</td>\n",
              "      <td>74.47K</td>\n",
              "      <td>-1.67</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2061</th>\n",
              "      <td>2009年03月06日</td>\n",
              "      <td>7170.0</td>\n",
              "      <td>7400.0</td>\n",
              "      <td>7440.0</td>\n",
              "      <td>7150.0</td>\n",
              "      <td>91.24K</td>\n",
              "      <td>-3.24</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2062</th>\n",
              "      <td>2009年03月05日</td>\n",
              "      <td>7410.0</td>\n",
              "      <td>7310.0</td>\n",
              "      <td>7540.0</td>\n",
              "      <td>7290.0</td>\n",
              "      <td>114.64K</td>\n",
              "      <td>1.93</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2063</th>\n",
              "      <td>2009年03月04日</td>\n",
              "      <td>7270.0</td>\n",
              "      <td>7200.0</td>\n",
              "      <td>7320.0</td>\n",
              "      <td>7090.0</td>\n",
              "      <td>79.49K</td>\n",
              "      <td>0.83</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>2064 rows × 7 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-ad87216b-0864-4911-9136-e68224b66494')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-ad87216b-0864-4911-9136-e68224b66494 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-ad87216b-0864-4911-9136-e68224b66494');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "              日付け       終値       始値       高値       安値      出来高  前日比%\n",
              "0     2017年01月31日  18980.0  19350.0  19390.0  18970.0   82.98K -1.81\n",
              "1     2017年01月30日  19330.0  19500.0  19560.0  19280.0   48.95K -0.67\n",
              "2     2017年01月27日  19460.0  19450.0  19560.0  19390.0   63.14K  0.00\n",
              "3     2017年01月26日  19460.0  19070.0  19460.0  19050.0   74.06K  2.05\n",
              "4     2017年01月25日  19070.0  18850.0  19130.0  18830.0   61.89K  1.22\n",
              "...           ...      ...      ...      ...      ...      ...   ...\n",
              "2059  2009年03月10日   7040.0   7050.0   7110.0   6960.0   85.37K -0.14\n",
              "2060  2009年03月09日   7050.0   7180.0   7230.0   7020.0   74.47K -1.67\n",
              "2061  2009年03月06日   7170.0   7400.0   7440.0   7150.0   91.24K -3.24\n",
              "2062  2009年03月05日   7410.0   7310.0   7540.0   7290.0  114.64K  1.93\n",
              "2063  2009年03月04日   7270.0   7200.0   7320.0   7090.0   79.49K  0.83\n",
              "\n",
              "[2064 rows x 7 columns]"
            ]
          },
          "execution_count": 3,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# データセットの読み込み、かつ末尾をカット。\n",
        "df = pd.read_csv('/content/nikkei225.csv',\n",
        "                        # usecols=[1],              #読み込む列を指定\n",
        "                        engine='python', \n",
        "                        skipfooter=2                # ここが大事。\n",
        "                        )\n",
        "df"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_KODwAt2Xz9C",
        "outputId": "1dd0da76-0501-4b67-e2e0-a10fe0806c6a"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "日付け     0\n",
              "終値      0\n",
              "始値      0\n",
              "高値      0\n",
              "安値      0\n",
              "出来高     0\n",
              "前日比%    0\n",
              "dtype: int64"
            ]
          },
          "execution_count": 4,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# 欠損値の確認⇒平気そう。\n",
        "df.dtypes\n",
        "df.isnull().sum()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0sW3g_AkZh2z"
      },
      "source": [
        "#### indexに日付をdfにしてぶち込んでみますか。\n",
        "\n",
        "⇒結論としてはいらなかった。忘備録として残します。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "v2Mu22Ni8-yr",
        "outputId": "636031f8-8baf-42d4-b622-09485344408f"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "2017-01-31\n",
            "2017-01-30\n",
            "2017-01-27\n",
            "2017-01-26\n",
            "2017-01-25\n",
            "2017-01-24\n",
            "2017-01-23\n",
            "2017-01-20\n",
            "2017-01-19\n",
            "2017-01-18\n",
            "2017-01-17\n",
            "2017-01-16\n",
            "2017-01-13\n",
            "2017-01-12\n",
            "2017-01-11\n",
            "2017-01-10\n",
            "2017-01-09\n",
            "2017-01-06\n",
            "2017-01-05\n",
            "2017-01-04\n",
            "2017-01-03\n",
            "2016-12-30\n",
            "2016-12-29\n",
            "2016-12-28\n",
            "2016-12-27\n",
            "2016-12-26\n",
            "2016-12-23\n",
            "2016-12-22\n",
            "2016-12-21\n",
            "2016-12-20\n",
            "2016-12-19\n",
            "2016-12-16\n",
            "2016-12-15\n",
            "2016-12-14\n",
            "2016-12-13\n",
            "2016-12-12\n",
            "2016-12-09\n",
            "2016-12-08\n",
            "2016-12-07\n",
            "2016-12-06\n",
            "2016-12-05\n",
            "2016-12-02\n",
            "2016-12-01\n",
            "2016-11-30\n",
            "2016-11-29\n",
            "2016-11-28\n",
            "2016-11-25\n",
            "2016-11-24\n",
            "2016-11-23\n",
            "2016-11-22\n",
            "2016-11-21\n",
            "2016-11-18\n",
            "2016-11-17\n",
            "2016-11-16\n",
            "2016-11-15\n",
            "2016-11-14\n",
            "2016-11-11\n",
            "2016-11-10\n",
            "2016-11-09\n",
            "2016-11-08\n",
            "2016-11-07\n",
            "2016-11-04\n",
            "2016-11-03\n",
            "2016-11-02\n",
            "2016-11-01\n",
            "2016-10-31\n",
            "2016-10-28\n",
            "2016-10-27\n",
            "2016-10-26\n",
            "2016-10-25\n",
            "2016-10-24\n",
            "2016-10-21\n",
            "2016-10-20\n",
            "2016-10-19\n",
            "2016-10-18\n",
            "2016-10-17\n",
            "2016-10-14\n",
            "2016-10-13\n",
            "2016-10-12\n",
            "2016-10-11\n",
            "2016-10-10\n",
            "2016-10-09\n",
            "2016-10-07\n",
            "2016-10-06\n",
            "2016-10-05\n",
            "2016-10-04\n",
            "2016-10-03\n",
            "2016-09-30\n",
            "2016-09-29\n",
            "2016-09-28\n",
            "2016-09-27\n",
            "2016-09-26\n",
            "2016-09-23\n",
            "2016-09-22\n",
            "2016-09-21\n",
            "2016-09-20\n",
            "2016-09-19\n",
            "2016-09-18\n",
            "2016-09-16\n",
            "2016-09-15\n",
            "2016-09-14\n",
            "2016-09-13\n",
            "2016-09-12\n",
            "2016-09-09\n",
            "2016-09-08\n",
            "2016-09-07\n",
            "2016-09-06\n",
            "2016-09-05\n",
            "2016-09-02\n",
            "2016-09-01\n",
            "2016-08-31\n",
            "2016-08-30\n",
            "2016-08-29\n",
            "2016-08-26\n",
            "2016-08-25\n",
            "2016-08-24\n",
            "2016-08-23\n",
            "2016-08-22\n",
            "2016-08-19\n",
            "2016-08-18\n",
            "2016-08-17\n",
            "2016-08-16\n",
            "2016-08-15\n",
            "2016-08-12\n",
            "2016-08-11\n",
            "2016-08-10\n",
            "2016-08-09\n",
            "2016-08-08\n",
            "2016-08-05\n",
            "2016-08-04\n",
            "2016-08-03\n",
            "2016-08-02\n",
            "2016-08-01\n",
            "2016-07-29\n",
            "2016-07-28\n",
            "2016-07-27\n",
            "2016-07-26\n",
            "2016-07-25\n",
            "2016-07-22\n",
            "2016-07-21\n",
            "2016-07-20\n",
            "2016-07-19\n",
            "2016-07-18\n",
            "2016-07-17\n",
            "2016-07-15\n",
            "2016-07-14\n",
            "2016-07-13\n",
            "2016-07-12\n",
            "2016-07-11\n",
            "2016-07-08\n",
            "2016-07-07\n",
            "2016-07-06\n",
            "2016-07-05\n",
            "2016-07-04\n",
            "2016-07-01\n",
            "2016-06-30\n",
            "2016-06-29\n",
            "2016-06-28\n",
            "2016-06-27\n",
            "2016-06-24\n",
            "2016-06-23\n",
            "2016-06-22\n",
            "2016-06-21\n",
            "2016-06-20\n",
            "2016-06-17\n",
            "2016-06-16\n",
            "2016-06-15\n",
            "2016-06-14\n",
            "2016-06-13\n",
            "2016-06-10\n",
            "2016-06-09\n",
            "2016-06-08\n",
            "2016-06-07\n",
            "2016-06-06\n",
            "2016-06-03\n",
            "2016-06-02\n",
            "2016-06-01\n",
            "2016-05-31\n",
            "2016-05-30\n",
            "2016-05-27\n",
            "2016-05-26\n",
            "2016-05-25\n",
            "2016-05-24\n",
            "2016-05-23\n",
            "2016-05-20\n",
            "2016-05-19\n",
            "2016-05-18\n",
            "2016-05-17\n",
            "2016-05-16\n",
            "2016-05-13\n",
            "2016-05-12\n",
            "2016-05-11\n",
            "2016-05-10\n",
            "2016-05-09\n",
            "2016-05-06\n",
            "2016-05-05\n",
            "2016-05-04\n",
            "2016-05-03\n",
            "2016-05-02\n",
            "2016-04-29\n",
            "2016-04-28\n",
            "2016-04-27\n",
            "2016-04-26\n",
            "2016-04-25\n",
            "2016-04-22\n",
            "2016-04-21\n",
            "2016-04-20\n",
            "2016-04-19\n",
            "2016-04-18\n",
            "2016-04-15\n",
            "2016-04-14\n",
            "2016-04-13\n",
            "2016-04-12\n",
            "2016-04-11\n",
            "2016-04-08\n",
            "2016-04-07\n",
            "2016-04-06\n",
            "2016-04-05\n",
            "2016-04-04\n",
            "2016-04-01\n",
            "2016-03-31\n",
            "2016-03-30\n",
            "2016-03-29\n",
            "2016-03-28\n",
            "2016-03-25\n",
            "2016-03-24\n",
            "2016-03-23\n",
            "2016-03-22\n",
            "2016-03-21\n",
            "2016-03-20\n",
            "2016-03-18\n",
            "2016-03-17\n",
            "2016-03-16\n",
            "2016-03-15\n",
            "2016-03-14\n",
            "2016-03-11\n",
            "2016-03-10\n",
            "2016-03-09\n",
            "2016-03-08\n",
            "2016-03-07\n",
            "2016-03-04\n",
            "2016-03-03\n",
            "2016-03-02\n",
            "2016-03-01\n",
            "2016-02-29\n",
            "2016-02-26\n",
            "2016-02-25\n",
            "2016-02-24\n",
            "2016-02-23\n",
            "2016-02-22\n",
            "2016-02-19\n",
            "2016-02-18\n",
            "2016-02-17\n",
            "2016-02-16\n",
            "2016-02-15\n",
            "2016-02-12\n",
            "2016-02-11\n",
            "2016-02-10\n",
            "2016-02-09\n",
            "2016-02-08\n",
            "2016-02-05\n",
            "2016-02-04\n",
            "2016-02-03\n",
            "2016-02-02\n",
            "2016-02-01\n",
            "2016-01-29\n",
            "2016-01-28\n",
            "2016-01-27\n",
            "2016-01-26\n",
            "2016-01-25\n",
            "2016-01-22\n",
            "2016-01-21\n",
            "2016-01-20\n",
            "2016-01-19\n",
            "2016-01-18\n",
            "2016-01-15\n",
            "2016-01-14\n",
            "2016-01-13\n",
            "2016-01-12\n",
            "2016-01-11\n",
            "2016-01-10\n",
            "2016-01-08\n",
            "2016-01-07\n",
            "2016-01-06\n",
            "2016-01-05\n",
            "2016-01-04\n",
            "2015-12-31\n",
            "2015-12-30\n",
            "2015-12-29\n",
            "2015-12-28\n",
            "2015-12-25\n",
            "2015-12-24\n",
            "2015-12-23\n",
            "2015-12-22\n",
            "2015-12-21\n",
            "2015-12-18\n",
            "2015-12-17\n",
            "2015-12-16\n",
            "2015-12-15\n",
            "2015-12-14\n",
            "2015-12-11\n",
            "2015-12-10\n",
            "2015-12-09\n",
            "2015-12-08\n",
            "2015-12-07\n",
            "2015-12-04\n",
            "2015-12-03\n",
            "2015-12-02\n",
            "2015-12-01\n",
            "2015-11-30\n",
            "2015-11-27\n",
            "2015-11-26\n",
            "2015-11-25\n",
            "2015-11-24\n",
            "2015-11-23\n",
            "2015-11-22\n",
            "2015-11-20\n",
            "2015-11-19\n",
            "2015-11-18\n",
            "2015-11-17\n",
            "2015-11-16\n",
            "2015-11-13\n",
            "2015-11-12\n",
            "2015-11-11\n",
            "2015-11-10\n",
            "2015-11-09\n",
            "2015-11-06\n",
            "2015-11-05\n",
            "2015-11-04\n",
            "2015-11-03\n",
            "2015-11-02\n",
            "2015-10-30\n",
            "2015-10-29\n",
            "2015-10-28\n",
            "2015-10-27\n",
            "2015-10-26\n",
            "2015-10-23\n",
            "2015-10-22\n",
            "2015-10-21\n",
            "2015-10-20\n",
            "2015-10-19\n",
            "2015-10-16\n",
            "2015-10-15\n",
            "2015-10-14\n",
            "2015-10-13\n",
            "2015-10-12\n",
            "2015-10-11\n",
            "2015-10-09\n",
            "2015-10-08\n",
            "2015-10-07\n",
            "2015-10-06\n",
            "2015-10-05\n",
            "2015-10-02\n",
            "2015-10-01\n",
            "2015-09-30\n",
            "2015-09-29\n",
            "2015-09-28\n",
            "2015-09-25\n",
            "2015-09-24\n",
            "2015-09-23\n",
            "2015-09-22\n",
            "2015-09-21\n",
            "2015-09-20\n",
            "2015-09-18\n",
            "2015-09-17\n",
            "2015-09-16\n",
            "2015-09-15\n",
            "2015-09-14\n",
            "2015-09-11\n",
            "2015-09-10\n",
            "2015-09-09\n",
            "2015-09-08\n",
            "2015-09-07\n",
            "2015-09-04\n",
            "2015-09-03\n",
            "2015-09-02\n",
            "2015-09-01\n",
            "2015-08-31\n",
            "2015-08-28\n",
            "2015-08-27\n",
            "2015-08-26\n",
            "2015-08-25\n",
            "2015-08-24\n",
            "2015-08-21\n",
            "2015-08-20\n",
            "2015-08-19\n",
            "2015-08-18\n",
            "2015-08-17\n",
            "2015-08-14\n",
            "2015-08-13\n",
            "2015-08-12\n",
            "2015-08-11\n",
            "2015-08-10\n",
            "2015-08-07\n",
            "2015-08-06\n",
            "2015-08-05\n",
            "2015-08-04\n",
            "2015-08-03\n",
            "2015-07-31\n",
            "2015-07-30\n",
            "2015-07-29\n",
            "2015-07-28\n",
            "2015-07-27\n",
            "2015-07-24\n",
            "2015-07-23\n",
            "2015-07-22\n",
            "2015-07-21\n",
            "2015-07-20\n",
            "2015-07-19\n",
            "2015-07-17\n",
            "2015-07-16\n",
            "2015-07-15\n",
            "2015-07-14\n",
            "2015-07-13\n",
            "2015-07-10\n",
            "2015-07-09\n",
            "2015-07-08\n",
            "2015-07-07\n",
            "2015-07-06\n",
            "2015-07-03\n",
            "2015-07-02\n",
            "2015-07-01\n",
            "2015-06-30\n",
            "2015-06-29\n",
            "2015-06-26\n",
            "2015-06-25\n",
            "2015-06-24\n",
            "2015-06-23\n",
            "2015-06-22\n",
            "2015-06-19\n",
            "2015-06-18\n",
            "2015-06-17\n",
            "2015-06-16\n",
            "2015-06-15\n",
            "2015-06-12\n",
            "2015-06-11\n",
            "2015-06-10\n",
            "2015-06-09\n",
            "2015-06-08\n",
            "2015-06-05\n",
            "2015-06-04\n",
            "2015-06-03\n",
            "2015-06-02\n",
            "2015-06-01\n",
            "2015-05-29\n",
            "2015-05-28\n",
            "2015-05-27\n",
            "2015-05-26\n",
            "2015-05-25\n",
            "2015-05-22\n",
            "2015-05-21\n",
            "2015-05-20\n",
            "2015-05-19\n",
            "2015-05-18\n",
            "2015-05-15\n",
            "2015-05-14\n",
            "2015-05-13\n",
            "2015-05-12\n",
            "2015-05-11\n",
            "2015-05-08\n",
            "2015-05-07\n",
            "2015-05-06\n",
            "2015-05-05\n",
            "2015-05-04\n",
            "2015-05-03\n",
            "2015-05-01\n",
            "2015-04-30\n",
            "2015-04-29\n",
            "2015-04-28\n",
            "2015-04-27\n",
            "2015-04-24\n",
            "2015-04-23\n",
            "2015-04-22\n",
            "2015-04-21\n",
            "2015-04-20\n",
            "2015-04-17\n",
            "2015-04-16\n",
            "2015-04-15\n",
            "2015-04-14\n",
            "2015-04-13\n",
            "2015-04-10\n",
            "2015-04-09\n",
            "2015-04-08\n",
            "2015-04-07\n",
            "2015-04-06\n",
            "2015-04-03\n",
            "2015-04-02\n",
            "2015-04-01\n",
            "2015-03-31\n",
            "2015-03-30\n",
            "2015-03-27\n",
            "2015-03-26\n",
            "2015-03-25\n",
            "2015-03-24\n",
            "2015-03-23\n",
            "2015-03-20\n",
            "2015-03-19\n",
            "2015-03-18\n",
            "2015-03-17\n",
            "2015-03-16\n",
            "2015-03-13\n",
            "2015-03-12\n",
            "2015-03-11\n",
            "2015-03-10\n",
            "2015-03-09\n",
            "2015-03-06\n",
            "2015-03-05\n",
            "2015-03-04\n",
            "2015-03-03\n",
            "2015-03-02\n",
            "2015-02-27\n",
            "2015-02-26\n",
            "2015-02-25\n",
            "2015-02-24\n",
            "2015-02-23\n",
            "2015-02-20\n",
            "2015-02-19\n",
            "2015-02-18\n",
            "2015-02-17\n",
            "2015-02-16\n",
            "2015-02-13\n",
            "2015-02-12\n",
            "2015-02-11\n",
            "2015-02-10\n",
            "2015-02-09\n",
            "2015-02-06\n",
            "2015-02-05\n",
            "2015-02-04\n",
            "2015-02-03\n",
            "2015-02-02\n",
            "2015-01-30\n",
            "2015-01-29\n",
            "2015-01-28\n",
            "2015-01-27\n",
            "2015-01-26\n",
            "2015-01-23\n",
            "2015-01-22\n",
            "2015-01-21\n",
            "2015-01-20\n",
            "2015-01-19\n",
            "2015-01-16\n",
            "2015-01-15\n",
            "2015-01-14\n",
            "2015-01-13\n",
            "2015-01-12\n",
            "2015-01-11\n",
            "2015-01-09\n",
            "2015-01-08\n",
            "2015-01-07\n",
            "2015-01-06\n",
            "2015-01-05\n",
            "2015-01-02\n",
            "2015-01-01\n",
            "2014-12-31\n",
            "2014-12-30\n",
            "2014-12-29\n",
            "2014-12-26\n",
            "2014-12-25\n",
            "2014-12-24\n",
            "2014-12-23\n",
            "2014-12-22\n",
            "2014-12-19\n",
            "2014-12-18\n",
            "2014-12-17\n",
            "2014-12-16\n",
            "2014-12-15\n",
            "2014-12-12\n",
            "2014-12-11\n",
            "2014-12-10\n",
            "2014-12-09\n",
            "2014-12-08\n",
            "2014-12-05\n",
            "2014-12-04\n",
            "2014-12-03\n",
            "2014-12-02\n",
            "2014-12-01\n",
            "2014-11-28\n",
            "2014-11-27\n",
            "2014-11-26\n",
            "2014-11-25\n",
            "2014-11-24\n",
            "2014-11-23\n",
            "2014-11-21\n",
            "2014-11-20\n",
            "2014-11-19\n",
            "2014-11-18\n",
            "2014-11-17\n",
            "2014-11-14\n",
            "2014-11-13\n",
            "2014-11-12\n",
            "2014-11-11\n",
            "2014-11-10\n",
            "2014-11-07\n",
            "2014-11-06\n",
            "2014-11-05\n",
            "2014-11-04\n",
            "2014-11-03\n",
            "2014-10-31\n",
            "2014-10-30\n",
            "2014-10-29\n",
            "2014-10-28\n",
            "2014-10-27\n",
            "2014-10-24\n",
            "2014-10-23\n",
            "2014-10-22\n",
            "2014-10-21\n",
            "2014-10-20\n",
            "2014-10-17\n",
            "2014-10-16\n",
            "2014-10-15\n",
            "2014-10-14\n",
            "2014-10-13\n",
            "2014-10-10\n",
            "2014-10-09\n",
            "2014-10-08\n",
            "2014-10-07\n",
            "2014-10-06\n",
            "2014-10-03\n",
            "2014-10-02\n",
            "2014-10-01\n",
            "2014-09-30\n",
            "2014-09-29\n",
            "2014-09-26\n",
            "2014-09-25\n",
            "2014-09-24\n",
            "2014-09-23\n",
            "2014-09-22\n",
            "2014-09-19\n",
            "2014-09-18\n",
            "2014-09-17\n",
            "2014-09-16\n",
            "2014-09-15\n",
            "2014-09-12\n",
            "2014-09-11\n",
            "2014-09-10\n",
            "2014-09-09\n",
            "2014-09-08\n",
            "2014-09-05\n",
            "2014-09-04\n",
            "2014-09-03\n",
            "2014-09-02\n",
            "2014-09-01\n",
            "2014-08-29\n",
            "2014-08-28\n",
            "2014-08-27\n",
