VK_Energie_LSTM_Auto_Regression_(Heat_Demand)_.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "VK Energie - LSTM Auto Regression (Heat Demand) .ipynb",
      "provenance": [],
      "collapsed_sections": [
        "uph2zf8vc0fQ"
      ],
      "toc_visible": true,
      "authorship_tag": "ABX9TyM0rF6+LjggdRBDZM2ZO+Mn",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/victor-armegioiu/Time-Series-Prediction/blob/main/VK_Energie_LSTM_Auto_Regression_(Heat_Demand)_.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uph2zf8vc0fQ"
      },
      "source": [
        "# Praktikum Project (VK Energie - Forecasting Heat Demand)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "D4XNcEHNdAUu"
      },
      "source": [
        "# Boilerplate \n",
        "\n",
        "This section does necesary imports, loads/parses the data."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WxS_3Li63y8S"
      },
      "source": [
        "# Non-neural based math imports.\n",
        "import numpy as np\n",
        "\n",
        "import scipy \n",
        "from scipy import stats\n",
        "\n",
        "from sklearn.preprocessing import StandardScaler, MinMaxScaler\n",
        "from sklearn.model_selection import train_test_split"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SE2Wellp38cv"
      },
      "source": [
        "# Data loading\n",
        "import pandas as pd\n",
        "import csv"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oHTXJGQjO2pA"
      },
      "source": [
        "# Drawing stuff.\n",
        "import matplotlib\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "#matplotlib.style.use('ggplot')"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lZFD7pYxdjmZ"
      },
      "source": [
        "# Neural networks stuff.\n",
        "import torch\n",
        "from torch.autograd import Variable\n",
        "import torch.nn.functional as F\n",
        "import torch.utils.data as Data"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-HdkIun_PXQb"
      },
      "source": [
        "df = pd.DataFrame()\n",
        "reader = csv.reader(\n",
        "    open(\"Oberbayern.csv\"), delimiter=\";\")\n",
        "\n",
        "rows = [row for row in reader]\n",
        "df = df.append(rows[1:])"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ir9lyF5pQmNG"
      },
      "source": [
        "df.columns = ['TimestampUtc', 'Heat Demand', 'Ambient Temperature',  \n",
        "              'Solar Irradiance']"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QfRsrPw3VHDJ"
      },
      "source": [
        "for column in df.columns[1:]:\n",
        "  df[column] = df[column].astype(float)"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bkCOz0_jdLis"
      },
      "source": [
        "# Data Analysis\n",
        "\n",
        "This section covers the following analyses on the dataset:\n",
        "\n",
        "\n",
        "*   Peaking at samples from the data.\n",
        "*   Missing values count\n",
        "*   Overall statistics on the regressors/regressands (mean, max, std, etc). Plots for the histograms.\n",
        "*   Correlation analysis - Making sure the correlations between the features and the targets make sense, as well as ensuring we don't have too strong a linear correlation between the features. \n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7xDrqvnJQexE",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "0dff63c4-90fc-4ac5-fbff-b10018036396"
      },
      "source": [
        "df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>TimestampUtc</th>\n",
              "      <th>Heat Demand</th>\n",
              "      <th>Ambient Temperature</th>\n",
              "      <th>Solar Irradiance</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2020-02-26 17:00:00+01:00</td>\n",