            "2014-08-26\n",
            "2014-08-25\n",
            "2014-08-22\n",
            "2014-08-21\n",
            "2014-08-20\n",
            "2014-08-19\n",
            "2014-08-18\n",
            "2014-08-15\n",
            "2014-08-14\n",
            "2014-08-13\n",
            "2014-08-12\n",
            "2014-08-11\n",
            "2014-08-08\n",
            "2014-08-07\n",
            "2014-08-06\n",
            "2014-08-05\n",
            "2014-08-04\n",
            "2014-08-01\n",
            "2014-07-31\n",
            "2014-07-30\n",
            "2014-07-29\n",
            "2014-07-28\n",
            "2014-07-25\n",
            "2014-07-24\n",
            "2014-07-23\n",
            "2014-07-22\n",
            "2014-07-21\n",
            "2014-07-18\n",
            "2014-07-17\n",
            "2014-07-16\n",
            "2014-07-15\n",
            "2014-07-14\n",
            "2014-07-11\n",
            "2014-07-10\n",
            "2014-07-09\n",
            "2014-07-08\n",
            "2014-07-07\n",
            "2014-07-04\n",
            "2014-07-03\n",
            "2014-07-02\n",
            "2014-07-01\n",
            "2014-06-30\n",
            "2014-06-27\n",
            "2014-06-26\n",
            "2014-06-25\n",
            "2014-06-24\n",
            "2014-06-23\n",
            "2014-06-20\n",
            "2014-06-19\n",
            "2014-06-18\n",
            "2014-06-17\n",
            "2014-06-16\n",
            "2014-06-13\n",
            "2014-06-12\n",
            "2014-06-11\n",
            "2014-06-10\n",
            "2014-06-09\n",
            "2014-06-06\n",
            "2014-06-05\n",
            "2014-06-04\n",
            "2014-06-03\n",
            "2014-06-02\n",
            "2014-05-30\n",
            "2014-05-29\n",
            "2014-05-28\n",
            "2014-05-27\n",
            "2014-05-26\n",
            "2014-05-23\n",
            "2014-05-22\n",
            "2014-05-21\n",
            "2014-05-20\n",
            "2014-05-19\n",
            "2014-05-16\n",
            "2014-05-15\n",
            "2014-05-14\n",
            "2014-05-13\n",
            "2014-05-12\n",
            "2014-05-09\n",
            "2014-05-08\n",
            "2014-05-07\n",
            "2014-05-06\n",
            "2014-05-05\n",
            "2014-05-02\n",
            "2014-05-01\n",
            "2014-04-30\n",
            "2014-04-29\n",
            "2014-04-28\n",
            "2014-04-25\n",
            "2014-04-24\n",
            "2014-04-23\n",
            "2014-04-22\n",
            "2014-04-21\n",
            "2014-04-18\n",
            "2014-04-17\n",
            "2014-04-16\n",
            "2014-04-15\n",
            "2014-04-14\n",
            "2014-04-11\n",
            "2014-04-10\n",
            "2014-04-09\n",
            "2014-04-08\n",
            "2014-04-07\n",
            "2014-04-04\n",
            "2014-04-03\n",
            "2014-04-02\n",
            "2014-04-01\n",
            "2014-03-31\n",
            "2014-03-28\n",
            "2014-03-27\n",
            "2014-03-26\n",
            "2014-03-25\n",
            "2014-03-24\n",
            "2014-03-21\n",
            "2014-03-20\n",
            "2014-03-19\n",
            "2014-03-18\n",
            "2014-03-17\n",
            "2014-03-14\n",
            "2014-03-13\n",
            "2014-03-12\n",
            "2014-03-11\n",
            "2014-03-10\n",
            "2014-03-07\n",
            "2014-03-06\n",
            "2014-03-05\n",
            "2014-03-04\n",
            "2014-03-03\n",
            "2014-02-28\n",
            "2014-02-27\n",
            "2014-02-26\n",
            "2014-02-25\n",
            "2014-02-24\n",
            "2014-02-21\n",
            "2014-02-20\n",
            "2014-02-19\n",
            "2014-02-18\n",
            "2014-02-17\n",
            "2014-02-14\n",
            "2014-02-13\n",
            "2014-02-12\n",
            "2014-02-11\n",
            "2014-02-10\n",
            "2014-02-07\n",
            "2014-02-06\n",
            "2014-02-05\n",
            "2014-02-04\n",
            "2014-02-03\n",
            "2014-01-31\n",
            "2014-01-30\n",
            "2014-01-29\n",
            "2014-01-28\n",
            "2014-01-27\n",
            "2014-01-24\n",
            "2014-01-23\n",
            "2014-01-22\n",
            "2014-01-21\n",
            "2014-01-20\n",
            "2014-01-17\n",
            "2014-01-16\n",
            "2014-01-15\n",
            "2014-01-14\n",
            "2014-01-13\n",
            "2014-01-10\n",
            "2014-01-09\n",
            "2014-01-08\n",
            "2014-01-07\n",
            "2014-01-06\n",
            "2014-01-03\n",
            "2014-01-02\n",
            "2014-01-01\n",
            "2013-12-31\n",
            "2013-12-30\n",
            "2013-12-27\n",
            "2013-12-26\n",
            "2013-12-25\n",
            "2013-12-24\n",
            "2013-12-23\n",
            "2013-12-20\n",
            "2013-12-19\n",
            "2013-12-18\n",
            "2013-12-17\n",
            "2013-12-16\n",
            "2013-12-13\n",
            "2013-12-12\n",
            "2013-12-11\n",
            "2013-12-10\n",
            "2013-12-09\n",
            "2013-12-06\n",
            "2013-12-05\n",
            "2013-12-04\n",
            "2013-12-03\n",
            "2013-12-02\n",
            "2013-11-29\n",
            "2013-11-28\n",
            "2013-11-27\n",
            "2013-11-26\n",
            "2013-11-25\n",
            "2013-11-22\n",
            "2013-11-21\n",
            "2013-11-20\n",
            "2013-11-19\n",
            "2013-11-18\n",
            "2013-11-15\n",
            "2013-11-14\n",
            "2013-11-13\n",
            "2013-11-12\n",
            "2013-11-11\n",
            "2013-11-08\n",
            "2013-11-07\n",
            "2013-11-06\n",
            "2013-11-05\n",
            "2013-11-04\n",
            "2013-11-01\n",
            "2013-10-31\n",
            "2013-10-30\n",
            "2013-10-29\n",
            "2013-10-28\n",
            "2013-10-25\n",
            "2013-10-24\n",
            "2013-10-23\n",
            "2013-10-22\n",
            "2013-10-21\n",
            "2013-10-18\n",
            "2013-10-17\n",
            "2013-10-16\n",
            "2013-10-15\n",
            "2013-10-14\n",
            "2013-10-11\n",
            "2013-10-10\n",
            "2013-10-09\n",
            "2013-10-08\n",
            "2013-10-07\n",
            "2013-10-04\n",
            "2013-10-03\n",
            "2013-10-02\n",
            "2013-10-01\n",
            "2013-09-30\n",
            "2013-09-27\n",
            "2013-09-26\n",
            "2013-09-25\n",
            "2013-09-24\n",
            "2013-09-23\n",
            "2013-09-20\n",
            "2013-09-19\n",
            "2013-09-18\n",
            "2013-09-17\n",
            "2013-09-16\n",
            "2013-09-13\n",
            "2013-09-12\n",
            "2013-09-11\n",
            "2013-09-10\n",
            "2013-09-09\n",
            "2013-09-06\n",
            "2013-09-05\n",
            "2013-09-04\n",
            "2013-09-03\n",
            "2013-09-02\n",
            "2013-08-30\n",
            "2013-08-29\n",
            "2013-08-28\n",
            "2013-08-27\n",
            "2013-08-26\n",
            "2013-08-23\n",
            "2013-08-22\n",
            "2013-08-21\n",
            "2013-08-20\n",
            "2013-08-19\n",
            "2013-08-16\n",
            "2013-08-15\n",
            "2013-08-14\n",
            "2013-08-13\n",
            "2013-08-12\n",
            "2013-08-09\n",
            "2013-08-08\n",
            "2013-08-07\n",
            "2013-08-06\n",
            "2013-08-05\n",
            "2013-08-02\n",
            "2013-08-01\n",
            "2013-07-31\n",
            "2013-07-30\n",
            "2013-07-29\n",
            "2013-07-26\n",
            "2013-07-25\n",
            "2013-07-24\n",
            "2013-07-23\n",
            "2013-07-22\n",
            "2013-07-19\n",
            "2013-07-18\n",
            "2013-07-17\n",
            "2013-07-16\n",
            "2013-07-15\n",
            "2013-07-12\n",
            "2013-07-11\n",
            "2013-07-10\n",
            "2013-07-09\n",
            "2013-07-08\n",
            "2013-07-05\n",
            "2013-07-04\n",
            "2013-07-03\n",
            "2013-07-02\n",
            "2013-07-01\n",
            "2013-06-28\n",
            "2013-06-27\n",
            "2013-06-26\n",
            "2013-06-25\n",
            "2013-06-24\n",
            "2013-06-21\n",
            "2013-06-20\n",
            "2013-06-19\n",
            "2013-06-18\n",
            "2013-06-17\n",
            "2013-06-14\n",
            "2013-06-13\n",
            "2013-06-12\n",
            "2013-06-11\n",
            "2013-06-10\n",
            "2013-06-07\n",
            "2013-06-06\n",
            "2013-06-05\n",
            "2013-06-04\n",
            "2013-06-03\n",
            "2013-05-31\n",
            "2013-05-30\n",
            "2013-05-29\n",
            "2013-05-28\n",
            "2013-05-27\n",
            "2013-05-24\n",
            "2013-05-23\n",
            "2013-05-22\n",
            "2013-05-21\n",
            "2013-05-20\n",
            "2013-05-17\n",
            "2013-05-16\n",
            "2013-05-15\n",
            "2013-05-14\n",
            "2013-05-13\n",
            "2013-05-10\n",
            "2013-05-09\n",
            "2013-05-08\n",
            "2013-05-07\n",
            "2013-05-06\n",
            "2013-05-03\n",
            "2013-05-02\n",
            "2013-05-01\n",
            "2013-04-30\n",
            "2013-04-29\n",
            "2013-04-26\n",
            "2013-04-25\n",
            "2013-04-24\n",
            "2013-04-23\n",
            "2013-04-22\n",
            "2013-04-19\n",
            "2013-04-18\n",
            "2013-04-17\n",
            "2013-04-16\n",
            "2013-04-15\n",
            "2013-04-12\n",
            "2013-04-11\n",
            "2013-04-10\n",
            "2013-04-09\n",
            "2013-04-08\n",
            "2013-04-05\n",
            "2013-04-04\n",
            "2013-04-03\n",
            "2013-04-02\n",
            "2013-04-01\n",
            "2013-03-29\n",
            "2013-03-28\n",
            "2013-03-27\n",
            "2013-03-26\n",
            "2013-03-25\n",
            "2013-03-22\n",
            "2013-03-21\n",
            "2013-03-20\n",
            "2013-03-19\n",
            "2013-03-18\n",
            "2013-03-15\n",
            "2013-03-14\n",
            "2013-03-13\n",
            "2013-03-12\n",
            "2013-03-11\n",
            "2013-03-08\n",
            "2013-03-07\n",
            "2013-03-06\n",
            "2013-03-05\n",
            "2013-03-04\n",
            "2013-03-01\n",
            "2013-02-28\n",
            "2013-02-27\n",
            "2013-02-26\n",
            "2013-02-25\n",
            "2013-02-22\n",
            "2013-02-21\n",
            "2013-02-20\n",
            "2013-02-19\n",
            "2013-02-18\n",
            "2013-02-15\n",
            "2013-02-14\n",
            "2013-02-13\n",
            "2013-02-12\n",
            "2013-02-11\n",
            "2013-02-08\n",
            "2013-02-07\n",
            "2013-02-06\n",
            "2013-02-05\n",
            "2013-02-04\n",
            "2013-02-01\n",
            "2013-01-31\n",
            "2013-01-30\n",
            "2013-01-29\n",
            "2013-01-28\n",
            "2013-01-25\n",
            "2013-01-24\n",
            "2013-01-23\n",
            "2013-01-22\n",
            "2013-01-21\n",
            "2013-01-18\n",
            "2013-01-17\n",
            "2013-01-16\n",
            "2013-01-15\n",
            "2013-01-14\n",
            "2013-01-11\n",
            "2013-01-10\n",
            "2013-01-09\n",
            "2013-01-08\n",
            "2013-01-07\n",
            "2013-01-04\n",
            "2013-01-03\n",
            "2013-01-02\n",
            "2013-01-01\n",
            "2012-12-31\n",
            "2012-12-28\n",
            "2012-12-27\n",
            "2012-12-26\n",
            "2012-12-25\n",
            "2012-12-24\n",
            "2012-12-21\n",
            "2012-12-20\n",
            "2012-12-19\n",
            "2012-12-18\n",
            "2012-12-17\n",
            "2012-12-14\n",
            "2012-12-13\n",
            "2012-12-12\n",
            "2012-12-11\n",
            "2012-12-10\n",
            "2012-12-07\n",
            "2012-12-06\n",
            "2012-12-05\n",
            "2012-12-04\n",
            "2012-12-03\n",
            "2012-11-30\n",
            "2012-11-29\n",
            "2012-11-28\n",
            "2012-11-27\n",
            "2012-11-26\n",
            "2012-11-23\n",
            "2012-11-22\n",
            "2012-11-21\n",
            "2012-11-20\n",
            "2012-11-19\n",
            "2012-11-16\n",
            "2012-11-15\n",
            "2012-11-14\n",
            "2012-11-13\n",
            "2012-11-12\n",
            "2012-11-09\n",
            "2012-11-08\n",
            "2012-11-07\n",
            "2012-11-06\n",
            "2012-11-05\n",
            "2012-11-02\n",
            "2012-11-01\n",
            "2012-10-31\n",
            "2012-10-30\n",
            "2012-10-29\n",
            "2012-10-26\n",
            "2012-10-25\n",
            "2012-10-24\n",
            "2012-10-23\n",
            "2012-10-22\n",
            "2012-10-19\n",
            "2012-10-18\n",
            "2012-10-17\n",
            "2012-10-16\n",
            "2012-10-15\n",
            "2012-10-12\n",
            "2012-10-11\n",
            "2012-10-10\n",
            "2012-10-09\n",
            "2012-10-08\n",
            "2012-10-05\n",
            "2012-10-04\n",
            "2012-10-03\n",
            "2012-10-02\n",
            "2012-10-01\n",
            "2012-09-28\n",
            "2012-09-27\n",
            "2012-09-26\n",
            "2012-09-25\n",
            "2012-09-24\n",
            "2012-09-21\n",
            "2012-09-20\n",
            "2012-09-19\n",
            "2012-09-18\n",
            "2012-09-17\n",
            "2012-09-14\n",
            "2012-09-13\n",
            "2012-09-12\n",
            "2012-09-11\n",
            "2012-09-10\n",
            "2012-09-07\n",
            "2012-09-06\n",
            "2012-09-05\n",
            "2012-09-04\n",
            "2012-09-03\n",
            "2012-08-31\n",
            "2012-08-30\n",
            "2012-08-29\n",
            "2012-08-28\n",
            "2012-08-27\n",
            "2012-08-24\n",
            "2012-08-23\n",
            "2012-08-22\n",
            "2012-08-21\n",
            "2012-08-20\n",
            "2012-08-17\n",
            "2012-08-16\n",
            "2012-08-15\n",
            "2012-08-14\n",
            "2012-08-13\n",
            "2012-08-10\n",
            "2012-08-09\n",
            "2012-08-08\n",
            "2012-08-07\n",
            "2012-08-06\n",
            "2012-08-03\n",
            "2012-08-02\n",
            "2012-08-01\n",
            "2012-07-31\n",
            "2012-07-30\n",
            "2012-07-27\n",
            "2012-07-26\n",
            "2012-07-25\n",
            "2012-07-24\n",
            "2012-07-23\n",
            "2012-07-20\n",
            "2012-07-19\n",
            "2012-07-18\n",
            "2012-07-17\n",
            "2012-07-16\n",
            "2012-07-13\n",
            "2012-07-12\n",
            "2012-07-11\n",
            "2012-07-10\n",
            "2012-07-09\n",
            "2012-07-06\n",
            "2012-07-05\n",
            "2012-07-04\n",
            "2012-07-03\n",
            "2012-07-02\n",
            "2012-06-29\n",
            "2012-06-28\n",
            "2012-06-27\n",
            "2012-06-26\n",
            "2012-06-25\n",
            "2012-06-22\n",
            "2012-06-21\n",
            "2012-06-20\n",
            "2012-06-19\n",
            "2012-06-18\n",
            "2012-06-15\n",
            "2012-06-14\n",
            "2012-06-13\n",
            "2012-06-12\n",
            "2012-06-11\n",
            "2012-06-08\n",
            "2012-06-07\n",
            "2012-06-06\n",
            "2012-06-05\n",
            "2012-06-04\n",
            "2012-06-01\n",
            "2012-05-31\n",
            "2012-05-30\n",
            "2012-05-29\n",
            "2012-05-28\n",
            "2012-05-25\n",
            "2012-05-24\n",
            "2012-05-23\n",
            "2012-05-22\n",
            "2012-05-21\n",
            "2012-05-18\n",
            "2012-05-17\n",
            "2012-05-16\n",
            "2012-05-15\n",
            "2012-05-14\n",
            "2012-05-11\n",
            "2012-05-10\n",
            "2012-05-09\n",
            "2012-05-08\n",
            "2012-05-07\n",
            "2012-05-04\n",
            "2012-05-03\n",
            "2012-05-02\n",
            "2012-05-01\n",
            "2012-04-30\n",
            "2012-04-27\n",
            "2012-04-26\n",
            "2012-04-25\n",
            "2012-04-24\n",
            "2012-04-23\n",
            "2012-04-20\n",
            "2012-04-19\n",
            "2012-04-18\n",
            "2012-04-17\n",
            "2012-04-16\n",
            "2012-04-13\n",
            "2012-04-12\n",
            "2012-04-11\n",
            "2012-04-10\n",
            "2012-04-09\n",
            "2012-04-06\n",
            "2012-04-05\n",
            "2012-04-04\n",
            "2012-04-03\n",
            "2012-04-02\n",
            "2012-03-30\n",
            "2012-03-29\n",
            "2012-03-28\n",
            "2012-03-27\n",
            "2012-03-26\n",
            "2012-03-23\n",
            "2012-03-22\n",
            "2012-03-21\n",
            "2012-03-20\n",
            "2012-03-19\n",
            "2012-03-16\n",
            "2012-03-15\n",
            "2012-03-14\n",
            "2012-03-13\n",
            "2012-03-12\n",
            "2012-03-09\n",
            "2012-03-08\n",
            "2012-03-07\n",
            "2012-03-06\n",
            "2012-03-05\n",
            "2012-03-02\n",
            "2012-03-01\n",
            "2012-02-29\n",
            "2012-02-28\n",
            "2012-02-27\n",
            "2012-02-24\n",
            "2012-02-23\n",
            "2012-02-22\n",
            "2012-02-21\n",
            "2012-02-20\n",
            "2012-02-17\n",
            "2012-02-16\n",
            "2012-02-15\n",
            "2012-02-14\n",
            "2012-02-13\n",
            "2012-02-10\n",
            "2012-02-09\n",
            "2012-02-08\n",
            "2012-02-07\n",
            "2012-02-06\n",
            "2012-02-03\n",
            "2012-02-02\n",
            "2012-02-01\n",
            "2012-01-31\n",
            "2012-01-30\n",
            "2012-01-27\n",
            "2012-01-26\n",
            "2012-01-25\n",
            "2012-01-24\n",
            "2012-01-23\n",
            "2012-01-20\n",
            "2012-01-19\n",
            "2012-01-18\n",
            "2012-01-17\n",
            "2012-01-16\n",
            "2012-01-13\n",
            "2012-01-12\n",
            "2012-01-11\n",
            "2012-01-10\n",
            "2012-01-09\n",
            "2012-01-06\n",
            "2012-01-05\n",
            "2012-01-04\n",
            "2012-01-03\n",
            "2012-01-02\n",
            "2011-12-30\n",
            "2011-12-29\n",
            "2011-12-28\n",
            "2011-12-27\n",
            "2011-12-26\n",
            "2011-12-23\n",
            "2011-12-22\n",
            "2011-12-21\n",
            "2011-12-20\n",
            "2011-12-19\n",
            "2011-12-16\n",
            "2011-12-15\n",
            "2011-12-14\n",
            "2011-12-13\n",
            "2011-12-12\n",
            "2011-12-09\n",
            "2011-12-08\n",
            "2011-12-07\n",
            "2011-12-06\n",
            "2011-12-05\n",
            "2011-12-02\n",
            "2011-12-01\n",
            "2011-11-30\n",
            "2011-11-29\n",
            "2011-11-28\n",
            "2011-11-25\n",
            "2011-11-24\n",
            "2011-11-23\n",
            "2011-11-22\n",
            "2011-11-21\n",
            "2011-11-18\n",
            "2011-11-17\n",
            "2011-11-16\n",
            "2011-11-15\n",
            "2011-11-14\n",
            "2011-11-11\n",
            "2011-11-10\n",
            "2011-11-09\n",
            "2011-11-08\n",
            "2011-11-07\n",
            "2011-11-04\n",
            "2011-11-03\n",
            "2011-11-02\n",
            "2011-11-01\n",
            "2011-10-31\n",
            "2011-10-28\n",
            "2011-10-27\n",
            "2011-10-26\n",
            "2011-10-25\n",
            "2011-10-24\n",
            "2011-10-21\n",
            "2011-10-20\n",
            "2011-10-19\n",
            "2011-10-18\n",
            "2011-10-17\n",
            "2011-10-14\n",
            "2011-10-13\n",
            "2011-10-12\n",
            "2011-10-11\n",
            "2011-10-10\n",
            "2011-10-07\n",
            "2011-10-06\n",
            "2011-10-05\n",
            "2011-10-04\n",
            "2011-10-03\n",
            "2011-09-30\n",
            "2011-09-29\n",
            "2011-09-28\n",
            "2011-09-27\n",
            "2011-09-26\n",
            "2011-09-23\n",
            "2011-09-22\n",
            "2011-09-21\n",
            "2011-09-20\n",
            "2011-09-19\n",
            "2011-09-16\n",
            "2011-09-15\n",
            "2011-09-14\n",
            "2011-09-13\n",
            "2011-09-12\n",
            "2011-09-09\n",
            "2011-09-08\n",
            "2011-09-07\n",
            "2011-09-06\n",
            "2011-09-05\n",
            "2011-09-02\n",
            "2011-09-01\n",
            "2011-08-31\n",
            "2011-08-30\n",
            "2011-08-29\n",
            "2011-08-26\n",
            "2011-08-25\n",
            "2011-08-24\n",
            "2011-08-23\n",
            "2011-08-22\n",
            "2011-08-19\n",
            "2011-08-18\n",
            "2011-08-17\n",
            "2011-08-16\n",
            "2011-08-15\n",
            "2011-08-12\n",
            "2011-08-11\n",
            "2011-08-10\n",
            "2011-08-09\n",
            "2011-08-08\n",
            "2011-08-05\n",
            "2011-08-04\n",
            "2011-08-03\n",
            "2011-08-02\n",
            "2011-08-01\n",
            "2011-07-29\n",
            "2011-07-28\n",
            "2011-07-27\n",
            "2011-07-26\n",
            "2011-07-25\n",
            "2011-07-22\n",
            "2011-07-21\n",
            "2011-07-20\n",
            "2011-07-19\n",
            "2011-07-18\n",
            "2011-07-15\n",
            "2011-07-14\n",
            "2011-07-13\n",
            "2011-07-12\n",
            "2011-07-11\n",
            "2011-07-08\n",
            "2011-07-07\n",
            "2011-07-06\n",
            "2011-07-05\n",
            "2011-07-04\n",
            "2011-07-01\n",
            "2011-06-30\n",
            "2011-06-29\n",
            "2011-06-28\n",
            "2011-06-27\n",
            "2011-06-24\n",
            "2011-06-23\n",
            "2011-06-22\n",
            "2011-06-21\n",
            "2011-06-20\n",
            "2011-06-17\n",
            "2011-06-16\n",
            "2011-06-15\n",
            "2011-06-14\n",
            "2011-06-13\n",
            "2011-06-10\n",
            "2011-06-09\n",
            "2011-06-08\n",
            "2011-06-07\n",
            "2011-06-06\n",
            "2011-06-03\n",
            "2011-06-02\n",
            "2011-06-01\n",
            "2011-05-31\n",
            "2011-05-30\n",
            "2011-05-27\n",
            "2011-05-26\n",
            "2011-05-25\n",
            "2011-05-24\n",
            "2011-05-23\n",
            "2011-05-20\n",
            "2011-05-19\n",
            "2011-05-18\n",
            "2011-05-17\n",
            "2011-05-16\n",
            "2011-05-13\n",
            "2011-05-12\n",
            "2011-05-11\n",
            "2011-05-10\n",
            "2011-05-09\n",
            "2011-05-06\n",
            "2011-05-05\n",
            "2011-05-04\n",
            "2011-05-03\n",
            "2011-05-02\n",
            "2011-04-29\n",
            "2011-04-28\n",
            "2011-04-27\n",
            "2011-04-26\n",
            "2011-04-25\n",
            "2011-04-22\n",
            "2011-04-21\n",
            "2011-04-20\n",
            "2011-04-19\n",
            "2011-04-18\n",
            "2011-04-15\n",
            "2011-04-14\n",
            "2011-04-13\n",
            "2011-04-12\n",
            "2011-04-11\n",
            "2011-04-08\n",
            "2011-04-07\n",
            "2011-04-06\n",
            "2011-04-05\n",
            "2011-04-04\n",
            "2011-04-01\n",
            "2011-03-31\n",
            "2011-03-30\n",
            "2011-03-29\n",
            "2011-03-28\n",
            "2011-03-25\n",
            "2011-03-24\n",
            "2011-03-23\n",
            "2011-03-22\n",
            "2011-03-21\n",
            "2011-03-18\n",
            "2011-03-17\n",
            "2011-03-16\n",
            "2011-03-15\n",
            "2011-03-14\n",
            "2011-03-11\n",
            "2011-03-10\n",
            "2011-03-09\n",
            "2011-03-08\n",
            "2011-03-07\n",
            "2011-03-04\n",
            "2011-03-03\n",
            "2011-03-02\n",
            "2011-03-01\n",
            "2011-02-28\n",
            "2011-02-25\n",
            "2011-02-24\n",
            "2011-02-23\n",
            "2011-02-22\n",
            "2011-02-21\n",
            "2011-02-18\n",
            "2011-02-17\n",
            "2011-02-16\n",
            "2011-02-15\n",
            "2011-02-14\n",
            "2011-02-11\n",
            "2011-02-10\n",
            "2011-02-09\n",
            "2011-02-08\n",
            "2011-02-07\n",
            "2011-02-04\n",
            "2011-02-03\n",
            "2011-02-02\n",
            "2011-02-01\n",
            "2011-01-31\n",
            "2011-01-28\n",
            "2011-01-27\n",
            "2011-01-26\n",
            "2011-01-25\n",
            "2011-01-24\n",
            "2011-01-21\n",
            "2011-01-20\n",
            "2011-01-19\n",
            "2011-01-18\n",
            "2011-01-17\n",
            "2011-01-14\n",
            "2011-01-13\n",
            "2011-01-12\n",
            "2011-01-11\n",
            "2011-01-10\n",
            "2011-01-07\n",
            "2011-01-06\n",
            "2011-01-05\n",
            "2011-01-04\n",
            "2011-01-03\n",
            "2010-12-31\n",
            "2010-12-30\n",
            "2010-12-29\n",
            "2010-12-28\n",
            "2010-12-27\n",
            "2010-12-24\n",
            "2010-12-23\n",
            "2010-12-22\n",
            "2010-12-21\n",
            "2010-12-20\n",
            "2010-12-17\n",
            "2010-12-16\n",
            "2010-12-15\n",
            "2010-12-14\n",
            "2010-12-13\n",
            "2010-12-10\n",
            "2010-12-09\n",
            "2010-12-08\n",
            "2010-12-07\n",
            "2010-12-06\n",
            "2010-12-03\n",
            "2010-12-02\n",
            "2010-12-01\n",
            "2010-11-30\n",
            "2010-11-29\n",
            "2010-11-26\n",
            "2010-11-25\n",
            "2010-11-24\n",
            "2010-11-23\n",
            "2010-11-22\n",
            "2010-11-19\n",
            "2010-11-18\n",
            "2010-11-17\n",
            "2010-11-16\n",
            "2010-11-15\n",
            "2010-11-12\n",
            "2010-11-11\n",
            "2010-11-10\n",
            "2010-11-09\n",
            "2010-11-08\n",
            "2010-11-05\n",
            "2010-11-04\n",
            "2010-11-03\n",
            "2010-11-02\n",
            "2010-11-01\n",
            "2010-10-29\n",
            "2010-10-28\n",
            "2010-10-27\n",
            "2010-10-26\n",
            "2010-10-25\n",
            "2010-10-22\n",
            "2010-10-21\n",
            "2010-10-20\n",
            "2010-10-19\n",
            "2010-10-18\n",
            "2010-10-15\n",
            "2010-10-14\n",
            "2010-10-13\n",
            "2010-10-12\n",
            "2010-10-11\n",
            "2010-10-08\n",
            "2010-10-07\n",
            "2010-10-06\n",
            "2010-10-05\n",
            "2010-10-04\n",
            "2010-10-01\n",
            "2010-09-30\n",
            "2010-09-29\n",
            "2010-09-28\n",
            "2010-09-27\n",
            "2010-09-24\n",
            "2010-09-23\n",
            "2010-09-22\n",
            "2010-09-21\n",
            "2010-09-20\n",
            "2010-09-17\n",
            "2010-09-16\n",
            "2010-09-15\n",
            "2010-09-14\n",
            "2010-09-13\n",
            "2010-09-10\n",
            "2010-09-09\n",
            "2010-09-08\n",
            "2010-09-07\n",
            "2010-09-06\n",
            "2010-09-03\n",
            "2010-09-02\n",
            "2010-09-01\n",
            "2010-08-31\n",
            "2010-08-30\n",
            "2010-08-27\n",
            "2010-08-26\n",
            "2010-08-25\n",
            "2010-08-24\n",
            "2010-08-23\n",
            "2010-08-20\n",
            "2010-08-19\n",
            "2010-08-18\n",
            "2010-08-17\n",
            "2010-08-16\n",
            "2010-08-13\n",
            "2010-08-12\n",
            "2010-08-11\n",
            "2010-08-10\n",
            "2010-08-09\n",
            "2010-08-06\n",
            "2010-08-05\n",
            "2010-08-04\n",
            "2010-08-03\n",
            "2010-08-02\n",
            "2010-07-30\n",
            "2010-07-29\n",
            "2010-07-28\n",
            "2010-07-27\n",
            "2010-07-26\n",
            "2010-07-23\n",
            "2010-07-22\n",
            "2010-07-21\n",
            "2010-07-20\n",
            "2010-07-19\n",
            "2010-07-16\n",
            "2010-07-15\n",
            "2010-07-14\n",
            "2010-07-13\n",
            "2010-07-12\n",
            "2010-07-09\n",
            "2010-07-08\n",
            "2010-07-07\n",
            "2010-07-06\n",
            "2010-07-05\n",
            "2010-07-02\n",
            "2010-07-01\n",
            "2010-06-30\n",
            "2010-06-29\n",
            "2010-06-28\n",
            "2010-06-25\n",
            "2010-06-24\n",
            "2010-06-23\n",
            "2010-06-22\n",
            "2010-06-21\n",
            "2010-06-18\n",
            "2010-06-17\n",
            "2010-06-16\n",
            "2010-06-15\n",
            "2010-06-14\n",
            "2010-06-11\n",
            "2010-06-10\n",
            "2010-06-09\n",
            "2010-06-08\n",
            "2010-06-07\n",
            "2010-06-04\n",
            "2010-06-03\n",
            "2010-06-02\n",
            "2010-06-01\n",
            "2010-05-31\n",
            "2010-05-28\n",
            "2010-05-27\n",
            "2010-05-26\n",
            "2010-05-25\n",
            "2010-05-24\n",
            "2010-05-21\n",
            "2010-05-20\n",
            "2010-05-19\n",
            "2010-05-18\n",
            "2010-05-17\n",
            "2010-05-14\n",
            "2010-05-13\n",
            "2010-05-12\n",
            "2010-05-11\n",
            "2010-05-10\n",
            "2010-05-07\n",
            "2010-05-06\n",
            "2010-05-05\n",
            "2010-05-04\n",
            "2010-05-03\n",
            "2010-04-30\n",
            "2010-04-29\n",
            "2010-04-28\n",
            "2010-04-27\n",
            "2010-04-26\n",
            "2010-04-23\n",
            "2010-04-22\n",
            "2010-04-21\n",
            "2010-04-20\n",
            "2010-04-19\n",
            "2010-04-16\n",
            "2010-04-15\n",
            "2010-04-14\n",
            "2010-04-13\n",
            "2010-04-12\n",
            "2010-04-09\n",
            "2010-04-08\n",
            "2010-04-07\n",
            "2010-04-06\n",
            "2010-04-05\n",
            "2010-04-02\n",
            "2010-04-01\n",
            "2010-03-31\n",
            "2010-03-30\n",
            "2010-03-29\n",
            "2010-03-26\n",
            "2010-03-25\n",
            "2010-03-24\n",
            "2010-03-23\n",
            "2010-03-22\n",
            "2010-03-19\n",
            "2010-03-18\n",
            "2010-03-17\n",
            "2010-03-16\n",
            "2010-03-15\n",
            "2010-03-12\n",
            "2010-03-11\n",
            "2010-03-10\n",
            "2010-03-09\n",
            "2010-03-08\n",
            "2010-03-05\n",
            "2010-03-04\n",
            "2010-03-03\n",
            "2010-03-02\n",
            "2010-03-01\n",
            "2010-02-26\n",
            "2010-02-25\n",
            "2010-02-24\n",
            "2010-02-23\n",
            "2010-02-22\n",
            "2010-02-19\n",
            "2010-02-18\n",
            "2010-02-17\n",
            "2010-02-16\n",
            "2010-02-15\n",
            "2010-02-12\n",
            "2010-02-11\n",
            "2010-02-10\n",
            "2010-02-09\n",
            "2010-02-08\n",
            "2010-02-05\n",
            "2010-02-04\n",
            "2010-02-03\n",
            "2010-02-02\n",
            "2010-02-01\n",
            "2010-01-29\n",
            "2010-01-28\n",
            "2010-01-27\n",
            "2010-01-26\n",
            "2010-01-25\n",
            "2010-01-22\n",
            "2010-01-21\n",
            "2010-01-20\n",
            "2010-01-19\n",
            "2010-01-18\n",
            "2010-01-15\n",
            "2010-01-14\n",
            "2010-01-13\n",
            "2010-01-12\n",
            "2010-01-11\n",
            "2010-01-08\n",
            "2010-01-07\n",
            "2010-01-06\n",
            "2010-01-05\n",
            "2010-01-04\n",
            "2010-01-01\n",
            "2009-12-31\n",
            "2009-12-30\n",
            "2009-12-29\n",
            "2009-12-28\n",
            "2009-12-25\n",
            "2009-12-24\n",
            "2009-12-23\n",
            "2009-12-22\n",
            "2009-12-21\n",
            "2009-12-18\n",
            "2009-12-17\n",
            "2009-12-16\n",
            "2009-12-15\n",
            "2009-12-14\n",
            "2009-12-11\n",
            "2009-12-10\n",
            "2009-12-09\n",
            "2009-12-08\n",
            "2009-12-07\n",
            "2009-12-04\n",
            "2009-12-03\n",
            "2009-12-02\n",
            "2009-12-01\n",
            "2009-11-30\n",
            "2009-11-27\n",
            "2009-11-26\n",
            "2009-11-25\n",
            "2009-11-24\n",
            "2009-11-23\n",
            "2009-11-20\n",
            "2009-11-19\n",
            "2009-11-18\n",
            "2009-11-17\n",
            "2009-11-16\n",
            "2009-11-13\n",
            "2009-11-12\n",
            "2009-11-11\n",
            "2009-11-10\n",
            "2009-11-09\n",
            "2009-11-06\n",
            "2009-11-05\n",
            "2009-11-04\n",
            "2009-11-02\n",
            "2009-10-30\n",
            "2009-10-29\n",
            "2009-10-28\n",
            "2009-10-27\n",
            "2009-10-26\n",
            "2009-10-23\n",
            "2009-10-22\n",
            "2009-10-21\n",
            "2009-10-20\n",
            "2009-10-19\n",
            "2009-10-16\n",
            "2009-10-15\n",
            "2009-10-14\n",
            "2009-10-13\n",
            "2009-10-09\n",
            "2009-10-08\n",
            "2009-10-07\n",
            "2009-10-06\n",
            "2009-10-05\n",
            "2009-10-02\n",
            "2009-10-01\n",
            "2009-09-30\n",
            "2009-09-29\n",
            "2009-09-28\n",
            "2009-09-25\n",
            "2009-09-24\n",
            "2009-09-18\n",
            "2009-09-17\n",
            "2009-09-16\n",
            "2009-09-15\n",
            "2009-09-14\n",
            "2009-09-11\n",
            "2009-09-10\n",
            "2009-09-09\n",
            "2009-09-08\n",
            "2009-09-07\n",
            "2009-09-04\n",
            "2009-09-03\n",
            "2009-09-02\n",
            "2009-09-01\n",
            "2009-08-31\n",
            "2009-08-28\n",
            "2009-08-27\n",
            "2009-08-26\n",
            "2009-08-25\n",
            "2009-08-24\n",
            "2009-08-21\n",
            "2009-08-20\n",
            "2009-08-19\n",
            "2009-08-18\n",
            "2009-08-17\n",
            "2009-08-14\n",
            "2009-08-13\n",
            "2009-08-12\n",
            "2009-08-11\n",
            "2009-08-10\n",
            "2009-08-07\n",
            "2009-08-06\n",
            "2009-08-05\n",
            "2009-08-04\n",
            "2009-08-03\n",
            "2009-07-31\n",
            "2009-07-30\n",
            "2009-07-29\n",
            "2009-07-28\n",
            "2009-07-27\n",
            "2009-07-24\n",
            "2009-07-23\n",
            "2009-07-22\n",
            "2009-07-21\n",
            "2009-07-17\n",
            "2009-07-16\n",
            "2009-07-15\n",
            "2009-07-14\n",
            "2009-07-13\n",
            "2009-07-10\n",
            "2009-07-09\n",
            "2009-07-08\n",
            "2009-07-07\n",
            "2009-07-06\n",
            "2009-07-03\n",
            "2009-07-02\n",
            "2009-07-01\n",
            "2009-06-30\n",
            "2009-06-29\n",
            "2009-06-26\n",
            "2009-06-25\n",
            "2009-06-24\n",
            "2009-06-23\n",
            "2009-06-22\n",
            "2009-06-19\n",
            "2009-06-18\n",
            "2009-06-17\n",
            "2009-06-16\n",
            "2009-06-15\n",
            "2009-06-12\n",
            "2009-06-11\n",
            "2009-06-10\n",
            "2009-06-09\n",
            "2009-06-08\n",
            "2009-06-05\n",
            "2009-06-04\n",
            "2009-06-03\n",
            "2009-06-02\n",
            "2009-06-01\n",
            "2009-05-29\n",
            "2009-05-28\n",
            "2009-05-27\n",
            "2009-05-26\n",
            "2009-05-25\n",
            "2009-05-22\n",
            "2009-05-21\n",
            "2009-05-20\n",
            "2009-05-19\n",
            "2009-05-18\n",
            "2009-05-15\n",
            "2009-05-14\n",
            "2009-05-13\n",
            "2009-05-12\n",
            "2009-05-11\n",
            "2009-05-08\n",
            "2009-05-07\n",
            "2009-05-01\n",
            "2009-04-30\n",
            "2009-04-28\n",
            "2009-04-27\n",
            "2009-04-24\n",
            "2009-04-23\n",
            "2009-04-22\n",
            "2009-04-21\n",
            "2009-04-20\n",
            "2009-04-17\n",
            "2009-04-16\n",
            "2009-04-15\n",
            "2009-04-14\n",
            "2009-04-13\n",
            "2009-04-10\n",
            "2009-04-09\n",
            "2009-04-08\n",
            "2009-04-07\n",
            "2009-04-06\n",
            "2009-04-03\n",
            "2009-04-02\n",
            "2009-04-01\n",
            "2009-03-31\n",
            "2009-03-30\n",
            "2009-03-27\n",
            "2009-03-26\n",
            "2009-03-25\n",
            "2009-03-24\n",
            "2009-03-23\n",
            "2009-03-19\n",
            "2009-03-18\n",
            "2009-03-17\n",
            "2009-03-16\n",
            "2009-03-13\n",
            "2009-03-12\n",
            "2009-03-11\n",
            "2009-03-10\n",
            "2009-03-09\n",
            "2009-03-06\n",
            "2009-03-05\n",
            "2009-03-04\n"
          ]
        },
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-8f14e58c-067d-460e-95ef-535676498ef6\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>終値</th>\n",
              "      <th>始値</th>\n",
              "      <th>高値</th>\n",
              "      <th>安値</th>\n",
              "      <th>出来高</th>\n",
              "      <th>前日比%</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>日付け</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>2017-01-31</th>\n",
              "      <td>18980.0</td>\n",
              "      <td>19350.0</td>\n",
              "      <td>19390.0</td>\n",
              "      <td>18970.0</td>\n",
              "      <td>82.98K</td>\n",
              "      <td>-1.81</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2017-01-30</th>\n",
              "      <td>19330.0</td>\n",
              "      <td>19500.0</td>\n",
              "      <td>19560.0</td>\n",
              "      <td>19280.0</td>\n",
              "      <td>48.95K</td>\n",
              "      <td>-0.67</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2017-01-27</th>\n",
              "      <td>19460.0</td>\n",
              "      <td>19450.0</td>\n",
              "      <td>19560.0</td>\n",
              "      <td>19390.0</td>\n",
              "      <td>63.14K</td>\n",
              "      <td>0.00</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-8f14e58c-067d-460e-95ef-535676498ef6')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-8f14e58c-067d-460e-95ef-535676498ef6 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-8f14e58c-067d-460e-95ef-535676498ef6');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "                 終値       始値       高値       安値     出来高  前日比%\n",
              "日付け                                                         \n",
              "2017-01-31  18980.0  19350.0  19390.0  18970.0  82.98K -1.81\n",
              "2017-01-30  19330.0  19500.0  19560.0  19280.0  48.95K -0.67\n",
              "2017-01-27  19460.0  19450.0  19560.0  19390.0  63.14K  0.00"
            ]
          },
          "execution_count": 5,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# 年月日を除いてハイフン(-)で繋げる\n",
        "date = []\n",
        "for i in df[\"日付け\"]:\n",
        "  p = i[0:4] + \"-\" + i[5:7] + \"-\" + i[8:10]\n",
        "  print(p)\n",
        "  date.append(p)\n",
        "\n",
        "# 上記処理後、データフレーム化および型(datetime)に変換\n",
        "date = pd.DataFrame({\"日付け\":date})\n",
        "date = pd.to_datetime(date[\"日付け\"])\n",
        "\n",
        "# 元データからカラム(日付)を削除\n",
        "df = df.drop(columns = [\"日付け\"])\n",
        "# 横結合\n",
        "df2 = pd.concat([df, date], axis=1)\n",
        "# 日付のインデックス化\n",
        "df3 = df2.set_index(\"日付け\")\n",
        "df3.head(3)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 175
        },
        "id": "j_G1pOzh9YOh",
        "outputId": "81ec2a7b-34ca-4096-fb38-a7fd07960ee6"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-59baf4ae-c823-4e4d-ad5b-91fa0ab4a256\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>終値</th>\n",
              "      <th>始値</th>\n",
              "      <th>高値</th>\n",
              "      <th>安値</th>\n",
              "      <th>出来高</th>\n",
              "      <th>前日比%</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>日付け</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>2017-01-31</th>\n",
              "      <td>18980.0</td>\n",
              "      <td>19350.0</td>\n",
              "      <td>19390.0</td>\n",
              "      <td>18970.0</td>\n",
              "      <td>82.98K</td>\n",
              "      <td>-1.81</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2017-01-30</th>\n",
              "      <td>19330.0</td>\n",
              "      <td>19500.0</td>\n",
              "      <td>19560.0</td>\n",
              "      <td>19280.0</td>\n",
              "      <td>48.95K</td>\n",
              "      <td>-0.67</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2017-01-27</th>\n",
              "      <td>19460.0</td>\n",
              "      <td>19450.0</td>\n",
              "      <td>19560.0</td>\n",
              "      <td>19390.0</td>\n",
              "      <td>63.14K</td>\n",