              "      <td>5715.7325</td>\n",
              "      <td>1.370000</td>\n",
              "      <td>0.050000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2020-02-26 18:00:00+01:00</td>\n",
              "      <td>5810.5000</td>\n",
              "      <td>0.880000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2020-02-26 19:00:00+01:00</td>\n",
              "      <td>6059.9000</td>\n",
              "      <td>0.670000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2020-02-26 20:00:00+01:00</td>\n",
              "      <td>6114.0850</td>\n",
              "      <td>0.600000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2020-02-26 21:00:00+01:00</td>\n",
              "      <td>6027.0525</td>\n",
              "      <td>0.540000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5941</th>\n",
              "      <td>2020-11-02 12:00:00+01:00</td>\n",
              "      <td>3431.5000</td>\n",
              "      <td>16.787415</td>\n",
              "      <td>5.718750</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5942</th>\n",
              "      <td>2020-11-02 13:00:00+01:00</td>\n",
              "      <td>3221.2500</td>\n",
              "      <td>16.833160</td>\n",
              "      <td>7.406250</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5943</th>\n",
              "      <td>2020-11-02 14:00:00+01:00</td>\n",
              "      <td>3153.7500</td>\n",
              "      <td>16.731232</td>\n",
              "      <td>2.312500</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5944</th>\n",
              "      <td>2020-11-02 15:00:00+01:00</td>\n",
              "      <td>2795.1250</td>\n",
              "      <td>16.396759</td>\n",
              "      <td>7.906250</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5945</th>\n",
              "      <td>2020-11-02 16:00:00+01:00</td>\n",
              "      <td>1974.0000</td>\n",
              "      <td>16.483002</td>\n",
              "      <td>1.140625</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5946 rows × 4 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                   TimestampUtc  ...  Solar Irradiance\n",
              "0     2020-02-26 17:00:00+01:00  ...          0.050000\n",
              "1     2020-02-26 18:00:00+01:00  ...          0.000000\n",
              "2     2020-02-26 19:00:00+01:00  ...          0.000000\n",
              "3     2020-02-26 20:00:00+01:00  ...          0.000000\n",
              "4     2020-02-26 21:00:00+01:00  ...          0.000000\n",
              "...                         ...  ...               ...\n",
              "5941  2020-11-02 12:00:00+01:00  ...          5.718750\n",
              "5942  2020-11-02 13:00:00+01:00  ...          7.406250\n",
              "5943  2020-11-02 14:00:00+01:00  ...          2.312500\n",
              "5944  2020-11-02 15:00:00+01:00  ...          7.906250\n",
              "5945  2020-11-02 16:00:00+01:00  ...          1.140625\n",
              "\n",
              "[5946 rows x 4 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "22k6oXE1RiyA",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "91b52d92-def7-446f-d49d-00bbf73e9525"
      },
      "source": [
        "df.isnull().sum()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "TimestampUtc           0\n",
              "Heat Demand            0\n",
              "Ambient Temperature    0\n",
              "Solar Irradiance       0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "48QbG6f1X4ma",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "77083c6a-50ab-475c-910e-18432bc0dc2a"
      },
      "source": [
        "df.columns"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['TimestampUtc', 'Heat Demand', 'Ambient Temperature',\n",