              "      <td>0.00</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-59baf4ae-c823-4e4d-ad5b-91fa0ab4a256')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-59baf4ae-c823-4e4d-ad5b-91fa0ab4a256 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-59baf4ae-c823-4e4d-ad5b-91fa0ab4a256');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "                 終値       始値       高値       安値     出来高  前日比%\n",
              "日付け                                                         \n",
              "2017-01-31  18980.0  19350.0  19390.0  18970.0  82.98K -1.81\n",
              "2017-01-30  19330.0  19500.0  19560.0  19280.0  48.95K -0.67\n",
              "2017-01-27  19460.0  19450.0  19560.0  19390.0  63.14K  0.00"
            ]
          },
          "execution_count": 6,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "df3.head(3)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 455
        },
        "id": "fOhMnEDYZzH3",
        "outputId": "3460b492-5051-4084-cc06-94700ed51d3e"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-011cbd37-1f49-44bf-a9ff-1fb3a3e9dc1b\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>終値</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>日付け</th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>2017-01-31</th>\n",
              "      <td>18980.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2017-01-30</th>\n",
              "      <td>19330.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2017-01-27</th>\n",
              "      <td>19460.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2017-01-26</th>\n",
              "      <td>19460.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2017-01-25</th>\n",
              "      <td>19070.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2009-03-10</th>\n",
              "      <td>7040.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2009-03-09</th>\n",
              "      <td>7050.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2009-03-06</th>\n",
              "      <td>7170.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2009-03-05</th>\n",
              "      <td>7410.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2009-03-04</th>\n",
              "      <td>7270.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>2064 rows × 1 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-011cbd37-1f49-44bf-a9ff-1fb3a3e9dc1b')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-011cbd37-1f49-44bf-a9ff-1fb3a3e9dc1b button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-011cbd37-1f49-44bf-a9ff-1fb3a3e9dc1b');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "                 終値\n",
              "日付け                \n",
              "2017-01-31  18980.0\n",
              "2017-01-30  19330.0\n",
              "2017-01-27  19460.0\n",
              "2017-01-26  19460.0\n",
              "2017-01-25  19070.0\n",
              "...             ...\n",
              "2009-03-10   7040.0\n",
              "2009-03-09   7050.0\n",
              "2009-03-06   7170.0\n",
              "2009-03-05   7410.0\n",
              "2009-03-04   7270.0\n",
              "\n",
              "[2064 rows x 1 columns]"
            ]
          },
          "execution_count": 7,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "df = df3[[\"終値\"]]\n",
        "df"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HMiB6n-jb1GO"
      },
      "source": [
        "ひとまずdfで進むとする。\n",
        "\n",
        "これでfloatに変換.⇒array形式になってしまった。\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "M81sPg1Pb4j1",
        "outputId": "540badb7-cb7b-4fff-e8b5-fe22389018aa"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([[18980.],\n",
              "       [19330.],\n",
              "       [19460.],\n",
              "       ...,\n",
              "       [ 7170.],\n",
              "       [ 7410.],\n",
              "       [ 7270.]], dtype=float32)"
            ]
          },
          "execution_count": 8,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "df = df.values          # 辞書の値のみ取得するのが.values\n",
        "df = df.astype('float32')\n",
        "df"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4oWfP4Xz3oRf"
      },
      "source": [
        "### [データ分割](https://premium.aidemy.jp/courses/6160/exercises/ZkSMBMEp7K)\n",
        "前半67%(2/3)を訓練データに、後半33%(1/3)を検証データに分割"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IL1_a1vZ37iG"
      },
      "source": [
        "訓練データは機械学習モデルの学習（調整）を行うためのデータ。\n",
        "\n",
        "検証データはモデルの予測結果の精度を評価するデータ。\n",
        "\n",
        "ランダムにデータを分けてはいけません。⇒シャッフルフォルス\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "afBG86Kdcj9c"
      },
      "source": [
        "今回は<font color=\"red\">validation</font>を用いません。私があまり理解していないから。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "g0EgYw_-c1aO",
        "outputId": "786207dd-ec9d-46a4-b042-c65c28490252"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "2064 1382 682\n"
          ]
        }
      ],
      "source": [
        "#データ分割。シンプルに２つに分ける。\n",
        "#訓練データ、テストデータは2:1になるように。だとよ。\n",
        "\n",
        "train =df[:int(len(df)*0.67)] # この長さまで。\n",
        "test  =df[int(len(df)*0.67):] # この長さ以降。\n",
        "\n",
        "print(len(df), len(train), len(test))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OeZJk_nC6vAM"
      },
      "source": [
        "### [データスケーリング](https://premium.aidemy.jp/courses/6160/exercises/NpMLLgDLXX)\n",
        "\n",
        "単位や値の範囲が異なるデータをある基準で均一化して尺度を揃える\n",
        "\n",
        "今回は正規化"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TeBMK7xp7GOd"
      },
      "source": [
        "年齢と年収の場合で、\n",
        "\n",
        "年齢：0〜100程度の値\n",
        "\n",
        "年収は0〜1億など広範囲の値。\n",
        "\n",
        "これらの値をそのまま使ってしまうと年齢に対して年収の値が圧倒的に大きな影響を持ってしまう。⇒うまく学習できないといった現象に。\n",
        "　⇒これを防ぐための<font color=\"red\">スケーリング</font>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cr04RKIF8Ez_"
      },
      "source": [
        "スケーリングの手段⇒正規化と標準化"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ROEclj0i8J8d"
      },
      "source": [
        "<font color=\"red\">正規化</font>（MinMaxScaler）：最低値が0、最高値が1になるように値を変換。\n",
        "\n",
        "<font color=\"red\">標準化</font>（StandardScaler）は平均が0、標準偏差が1になるように元の値を変換。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kxHLPpVQ88wM"
      },
      "source": [
        "機械学習ではスケーリングは<font color=\"red\">訓練データのみ</font>を基準に行います。\n",
        "\n",
        "下記の例を参照。トレインデータを基準として、トレインとテストを正規化。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 143
        },
        "id": "pTP0G6kc9VOG",
        "outputId": "1236a48c-45f2-4d83-9608-e8da037e563f"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-5a45792e-eb88-4183-8193-f440209fe819\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>0</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.070189</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.059937</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.044953</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-5a45792e-eb88-4183-8193-f440209fe819')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-5a45792e-eb88-4183-8193-f440209fe819 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-5a45792e-eb88-4183-8193-f440209fe819');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "          0\n",
              "0  0.070189\n",
              "1  0.059937\n",
              "2  0.044953"
            ]
          },
          "execution_count": 10,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# データのスケーリング（正規化）⇒最小値が0, 最大値が1となるようにする。\n",
        "# MinMaxScaler()は、データをfeature_rangeで指定した範囲内(今回は0～1)に正規化するモジュール\n",
        "\n",
        "from sklearn.preprocessing import MinMaxScaler\n",
        "scaler = MinMaxScaler(feature_range=(0, 1))\n",
        "\n",
        "\n",
        "# トレインデータを基準として、トレインとテストを正規化。\n",
        "scaler_train = scaler.fit(train)\n",
        "# パラメータを用いて`train`データをスケーリングします\n",
        "train = scaler_train.transform(train)\n",
        "# パラメータを用いて`test`データをスケーリングします\n",
        "test = scaler_train.transform(test)\n",
        "\n",
        "\n",
        "df_train_scale = pd.DataFrame(train)\n",
        "df_test_scale = pd.DataFrame(test)\n",
        "\n",
        "df_train_scale.head(3)\n",
        "df_test_scale.head(3)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LWAtGLSVB-pR"
      },
      "source": [
        "### [入力データ・正解ラベルの作成](https://premium.aidemy.jp/courses/6160/exercises/SOTuOSnQ2Y)\n",
        "<font color=\"red\">look_back = nのn</font>がいくつ前のデータまで利用するか。\n",
        "　どのくらい前までのデータを利用するのですか？という事。\n",
        "\n",
        "この例だと、基準点含めて３つ前までのデータを利用して次を予測します。という事\n",
        "\n",
        "train⇒train_X, train_Y\n",
        "\n",
        "test ⇒test_X, test_Y\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WzBImmpLCRz9"
      },
      "source": [
        "![2022-08-20_11h17_07.png](data:image/png;base64,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)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 143
        },
        "id": "sPZfag6TDf3L",
        "outputId": "025e7158-971a-43a2-8d04-43ffd66d88bb"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-8e656a08-5ee4-4baf-9760-e86ae5c88824\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>0</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.070189</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.059937</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.044953</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-8e656a08-5ee4-4baf-9760-e86ae5c88824')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-8e656a08-5ee4-4baf-9760-e86ae5c88824 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-8e656a08-5ee4-4baf-9760-e86ae5c88824');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "          0\n",
              "0  0.070189\n",
              "1  0.059937\n",
              "2  0.044953"
            ]
          },
          "execution_count": 11,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# trainとtestデータ\n",
        "df_train_scale.head(3)\n",
        "df_test_scale.head(3)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "id": "cljKj2HlDX7s"
      },
      "outputs": [],
      "source": [
        "# LSTn用の入力データ・正解ラベルの作成　独自関数を作成。\n",
        "#　引数に look_back,dataset\n",
        "import numpy as np\n",
        "\n",
        "def create_dataset(dataset, look_back):\n",
        "    data_X, data_Y = [], []\n",
        "    for i in range(look_back, len(dataset)):\n",
        "        data_X.append(dataset[i-look_back:i, 0])\n",
        "        data_Y.append(dataset[i, 0])\n",
        "    return np.array(data_X), np.array(data_Y)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "s20fmFU4E4Se",
        "outputId": "020b09a1-3e36-4f14-f481-d0523e56d118"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-48e25df3-d42c-4adf-9f29-e0c563926c0f\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>0</th>\n",
              "      <th>1</th>\n",
              "      <th>2</th>\n",
              "      <th>3</th>\n",
              "      <th>4</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.853312</td>\n",
              "      <td>0.880915</td>\n",
              "      <td>0.891167</td>\n",
              "      <td>0.891167</td>\n",
              "      <td>0.860410</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.880915</td>\n",
              "      <td>0.891167</td>\n",
              "      <td>0.891167</td>\n",
              "      <td>0.860410</td>\n",
              "      <td>0.842271</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.891167</td>\n",
              "      <td>0.891167</td>\n",
              "      <td>0.860410</td>\n",
              "      <td>0.842271</td>\n",
              "      <td>0.848580</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.891167</td>\n",
              "      <td>0.860410</td>\n",
              "      <td>0.842271</td>\n",
              "      <td>0.848580</td>\n",
              "      <td>0.867508</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.860410</td>\n",
              "      <td>0.842271</td>\n",
              "      <td>0.848580</td>\n",
              "      <td>0.867508</td>\n",
              "      <td>0.862776</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1372</th>\n",
              "      <td>0.026814</td>\n",
              "      <td>0.026025</td>\n",
              "      <td>0.044953</td>\n",
              "      <td>0.040221</td>\n",
              "      <td>0.047319</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1373</th>\n",
              "      <td>0.026025</td>\n",
              "      <td>0.044953</td>\n",
              "      <td>0.040221</td>\n",
              "      <td>0.047319</td>\n",
              "      <td>0.050473</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1374</th>\n",
              "      <td>0.044953</td>\n",
              "      <td>0.040221</td>\n",
              "      <td>0.047319</td>\n",
              "      <td>0.050473</td>\n",
              "      <td>0.049487</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1375</th>\n",
              "      <td>0.040221</td>\n",
              "      <td>0.047319</td>\n",
              "      <td>0.050473</td>\n",
              "      <td>0.049487</td>\n",
              "      <td>0.041798</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1376</th>\n",
              "      <td>0.047319</td>\n",
              "      <td>0.050473</td>\n",
              "      <td>0.049487</td>\n",
              "      <td>0.041798</td>\n",
              "      <td>0.054416</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>1377 rows × 5 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-48e25df3-d42c-4adf-9f29-e0c563926c0f')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-48e25df3-d42c-4adf-9f29-e0c563926c0f button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-48e25df3-d42c-4adf-9f29-e0c563926c0f');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "             0         1         2         3         4\n",
              "0     0.853312  0.880915  0.891167  0.891167  0.860410\n",
              "1     0.880915  0.891167  0.891167  0.860410  0.842271\n",
              "2     0.891167  0.891167  0.860410  0.842271  0.848580\n",
              "3     0.891167  0.860410  0.842271  0.848580  0.867508\n",
              "4     0.860410  0.842271  0.848580  0.867508  0.862776\n",
              "...        ...       ...       ...       ...       ...\n",
              "1372  0.026814  0.026025  0.044953  0.040221  0.047319\n",
              "1373  0.026025  0.044953  0.040221  0.047319  0.050473\n",
              "1374  0.044953  0.040221  0.047319  0.050473  0.049487\n",
              "1375  0.040221  0.047319  0.050473  0.049487  0.041798\n",
              "1376  0.047319  0.050473  0.049487  0.041798  0.054416\n",
              "\n",
              "[1377 rows x 5 columns]"
            ]
          },
          "execution_count": 13,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# その独自関数の呼び出し\n",
        "look_back = 5\n",
        "\n",
        "train_X, train_Y = create_dataset(train, look_back)\n",
        "test_X, test_Y = create_dataset(test, look_back)\n",
        "\n",
        "train_X_df = pd.DataFrame(train_X)\n",
        "train_X_df"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Vv-7CKsuc_C-"
      },
      "source": [
        "look backが5つなんで、5列あるという訳ですね。\n",
        "\n",
        "そして時系列なので斜めが同じ数値になっている。⇒一つづつズレている。という事に着目。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1Bc2WwpeFPNv",
        "outputId": "2be36435-614e-43b6-faa4-24dec19bc8bd"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "train_X\n",
            "[[0.85331225 0.8809148  0.89116716 0.89116716 0.8604101 ]\n",
            " [0.8809148  0.89116716 0.89116716 0.8604101  0.8422713 ]\n",
            " [0.89116716 0.89116716 0.8604101  0.8422713  0.8485805 ]\n",
            " ...\n",
            " [0.04495269 0.0402208  0.04731864 0.05047321 0.04948741]\n",
            " [0.0402208  0.04731864 0.05047321 0.04948741 0.04179811]\n",
            " [0.04731864 0.05047321 0.04948741 0.04179811 0.05441642]]\n",
            "\n",
            "train_Y\n",
            "[0.8422713  0.8485805  0.8675078  ... 0.04179811 0.05441642 0.06309146]\n",
            "\n",
            "test_X\n",
            "[[ 0.0701893   0.05993688  0.04495269  0.04810727  0.05362779]\n",
            " [ 0.05993688  0.04495269  0.04810727  0.05362779  0.04100949]\n",
            " [ 0.04495269  0.04810727  0.05362779  0.04100949  0.0402208 ]\n",
            " ...\n",
            " [-0.05126184 -0.08438486 -0.06072557 -0.08832806 -0.08753943]\n",
            " [-0.08438486 -0.06072557 -0.08832806 -0.08753943 -0.07807571]\n",
            " [-0.06072557 -0.08832806 -0.08753943 -0.07807571 -0.05914825]]\n",
            "\n",
            "test_Y\n",
            "[ 0.04100949  0.0402208   0.04968452  0.04652995  0.05835962  0.04810727\n",
            "  0.05047321  0.04574132  0.04810727  0.0498817   0.03391165  0.02917981\n",
            "  0.01735014  0.02287066  0.02917981  0.04258674  0.04337537  0.03785491\n",
            "  0.03312302  0.01025236  0.02251577  0.02602524  0.04100949  0.03864354\n",
            "  0.04160094  0.04968452  0.03470033  0.02365929  0.03233439  0.02523661\n",
            "  0.03864354  0.04495269  0.0488959   0.03548896  0.04810727  0.0615142\n",
            "  0.07097793  0.06309146  0.06230283  0.05599368  0.04968452  0.0488959\n",
            "  0.03943217  0.04574132  0.03627759  0.04337537  0.0615142   0.06940061\n",
            "  0.07176656  0.07097793  0.06309146  0.06466877  0.06782335  0.06624603\n",
            "  0.07570976  0.08832806  0.11908519  0.11750787  0.1317035   0.14274448\n",
            "  0.1317035   0.13722396  0.148265    0.15378547  0.14905363  0.15615141\n",
            "  0.14589906  0.14668769  0.13643533  0.14018136  0.14274448  0.1395899\n",
            "  0.14116716  0.13880128  0.15141958  0.15772873  0.15141958  0.15141958\n",
            "  0.14195585  0.14195585  0.13328075  0.1309148   0.12854892  0.11750787\n",
            "  0.11277604  0.11908519  0.11356467  0.1143533   0.101735    0.09384859\n",
            "  0.09305996  0.09858042  0.11198735  0.10804415  0.101735    0.10567826\n",
            "  0.10410094  0.10252368  0.101735    0.09779179  0.10725552  0.10962147\n",
            "  0.12223977  0.12066245  0.10725552  0.10804415  0.10962147  0.09858042\n",
            "  0.10410094  0.10252368  0.11514193  0.11593062  0.11908519  0.11198735\n",
            "  0.11041009  0.11908519  0.12145108  0.13406938  0.1309148   0.13012618\n",
            "  0.13328075  0.12835175  0.13347793  0.13978708  0.14747632  0.13978708\n",
            "  0.13485807  0.12223977  0.10962147  0.11987382  0.11987382  0.11987382\n",
            "  0.1143533   0.10094637  0.11041009  0.11041009  0.11750787  0.11671925\n",
            "  0.11041009  0.1230284   0.1269716   0.11277604  0.1143533   0.11593062\n",
            "  0.12223977  0.12145108  0.12618297  0.12223977  0.10410094  0.09858042\n",
            "  0.101735    0.09384859  0.09621453  0.10804415  0.10429811  0.07649845\n",
            "  0.06072557  0.06624603  0.03785491  0.10252368  0.15851736  0.17429024\n",
            "  0.192429    0.18611985  0.18690854  0.20031548  0.19164038  0.18375397\n",
            "  0.20425868  0.19479495  0.18533123  0.18217665  0.19006312  0.1971609\n",
            "  0.21372241  0.21214509  0.21135646  0.20820189  0.20347005  0.20268136\n",
            "  0.20074922  0.19400632  0.19400632  0.19479495  0.19164038  0.18690854\n",
            "  0.1798107   0.18138802  0.16640377  0.1632492   0.17429024  0.18296528\n",
            "  0.17823344  0.18217665  0.1719243   0.1671924   0.1798107   0.19006312\n",
            "  0.18611985  0.1845426   0.1845426   0.18927443  0.18611985  0.18533123\n",
            "  0.18119085  0.1884858   0.18533123  0.17507887  0.17507887  0.1729101\n",
            "  0.16068614  0.16167194  0.17350155  0.17034698  0.17271292  0.1671924\n",
            "  0.16502368  0.17113566  0.17271292  0.1632492   0.16798109  0.16876972\n",
            "  0.16876972  0.16955835  0.1671924   0.16167194  0.16640377  0.16403782\n",
            "  0.15694004  0.15930599  0.16009462  0.15772873  0.14511043  0.14116716\n",
            "  0.15457416  0.14984226  0.15141958  0.148265    0.13978708  0.15457416\n",
            "  0.14905363  0.14353311  0.1317035   0.12776023  0.1309148   0.12460566\n",
            "  0.13406938  0.13012618  0.12066245  0.1230284   0.11750787  0.09463722\n",
            "  0.08970821  0.07965302  0.07886434  0.08123028  0.09542584  0.09858042\n",
            "  0.09463722  0.10015774  0.09936911  0.09542584  0.09621453  0.10804415\n",
            "  0.10488957  0.10725552  0.11119872  0.09858042  0.09779179  0.11770505\n",
            "  0.11198735  0.12066245  0.12145108  0.10725552  0.09384859  0.09700316\n",
            "  0.09463722  0.11041009  0.10488957  0.10883284  0.09779179  0.09444004\n",
            "  0.10725552  0.10804415  0.11652207  0.11119872  0.10015774  0.10331231\n",
            "  0.08596212  0.08675081  0.08044165  0.06861198  0.06703472  0.08359623\n",
            "  0.08911669  0.0741325   0.06940061  0.06072557  0.05047321  0.07649845\n",
            "  0.06388015  0.05757099  0.0528391   0.06388015  0.0741325   0.07886434\n",
            "  0.09463722  0.08517349  0.07965302  0.08044165  0.08753943  0.08123028\n",
            "  0.08832806  0.10883284  0.11041009  0.11750787  0.11514193  0.10488957\n",
            "  0.11987382  0.11041009  0.10804415  0.12066245  0.12381703  0.10567826\n",
            "  0.10488957  0.10094637  0.08201891  0.08832806  0.08990538  0.085765\n",
            "  0.09779179  0.11908519  0.12854892  0.10804415  0.10962147  0.11198735\n",
            "  0.10883284  0.08911669  0.09384859  0.08675081  0.0828076   0.08044165\n",
            "  0.09463722  0.10962147  0.12066245  0.12460566  0.13801259  0.13722396\n",