              "       'Solar Irradiance'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 36
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_2CO-n7qQj_G"
      },
      "source": [
        "# Compute and store pairwise correlations between features and\n",
        "# targets.\n",
        "correlations = {}\n",
        "for col1 in df.columns[1:]:\n",
        "  for col2 in df.columns[1:]:\n",
        "    if col1 != col2:\n",
        "      r2_correlation_coeff = stats.pearsonr(df[col1], df[col2])[0]\n",
        "      correlations[col1, col2] = r2_correlation_coeff"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4tMqblL4Yk3Z",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 851
        },
        "outputId": "2cc2107d-22c6-461a-9441-92f2926fdee6"
      },
      "source": [
        "# Plot all pairwise features/targets/\n",
        "relevant_columns = df.columns[1:] \n",
        "\n",
        "for i in range(len(relevant_columns)):\n",
        "  for j in range(i + 1, len(relevant_columns)):\n",
        "    col_i, col_j = relevant_columns[i], relevant_columns[j]\n",
        "    plt.scatter(df[col_i], df[col_j])\n",
        "    plt.xlabel(col_i)\n",
        "    plt.ylabel(col_j)\n",
        "\n",
        "    r2_correlation_coeff = round(correlations[col_i, col_j], 3)\n",
        "    plt.title('R2 correlation coefficient: {}'.format(r2_correlation_coeff))\n",
        "    plt.show()\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wnUC6mU6Vl9y",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "d7394a3b-025e-4043-ee67-24aacea52ef6"
      },
      "source": [
        "for column in df.columns[1:]:\n",
        "  print(df[column].describe())\n",
        "  plt.hist(df[column], bins=500)\n",
        "  plt.xlabel(column)\n",
        "  plt.ylabel('Counts')\n",
        "  plt.show()\n",
        "  print()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "count    5946.000000\n",
            "mean     2753.058015\n",
            "std      1530.114730\n",
            "min        43.000000\n",
            "25%      1514.583328\n",
            "50%      2222.491250\n",
            "75%      3754.158808\n",
            "max      9681.533203\n",
            "Name: Heat Demand, dtype: float64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAEICAYAAACqMQjAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAXb0lEQVR4nO3dfZBldX3n8fdHRgRRYdDeqQnozrBSGMrgoC2BEC3lwSC6QrIsQqw4ZnGnNq5G41bMsNZG3UrtomX5tO6qEzGZ7BoeRAgIBEJG1KRijTaC8hwGBIUCpnVFjHEV9Lt/3NNy7erTffvh9L3d/X5V3epzfufpd+7pmU+f3znnd1JVSJI0kycNuwKSpNFlSEiSWhkSkqRWhoQkqZUhIUlqZUhIklp1GhJJ/iDJrUluSXJBkv2SbE6yO8meJBcl2bfLOkiSFi5dPSeR5BDg74Ejq+pHSS4GrgZOBS6tqguTfBz4elV9bLZ1PetZz6pNmzZ1Uk9JWq1uuOGG71TV2GLWsW6pKjPL+vdP8hjwVOBB4ATgt5vpO4F3A7OGxKZNm5iYmOiwmpK0+iS5b7Hr6Ky5qaoeAN4PfIteOHwfuAF4pKoeb2a7HzikqzpIkhans5BIsh44DdgM/BJwAHDKPJbflmQiycTk5GRHtZQkzabLC9cnAd+sqsmqegy4FDgeOCjJVDPXocADMy1cVTuqaryqxsfGFtWkJklaoC5D4lvAsUmemiTAicBtwPXAGc08W4HLO6yDJGkRurwmsRu4BPgacHOzrR3AHwFvT7IHeCZwfld1kCQtTqd3N1XVu4B3TSu+Bzimy+1KkpaGT1xLkloZEpKkVoaEJKmVISFJamVISJJaGRKSpFaGhCSplSEhSWplSEiSWhkSkqRWhoQkqZUhIUlqZUhIkloZEpKkVoaEJKmVISFJamVISJJaGRKSpFadhUSSI5Lc1Pd5NMnbkhyc5LokdzU/13dVB0nS4nQWElV1Z1VtqaotwIuAfwYuA7YDu6rqcGBXMy5JGkHL1dx0InB3Vd0HnAbsbMp3AqcvUx0kSfO0XCFxFnBBM7yhqh5shh8CNixTHSRJ89R5SCTZF3gN8Jnp06qqgGpZbluSiSQTk5OTHddSkjST5TiTeCXwtap6uBl/OMlGgObn3pkWqqodVTVeVeNjY2PLUE1J0nTLERJn80RTE8AVwNZmeCtw+TLUYWRs2n7VsKsgSQPrNCSSHACcDFzaV3wecHKSu4CTmnFJ0gha1+XKq+qHwDOnlX2X3t1OkqQR5xPXkqRWhoQkqZUhIUlqZUgsE+9qkrQSGRKSpFaGhCSplSEhSWplSEiSWhkSkqRWhoQkqZUhIUlqZUgMic9NSFoJDAlJUitDYhl59iBppTEkJEmtDAlJUitDQpLUypCQJLXq+h3XByW5JMkdSW5PclySg5Ncl+Su5uf6LusgSVq4rs8kPgxcU1XPA14A3A5sB3ZV1eHArmZckjSCOguJJAcCLwXOB6iqn1TVI8BpwM5mtp3A6V3VQZK0OF2eSWwGJoE/S3Jjkk8mOQDYUFUPNvM8BGyYaeEk25JMJJmYnJzssJqSpDZdhsQ64IXAx6rqaOCHTGtaqqoCaqaFq2pHVY1X1fjY2FiH1ZQktekyJO4H7q+q3c34JfRC4+EkGwGan3s7rIMkaRE6C4mqegj4dpIjmqITgduAK4CtTdlW4PKu6iBJWpx1Ha//LcCnk+wL3AP8Lr1gujjJOcB9wJkd10GStECdhkRV3QSMzzDpxC63K0laGj5xLUlqZUhIkloZEpKkVobEECzFy4d8gZGk5WBISJJaGRKSpFaGhCSplSEhSWplSEiSWhkSkqRWhoQkqZUhMWJ8/kHSKDEkJEmtDIlVwjMQSV0wJCRJrQwJSVIrQ0KS1MqQkCS16vT1pUnuBX4A/BR4vKrGkxwMXARsAu4Fzqyq73VZD0nSwizHmcTLq2pLVU2963o7sKuqDgd2NeNrlnclSRplw2huOg3Y2QzvBE4fQh0kSQPoOiQK+JskNyTZ1pRtqKoHm+GHgA0zLZhkW5KJJBOTk5MdV7M7nilIWsk6vSYB/HpVPZDkXwDXJbmjf2JVVZKaacGq2gHsABgfH59xHklStzo9k6iqB5qfe4HLgGOAh5NsBGh+7u2yDpKkhessJJIckOTpU8PAK4BbgCuArc1sW4HLu6rDqBtmU5TNYJIG0WVz0wbgsiRT2/nLqromyVeBi5OcA9wHnNlhHSRJi9BZSFTVPcALZij/LnBiV9uVJC0dn7geEUvZ/GNTkqSlYkhIkloZEiucZw2SumRISJJaGRKSpFbzDokk65Mc1UVlJEmjZaCQSPKFJM9ouvn+GvCnST7QbdUkScM26JnEgVX1KPBbwF9U1a8CJ3VXLUnSKBg0JNY1/SydCVzZYX1Wpa7uQPLOJkldGzQk3gNcC+ypqq8mOQy4q7tqSZJGwaAh8WBVHVVVb4Kfd7nhNYmOeaYgadgGDYn/MWCZJGkVmbWDvyTHAb8GjCV5e9+kZwD7dFkxSdLwzXUmsS/wNHph8vS+z6PAGd1Wbe2aq5nJZihJy2XWM4mq+iLwxSR/XlX3LVOdJEkjYtD3STwlyQ5gU/8yVXVCF5WSJI2GQUPiM8DHgU8CP+2uOpqLTU2SltOgIfF4VX2s05pIkkbOoLfAfi7Jm5JsTHLw1GeQBZPsk+TGJFc245uT7E6yJ8lFSfZdcO0lSZ0aNCS2An8I/ANwQ/OZGHDZtwK3942/F/hgVT0X+B5wzoDrWVPm06zUP6/NUZKW0kAhUVWbZ/gcNtdySQ4FXkXvWgZJApwAXNLMshM4fWFVlyR1baBrEkleP1N5Vf3FHIt+CHgHvWcrAJ4JPFJVjzfj9wOHtGxzG7AN4DnPec4g1VzxZjsLmO8ZwqbtV3Hvea+acR3TyyWpzaAXrl/cN7wfcCK990q0hkSSVwN7q+qGJC+bb8WqagewA2B8fLzmu7wkafEGComqekv/eJKDgAvnWOx44DVJTqUXLM8APgwclGRdczZxKPDAvGstSVoWC33H9Q+BzbPNUFXnVtWhVbUJOAv4fFW9DrieJ7r02ApcvsA6jLSlvIDsxWhJwzLoNYnPAVNNPvsAvwxcvMBt/hFwYZI/AW4Ezl/geiRJHRv0msT7+4YfB+6rqvsH3UhVfQH4QjN8D3DMoMtKkoZn0FtgvwjcQe8upfXAT7qslBZmpmYpm6okLcZAIZHkTOArwL+l957r3UnsKlySVrlBm5veCby4qvYCJBkD/pYnHoqTJK1Cg97d9KSpgGh8dx7LSpJWqEHPJK5Jci1wQTP+WuDqbqokSRoVs54NJHlukuOr6g+BTwBHNZ8v0zwNrXajctF4VOohaeWZ60ziQ8C5AFV1KXApQJJfaab9605rJ0kaqrmuK2yoqpunFzZlmzqpkSRpZMwVEgfNMm3/payIJGn0zBUSE0n+/fTCJG+k9+IhSdIqNtc1ibcBlyV5HU+EwjiwL/CbXVZsLVnK90hI0lKaNSSq6mHg15K8HHh+U3xVVX2+85pJkoZu0PdJXE+vi29J0hriU9OrWFtTlU1YkgZlSEiSWhkSkqRWhsQaZrOTpLkYEpKkVp2FRJL9knwlydeT3JrkPU355iS7k+xJclGSfbuqg57gWYOkhejyTOLHwAlV9QJgC3BKkmOB9wIfrKrnAt8DzumwDpKkRegsJKrnn5rRJzefAk7giTfa7QRO76oOkqTF6fSaRJJ9ktwE7AWuA+4GHqmqx5tZ7gcOaVl2W5KJJBOTk5NdVlOS1KLTkKiqn1bVFuBQ4BjgefNYdkdVjVfV+NjYWGd1lCS1W5a7m6rqEXrdehwHHJRkqjuQQ4EHlqMOkqT56/LuprEkBzXD+wMnA7fTC4szmtm2Apd3VQctjHdCSZoyUAd/C7QR2JlkH3phdHFVXZnkNuDCJH8C3Aic32EdJEmL0FlIVNU3gKNnKL+H3vUJjYhN26/i3vNeNexqSBpBPnEtSWplSEiSWhkSmpUXsaW1zZCQJLUyJCRJrQwJjRSbt6TRYkhIkloZEpKkVoaERp5NUNLwGBKSpFaGRAdW4l++c9V5Je6TpMUzJCRJrQwJSVIrQ0I/NypNSqNSD0mGhCRpFoaEJKmVIaFf0HVTj01J0spiSEiSWnUWEkmeneT6JLcluTXJW5vyg5Ncl+Su5uf6ruqgua3Uv+y7qPdK/S6kLnV5JvE48J+q6kjgWOA/JjkS2A7sqqrDgV3NuCRpBHUWElX1YFV9rRn+AXA7cAhwGrCzmW0ncHpXdZAkLc6yXJNIsgk4GtgNbKiqB5tJDwEbWpbZlmQiycTk5ORyVFPztNTNMzOtzyYgabg6D4kkTwM+C7ytqh7tn1ZVBdRMy1XVjqoar6rxsbGxrqspSZpBpyGR5Mn0AuLTVXVpU/xwko3N9I3A3i7rIElauC7vbgpwPnB7VX2gb9IVwNZmeCtweVd10Mo1n2Ymn+2QurOuw3UfD/wOcHOSm5qy/wycB1yc5BzgPuDMDusgSVqELu9u+vuqSlUdVVVbms/VVfXdqjqxqg6vqpOq6v92VQctXP9fz23Ds5UtdP1LbTnPAjzj0GrkE9eSpFaGhCSplSGhBVlM08pCmqzmWmZqeL71GkYTkc1SWkkMCUlSK0NCktTKkFCrxd61tNQ2bb9qoPUvZR2WYl02L2klMyQkSa0MCS1K/wXj6X8xD/qX/1zrHrR8oeuT1M6QkCS1MiQkSa0MCc3bami2Wap9WA3fhTQbQ0KS1MqQkCS1MiQ0sEGbVubT7caoPNMw051ZXW97rm3Ots7l6kVXMiQkSa0MCUlSK0NCnRiF5pCV8DDeIA8c2pykYeryHdefSrI3yS19ZQcnuS7JXc3P9V1tX5K0eF2eSfw5cMq0su3Arqo6HNjVjGuNWClnFLO9m2KQMv/y12rS5TuuvwRMf3/1acDOZngncHpX25ckLd5yX5PYUFUPNsMPARuWefuSpHkY2oXrqiqg2qYn2ZZkIsnE5OTkMtZMK8lyve9hIc+ILOQ1rdKoWe6QeDjJRoDm5962GatqR1WNV9X42NjYslVQkvSE5Q6JK4CtzfBW4PJl3r4kaR66vAX2AuDLwBFJ7k9yDnAecHKSu4CTmnGtUcNoelmKZxLmU++l7nZkFJ4/0dqyrqsVV9XZLZNO7GqbkqSl5RPXS8i/7EbHchyLQTvgmxqfz/zSqDAkJEmtDAlJUitDQhqCxXQ+ONPFa5ur1BVDQpLUypCQJLUyJLQmLHVzTJfNO/13Qs23OWkxr0SVZmJISJJaGRJLzL/UNCrmcxYx399bf8/XDkNCktTKkJAktTIkloCn3hoFs3X+t9BXuEqGhCSplSEhSWplSCwRT8+10iy0C5C2aQv9N+C/ndFmSEiSWnX20qFRsWn7Vdx73qtW3LqlpbKUT2HP99mL2f59eAaxMngmIUlqNZSQSHJKkjuT7EmyfRh1kCTNbdlDIsk+wP8EXgkcCZyd5Mjlrsd8zHSBzn78NaoWcwF5se+qmGkdg/ybmWmeuV4BO8x/e3NtezX9vzCMM4ljgD1VdU9V/QS4EDhtCPWQJM1hGCFxCPDtvvH7mzJJ0ohJVS3vBpMzgFOq6o3N+O8Av1pVb5423zZgWzN6BHDnAjb3LOA7i6juSua+r03u+9rUtu//sqrGFrPiYdwC+wDw7L7xQ5uyX1BVO4Adi9lQkomqGl/MOlYq9919X2vc9272fRjNTV8FDk+yOcm+wFnAFUOohyRpDst+JlFVjyd5M3AtsA/wqaq6dbnrIUma21CeuK6qq4Grl2FTi2quWuHc97XJfV+bOtv3Zb9wLUlaOeyWQ5LUalWGxGrs9iPJs5Ncn+S2JLcmeWtTfnCS65Lc1fxc35QnyUea7+AbSV7Yt66tzfx3Jdk6rH2aryT7JLkxyZXN+OYku5t9vKi5EYIkT2nG9zTTN/Wt49ym/M4kvzGcPZmfJAcluSTJHUluT3LcWjnuSf6g+X2/JckFSfZbzcc9yaeS7E1yS1/Zkh3rJC9KcnOzzEeSZM5KVdWq+tC7GH43cBiwL/B14Mhh12sJ9msj8MJm+OnAP9Lr1uR9wPamfDvw3mb4VOCvgQDHArub8oOBe5qf65vh9cPevwG/g7cDfwlc2YxfDJzVDH8c+L1m+E3Ax5vhs4CLmuEjm9+HpwCbm9+TfYa9XwPs907gjc3wvsBBa+G403vI9pvA/n3H+w2r+bgDLwVeCNzSV7Zkxxr4SjNvmmVfOWedhv2ldPAlHwdc2zd+LnDusOvVwX5eDpxM7yHDjU3ZRuDOZvgTwNl989/ZTD8b+ERf+S/MN6ofes/T7AJOAK5sfsm/A6ybftzp3Tl3XDO8rpkv038X+ucb1Q9wYPMfZaaVr/rjzhO9MxzcHMcrgd9Y7ccd2DQtJJbkWDfT7ugr/4X52j6rsblp1Xf70ZxGHw3sBjZU1YPNpIeADc1w2/ewUr+fDwHvAH7WjD8TeKSqHm/G+/fj5/vYTP9+M/9K3PfNwCTwZ01T2yeTHMAaOO5V9QDwfuBbwIP0juMNrI3j3m+pjvUhzfD08lmtxpBY1ZI8Dfgs8LaqerR/WvX+PFh1t6sleTWwt6puGHZdhmAdveaHj1XV0cAP6TU5/NwqPu7r6XX+uRn4JeAA4JShVmrIhnGsV2NIDNTtx0qU5Mn0AuLTVXVpU/xwko3N9I3A3qa87XtYid/P8cBrktxLr9fgE4APAwclmXrWp38/fr6PzfQDge+yMvf9fuD+qtrdjF9CLzTWwnE/CfhmVU1W1WPApfR+F9bCce+3VMf6gWZ4evmsVmNIrMpuP5q7EM4Hbq+qD/RNugKYunthK71rFVPlr2/ugDgW+H5zynot8Iok65u/1F7RlI2sqjq3qg6tqk30jufnq+p1wPXAGc1s0/d96js5o5m/mvKzmrtgNgOH07uQN7Kq6iHg20mOaIpOBG5jDRx3es1MxyZ5avP7P7Xvq/64T7Mkx7qZ9miSY5vv8/V962o37Is0HV34OZXe3T93A+8cdn2WaJ9+nd5p5jeAm5rPqfTaXHcBdwF/CxzczB96L3e6G7gZGO9b178D9jSf3x32vs3ze3gZT9zddBi9f+x7gM8AT2nK92vG9zTTD+tb/p3Nd3InA9zZMQofYAsw0Rz7v6J3x8qaOO7Ae4A7gFuA/03vDqVVe9yBC+hdf3mM3lnkOUt5rIHx5ru8G/go026ImOnjE9eSpFarsblJkrREDAlJUitDQpLUypCQJLUyJCRJrQwJrVhJ/mna+BuSfHSB69qS5NSWaS9L8v2mW4w7k3ypeQp8JCT5QpI1+W5ndW8ob6aTRtAWeveQt70x8e+q6tXQCxTgr5L8qKp2LVcFpWHwTEKrUpKxJJ9N8tXmc3xTfkySLzdnBf+Q5Ijmyfz/Crw2yU1JXjvbuqvqpmb+N8+xrXcn2Znk75Lcl+S3kryv6c//mqabFZL8cbPcLUl2TPXx35whvDfJV5L8Y5KXNOX7J7kwvXdLXAbs39HXKBkSWtH2b/5TvynJ1H/cUz4MfLCqXgz8G+CTTfkdwEuq11neHwP/rap+0gxfVFVbquqiAbb9NeB5c2wL4F/R62vqNcD/Aa6vql8BfgS8qpnno1X14qp6Pr3/8PubstZV1THA24B3NWW/B/xzVf1yU/aiAeorLYjNTVrJflRVW6ZGkryBXpMR9DqHO7LvxVvPaHrQPRDYmeRwet2cPHmB2+5/o1fbtgD+uqoeS3IzvRdiXdOU30zvvQEAL0/yDuCp9N6dcCvwuWbaVEeON/TN/1LgIwBV9Y0k31jgPkhzMiS0Wj0JOLaq/l9/YXNh+/qq+s303svxhQWu/2jg9jm2BfBjgKr6WZLH6ol+cH4GrEuyH/C/6PW78+0k76bXB9GUHzc/f4r/XjUENjdptfob4C1TI83FZuidSUx1j/yGvvl/QO+1sHNKchTwX+h1rjbbtgYxFQjfac4+zpht5saXgN9utvV84Kh5bE+aF0NCq9XvA+PpvSD+NuA/NOXvA/57khv5xb/Mr6fXZNR24folU7fA0guH3++7s6ltW3OqqkeAP6XXM+e19Lq6n8vHgKcluZ3edZi1+DImLRN7gZUktfJMQpLUypCQJLUyJCRJrQwJSVIrQ0KS1MqQkCS1MiQkSa0MCUlSq/8PRDiiJ3BSVGgAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "count    5946.000000\n",
            "mean       13.007976\n",
            "std         6.339653\n",
            "min        -5.020000\n",
            "25%         8.600000\n",
            "50%        13.198389\n",
            "75%        17.370000\n",
            "max        31.371857\n",
            "Name: Ambient Temperature, dtype: float64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "count    5946.000000\n",
            "mean      105.020707\n",
            "std       184.026273\n",
            "min        -0.260000\n",
            "25%         0.000000\n",
            "50%         1.812500\n",
            "75%       131.907500\n",
            "max       787.470000\n",
            "Name: Solar Irradiance, dtype: float64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "duPE-KIhd9Cj"
      },
      "source": [
        "# Building datasets for training\n",
        "\n",
        "This section covers the setup needed for creating datasets than can be fed to Pytorch NNs. We also do some preprocessing shift/scale our inputs."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SCBebg5ZoDvN"
      },