            "  0.15378547  0.1632492   0.14511043  0.14589906  0.15220821  0.13249213\n",
            "  0.1356467   0.12223977  0.10646689  0.10410094  0.10804415  0.10646689\n",
            "  0.1395899   0.14037853  0.1143533   0.12460566  0.12618297  0.12933755\n",
            "  0.11750787  0.10488957  0.10094637  0.12618297  0.12381703  0.148265\n",
            "  0.15615141  0.16561514  0.16403782  0.18059939  0.19558358  0.17429024\n",
            "  0.17665613  0.18769717  0.17902207  0.19873816  0.20130128  0.20130128\n",
            "  0.23229498  0.22791797  0.23008674  0.21924287  0.24053627  0.23659307\n",
            "  0.22003156  0.22003156  0.23186117  0.21608835  0.21687698  0.23107255\n",
            "  0.2452681   0.24053627  0.23580444  0.24447948  0.24053627  0.2373817\n",
            "  0.24684542  0.2460568   0.25236595  0.24842268  0.24211359  0.23186117\n",
            "  0.23343849  0.22555202  0.21845424  0.20583594  0.20347005  0.20189273\n",
            "  0.20248425  0.20425868  0.19952679  0.20662463  0.19794953  0.19794953\n",
            "  0.19873816  0.19321764  0.19006312  0.18927443  0.19085175  0.17429024\n",
            "  0.15615141  0.1632492   0.1632492   0.15851736  0.15299684  0.15299684\n",
            "  0.16088331  0.17271292  0.17429024  0.15615141  0.17113566  0.16798109\n",
            "  0.14668769  0.14589906  0.15299684  0.15043378  0.14432174  0.14037853\n",
            "  0.14195585  0.14984226  0.17271292  0.17823344  0.17429024  0.16009462\n",
            "  0.15772873  0.17823344  0.16640377  0.1719243   0.18533123  0.19164038\n",
            "  0.21135646  0.2011041   0.20504731  0.21135646  0.22160882  0.21529967\n",
            "  0.20189273  0.21529967  0.20485014  0.20899051  0.20031548  0.2011041\n",
            "  0.19952679  0.19558358  0.19302052  0.1932571   0.18769717  0.19637221\n",
            "  0.19479495  0.18533123  0.18533123  0.18237382  0.17271292  0.16009462\n",
            "  0.15615141  0.15694004  0.15930599  0.15063089  0.15299684  0.15299684\n",
            "  0.1317035   0.14511043  0.15457416  0.15694004  0.14511043  0.14116716\n",
            "  0.11593062  0.10962147  0.09227127  0.07176656  0.09700316  0.101735\n",
            "  0.09700316  0.10666406  0.10488957  0.10962147  0.11908519  0.12460566\n",
            "  0.12933755  0.12539434  0.12854892  0.1356467   0.13485807  0.12933755\n",
            "  0.12776023  0.12145108  0.1309148   0.1317035   0.14589906  0.1356467\n",
            "  0.15141958  0.16482651  0.17350155  0.16798109  0.16640377  0.17113566\n",
            "  0.17113566  0.16482651  0.16640377  0.16403782  0.15141958  0.15220821\n",
            "  0.14589906  0.13406938  0.12854892  0.11987382  0.11987382  0.12381703\n",
            "  0.14353311  0.15378547  0.15299684  0.14747632  0.16955835  0.18138802\n",
            "  0.1719243   0.17507887  0.16167194  0.16009462  0.15930599  0.17665613\n",
            "  0.18217665  0.1719243   0.17744482  0.16955835  0.16009462  0.16088331\n",
            "  0.16955835  0.1845426   0.18059939  0.18690854  0.18533123  0.19558358\n",
            "  0.1845426   0.1884858   0.1671924   0.17350155  0.16167194  0.16876972\n",
            "  0.16482651  0.19164038  0.18611985  0.1798107   0.19085175  0.1884858\n",
            "  0.17823344  0.17665613  0.16482651  0.17665613  0.17350155  0.17429024\n",
            "  0.16088331  0.15378547  0.15141958  0.15141958  0.14037853  0.12776023\n",
            "  0.12381703  0.11750787  0.09621453  0.09227127  0.08911669  0.08596212\n",
            "  0.06940061  0.08596212  0.09069401  0.09858042  0.11671925  0.12145108\n",
            "  0.13249213  0.13643533  0.1395899   0.1395899   0.1309148   0.13722396\n",
            "  0.12854892  0.11119872  0.10883284  0.1317035   0.1269716   0.12381703\n",
            "  0.1317035   0.1269716   0.148265    0.15615141  0.14668769  0.14432174\n",
            "  0.12776023  0.13328075  0.12776023  0.12066245  0.12618297  0.12223977\n",
            "  0.12066245  0.10646689  0.101735    0.10015774  0.09305996  0.09305996\n",
            "  0.08438486  0.08832806  0.09227127  0.08990538  0.07255518  0.08753943\n",
            "  0.07649845  0.09463722  0.09148264  0.101735    0.10252368  0.09542584\n",
            "  0.06861198  0.05520505  0.02760255  0.04652995  0.04574132  0.0528391\n",
            "  0.04652995  0.04810727  0.06072557  0.0615142   0.04652995  0.04574132\n",
            "  0.05441642  0.05757099  0.06388015  0.06072557  0.03312302  0.05520505\n",
            "  0.05757099  0.04495269  0.04179811  0.01735014 -0.00315458  0.00315458\n",
            "  0.03706622  0.04337537  0.02208203  0.01892745  0.00157732 -0.02365929\n",
            " -0.01892745 -0.0212934  -0.03785491 -0.05126184 -0.08438486 -0.06072557\n",
            " -0.08832806 -0.08753943 -0.07807571 -0.05914825 -0.0701893 ]\n"
          ]
        }
      ],
      "source": [
        "# データの確認\n",
        "print('train_X')\n",
        "print(train_X)\n",
        "print()\n",
        "print('train_Y')\n",
        "print(train_Y)\n",
        "print()\n",
        "print('test_X')\n",
        "print(test_X)\n",
        "print()\n",
        "print('test_Y')\n",
        "print(test_Y)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "orPxGgyoOFmx"
      },
      "source": [
        "### [データ整形](https://premium.aidemy.jp/courses/6160/exercises/3xXX8qwu7-)\n",
        "作成したデータはLSTMで分析できるデータ形式ではない⇒入力データの整形。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zRmh7-IBOMiy"
      },
      "source": [
        "入力データを<font color=\"red\">行数×変数数×カラム数</font>の3次元の行列に変換、LSTMで分析できるデータ形式に整形する。\n",
        "\n",
        "- 行数：データの総数 　　　　　　　　　　　　　入力データの総数61\n",
        "- 変数数：1セットのデータに含まれる要素数　　１セットの入力データに含まれる要素数look_backで指定した3\n",
        "- カラム数：扱うデータの種別数　　　　　　　　売上データのみを扱うので1"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "L-7YalzOOidR"
      },
      "source": [
        "reshapeメソッドを用いて、データを指定した形式の行列に変換します。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "N72B3_ZZPVlA"
      },
      "outputs": [],
      "source": [
        "# 3次元のnumpy.ndarrayに変換します⇒ reshapeでLSTN用に変換している。\n",
        "train_X = train_X.reshape(train_X.shape[0], train_X.shape[1], 1)\n",
        "test_X = test_X.reshape(test_X.shape[0], test_X.shape[1], 1)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aOn6IO_15pE7"
      },
      "source": [
        "### [LSTMネットワークの構築と訓練](https://premium.aidemy.jp/courses/6160/exercises/mRBWTDAXqs)\n",
        "\n",
        "、LSTMネットワークを構築し、これまでに用意したデータを用いてモデルの訓練（学習）と予測、評価\n",
        "\n",
        "忘備録として、私の進捗を残しましたが、4回目のみ実装して頂ければ平気かと。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PTscOUIICOGL"
      },
      "source": [
        "#### 1回目の学習"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NRXz7TUM7Sdt"
      },
      "source": [
        "6行目のレイヤーはモデルの最初のレイヤーのため、引数input_shapeを指定します。以降の層では、引数input_shapeは指定しない。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZHe5RsidAQ4d"
      },
      "source": [
        "出力層：モデル最後のレイヤー：出力結果を吐き出す物。\n",
        "1つのノードを持つ全結合層を追加します。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZMioxDloAsGT"
      },
      "source": [
        "訓練過程における損失関数（loss）と最適化アルゴリズム（optimizer）の定義"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7VqdEaNDAzFb"
      },
      "source": [
        "正解ラベルと出力の誤差がより小さくなるように。⇒損失関数の値が小さいほど予測が正確。\n",
        "\n",
        "分類：categorical_crossentropy\n",
        "\n",
        "回帰：mean_squared_error"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ttbk5kLnBP1D"
      },
      "source": [
        "最適化アルゴリズムとは、損失関数の値を減少させるように、重みを更新するアルゴリズム。\n",
        "\n",
        "どれが有効化は分からないので、毎度確認していく。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rrsr-8-MBjp8",
        "outputId": "1424f580-c31e-4f8d-c9a2-f5880cb1bd69"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "1377/1377 - 8s - loss: 0.0269 - 8s/epoch - 6ms/step\n",
            "Epoch 2/20\n",
            "1377/1377 - 4s - loss: 8.7941e-04 - 4s/epoch - 3ms/step\n",
            "Epoch 3/20\n",
            "1377/1377 - 5s - loss: 8.9369e-04 - 5s/epoch - 3ms/step\n",
            "Epoch 4/20\n",
            "1377/1377 - 3s - loss: 8.1830e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 5/20\n",
            "1377/1377 - 3s - loss: 7.7729e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 6/20\n",
            "1377/1377 - 3s - loss: 7.1320e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 7/20\n",
            "1377/1377 - 3s - loss: 6.5403e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 8/20\n",
            "1377/1377 - 3s - loss: 6.1782e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 9/20\n",
            "1377/1377 - 3s - loss: 5.7806e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 10/20\n",
            "1377/1377 - 3s - loss: 5.4898e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 11/20\n",
            "1377/1377 - 3s - loss: 4.8279e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 12/20\n",
            "1377/1377 - 3s - loss: 4.1269e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 13/20\n",
            "1377/1377 - 3s - loss: 3.9308e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 14/20\n",
            "1377/1377 - 3s - loss: 4.0327e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 15/20\n",
            "1377/1377 - 3s - loss: 4.0471e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 16/20\n",
            "1377/1377 - 3s - loss: 3.6757e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 17/20\n",
            "1377/1377 - 3s - loss: 3.8566e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 18/20\n",
            "1377/1377 - 3s - loss: 3.7665e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 19/20\n",
            "1377/1377 - 3s - loss: 3.6709e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 20/20\n",
            "1377/1377 - 3s - loss: 3.6523e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 1/20\n",
            "1377/1377 - 3s - loss: 3.5902e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 2/20\n",
            "1377/1377 - 3s - loss: 3.7483e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 3/20\n",
            "1377/1377 - 3s - loss: 3.5780e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 4/20\n",
            "1377/1377 - 3s - loss: 3.5885e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 5/20\n",
            "1377/1377 - 3s - loss: 3.4881e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 6/20\n",
            "1377/1377 - 3s - loss: 3.5192e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 7/20\n",
            "1377/1377 - 4s - loss: 3.5570e-04 - 4s/epoch - 3ms/step\n",
            "Epoch 8/20\n",
            "1377/1377 - 3s - loss: 3.5487e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 9/20\n",
            "1377/1377 - 3s - loss: 3.5150e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 10/20\n",
            "1377/1377 - 3s - loss: 3.4218e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 11/20\n",
            "1377/1377 - 3s - loss: 3.3676e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 12/20\n",
            "1377/1377 - 3s - loss: 3.4176e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 13/20\n",
            "1377/1377 - 3s - loss: 3.2948e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 14/20\n",
            "1377/1377 - 3s - loss: 3.4592e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 15/20\n",
            "1377/1377 - 3s - loss: 3.5739e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 16/20\n",
            "1377/1377 - 3s - loss: 3.4264e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 17/20\n",
            "1377/1377 - 3s - loss: 3.4540e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 18/20\n",
            "1377/1377 - 3s - loss: 3.4333e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 19/20\n",
            "1377/1377 - 3s - loss: 3.4588e-04 - 3s/epoch - 2ms/step\n",
            "Epoch 20/20\n",
            "1377/1377 - 3s - loss: 3.4520e-04 - 3s/epoch - 2ms/step\n"
          ]
        }
      ],
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "\n",
        "# LSTMモデルの作成｡\n",
        "model = keras.Sequential()                                                      \n",
        "model.add(layers.LSTM(8, input_shape=(look_back, 1))) # 8つのLSTMブロックからなるレイヤー(層)を追加。\n",
        "model.add(layers.Dense(1))                                                   # 出力層だから今回は１固定。\n",
        "\n",
        "model.compile(loss='mean_squared_error', optimizer='adam')\n",
        "\n",
        "\n",
        "# LSTMネットワークの訓練（学習）\n",
        "model.fit(train_X, train_Y, epochs=20, batch_size=1, verbose=2)\n",
        "\n",
        "history = model.fit(train_X, train_Y, epochs=20, batch_size=1, verbose=2)\n",
        "\n",
        "#loss: loss: 8.5483e-04 "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "id": "mW4CHR76kyW1",
        "outputId": "e6c25801-a4e8-45cf-f03d-8768e757541f"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fe86faa2f50>"
            ]
          },
          "execution_count": 17,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "# 学習過程のプロット\n",
        "plt.plot(history.epoch, history.history[\"loss\"], label=\"loss\")\n",
        "plt.legend()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BWCnCJEYCUZi"
      },
      "source": [
        "#### 1回目を考慮した2回目の学習"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jDtoRRVuCaWy"
      },
      "source": [
        "これはAidemyの教材通り"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QoOUI1FVC5aS"
      },
      "source": [
        "64,32という数字はどこから来たのか？何を持って決めたのか？⇒\n",
        "2の乗数で決まる。\n",
        "\n",
        "batch_sizeも何をもって決めたのか？"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "t-OcY041HJJC"
      },
      "source": [
        "return_sequences=True ⇒LSTNの時系列データで使用する際に引数として使用する。\n",
        "\n",
        "最後の出力層で model.add(layers.Dense(1)) になるじゃないですか。\n",
        "\n",
        "それと<font color=\"red\">**くっつくレイヤーにはreturn_sequences=Trueは不要**</font>です。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qCikJgvrCdJq",
        "outputId": "199a8a5b-a17f-45c2-9fce-80a3f195bcae"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/15\n",
            "1377/1377 - 15s - loss: 0.0063 - 15s/epoch - 11ms/step\n",
            "Epoch 2/15\n",
            "1377/1377 - 9s - loss: 0.0017 - 9s/epoch - 7ms/step\n",
            "Epoch 3/15\n",
            "1377/1377 - 9s - loss: 0.0015 - 9s/epoch - 7ms/step\n",
            "Epoch 4/15\n",
            "1377/1377 - 9s - loss: 0.0015 - 9s/epoch - 7ms/step\n",
            "Epoch 5/15\n",
            "1377/1377 - 9s - loss: 0.0013 - 9s/epoch - 7ms/step\n",
            "Epoch 6/15\n",
            "1377/1377 - 9s - loss: 0.0013 - 9s/epoch - 7ms/step\n",
            "Epoch 7/15\n",
            "1377/1377 - 9s - loss: 0.0013 - 9s/epoch - 7ms/step\n",
            "Epoch 8/15\n",
            "1377/1377 - 9s - loss: 0.0011 - 9s/epoch - 7ms/step\n",
            "Epoch 9/15\n",
            "1377/1377 - 10s - loss: 0.0011 - 10s/epoch - 7ms/step\n",
            "Epoch 10/15\n",
            "1377/1377 - 10s - loss: 0.0010 - 10s/epoch - 7ms/step\n",
            "Epoch 11/15\n",
            "1377/1377 - 10s - loss: 7.6424e-04 - 10s/epoch - 7ms/step\n",
            "Epoch 12/15\n",
            "1377/1377 - 9s - loss: 6.2076e-04 - 9s/epoch - 7ms/step\n",
            "Epoch 13/15\n",
            "1377/1377 - 9s - loss: 6.1408e-04 - 9s/epoch - 7ms/step\n",
            "Epoch 14/15\n",
            "1377/1377 - 10s - loss: 5.6047e-04 - 10s/epoch - 7ms/step\n",
            "Epoch 15/15\n",
            "1377/1377 - 10s - loss: 5.5316e-04 - 10s/epoch - 7ms/step\n"
          ]
        }
      ],
      "source": [
        "# LSTMモデルの作成\n",
        "model2 = keras.Sequential()\n",
        "\n",
        "model2.add(layers.LSTM(64, input_shape=(look_back, 1), return_sequences=True))\n",
        "model2.add(layers.LSTM(32,return_sequences=True))\n",
        "\n",
        "model2.add(layers.LSTM(16,return_sequences=True)) #私が追加しました。\n",
        "model2.add(layers.LSTM(8)) #私が追加しました。\n",
        "\n",
        "model2.add(layers.Dense(1))\n",
        "# モデルをコンパイルしてください\n",
        "model2.compile(loss='mean_squared_error', optimizer='adam')\n",
        "# 訓練を行ってください\n",
        "# model.fit(train_X, train_Y, epochs=15, batch_size=1, verbose=2)\n",
        "\n",
        "history2 = model2.fit(train_X, train_Y, epochs=15, batch_size=1, verbose=2)\n",
        "# loss: 5.9269e-04"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "id": "oNT4fVADFkc3",
        "outputId": "f8a32c01-42e4-44da-c545-a004c9a40776"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fe86fa78f50>"
            ]
          },
          "execution_count": 19,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "# 学習過程のプロット\n",
        "plt.plot(history2.epoch, history2.history[\"loss\"], label=\"loss\")\n",
        "plt.legend()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zCRq96xOJAJ7"
      },
      "source": [
        "#### 3回目の学習\n",
        "Aidemyの教材通り"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bIdY4tkoJGcx",
        "outputId": "d03a09df-18fc-46a4-c0f5-80c0e675aa0e"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/15\n",
            "1377/1377 - 8s - loss: 0.0041 - 8s/epoch - 6ms/step\n",
            "Epoch 2/15\n",
            "1377/1377 - 6s - loss: 0.0012 - 6s/epoch - 4ms/step\n",
            "Epoch 3/15\n",
            "1377/1377 - 6s - loss: 0.0011 - 6s/epoch - 4ms/step\n",
            "Epoch 4/15\n",
            "1377/1377 - 5s - loss: 0.0010 - 5s/epoch - 4ms/step\n",
            "Epoch 5/15\n",
            "1377/1377 - 6s - loss: 9.4933e-04 - 6s/epoch - 4ms/step\n",
            "Epoch 6/15\n",
            "1377/1377 - 5s - loss: 7.3662e-04 - 5s/epoch - 4ms/step\n",
            "Epoch 7/15\n",
            "1377/1377 - 5s - loss: 6.2203e-04 - 5s/epoch - 4ms/step\n",
            "Epoch 8/15\n",
            "1377/1377 - 5s - loss: 5.6603e-04 - 5s/epoch - 4ms/step\n",
            "Epoch 9/15\n",
            "1377/1377 - 5s - loss: 5.6985e-04 - 5s/epoch - 4ms/step\n",
            "Epoch 10/15\n",
            "1377/1377 - 6s - loss: 5.1503e-04 - 6s/epoch - 4ms/step\n",
            "Epoch 11/15\n",
            "1377/1377 - 5s - loss: 5.4040e-04 - 5s/epoch - 4ms/step\n",
            "Epoch 12/15\n",
            "1377/1377 - 6s - loss: 4.9741e-04 - 6s/epoch - 4ms/step\n",
            "Epoch 13/15\n",
            "1377/1377 - 6s - loss: 4.7429e-04 - 6s/epoch - 4ms/step\n",
            "Epoch 14/15\n",
            "1377/1377 - 5s - loss: 4.7932e-04 - 5s/epoch - 4ms/step\n",
            "Epoch 15/15\n",
            "1377/1377 - 5s - loss: 4.8156e-04 - 5s/epoch - 4ms/step\n"
          ]
        }
      ],
      "source": [
        "# LSTMモデルを作成してください\n",
        "model3 = keras.Sequential()\n",
        "model3.add(layers.LSTM(64, input_shape=(look_back, 1), return_sequences=True))\n",
        "model3.add(layers.LSTM(32))\n",
        "model3.add(layers.Dense(1))\n",
        "# モデルをコンパイルしてください\n",
        "model3.compile(loss='mean_squared_error', optimizer='adam')\n",
        "# 訓練を行ってください\n",
        "# model.fit(train_X, train_Y, epochs=15, batch_size=1, verbose=2)\n",
        "# 学習\n",
        "history3 = model3.fit(train_X, train_Y, epochs=15, batch_size=1, verbose=2)\n",
        "# loss: 4.3211e-04 "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "id": "hYNAjtgDjiK0",
        "outputId": "726dc848-a833-42af-d7dc-b3a778832a42"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fe865672d10>"
            ]
          },
          "execution_count": 22,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "# 学習過程のプロット\n",
        "plt.plot(history3.epoch, history3.history[\"loss\"], label=\"loss\")\n",
        "plt.legend()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "M1xeyo1lbwzU"
      },
      "source": [
        "#### 4回目の学習\n",
        "call backを用いたもの。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "id": "LGi316YafPuc"
      },
      "outputs": [],
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "\n",
        "# LSTMモデルの作成｡\n",
        "model4 = keras.Sequential()                                                      \n",
        "model4.add(layers.LSTM(8, input_shape=(look_back, 1))) # 8つのLSTMブロックからなるレイヤー(層)を追加。\n",
        "model4.add(layers.Dense(1))                                                   # 出力層だから今回は１固定。\n",
        "\n",
        "model4.compile(loss='mean_squared_error', optimizer='adam')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BH1ODZw7lQLj",
        "outputId": "e2be4050-f06e-4e46-9dc2-a59caa4971ab"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/300\n",
            "1366/1377 [============================>.] - ETA: 0s - loss: 0.0050"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 10s 5ms/step - loss: 0.0050 - val_loss: 3.2267e-04\n",
            "Epoch 2/300\n",
            "1361/1377 [============================>.] - ETA: 0s - loss: 6.8367e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 6.8162e-04 - val_loss: 3.1568e-04\n",
            "Epoch 3/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 6.7624e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 6.8231e-04 - val_loss: 3.0362e-04\n",
            "Epoch 4/300\n",
            "1377/1377 [==============================] - ETA: 0s - loss: 6.6180e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 6.6180e-04 - val_loss: 3.1730e-04\n",
            "Epoch 5/300\n",
            "1365/1377 [============================>.] - ETA: 0s - loss: 5.9772e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 5.9575e-04 - val_loss: 2.4325e-04\n",
            "Epoch 6/300\n",
            "1368/1377 [============================>.] - ETA: 0s - loss: 5.9868e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 5.9870e-04 - val_loss: 2.3528e-04\n",
            "Epoch 7/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 5.9810e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 6.0086e-04 - val_loss: 2.3414e-04\n",