      "source": [
        "def DfTrainTestSplitter(df, test_proportion=0.2, as_numpy=True):\n",
        "  \"\"\"Split dataframe into train and test data.\n",
        "  \n",
        "  In order to have the test distribution as close as possible to th\n",
        "  train distribution, we split the initial datafram into 4 chunks,\n",
        "  one corresponding from each season.\n",
        "\n",
        "  We build the test set by concatenating random samples from each season.\n",
        "  \"\"\"\n",
        "  cnt_test = int(test_proportion * len(df))\n",
        "  season_cnt = 4\n",
        "  df_train, df_test = None, None\n",
        "\n",
        "  for i in range(season_cnt):\n",
        "    left_idx = i * int(len(df) / season_cnt)\n",
        "    right_idx = min(int(left_idx + cnt_test // 4.0), len(df) - 2)\n",
        "    \n",
        "    if df_test is None:\n",
        "      df_test = df[left_idx: right_idx + 1]\n",
        "    else:\n",
        "      df_test = pd.concat([df_test, df[left_idx: right_idx + 1]])\n",
        "\n",
        "  df_train = pd.concat([df, df_test]).drop_duplicates(keep=False)\n",
        "  if as_numpy:\n",
        "   return df_train.to_numpy()[:, 1:],  df_test.to_numpy()[:, 1:]\n",
        "  return df_train, df_test"
      ],
      "execution_count": 50,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eT0h10RvgG2k"
      },
      "source": [
        "train_data, test_data = DfTrainTestSplitter(df, test_proportion=0.1)\n",
        "X_train, y_train = (train_data[:, [1, 2]].astype(float),\n",
        "                    train_data[:, 0].astype(float))\n",
        "\n",
        "X_test, y_test = (test_data[:, [1, 2]].astype(float),\n",
        "                  test_data[:, 0].astype(float))"
      ],
      "execution_count": 51,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qAhcM4vPgZBE",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "34784d85-bc14-4780-d9e2-326851dc2430"
      },
      "source": [
        "X_train.shape, X_test.shape"
      ],
      "execution_count": 52,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((5350, 2), (596, 2))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 52
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0WoIeHxbpzcX"
      },
      "source": [
        "scaler = MinMaxScaler().fit(X_train)"
      ],
      "execution_count": 53,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VQZtL6nxglLY"
      },
      "source": [
        "# Use the same scaler in order to ensure that we don't inject knowledge\n",
        "# about the test distribution in our training biases.\n",
        "X_train = scaler.transform(X_train)\n",
        "X_test = scaler.transform(X_test)"
      ],
      "execution_count": 54,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lIUpN7Vkwj8A",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 294
        },
        "outputId": "e7f4a3a3-cb93-4cf5-8ea1-c848c400c764"
      },
      "source": [
        "fig, (ax1, ax2) = plt.subplots(1, 2)\n",
        "fig.suptitle('Inputs Train/Test distributions')\n",
        "ax1.hist(X_train, bins=4)\n",
        "ax2.hist(X_test, bins=4)\n",
        "plt.show()"
      ],
      "execution_count": 55,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3ldHY_xK2Y9O"
      },
      "source": [
        "# Scale both dataset by `y_train_max`. Same reasoning\n",
        "# as in using the same scaler on inputs. This is done to avoid\n",
        "# exploding gradients.\n",
        "y_train_max = y_train.max()\n",
        "y_train /= y_train_max\n",
        "y_test /= y_train_max"
      ],
      "execution_count": 56,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VEdTaSi9w5iC",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 294
        },
        "outputId": "f955eaa3-9ec9-4ae4-d971-c012958ca8d4"