            "Epoch 8/300\n",
            "1375/1377 [============================>.] - ETA: 0s - loss: 5.3544e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 5.3497e-04 - val_loss: 2.1296e-04\n",
            "Epoch 9/300\n",
            "1366/1377 [============================>.] - ETA: 0s - loss: 5.1670e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 5.1465e-04 - val_loss: 2.2334e-04\n",
            "Epoch 10/300\n",
            "1371/1377 [============================>.] - ETA: 0s - loss: 4.8415e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 4.8317e-04 - val_loss: 2.0501e-04\n",
            "Epoch 11/300\n",
            "1377/1377 [==============================] - ETA: 0s - loss: 4.5688e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 4.5688e-04 - val_loss: 2.6624e-04\n",
            "Epoch 12/300\n",
            "1361/1377 [============================>.] - ETA: 0s - loss: 4.4639e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 4.4540e-04 - val_loss: 1.8243e-04\n",
            "Epoch 13/300\n",
            "1361/1377 [============================>.] - ETA: 0s - loss: 4.0903e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 4.0551e-04 - val_loss: 1.7823e-04\n",
            "Epoch 14/300\n",
            "1358/1377 [============================>.] - ETA: 0s - loss: 4.0408e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 4.0169e-04 - val_loss: 1.6919e-04\n",
            "Epoch 15/300\n",
            "1374/1377 [============================>.] - ETA: 0s - loss: 3.7819e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.7783e-04 - val_loss: 1.5746e-04\n",
            "Epoch 16/300\n",
            "1361/1377 [============================>.] - ETA: 0s - loss: 3.8053e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.7956e-04 - val_loss: 1.6040e-04\n",
            "Epoch 17/300\n",
            "1359/1377 [============================>.] - ETA: 0s - loss: 3.5838e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.5922e-04 - val_loss: 2.2501e-04\n",
            "Epoch 18/300\n",
            "1376/1377 [============================>.] - ETA: 0s - loss: 3.6983e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.6959e-04 - val_loss: 1.4576e-04\n",
            "Epoch 19/300\n",
            "1374/1377 [============================>.] - ETA: 0s - loss: 3.7791e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 6s 5ms/step - loss: 3.7737e-04 - val_loss: 1.4183e-04\n",
            "Epoch 20/300\n",
            "1376/1377 [============================>.] - ETA: 0s - loss: 3.6407e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.6381e-04 - val_loss: 1.5944e-04\n",
            "Epoch 21/300\n",
            "1375/1377 [============================>.] - ETA: 0s - loss: 3.5563e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.5537e-04 - val_loss: 1.5824e-04\n",
            "Epoch 22/300\n",
            "1356/1377 [============================>.] - ETA: 0s - loss: 3.6557e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.6266e-04 - val_loss: 1.6622e-04\n",
            "Epoch 23/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.7088e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.6814e-04 - val_loss: 1.3914e-04\n",
            "Epoch 24/300\n",
            "1372/1377 [============================>.] - ETA: 0s - loss: 3.6823e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.7235e-04 - val_loss: 1.4505e-04\n",
            "Epoch 25/300\n",
            "1363/1377 [============================>.] - ETA: 0s - loss: 3.6026e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.6026e-04 - val_loss: 1.4367e-04\n",
            "Epoch 26/300\n",
            "1360/1377 [============================>.] - ETA: 0s - loss: 3.5997e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.5908e-04 - val_loss: 2.1794e-04\n",
            "Epoch 27/300\n",
            "1376/1377 [============================>.] - ETA: 0s - loss: 3.4497e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.4481e-04 - val_loss: 1.3481e-04\n",
            "Epoch 28/300\n",
            "1361/1377 [============================>.] - ETA: 0s - loss: 3.7193e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.6906e-04 - val_loss: 1.8044e-04\n",
            "Epoch 29/300\n",
            "1360/1377 [============================>.] - ETA: 0s - loss: 3.5663e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.5908e-04 - val_loss: 1.9057e-04\n",
            "Epoch 30/300\n",
            "1369/1377 [============================>.] - ETA: 0s - loss: 3.4642e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4559e-04 - val_loss: 2.0366e-04\n",
            "Epoch 31/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 3.5359e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.5460e-04 - val_loss: 1.7365e-04\n",
            "Epoch 32/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 3.5005e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.4936e-04 - val_loss: 1.3450e-04\n",
            "Epoch 33/300\n",
            "1359/1377 [============================>.] - ETA: 0s - loss: 3.4877e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.4844e-04 - val_loss: 1.3449e-04\n",
            "Epoch 34/300\n",
            "1365/1377 [============================>.] - ETA: 0s - loss: 3.5805e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.5780e-04 - val_loss: 2.0605e-04\n",
            "Epoch 35/300\n",
            "1370/1377 [============================>.] - ETA: 0s - loss: 3.5634e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.5534e-04 - val_loss: 1.3925e-04\n",
            "Epoch 36/300\n",
            "1371/1377 [============================>.] - ETA: 0s - loss: 3.4785e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.4746e-04 - val_loss: 2.2240e-04\n",
            "Epoch 37/300\n",
            "1358/1377 [============================>.] - ETA: 0s - loss: 3.4245e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.4323e-04 - val_loss: 1.3355e-04\n",
            "Epoch 38/300\n",
            "1363/1377 [============================>.] - ETA: 0s - loss: 3.5255e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.5291e-04 - val_loss: 1.6735e-04\n",
            "Epoch 39/300\n",
            "1370/1377 [============================>.] - ETA: 0s - loss: 3.4295e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4173e-04 - val_loss: 1.5572e-04\n",
            "Epoch 40/300\n",
            "1373/1377 [============================>.] - ETA: 0s - loss: 3.4218e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4216e-04 - val_loss: 1.3050e-04\n",
            "Epoch 41/300\n",
            "1369/1377 [============================>.] - ETA: 0s - loss: 3.4684e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.4527e-04 - val_loss: 1.3136e-04\n",
            "Epoch 42/300\n",
            "1357/1377 [============================>.] - ETA: 0s - loss: 3.5553e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.5208e-04 - val_loss: 2.2255e-04\n",
            "Epoch 43/300\n",
            "1374/1377 [============================>.] - ETA: 0s - loss: 3.4572e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4553e-04 - val_loss: 1.5508e-04\n",
            "Epoch 44/300\n",
            "1366/1377 [============================>.] - ETA: 0s - loss: 3.5053e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.5060e-04 - val_loss: 1.3015e-04\n",
            "Epoch 45/300\n",
            "1375/1377 [============================>.] - ETA: 0s - loss: 3.5565e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.5551e-04 - val_loss: 1.4501e-04\n",
            "Epoch 46/300\n",
            "1373/1377 [============================>.] - ETA: 0s - loss: 3.3727e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.3671e-04 - val_loss: 1.3070e-04\n",
            "Epoch 47/300\n",
            "1370/1377 [============================>.] - ETA: 0s - loss: 3.4671e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4575e-04 - val_loss: 1.3743e-04\n",
            "Epoch 48/300\n",
            "1369/1377 [============================>.] - ETA: 0s - loss: 3.3530e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3414e-04 - val_loss: 1.4019e-04\n",
            "Epoch 49/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.2680e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2572e-04 - val_loss: 1.4108e-04\n",
            "Epoch 50/300\n",
            "1371/1377 [============================>.] - ETA: 0s - loss: 3.4783e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4699e-04 - val_loss: 2.0388e-04\n",
            "Epoch 51/300\n",
            "1361/1377 [============================>.] - ETA: 0s - loss: 3.2446e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3426e-04 - val_loss: 1.9232e-04\n",
            "Epoch 52/300\n",
            "1376/1377 [============================>.] - ETA: 0s - loss: 3.4056e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.4037e-04 - val_loss: 2.5107e-04\n",
            "Epoch 53/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.4231e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4075e-04 - val_loss: 1.3249e-04\n",
            "Epoch 54/300\n",
            "1360/1377 [============================>.] - ETA: 0s - loss: 3.4589e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4319e-04 - val_loss: 1.4303e-04\n",
            "Epoch 55/300\n",
            "1356/1377 [============================>.] - ETA: 0s - loss: 3.2707e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3289e-04 - val_loss: 1.2774e-04\n",
            "Epoch 56/300\n",
            "1365/1377 [============================>.] - ETA: 0s - loss: 3.3376e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3607e-04 - val_loss: 1.4145e-04\n",
            "Epoch 57/300\n",
            "1359/1377 [============================>.] - ETA: 0s - loss: 3.5062e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4796e-04 - val_loss: 1.3911e-04\n",
            "Epoch 58/300\n",
            "1375/1377 [============================>.] - ETA: 0s - loss: 3.3914e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3891e-04 - val_loss: 1.6881e-04\n",
            "Epoch 59/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 3.3226e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3147e-04 - val_loss: 1.4031e-04\n",
            "Epoch 60/300\n",
            "1364/1377 [============================>.] - ETA: 0s - loss: 3.4862e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4724e-04 - val_loss: 1.4140e-04\n",
            "Epoch 61/300\n",
            "1376/1377 [============================>.] - ETA: 0s - loss: 3.3371e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3392e-04 - val_loss: 1.6746e-04\n",
            "Epoch 62/300\n",
            "1369/1377 [============================>.] - ETA: 0s - loss: 3.3510e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.3519e-04 - val_loss: 1.2555e-04\n",
            "Epoch 63/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.3702e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3411e-04 - val_loss: 1.3539e-04\n",
            "Epoch 64/300\n",
            "1366/1377 [============================>.] - ETA: 0s - loss: 3.4084e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4086e-04 - val_loss: 1.3904e-04\n",
            "Epoch 65/300\n",
            "1358/1377 [============================>.] - ETA: 0s - loss: 3.2885e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3027e-04 - val_loss: 1.2885e-04\n",
            "Epoch 66/300\n",
            "1364/1377 [============================>.] - ETA: 0s - loss: 3.4140e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4109e-04 - val_loss: 1.8165e-04\n",
            "Epoch 67/300\n",
            "1372/1377 [============================>.] - ETA: 0s - loss: 3.3350e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3277e-04 - val_loss: 2.1749e-04\n",
            "Epoch 68/300\n",
            "1368/1377 [============================>.] - ETA: 0s - loss: 3.4460e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4375e-04 - val_loss: 1.2588e-04\n",
            "Epoch 69/300\n",
            "1374/1377 [============================>.] - ETA: 0s - loss: 3.3815e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3776e-04 - val_loss: 1.7238e-04\n",
            "Epoch 70/300\n",
            "1356/1377 [============================>.] - ETA: 0s - loss: 3.3885e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3657e-04 - val_loss: 1.2906e-04\n",
            "Epoch 71/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.3869e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3881e-04 - val_loss: 1.8124e-04\n",
            "Epoch 72/300\n",
            "1377/1377 [==============================] - ETA: 0s - loss: 3.2678e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2678e-04 - val_loss: 1.3409e-04\n",
            "Epoch 73/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 3.2024e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2195e-04 - val_loss: 1.3256e-04\n",
            "Epoch 74/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 3.3979e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.3908e-04 - val_loss: 1.7281e-04\n",
            "Epoch 75/300\n",
            "1359/1377 [============================>.] - ETA: 0s - loss: 3.2766e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3130e-04 - val_loss: 1.2880e-04\n",
            "Epoch 76/300\n",
            "1377/1377 [==============================] - ETA: 0s - loss: 3.3205e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3205e-04 - val_loss: 1.2351e-04\n",
            "Epoch 77/300\n",
            "1371/1377 [============================>.] - ETA: 0s - loss: 3.4864e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4774e-04 - val_loss: 1.4560e-04\n",
            "Epoch 78/300\n",
            "1356/1377 [============================>.] - ETA: 0s - loss: 3.4345e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4395e-04 - val_loss: 2.3560e-04\n",
            "Epoch 79/300\n",
            "1356/1377 [============================>.] - ETA: 0s - loss: 3.3174e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3068e-04 - val_loss: 1.7671e-04\n",
            "Epoch 80/300\n",
            "1364/1377 [============================>.] - ETA: 0s - loss: 3.3164e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3262e-04 - val_loss: 1.3087e-04\n",
            "Epoch 81/300\n",
            "1369/1377 [============================>.] - ETA: 0s - loss: 3.3472e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3552e-04 - val_loss: 2.3685e-04\n",
            "Epoch 82/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.2645e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3359e-04 - val_loss: 1.2254e-04\n",
            "Epoch 83/300\n",
            "1365/1377 [============================>.] - ETA: 0s - loss: 3.2354e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2279e-04 - val_loss: 1.2510e-04\n",
            "Epoch 84/300\n",
            "1369/1377 [============================>.] - ETA: 0s - loss: 3.2711e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3084e-04 - val_loss: 1.7616e-04\n",
            "Epoch 85/300\n",
            "1373/1377 [============================>.] - ETA: 0s - loss: 3.2632e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2607e-04 - val_loss: 1.7207e-04\n",
            "Epoch 86/300\n",
            "1357/1377 [============================>.] - ETA: 0s - loss: 3.3756e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3512e-04 - val_loss: 1.2250e-04\n",
            "Epoch 87/300\n",
            "1365/1377 [============================>.] - ETA: 0s - loss: 3.3442e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3478e-04 - val_loss: 1.2956e-04\n",
            "Epoch 88/300\n",
            "1371/1377 [============================>.] - ETA: 0s - loss: 3.3062e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3082e-04 - val_loss: 1.4194e-04\n",
            "Epoch 89/300\n",
            "1373/1377 [============================>.] - ETA: 0s - loss: 3.3688e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3763e-04 - val_loss: 1.7465e-04\n",
            "Epoch 90/300\n",
            "1356/1377 [============================>.] - ETA: 0s - loss: 3.3287e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3070e-04 - val_loss: 1.3385e-04\n",
            "Epoch 91/300\n",
            "1374/1377 [============================>.] - ETA: 0s - loss: 3.3740e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.3669e-04 - val_loss: 1.3543e-04\n",
            "Epoch 92/300\n",
            "1356/1377 [============================>.] - ETA: 0s - loss: 3.4025e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3790e-04 - val_loss: 1.2940e-04\n",
            "Epoch 93/300\n",
            "1372/1377 [============================>.] - ETA: 0s - loss: 3.4323e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4220e-04 - val_loss: 1.2239e-04\n",
            "Epoch 94/300\n",
            "1361/1377 [============================>.] - ETA: 0s - loss: 3.2611e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2387e-04 - val_loss: 1.3065e-04\n",
            "Epoch 95/300\n",
            "1371/1377 [============================>.] - ETA: 0s - loss: 3.2274e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2240e-04 - val_loss: 1.3954e-04\n",
            "Epoch 96/300\n",
            "1372/1377 [============================>.] - ETA: 0s - loss: 3.2281e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2262e-04 - val_loss: 1.2119e-04\n",
            "Epoch 97/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 3.2961e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2862e-04 - val_loss: 1.2921e-04\n",
            "Epoch 98/300\n",
            "1366/1377 [============================>.] - ETA: 0s - loss: 3.2948e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2937e-04 - val_loss: 1.6635e-04\n",
            "Epoch 99/300\n",
            "1371/1377 [============================>.] - ETA: 0s - loss: 3.2289e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2231e-04 - val_loss: 1.2356e-04\n",
            "Epoch 100/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 3.3017e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.3127e-04 - val_loss: 1.2055e-04\n",
            "Epoch 101/300\n",
            "1371/1377 [============================>.] - ETA: 0s - loss: 3.3108e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3004e-04 - val_loss: 1.2154e-04\n",
            "Epoch 102/300\n",
            "1365/1377 [============================>.] - ETA: 0s - loss: 3.3110e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2946e-04 - val_loss: 1.9435e-04\n",
            "Epoch 103/300\n",
            "1366/1377 [============================>.] - ETA: 0s - loss: 3.2653e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2551e-04 - val_loss: 1.3367e-04\n",
            "Epoch 104/300\n",
            "1373/1377 [============================>.] - ETA: 0s - loss: 3.2870e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2824e-04 - val_loss: 2.4246e-04\n",
            "Epoch 105/300\n",
            "1372/1377 [============================>.] - ETA: 0s - loss: 3.3495e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3424e-04 - val_loss: 1.1852e-04\n",
            "Epoch 106/300\n",
            "1375/1377 [============================>.] - ETA: 0s - loss: 3.2557e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2520e-04 - val_loss: 1.2592e-04\n",
            "Epoch 107/300\n",
            "1359/1377 [============================>.] - ETA: 0s - loss: 3.3219e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2882e-04 - val_loss: 1.4452e-04\n",
            "Epoch 108/300\n",
            "1376/1377 [============================>.] - ETA: 0s - loss: 3.2907e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2910e-04 - val_loss: 1.3443e-04\n",
            "Epoch 109/300\n",
            "1370/1377 [============================>.] - ETA: 0s - loss: 3.2747e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2777e-04 - val_loss: 1.7571e-04\n",
            "Epoch 110/300\n",
            "1364/1377 [============================>.] - ETA: 0s - loss: 3.2592e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2678e-04 - val_loss: 1.2205e-04\n",
            "Epoch 111/300\n",
            "1363/1377 [============================>.] - ETA: 0s - loss: 3.2752e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2485e-04 - val_loss: 1.2799e-04\n",
            "Epoch 112/300\n",
            "1373/1377 [============================>.] - ETA: 0s - loss: 3.2611e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2724e-04 - val_loss: 1.1846e-04\n",
            "Epoch 113/300\n",
            "1360/1377 [============================>.] - ETA: 0s - loss: 3.3660e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3529e-04 - val_loss: 1.2239e-04\n",
            "Epoch 114/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.1579e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.1823e-04 - val_loss: 1.4747e-04\n",
            "Epoch 115/300\n",
            "1377/1377 [==============================] - ETA: 0s - loss: 3.3178e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3178e-04 - val_loss: 1.4247e-04\n",
            "Epoch 116/300\n",
            "1376/1377 [============================>.] - ETA: 0s - loss: 3.1890e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.1878e-04 - val_loss: 1.1813e-04\n",
            "Epoch 117/300\n",
            "1363/1377 [============================>.] - ETA: 0s - loss: 3.2214e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2954e-04 - val_loss: 1.1915e-04\n",
            "Epoch 118/300\n",
            "1359/1377 [============================>.] - ETA: 0s - loss: 3.3533e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3601e-04 - val_loss: 1.2424e-04\n",
            "Epoch 119/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 3.4867e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.4783e-04 - val_loss: 1.5619e-04\n",
            "Epoch 120/300\n",
            "1372/1377 [============================>.] - ETA: 0s - loss: 3.2779e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2701e-04 - val_loss: 1.1885e-04\n",
            "Epoch 121/300\n",
            "1356/1377 [============================>.] - ETA: 0s - loss: 3.2351e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2510e-04 - val_loss: 1.2070e-04\n",
            "Epoch 122/300\n",
            "1372/1377 [============================>.] - ETA: 0s - loss: 3.1959e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2046e-04 - val_loss: 1.2533e-04\n",
            "Epoch 123/300\n",
            "1360/1377 [============================>.] - ETA: 0s - loss: 3.1998e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.1921e-04 - val_loss: 1.2375e-04\n",
            "Epoch 124/300\n",
            "1363/1377 [============================>.] - ETA: 0s - loss: 3.2935e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2957e-04 - val_loss: 1.2012e-04\n",
            "Epoch 125/300\n",
            "1367/1377 [============================>.] - ETA: 0s - loss: 3.2860e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2659e-04 - val_loss: 1.9064e-04\n",
            "Epoch 126/300\n",
            "1375/1377 [============================>.] - ETA: 0s - loss: 3.2802e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2788e-04 - val_loss: 1.2093e-04\n",
            "Epoch 127/300\n",
            "1364/1377 [============================>.] - ETA: 0s - loss: 3.2517e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2382e-04 - val_loss: 1.2322e-04\n",
            "Epoch 128/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.2475e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2342e-04 - val_loss: 1.7573e-04\n",
            "Epoch 129/300\n",
            "1374/1377 [============================>.] - ETA: 0s - loss: 3.1805e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.1815e-04 - val_loss: 1.3543e-04\n",
            "Epoch 130/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.3460e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3429e-04 - val_loss: 1.9314e-04\n",
            "Epoch 131/300\n",
            "1363/1377 [============================>.] - ETA: 0s - loss: 3.3021e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2784e-04 - val_loss: 1.2347e-04\n",
            "Epoch 132/300\n",
            "1376/1377 [============================>.] - ETA: 0s - loss: 3.2704e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2708e-04 - val_loss: 1.8684e-04\n",
            "Epoch 133/300\n",
            "1374/1377 [============================>.] - ETA: 0s - loss: 3.2627e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2592e-04 - val_loss: 1.2615e-04\n",
            "Epoch 134/300\n",
            "1364/1377 [============================>.] - ETA: 0s - loss: 3.3432e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3278e-04 - val_loss: 2.0035e-04\n",
            "Epoch 135/300\n",
            "1368/1377 [============================>.] - ETA: 0s - loss: 3.2795e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2695e-04 - val_loss: 1.1917e-04\n",
            "Epoch 136/300\n",
            "1358/1377 [============================>.] - ETA: 0s - loss: 3.3845e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.3954e-04 - val_loss: 1.1935e-04\n",
            "Epoch 137/300\n",
            "1368/1377 [============================>.] - ETA: 0s - loss: 3.3241e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.3074e-04 - val_loss: 1.2066e-04\n",
            "Epoch 138/300\n",
            "1360/1377 [============================>.] - ETA: 0s - loss: 3.2498e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 4s 3ms/step - loss: 3.2641e-04 - val_loss: 1.2345e-04\n",
            "Epoch 139/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.2605e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2355e-04 - val_loss: 1.2334e-04\n",
            "Epoch 140/300\n",
            "1377/1377 [==============================] - ETA: 0s - loss: 3.2673e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2673e-04 - val_loss: 1.2497e-04\n",
            "Epoch 141/300\n",
            "1376/1377 [============================>.] - ETA: 0s - loss: 3.2841e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.2827e-04 - val_loss: 1.4930e-04\n",
            "Epoch 142/300\n",
            "1372/1377 [============================>.] - ETA: 0s - loss: 3.2289e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2195e-04 - val_loss: 1.2879e-04\n",
            "Epoch 143/300\n",
            "1358/1377 [============================>.] - ETA: 0s - loss: 3.2855e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2942e-04 - val_loss: 1.2029e-04\n",
            "Epoch 144/300\n",
            "1362/1377 [============================>.] - ETA: 0s - loss: 3.3512e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 4ms/step - loss: 3.3418e-04 - val_loss: 1.2005e-04\n",
            "Epoch 145/300\n",
            "1363/1377 [============================>.] - ETA: 0s - loss: 3.1999e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2039e-04 - val_loss: 1.5860e-04\n",
            "Epoch 146/300\n",
            "1371/1377 [============================>.] - ETA: 0s - loss: 3.2262e-04"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Can save best model only with  available, skipping.\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r1377/1377 [==============================] - 5s 3ms/step - loss: 3.2279e-04 - val_loss: 1.6200e-04\n"
          ]
        }
      ],
      "source": [
        "from keras import callbacks                   # コールバック（1epoch終了毎に呼び出される）\n",
        "# モデルの学習\n",
        "history4 = model4.fit(\n",
        "    train_X,\n",
        "    train_Y,\n",
        "    batch_size=1,\n",
        "    epochs=300, \n",
        "    validation_data=(test_X, test_Y),        # コールバックする際はvalidation dataも準備する。\n",
        "    callbacks=[\n",
        "        callbacks.EarlyStopping(monitor='val_loss',patience=30),\n",
        "        callbacks.ModelCheckpoint(filepath = './model/best_model.h5',monitor='val_loss', save_best_only=True)\n",
        "        ]\n",
        ")\n",
        "# val_lossが最も優れている学習結果を自動保存致します。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 25,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "id": "kVzrzb_gmMRd",
        "outputId": "8a229b14-adc3-4207-93d4-466e848aad9c"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fe86a211190>"
            ]
          },
          "execution_count": 25,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "# 学習過程のプロット\n",
        "plt.plot(history4.epoch, history4.history[\"loss\"], label=\"loss\")\n",
        "plt.legend()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 26,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "16cwKgPWI2O5",
        "outputId": "00a55021-cb29-46a4-a448-96d5df97a168"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([[0.8422713 ],\n",
              "       [0.8485805 ],\n",
              "       [0.8675078 ],\n",
              "       ...,\n",
              "       [0.04179811],\n",
              "       [0.05441642],\n",
              "       [0.06309146]], dtype=float32)"
            ]
          },
          "execution_count": 26,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# history = model.fit(train_X, train_Y, epochs=15, batch_size=1, verbose=2)\n",
        "# loss: 4.3211e-04 \n",
        "train_X\n",
        "train_Y\n",
        "train_Y = train_Y.reshape(-1,1)\n",
        "train_Y"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YvqidX_kI0CB"
      },
      "source": [
        "### [データの予測と評価](https://premium.aidemy.jp/courses/6160/exercises/ZNnpEhme53)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zqRqyoeTL47Y"
      },
      "source": [
        "#### 正規化を元に戻す\n",
        "予測したデータや結果を正しく評価、可視化する為の処理ですね。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ygO5EJZUdzJU"
      },
      "source": [
        "4回目の学習がhistory4"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 27,
      "metadata": {
        "id": "lyjNS3ptJ3VS"
      },
      "outputs": [],
      "source": [
        "#予測データの作成\n",
        "train_predict =model.predict(train_X)\n",
        "test_predict = model.predict(test_X) \n",
        "\n",
        "train_predict2 =model2.predict(train_X)\n",
        "test_predict2 = model2.predict(test_X) \n",
        "\n",
        "train_predict3 =model3.predict(train_X)\n",
        "test_predict3 = model3.predict(test_X) \n",
        "\n",
        "train_predict4 =model4.predict(train_X)\n",
        "test_predict4 = model4.predict(test_X) "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fhikf_CPKHb6"
      },
      "source": [
        "\n",
        "予測データを作成しました。その予測データを正しく評価するには<font color=\"red\">スケーリングしたデータを元に戻す</font>必要があります。\n",
        "\n",
        "元に戻すには<font color=\"red\">inverse_transform()メソッド</font>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fcGje0QdLmOq"
      },
      "source": [
        "**正規化したらもとに戻す**という手順が必要。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dpHaCaF8nHxX"
      },
      "source": [
        "test_Yの整形 ⇒ ちゃんと比較対象を統一させましょう。677行1列なのか、1列677行なのか。\n",
        "\n",
        "rmseで使用"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TULcdXUFKNRT",
        "outputId": "b7af3b97-9a24-47f5-d1de-94bb26283cbb"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([[ 8680.    ],\n",
              "       [ 8670.    ],\n",
              "       [ 8790.    ],\n",
              "       [ 8750.    ],\n",
              "       [ 8900.    ],\n",
              "       [ 8770.    ],\n",
              "       [ 8800.    ],\n",
              "       [ 8740.    ],\n",
              "       [ 8770.    ],\n",
              "       [ 8792.5   ],\n",
              "       [ 8590.    ],\n",
              "       [ 8530.    ],\n",
              "       [ 8380.    ],\n",
              "       [ 8450.    ],\n",
              "       [ 8530.    ],\n",
              "       [ 8700.    ],\n",
              "       [ 8710.    ],\n",
              "       [ 8640.    ],\n",
              "       [ 8580.    ],\n",
              "       [ 8290.    ],\n",
              "       [ 8445.5   ],\n",
              "       [ 8490.    ],\n",
              "       [ 8680.    ],\n",
              "       [ 8650.    ],\n",
              "       [ 8687.5   ],\n",
              "       [ 8790.    ],\n",
              "       [ 8600.    ],\n",
              "       [ 8460.    ],\n",
              "       [ 8570.    ],\n",
              "       [ 8480.    ],\n",
              "       [ 8650.    ],\n",
              "       [ 8730.    ],\n",
              "       [ 8780.    ],\n",
              "       [ 8610.    ],\n",
              "       [ 8770.    ],\n",
              "       [ 8940.    ],\n",
              "       [ 9060.    ],\n",
              "       [ 8960.    ],\n",
              "       [ 8950.    ],\n",
              "       [ 8870.    ],\n",
              "       [ 8790.    ],\n",
              "       [ 8780.    ],\n",
              "       [ 8660.    ],\n",
              "       [ 8740.    ],\n",
              "       [ 8620.    ],\n",
              "       [ 8710.    ],\n",
              "       [ 8940.    ],\n",
              "       [ 9040.    ],\n",
              "       [ 9070.    ],\n",
              "       [ 9060.    ],\n",
              "       [ 8960.    ],\n",
              "       [ 8980.    ],\n",
              "       [ 9020.    ],\n",
              "       [ 9000.    ],\n",
              "       [ 9120.    ],\n",
              "       [ 9280.    ],\n",
              "       [ 9670.    ],\n",
              "       [ 9650.    ],\n",
              "       [ 9830.    ],\n",
              "       [ 9970.    ],\n",
              "       [ 9830.    ],\n",
              "       [ 9900.    ],\n",
              "       [10040.    ],\n",
              "       [10110.    ],\n",
              "       [10050.    ],\n",
              "       [10140.    ],\n",
              "       [10010.    ],\n",
              "       [10020.    ],\n",
              "       [ 9890.    ],\n",
              "       [ 9937.5   ],\n",
              "       [ 9970.    ],\n",
              "       [ 9930.    ],\n",
              "       [ 9950.    ],\n",
              "       [ 9920.    ],\n",
              "       [10080.    ],\n",
              "       [10160.    ],\n",
              "       [10080.    ],\n",
              "       [10080.    ],\n",
              "       [ 9960.    ],\n",
              "       [ 9960.    ],\n",
              "       [ 9850.    ],\n",
              "       [ 9820.    ],\n",
              "       [ 9790.    ],\n",
              "       [ 9650.    ],\n",
              "       [ 9590.    ],\n",
              "       [ 9670.    ],\n",
              "       [ 9600.    ],\n",
              "       [ 9610.    ],\n",
              "       [ 9450.    ],\n",
              "       [ 9350.    ],\n",
              "       [ 9340.    ],\n",
              "       [ 9410.    ],\n",
              "       [ 9580.    ],\n",
              "       [ 9530.    ],\n",
              "       [ 9450.    ],\n",
              "       [ 9500.    ],\n",
              "       [ 9480.    ],\n",
              "       [ 9460.    ],\n",
              "       [ 9450.    ],\n",
              "       [ 9400.    ],\n",
              "       [ 9520.    ],\n",
              "       [ 9550.    ],\n",
              "       [ 9710.    ],\n",
              "       [ 9690.    ],\n",
              "       [ 9520.    ],\n",
              "       [ 9530.    ],\n",
              "       [ 9550.    ],\n",
              "       [ 9410.    ],\n",
              "       [ 9480.    ],\n",
              "       [ 9460.    ],\n",
              "       [ 9620.    ],\n",
              "       [ 9630.    ],\n",
              "       [ 9670.    ],\n",
              "       [ 9580.    ],\n",
              "       [ 9560.    ],\n",
              "       [ 9670.    ],\n",
              "       [ 9700.    ],\n",
              "       [ 9860.    ],\n",
              "       [ 9820.    ],\n",
              "       [ 9810.    ],\n",
              "       [ 9850.    ],\n",
              "       [ 9787.5   ],\n",
              "       [ 9852.5   ],\n",
              "       [ 9932.5   ],\n",
              "       [10030.    ],\n",
              "       [ 9932.5   ],\n",
              "       [ 9870.    ],\n",
              "       [ 9710.    ],\n",
              "       [ 9550.    ],\n",
              "       [ 9680.    ],\n",
              "       [ 9680.    ],\n",
              "       [ 9680.    ],\n",
              "       [ 9610.    ],\n",
              "       [ 9440.    ],\n",
              "       [ 9560.    ],\n",
              "       [ 9560.    ],\n",
              "       [ 9650.    ],\n",
              "       [ 9640.    ],\n",
              "       [ 9560.    ],\n",
              "       [ 9720.    ],\n",
              "       [ 9770.    ],\n",
              "       [ 9590.    ],\n",
              "       [ 9610.    ],\n",
              "       [ 9630.    ],\n",
              "       [ 9710.    ],\n",
              "       [ 9700.    ],\n",
              "       [ 9760.    ],\n",
              "       [ 9710.    ],\n",
              "       [ 9480.    ],\n",
              "       [ 9410.    ],\n",
              "       [ 9450.    ],\n",
              "       [ 9350.    ],\n",
              "       [ 9380.    ],\n",
              "       [ 9530.    ],\n",
              "       [ 9482.5   ],\n",
              "       [ 9130.    ],\n",
              "       [ 8930.    ],\n",
              "       [ 9000.    ],\n",
              "       [ 8640.    ],\n",
              "       [ 9460.    ],\n",
              "       [10170.    ],\n",
              "       [10370.    ],\n",
              "       [10600.    ],\n",
              "       [10520.    ],\n",
              "       [10530.    ],\n",
              "       [10700.    ],\n",
              "       [10590.    ],\n",
              "       [10490.    ],\n",
              "       [10750.    ],\n",
              "       [10630.    ],\n",
              "       [10510.    ],\n",
              "       [10470.    ],\n",
              "       [10570.    ],\n",
              "       [10660.    ],\n",
              "       [10870.    ],\n",
              "       [10850.    ],\n",
              "       [10840.    ],\n",
              "       [10800.    ],\n",
              "       [10740.    ],\n",
              "       [10730.    ],\n",
              "       [10705.5   ],\n",
              "       [10620.    ],\n",
              "       [10620.    ],\n",
              "       [10630.    ],\n",
              "       [10590.    ],\n",
              "       [10530.    ],\n",
              "       [10440.    ],\n",
              "       [10460.    ],\n",
              "       [10270.    ],\n",
              "       [10230.    ],\n",
              "       [10370.    ],\n",
              "       [10480.    ],\n",
              "       [10420.    ],\n",
              "       [10470.    ],\n",
              "       [10340.    ],\n",
              "       [10280.    ],\n",
              "       [10440.    ],\n",
              "       [10570.    ],\n",
              "       [10520.    ],\n",
              "       [10500.    ],\n",
              "       [10500.    ],\n",
              "       [10560.    ],\n",
              "       [10520.    ],\n",
              "       [10510.    ],\n",
              "       [10457.5   ],\n",
              "       [10550.    ],\n",
              "       [10510.    ],\n",
              "       [10380.    ],\n",
              "       [10380.    ],\n",
              "       [10352.5   ],\n",
              "       [10197.5   ],\n",
              "       [10210.    ],\n",
              "       [10360.    ],\n",
              "       [10320.    ],\n",
              "       [10350.    ],\n",
              "       [10280.    ],\n",
              "       [10252.5   ],\n",
              "       [10330.    ],\n",
              "       [10350.    ],\n",
              "       [10230.    ],\n",
              "       [10290.    ],\n",
              "       [10300.    ],\n",
              "       [10300.    ],\n",
              "       [10310.    ],\n",
              "       [10280.    ],\n",
              "       [10210.    ],\n",
              "       [10270.    ],\n",
              "       [10240.    ],\n",
              "       [10150.    ],\n",
              "       [10180.    ],\n",
              "       [10190.    ],\n",
              "       [10160.    ],\n",
              "       [10000.    ],\n",
              "       [ 9950.    ],\n",
              "       [10120.    ],\n",
              "       [10060.    ],\n",
              "       [10080.    ],\n",
              "       [10040.    ],\n",
              "       [ 9932.5   ],\n",
              "       [10120.    ],\n",
              "       [10050.    ],\n",
              "       [ 9980.    ],\n",
              "       [ 9830.    ],\n",
              "       [ 9780.    ],\n",
              "       [ 9820.    ],\n",
              "       [ 9740.    ],\n",
              "       [ 9860.    ],\n",
              "       [ 9810.    ],\n",
              "       [ 9690.    ],\n",
              "       [ 9720.    ],\n",
              "       [ 9650.    ],\n",
              "       [ 9360.    ],\n",
              "       [ 9297.5   ],\n",
              "       [ 9170.    ],\n",
              "       [ 9160.    ],\n",
              "       [ 9190.    ],\n",
              "       [ 9370.    ],\n",
              "       [ 9410.    ],\n",
              "       [ 9360.    ],\n",
              "       [ 9430.    ],\n",
              "       [ 9420.    ],\n",
              "       [ 9370.    ],\n",
              "       [ 9380.    ],\n",
              "       [ 9530.    ],\n",
              "       [ 9490.    ],\n",
              "       [ 9520.    ],\n",
              "       [ 9570.    ],\n",
              "       [ 9410.    ],\n",
              "       [ 9400.    ],\n",
              "       [ 9652.5   ],\n",
              "       [ 9580.    ],\n",
              "       [ 9690.    ],\n",
              "       [ 9700.    ],\n",
              "       [ 9520.    ],\n",
              "       [ 9350.    ],\n",
              "       [ 9390.    ],\n",
              "       [ 9360.    ],\n",
              "       [ 9560.    ],\n",
              "       [ 9490.    ],\n",
              "       [ 9540.    ],\n",
              "       [ 9400.    ],\n",
              "       [ 9357.5   ],\n",
              "       [ 9520.    ],\n",
              "       [ 9530.    ],\n",
              "       [ 9637.5   ],\n",
              "       [ 9570.    ],\n",
              "       [ 9430.    ],\n",
              "       [ 9470.    ],\n",
              "       [ 9250.    ],\n",
              "       [ 9260.    ],\n",
              "       [ 9180.    ],\n",
              "       [ 9030.    ],\n",
              "       [ 9010.    ],\n",
              "       [ 9220.    ],\n",
              "       [ 9290.    ],\n",
              "       [ 9100.    ],\n",
              "       [ 9040.    ],\n",
              "       [ 8930.    ],\n",
              "       [ 8800.    ],\n",
              "       [ 9130.    ],\n",
              "       [ 8970.    ],\n",
              "       [ 8890.    ],\n",
              "       [ 8830.    ],\n",
              "       [ 8970.    ],\n",
              "       [ 9100.    ],\n",
              "       [ 9160.    ],\n",
              "       [ 9360.    ],\n",
              "       [ 9240.    ],\n",
              "       [ 9170.    ],\n",
              "       [ 9180.    ],\n",
              "       [ 9270.    ],\n",
              "       [ 9190.    ],\n",
              "       [ 9280.    ],\n",
              "       [ 9540.    ],\n",
              "       [ 9560.    ],\n",
              "       [ 9650.    ],\n",
              "       [ 9620.    ],\n",
              "       [ 9490.    ],\n",
              "       [ 9680.    ],\n",
              "       [ 9560.    ],\n",
              "       [ 9530.    ],\n",
              "       [ 9690.    ],\n",
              "       [ 9730.    ],\n",
              "       [ 9500.    ],\n",
              "       [ 9490.    ],\n",
              "       [ 9440.    ],\n",
              "       [ 9200.    ],\n",
              "       [ 9280.    ],\n",
              "       [ 9300.    ],\n",
              "       [ 9247.5   ],\n",
              "       [ 9400.    ],\n",
              "       [ 9670.    ],\n",
              "       [ 9790.    ],\n",
              "       [ 9530.    ],\n",
              "       [ 9550.    ],\n",
              "       [ 9580.    ],\n",
              "       [ 9540.    ],\n",
              "       [ 9290.    ],\n",
              "       [ 9350.    ],\n",
              "       [ 9260.    ],\n",
              "       [ 9210.    ],\n",
              "       [ 9180.    ],\n",
              "       [ 9360.    ],\n",
              "       [ 9550.    ],\n",
              "       [ 9690.    ],\n",
              "       [ 9740.    ],\n",
              "       [ 9910.    ],\n",
              "       [ 9900.    ],\n",
              "       [10110.    ],\n",
              "       [10230.    ],\n",
              "       [10000.    ],\n",
              "       [10010.    ],\n",
              "       [10090.    ],\n",
              "       [ 9840.    ],\n",
              "       [ 9880.    ],\n",
              "       [ 9710.    ],\n",
              "       [ 9510.    ],\n",
              "       [ 9480.    ],\n",
              "       [ 9530.    ],\n",
              "       [ 9510.    ],\n",
              "       [ 9930.    ],\n",
              "       [ 9940.    ],\n",
              "       [ 9610.    ],\n",
              "       [ 9740.    ],\n",
              "       [ 9760.    ],\n",
              "       [ 9800.    ],\n",
              "       [ 9650.    ],\n",
              "       [ 9490.    ],\n",
              "       [ 9440.    ],\n",
              "       [ 9760.    ],\n",
              "       [ 9730.    ],\n",
              "       [10040.    ],\n",
              "       [10140.    ],\n",
              "       [10260.    ],\n",
              "       [10240.    ],\n",
              "       [10450.    ],\n",
              "       [10640.    ],\n",
              "       [10370.    ],\n",
              "       [10400.    ],\n",
              "       [10540.    ],\n",
              "       [10430.    ],\n",
              "       [10680.    ],\n",
              "       [10712.5   ],\n",
              "       [10712.5   ],\n",
              "       [11105.5   ],\n",
              "       [11050.    ],\n",
              "       [11077.5   ],\n",
              "       [10940.    ],\n",
              "       [11210.    ],\n",
              "       [11160.    ],\n",
              "       [10950.    ],\n",
              "       [10950.    ],\n",
              "       [11100.    ],\n",
              "       [10900.    ],\n",
              "       [10910.    ],\n",
              "       [11090.    ],\n",
              "       [11270.    ],\n",
              "       [11210.    ],\n",
              "       [11150.    ],\n",
              "       [11260.    ],\n",
              "       [11210.    ],\n",
              "       [11170.    ],\n",
              "       [11290.    ],\n",
              "       [11280.    ],\n",
              "       [11360.    ],\n",
              "       [11310.    ],\n",
              "       [11230.    ],\n",
              "       [11100.    ],\n",
              "       [11120.    ],\n",
              "       [11020.    ],\n",
              "       [10930.    ],\n",
              "       [10770.    ],\n",
              "       [10740.    ],\n",
              "       [10720.    ],\n",
              "       [10727.5   ],\n",
              "       [10750.    ],\n",
              "       [10690.    ],\n",
              "       [10780.    ],\n",
              "       [10670.    ],\n",
              "       [10670.    ],\n",
              "       [10680.    ],\n",
              "       [10610.    ],\n",
              "       [10570.    ],\n",
              "       [10560.    ],\n",
              "       [10580.    ],\n",
              "       [10370.    ],\n",
              "       [10140.    ],\n",
              "       [10230.    ],\n",
              "       [10230.    ],\n",
              "       [10170.    ],\n",
              "       [10100.    ],\n",
              "       [10100.    ],\n",
              "       [10200.    ],\n",
              "       [10350.    ],\n",
              "       [10370.    ],\n",
              "       [10140.    ],\n",
              "       [10330.    ],\n",
              "       [10290.    ],\n",
              "       [10020.    ],\n",
              "       [10010.    ],\n",
              "       [10100.    ],\n",
              "       [10067.5   ],\n",
              "       [ 9990.    ],\n",
              "       [ 9940.    ],\n",
              "       [ 9960.    ],\n",
              "       [10060.    ],\n",
              "       [10350.    ],\n",
              "       [10420.    ],\n",
              "       [10370.    ],\n",
              "       [10190.    ],\n",
              "       [10160.    ],\n",
              "       [10420.    ],\n",
              "       [10270.    ],\n",
              "       [10340.    ],\n",
              "       [10510.    ],\n",
              "       [10590.    ],\n",
              "       [10840.    ],\n",
              "       [10710.    ],\n",
              "       [10760.    ],\n",
              "       [10840.    ],\n",
              "       [10970.    ],\n",
              "       [10890.    ],\n",
              "       [10720.    ],\n",
              "       [10890.    ],\n",
              "       [10757.5   ],\n",
              "       [10810.    ],\n",
              "       [10700.    ],\n",
              "       [10710.    ],\n",
              "       [10690.    ],\n",
              "       [10640.    ],\n",
              "       [10607.5   ],\n",
              "       [10610.5   ],\n",
              "       [10540.    ],\n",
              "       [10650.    ],\n",
              "       [10630.    ],\n",
              "       [10510.    ],\n",
              "       [10510.    ],\n",
              "       [10472.5   ],\n",
              "       [10350.    ],\n",
              "       [10190.    ],\n",
              "       [10140.    ],\n",
              "       [10150.    ],\n",
              "       [10180.    ],\n",
              "       [10070.    ],\n",
              "       [10100.    ],\n",
              "       [10100.    ],\n",
              "       [ 9830.    ],\n",
              "       [10000.    ],\n",
              "       [10120.    ],\n",
              "       [10150.    ],\n",
              "       [10000.    ],\n",
              "       [ 9950.    ],\n",
              "       [ 9630.    ],\n",
              "       [ 9550.    ],\n",
              "       [ 9330.    ],\n",
              "       [ 9070.    ],\n",
              "       [ 9390.    ],\n",
              "       [ 9450.    ],\n",
              "       [ 9390.    ],\n",
              "       [ 9512.5   ],\n",
              "       [ 9490.    ],\n",
              "       [ 9550.    ],\n",
              "       [ 9670.    ],\n",
              "       [ 9740.    ],\n",
              "       [ 9800.    ],\n",
              "       [ 9750.    ],\n",
              "       [ 9790.    ],\n",
              "       [ 9880.    ],\n",
              "       [ 9870.    ],\n",
              "       [ 9800.    ],\n",
              "       [ 9780.    ],\n",
              "       [ 9700.    ],\n",
              "       [ 9820.    ],\n",
              "       [ 9830.    ],\n",
              "       [10010.    ],\n",
              "       [ 9880.    ],\n",
              "       [10080.    ],\n",
              "       [10250.    ],\n",
              "       [10360.    ],\n",
              "       [10290.    ],\n",
              "       [10270.    ],\n",
              "       [10330.    ],\n",
              "       [10330.    ],\n",
              "       [10250.    ],\n",
              "       [10270.    ],\n",
              "       [10240.    ],\n",
              "       [10080.    ],\n",
              "       [10090.    ],\n",
              "       [10010.    ],\n",
              "       [ 9860.    ],\n",
              "       [ 9790.    ],\n",
              "       [ 9680.    ],\n",
              "       [ 9680.    ],\n",
              "       [ 9730.    ],\n",
              "       [ 9980.    ],\n",
              "       [10110.    ],\n",
              "       [10100.    ],\n",
              "       [10030.    ],\n",
              "       [10310.    ],\n",
              "       [10460.    ],\n",
              "       [10340.    ],\n",
              "       [10380.    ],\n",
              "       [10210.    ],\n",
              "       [10190.    ],\n",
              "       [10180.    ],\n",
              "       [10400.    ],\n",
              "       [10470.    ],\n",
              "       [10340.    ],\n",
              "       [10410.    ],\n",
              "       [10310.    ],\n",
              "       [10190.    ],\n",
              "       [10200.    ],\n",
              "       [10310.    ],\n",
              "       [10500.    ],\n",
              "       [10450.    ],\n",
              "       [10530.    ],\n",
              "       [10510.    ],\n",
              "       [10640.    ],\n",
              "       [10500.    ],\n",
              "       [10550.    ],\n",
              "       [10280.    ],\n",
              "       [10360.    ],\n",
              "       [10210.    ],\n",
              "       [10300.    ],\n",
              "       [10250.    ],\n",
              "       [10590.    ],\n",
              "       [10520.    ],\n",
              "       [10440.    ],\n",
              "       [10580.    ],\n",
              "       [10550.    ],\n",
              "       [10420.    ],\n",
              "       [10400.    ],\n",
              "       [10250.    ],\n",
              "       [10400.    ],\n",
              "       [10360.    ],\n",
              "       [10370.    ],\n",
              "       [10200.    ],\n",
              "       [10110.    ],\n",
              "       [10080.    ],\n",
              "       [10080.    ],\n",
              "       [ 9940.    ],\n",
              "       [ 9780.    ],\n",
              "       [ 9730.    ],\n",
              "       [ 9650.    ],\n",
              "       [ 9380.    ],\n",
              "       [ 9330.    ],\n",
              "       [ 9290.    ],\n",
              "       [ 9250.    ],\n",
              "       [ 9040.    ],\n",
              "       [ 9250.    ],\n",
              "       [ 9310.    ],\n",
              "       [ 9410.    ],\n",
              "       [ 9640.    ],\n",
              "       [ 9700.    ],\n",
              "       [ 9840.    ],\n",
              "       [ 9890.    ],\n",
              "       [ 9930.    ],\n",
              "       [ 9930.    ],\n",
              "       [ 9820.    ],\n",
              "       [ 9900.    ],\n",
              "       [ 9790.    ],\n",
              "       [ 9570.    ],\n",
              "       [ 9540.    ],\n",
              "       [ 9830.    ],\n",
              "       [ 9770.    ],\n",
              "       [ 9730.    ],\n",
              "       [ 9830.    ],\n",
              "       [ 9770.    ],\n",
              "       [10040.    ],\n",
              "       [10140.    ],\n",
              "       [10020.    ],\n",
              "       [ 9990.    ],\n",
              "       [ 9780.    ],\n",
              "       [ 9850.    ],\n",
              "       [ 9780.    ],\n",
              "       [ 9690.    ],\n",
              "       [ 9760.    ],\n",
              "       [ 9710.    ],\n",
              "       [ 9690.    ],\n",
              "       [ 9510.    ],\n",
              "       [ 9450.    ],\n",
              "       [ 9430.    ],\n",
              "       [ 9340.    ],\n",
              "       [ 9340.    ],\n",
              "       [ 9230.    ],\n",
              "       [ 9280.    ],\n",
              "       [ 9330.    ],\n",
              "       [ 9300.    ],\n",
              "       [ 9080.    ],\n",
              "       [ 9270.    ],\n",
              "       [ 9130.    ],\n",
              "       [ 9360.    ],\n",
              "       [ 9320.    ],\n",
              "       [ 9450.    ],\n",
              "       [ 9460.    ],\n",
              "       [ 9370.    ],\n",
              "       [ 9030.    ],\n",
              "       [ 8860.    ],\n",
              "       [ 8510.    ],\n",
              "       [ 8750.    ],\n",
              "       [ 8740.    ],\n",
              "       [ 8830.    ],\n",
              "       [ 8750.    ],\n",
              "       [ 8770.    ],\n",
              "       [ 8930.    ],\n",
              "       [ 8940.    ],\n",
              "       [ 8750.    ],\n",
              "       [ 8740.    ],\n",
              "       [ 8850.    ],\n",
              "       [ 8890.    ],\n",
              "       [ 8970.    ],\n",
              "       [ 8930.    ],\n",
              "       [ 8580.    ],\n",
              "       [ 8860.    ],\n",
              "       [ 8890.    ],\n",
              "       [ 8730.    ],\n",
              "       [ 8690.    ],\n",
              "       [ 8380.    ],\n",
              "       [ 8120.    ],\n",
              "       [ 8200.    ],\n",
              "       [ 8630.    ],\n",
              "       [ 8710.    ],\n",
              "       [ 8440.    ],\n",
              "       [ 8400.    ],\n",
              "       [ 8180.0005],\n",
              "       [ 7860.    ],\n",
              "       [ 7920.    ],\n",
              "       [ 7889.9995],\n",
              "       [ 7679.9995],\n",
              "       [ 7510.    ],\n",
              "       [ 7090.    ],\n",
              "       [ 7389.9995],\n",
              "       [ 7040.    ],\n",
              "       [ 7050.    ],\n",
              "       [ 7170.    ],\n",
              "       [ 7410.    ],\n",
              "       [ 7269.9995]], dtype=float32)"
            ]
          },
          "execution_count": 28,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# ちゃんと677行１列にする。\n",
        "\n",
        "test_Y.shape\n",
        "test_Y = test_Y.reshape(-1,1)\n",
        "test_Y.shape\n",
        "test_Y\n",
        "\n",
        "test_Y = scaler_train.inverse_transform(test_Y)\n",
        "test_Y"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "P8eKKVEKoQML"
      },
      "source": [
        "算出した予測値の整形"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 29,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2VIQ8nNdqrMl",
        "outputId": "b1b34dac-19de-417a-8af4-868b6a351e5e"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(677, 1)"
            ]
          },
          "execution_count": 29,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "test_predict\n",
        "test_predict.shape"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 30,
      "metadata": {
        "id": "fxyROhn0oeVb"
      },
      "outputs": [],
      "source": [
        "test_predict  = scaler_train.inverse_transform(test_predict)\n",
        "test_predict2 = scaler_train.inverse_transform(test_predict2)\n",
        "test_predict3 = scaler_train.inverse_transform(test_predict3)\n",
        "test_predict4 = scaler_train.inverse_transform(test_predict4)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5Rbnn47EoBpc"
      },
      "source": [
        "あまり"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d2Ecp8zHMHpY"
      },
      "source": [
        "#### rmseの算出\n",
        "ちゃんと双方のデータ形式をそろえる事。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ldEkyegyHOJO"
      },
      "source": [
        "まずはrmse比較する２つのデータの形状を確認して必要ならばそろえる。\n",
        "\n",
        "その必要があったので揃えます。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 31,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Cpvc1XVRG8fF",
        "outputId": "d7078452-c991-47ad-a64d-7ada34f052d5"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(677, 1)"
            ]
          },
          "execution_count": 31,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# まずはrmse比較する２つのデータの形状を確認して必要ならばそろえる。\n",
        "test_predict.shape\n",
        "\n",
        "# test_Y      # これがおかしい。1列29行になってる。　ちゃんと29行1列にしないと。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 32,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4SMYaHcBHTTd",
        "outputId": "aa4d3a1a-8ea8-4b90-d85f-30734f308d66"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(677, 1)\n",
            "(677, 1)\n",
            "(677, 1)\n",
            "(677, 1)\n",
            "(677, 1)\n"
          ]
        }
      ],
      "source": [
        "# これで変換\n",
        "import numpy as np\n",
        "# test_Y = test_Y.reshape(29,1)\n",
        "\n",
        "# 返還後は確認しましょう。\n",
        "print(test_predict.shape)\n",
        "print(test_predict2.shape)\n",
        "print(test_predict3.shape)\n",
        "print(test_predict4.shape)\n",
        "print(test_Y.shape)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 33,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "v_mAsz3DJA0P",
        "outputId": "20519924-d26e-4e27-f4a7-f4533289dc95"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "No1のRMSE 145.98297\n",
            "No2のRMSE 189.88597\n",
            "No3のRMSE 144.85132\n",
            "No4のRMSE 161.38802\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "from sklearn.metrics import mean_squared_error as MSE\n",
        "# y_trainと予測値のRMSEを出す\n",
        "\n",
        "print(\"No1のRMSE\",np.sqrt(MSE(test_predict,test_Y)))\n",
        "print(\"No2のRMSE\",np.sqrt(MSE(test_predict2,test_Y)))\n",
        "print(\"No3のRMSE\",np.sqrt(MSE(test_predict3,test_Y)))\n",
        "print(\"No4のRMSE\",np.sqrt(MSE(test_predict4,test_Y)))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5woJCu6JR1Bw"
      },
      "source": [
        "#### [データ可視化](https://premium.aidemy.jp/courses/6160/exercises/9ShLDO1Ukc)\n",
        "可視化する時もデータを整形する事。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "h0aJFlPaMGDF"
      },
      "source": [
        "こちら私が可視化したもの。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jphhGHcOps7F"
      },
      "source": [
        "プロット時にこのようにデータ出力範囲を限定させました。\n",
        "\n",
        "```\n",
        "plt.plot(df[1200:], label='dataset',color = \"blue\")\n",
        "```"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 34,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 621
        },
        "id": "MKJseOFWSkxM",
        "outputId": "07e969a2-3309-4ec4-ee06-b1ac2f20438c"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1440x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "plt.rc(\"figure\",figsize=(20,10))\n",
        "# 検証データから予測したデータの整形\n",
        "# 既存の配列`dataset`と同じ大きさ（行数・列数）、データ型で\n",
        "# 値を０に初期化したプロット用の空の配列を作成します\n",
        "test_predict_plot = np.empty_like(df)\n",
        "\n",
        "# 空の配列のすべての値を欠損値`nan`にします\n",
        "test_predict_plot[:, :] = np.nan\n",
        "\n",
        "# 訓練データの予測値と位置を合わせます\n",
        "test_predict_plot[len(train_predict)+(look_back*2):len(df), :] = test_predict4\n",
        "\n",
        "# データのプロット\n",
        "plt.title(\"monthly-champagne-sales-in-1000s\")\n",
        "plt.xlabel(\"time(month)\")\n",
        "plt.ylabel(\"Earnings\")\n",
        "\n",
        "# 読み込んだままのデータをプロットします\n",
        "# plt.plot(df, label='dataset',color = \"blue\")\n",
        "plt.plot(df[1200:], label='dataset',color = \"blue\")\n",
        "\n",
        "# 検証データから予測した値をプロットします\n",
        "plt.plot(test_predict_plot[1200:], label='predict',color = \"red\")\n",
        "\n",
        "plt.legend(loc='lower right') # 右下に表示しますと。\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g7wHo31xeMq7"
      },
      "source": [
        "決して悪くないんじゃないですか？　"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 35,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Xt6hTTGhqK87",
        "outputId": "8bd14ec3-d8ac-43a8-89d2-4faf6a6d9f53"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(677, 1)\n",
            "(677, 1)\n"
          ]
        }
      ],
      "source": [
        "print(test_predict4.shape)\n",
        "print(test_Y.shape)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 36,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fAHEvDZ-qXm8",
        "outputId": "51f6b916-0352-4182-ce4f-ac4112895838"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([[18980.],\n",
              "       [19330.],\n",
              "       [19460.],\n",
              "       ...,\n",
              "       [ 7170.],\n",
              "       [ 7410.],\n",
              "       [ 7270.]], dtype=float32)"
            ]
          },
          "execution_count": 36,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "test_Y\n",
        "df"
      ]
    }
  ],
  "metadata": {
    "celltoolbar": "Edit Metadata",
    "colab": {
      "collapsed_sections": [
        "-8A4PciAV4Y8"
      ],
      "include_colab_link": true,
      "name": "時系列株価予測git用.ipynb",
      "provenance": []
    },
    "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": "3.6.4"
    },
    "toc": {
      "base_numbering": 1,
      "nav_menu": {},
      "number_sections": true,
      "sideBar": true,
      "skip_h1_title": false,
      "title_cell": "Table of Contents",
      "title_sidebar": "Contents",
      "toc_cell": false,
      "toc_position": {},
      "toc_section_display": true,
      "toc_window_display": false
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}