      },
      "source": [
        "fig, (ax1, ax2) = plt.subplots(1, 2)\n",
        "fig.suptitle('Labels Train/Test distributions')\n",
        "ax1.hist(y_train, bins=100)\n",
        "ax2.hist(y_test, bins=100)\n",
        "plt.show()"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXsAAAEVCAYAAAABwEUhAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAcQUlEQVR4nO3de7gdVZnn8e8PgqBCm8Qc84QQOFyCGpxHcM4gjraiYAPB7uDTyoQZMLSxgwqOF+w2ot1GETt2K4yOSndsaALSSBp1SHeiY4QwaA8BDxgCATEhBEkMSbgEEi8MCe/8UetgsbPv93Pq93me/ezaq1ZVvbt2nfesvWpVbUUEZmY2tu3T6wDMzKzznOzNzArAyd7MrACc7M3MCsDJ3sysAJzszcwKwMnezKwAnOwLQNItkt7X7WXbQdIuSUf0avutkBSSjkrTfy/pr9q03kPTftk3vW7rZyTp+5LmtGt91h+c7EcRSRslndzrOKqR9IcpEe2S9OuU8HblHoc2sr6IODAiNjSw/TdI2lCyzUixjLz+wybeV0v7PiLeHxEXt2M7EfHLtF/2NBtPbnsLJH2rZP2nRcTiVtdt/WVcrwOwsSUifgwcCCBpEHgIGB8Ru0vrShpXrrxFpwNXRsTnc9sJ4LURsb7N2+q6Du0zKwC37McASRMk/Zuk7ZKeTNOHlFQ7UtIdkp6WdKOkibnlT5D0fyXtkHS3pBMrbOcoSf9H0lOSHpN0fYNxLpB0g6RvSXoaOFfS8ZJuS9veIulrkl6UWybfFXKVpK9LWiZpp6TbJR1ZspmZwPIqMewv6UuSfilpa+peeXGaNyntux2SnpD0Y0n7SLoGOBT41/TN4C8rrPsv0nv4laT3lsy7StLnG92OpMG0D+ZK+iVwc64s31gr+/lKOlHSppJYNko6WdKpwEXAf0nbuzvNf75bKMX1aUkPS9om6WpJL0vzRuKYk/bnY5I+ldvO8ZKGU0xbJV1a6XOxznOyHxv2Af4JOIwsWfwW+FpJnfcA7wWmALuBrwJImgosAz4PTAQ+DnxH0kCZ7VwM/BCYABwC/M8mYp0F3ACMB64F9gAfBSYBbwBOAj5YZfnZwGdTDOuBS0ZmSJoCTAZ+VmX5hcDRwLHAUcBU4K/TvAuBTcBAWs9FQETEOcAvgT9O3Sd/W7rSlDg/DrwdmA5U64ppZjtvAV4NnFJhnWU/32oi4gfAF4Dr0/ZeW6bauenxVuAIsm9tpcfWm4BXkn12fy3p1an8K8BXIuIPgCOBJbViss5xsh8DIuLxiPhORPwmInaSJcC3lFS7JiLujYhfA38FnKnsBN/ZwPKIWB4Rz0XECmCYrIVc6lmyfygHR8TvIuInTYR7W0T8r7St30bEnRGxKiJ2R8RG4B/KxJ73vYi4I3VlXEuWtEfMBH4QFe7uJ0nAPOCjEfFE2ldfIPsHMvL+pgCHRcSzEfHjSusq40zgn3L7eEGVus1sZ0FE/DoiflthfqXPt1X/Dbg0IjZExC7gk8Dskm8Vn02f5d3A3cDIP41ngaMkTYqIXRGxqg3xWJOc7McASS+R9A/pq/bTwK3A+JI/9kdy0w8D+5G1pg8D3p26FHZI2kHWUptSZlN/CQi4Q9La0q6KOuXjQNLRqUvj0RT7F1JclTyam/4N6fxAUrULh6wl/RLgztx7/UEqB/g7sm8LP1R2knd+Xe8oczB77+NKmtnOIw3Mz3++rTqYF76Xh8nO9U3OlVX6TOaSfYv6uaSfSnpHG+KxJjnZjw0Xkn2Nfn36yvzmVK5cnWm56UPJWl2PkSWJayJifO7x0ohYWLqRiHg0Iv48Ig4GzgO+MdKf3oDSFuzlwM+B6Sn2i0rirouk/ci+EayoUu0xsi6uY3Lv9WURcSBAROyMiAsj4gjgT4CPSTqpQtyltrD3Pi6rye3U2n6lz/fXZP/gAEgNgHwXXa31/oqsQZBf925ga43liIh1EXEW8Argi8ANkl5aaznrDCf70Wc/SQfkHuOAg8iS2I50Yu4zZZY7W9IMSS8BPgfckIbufQv4Y0mnSNo3rfNE7X2CF0nvzpU/SZYonmvx/RwEPA3skvQq4ANNrudNwJqIeLpShYh4DvgmcJmkV0B2zkLSKWn6HcpOQgt4iux8wsj720rWZ13JErITziP7uNxnQBu2U0mlz/cXwAGSTk//ED8N7J9bbiswKKlSLrgO+KikwyUdyO/7+GuOCJJ0tqSBtN93pOJWjxdrkpP96LOcLLGPPBYA/wN4MVlLbhVZ10Spa4CryL5yHwD8d4CIeITspOlFwHaylv5fUP7Y+E/A7ZJ2AUuBDzcyBr6CjwP/FdhJlogbGuGTczrVu3BGfIKsC2VV6jb6Edm3IshOrP4I2AXcBnwjIlameX8DfDp1/3y8dKUR8X2yz+HmtP6bq8TQ9HaqqPT5PkV2wvsfgc1kLf386Jx/Sc+PS7qrzHqvTOu+lWwY7e+AD9UZ06nA2nS8fAWYXeWcg3WY/EtVNhZIug94V0Tc1+tYzPqRW/Y26ikbl3+1E71ZZW7Zm5kVgFv2ZmYF4GRvZlYATvZmZgXgZG9mVgBO9mZmBeBkb2ZWAE72ZmYF4GRvZlYATvZmZgXgZG9mVgBO9mZmBeBkb2ZWAE72ZmYF4GRvZlYA42pX6bxJkybF4OBgr8OwMerOO+98LCIGatdsPx/b1kmNHNt9kewHBwcZHh7udRg2Rkl6uFfb9rFtndTIse1uHDOzAnCyNzMrACd7M7MCcLI3MysAJ3szswJwsjczKwAnezOzAnCyNzMrACd7M7MCGNPJfnD+MgbnL+t1GGYNa+S49TFu9RjTyd7MzDJO9mZmBeBkb2ZWAE72ZmYF4GRvZlYATvZmZgXgZG9mVgBO9mZmBeBkb2ZWAE72ZmYF4GRvZlYATvZmZgXgZG9mVgBO9mZmBVAz2Us6QNIdku6WtFbSZ1P54ZJul7Re0vWSXpTK90+v16f5g519C2ZmVks9LftngLdFxGuBY4FTJZ0AfBG4LCKOAp4E5qb6c4EnU/llqZ6ZmfVQzWQfmV3p5X7pEcDbgBtS+WLgjDQ9K70mzT9JktoWsZmZNayuPntJ+0paDWwDVgAPAjsiYneqsgmYmqanAo8ApPlPAS8vs855koYlDW/fvr21d2FmZlXVlewjYk9EHAscAhwPvKrVDUfEoogYioihgYGBVldn1jaSpklaKem+dJ7qw6l8gaTNklanx8xex2pWr3GNVI6IHZJWAm8Axksal1rvhwCbU7XNwDRgk6RxwMuAx9sYs1mn7QYujIi7JB0E3ClpRZp3WUR8qYexmTWlntE4A5LGp+kXA28H7gdWAu9K1eYAN6bppek1af7NERHtDNqskyJiS0TclaZ3kh3vU6svZdbf6unGmQKslLQG+CmwIiL+DfgE8DFJ68n65K9I9a8AXp7KPwbMb3/YZt2Rhg4fB9yeii6QtEbSlZImVFjG56Os79TsxomINWQHe2n5BrL++9Ly3wHvbkt0Zj0k6UDgO8BHIuJpSZcDF5ONRrsY+DLw3tLlImIRsAhgaGjI32qtL/gKWrMyJO1HluivjYjvAkTE1jRY4Tngm5Rp7Jj1Kyd7sxLpupArgPsj4tJc+ZRctXcC93Y7NrNmNTQax6wg3gicA9yTri8BuAg4S9KxZN04G4HzehOeWeOc7M1KRMRPgHJXfS/vdixm7eJuHDOzAnDL3myUGpy/rNch2Cjilr2ZWQE42ZuZFYCTvZlZATjZm5kVgJO9mVkBONmbmRWAk72ZWQEUItl7PLKZFV0hkr2ZWdE52ZuZFYCTvZlZATjZm5kVgJO9mVkBONmbmRWAk72ZWQE42ZuZFUDNZC9pmqSVku6TtFbSh1P5AkmbJa1Oj5m5ZT4pab2kBySd0sk3YGZmtdXzS1W7gQsj4i5JBwF3SlqR5l0WEV/KV5Y0A5gNHAMcDPxI0tERsaedgZuZWf1qtuwjYktE3JWmdwL3A1OrLDIL+HZEPBMRDwHrgePbEayZmTWnoT57SYPAccDtqegCSWskXSlpQiqbCjySW2wTZf45SJonaVjS8Pbt2xsO3MzM6ld3spd0IPAd4CMR8TRwOXAkcCywBfhyIxuOiEURMRQRQwMDA40samZmDaor2UvajyzRXxsR3wWIiK0RsScingO+ye+7ajYD03KLH5LKzMysR+oZjSPgCuD+iLg0Vz4lV+2dwL1peikwW9L+kg4HpgN3tC9kMzNrVD2jcd4InAPcI2l1KrsIOEvSsUAAG4HzACJiraQlwH1kI3nO90gcM7PeqpnsI+IngMrMWl5lmUuAS1qIy8zM2shX0JqZFYCTvVmJKleNT5S0QtK69Dyh1rrM+oWTvdneRq4anwGcAJyfrgyfD9wUEdOBm9Jrs1HByd6sRJWrxmcBi1O1xcAZvYnQrHFO9mZVlFw1PjkitqRZjwKTKyzjq8Ot7zjZm1VQ5qrx50VEkA073ouvDrd+5GRvVka5q8aBrSMXE6bnbb2Kz6xRTvZmJSpdNU52dficND0HuLHbsZk1q54raM2KptJV4wuBJZLmAg8DZ/YoPrOGOdmblahy1TjASd2Mxaxd3I1jZlYATvZmfWpw/rJeh2BjiJO9mVkBONmbmRWAk72ZWQE42ZuZFYCTvZlZATjZm5kVgJO9mVkB+Apasz7isfXWKW7Zm5kVgJO9mVkBONmbmRVAzWQvaZqklZLuk7RW0odT+URJKyStS88TUrkkfVXSeklrJL2u02+inNK+T/eFmlmR1dOy3w1cGBEzgBOA8yXNAOYDN0XEdOCm9BrgNGB6eswDLm971GZm1pCayT4itkTEXWl6J3A/MBWYBSxO1RYDZ6TpWcDVkVkFjB/5KTczM+uNhvrsJQ0CxwG3A5MjYkua9SgwOU1PBR7JLbYplZWua56kYUnD27dvbzBss+IZnL+sYnekuymtlrqTvaQDyX6A+SMR8XR+XkQEEI1sOCIWRcRQRAwNDAw0sqiZmTWormQvaT+yRH9tRHw3FW8d6Z5Jz9tS+WZgWm7xQ1KZmZn1SD2jcQRcAdwfEZfmZi0F5qTpOcCNufL3pFE5JwBP5bp7+o6//ppZEdTTsn8jcA7wNkmr02MmsBB4u6R1wMnpNcByYAOwHvgm8MH2h92cSn2eTvhmNtbVvDdORPwEUIXZJ5WpH8D5LcZlZmZt5CtozcwKwMnebBRwV6O1ysnezKwAnOzNzArAyd7MrACc7JNql6KbmY12TvYlnPBN0pWStkm6N1e2QNLmkmtNzEaNQiZ7J3Sr4Srg1DLll0XEsemxvMsxmbWkkMnerJqIuBV4otdxmLWTk71Z/S5Iv7525cgvs5XTztt399O5pH6Jw5pT2GTfT39ENipcDhwJHAtsAb5cqaJv3239qLDJ3qwREbE1IvZExHNkN/g7vtcxmTXCyd6sDiU/rflO4N5Kdc36kZN9Fe7mKSZJ1wG3Aa+UtEnSXOBvJd0jaQ3wVuCjPQ3SrEE1b3FsVjQRcVaZ4iu6HohZG7llb2ZWAE72ZvYCHqk2NjnZm5kVgPvsy3CrxszGGrfszcwKwMnezKwAnOzNzAqgZrJv9N7ekj4pab2kBySd0qnAzcysfvW07K+iznt7S5oBzAaOSct8Q9K+7QrWzMyaUzPZN3hv71nAtyPimYh4CFiPbxhl1lYeLWbNaKXPvty9vacCj+TqbEple2nnPb/NzKy6ZpN93ff2rsT3/DYz656mkn2Ve3tvBqblqh6SyszMrIeaSvZV7u29FJgtaX9JhwPTgTtaC9HMzFpV83YJ6d7eJwKTJG0CPgOcKOlYIICNwHkAEbFW0hLgPmA3cH5E7OlM6GZmVq+ayb7Re3tHxCXAJa0EZWZm7eUraM1GIQ+/tEY52ZuZFYCTfQ1uQZnZWOBkb2ZWAE72ZmYF4GRvZlYATvZmZgXg36A16xOdGgyQX+/Ghac//3rjwtM7sj3rT27Zm5kVgJO9mVkBONmbmRWAk71ZiQq/uzxR0gpJ69LzhGrrMOs3TvZme7uKvX93eT5wU0RMB25Kr81GDSd7sxIVfnd5FrA4TS8GzuhqUGYtcrI3q8/kiNiSph8FJvcyGLNGOdmbNSgiguyHe8qSNE/SsKTh7du3dy2uwfnLenbjPt8wsP852dehl39E1je2jvwcZ3reVqliRCyKiKGIGBoYGOhagGbVONmb1WcpMCdNzwFu7GEsZg1zsjcrkX53+TbglZI2SZoLLATeLmkdcHJ6bTZq+N44ZiUq/O4ywEldDcSsjdyyNzMrACf7BvgkrZmNVjW7cSRdCbwD2BYRr0llE4HrgUFgI3BmRDwpScBXgJnAb4BzI+KuzoRuZpW0o2EyOH/ZqL4Nsm/l/EL1tOyvov5Lx08DpqfHPODy9oRpZmatqJnsG7x0fBZwdWRWAeNHxiabmVnvNNtnX+nS8anAI7l6m1KZmZn1UMsnaGtdOl5Jry4pNzMromaTfaVLxzcD03L1Dklle/El5WZm3dNssq906fhS4D3KnAA8levuKSwP2TSzXqtn6OV1wInAJEmbgM+QXSq+JF1G/jBwZqq+nGzY5XqyoZd/1oGYzcysQTWTfSOXjqf++/NbDcrMOqPct8xq49GbnTeajPbrCerlK2jNzArAyb6LfF98M+sVJ3szswIYk8nerWczsxcak8nezMxeyMnezKwAnOwb5C4i63eNHKP11h1tx32lIab1vI/R9l7r5WTfQx6dY2bd4mTfJfmk7gRvZt3mZN8jTvhm1k1O9mZmBeBk3wT3tZvZaFPzRmjWPP9DMLN+4WRv1kO9bBC0a9sj6+nEXTC7dUfKsXIHz2rcjWNmVgBO9mZmBeBk34JqX4Nb/Yrsk8Bm1k7us29REfr67PckbQR2AnuA3REx1NuIzOrjZG/WuLdGxGO9DsKsEe7GMTMrACd7s8YE8ENJd0qa1+tgzOrlZN9mzZxYrVXfJ2v7ypsi4nXAacD5kt5cWkHSPEnDkoa3b9/e/Qh7rNljtZ6/g0p1R/5G6r2FcaW6rdwaud852Zs1ICI2p+dtwPeA48vUWRQRQxExNDAw0O0QzcpqKdlL2ijpHkmrJQ2nsomSVkhal54ntCfU/tap//xjoUUxVkh6qaSDRqaBPwLu7W1UZvVpx2ic0pEJ84GbImKhpPnp9SfasJ0xLT+Es5Gvsx7y2VWTge9Jguxv558j4ge9DcmsPp0YejkLODFNLwZuwcnexoCI2AC8ttdxmDWj1T77ciMTJkfEljT9KFlraC9FP4lVia+8NbNOaLVl/6aI2CzpFcAKST/Pz4yIkBTlFoyIRcAigKGhobJ1RhsnWTPrVy217CuMTNgqaQpAet7WapBWH/+zsVa0OmyxUt1Wt9nPRtM36aZb9mk0wj4RsTM3MuFzwFJgDrAwPd/YjkCtvNFyoJlZb7XSjVN2ZIKknwJLJM0FHgbObD1MG+HkbmbNaDrZVxqZEBGPAye1EpSV10qi79Yv/phZf/IVtGZmBeBkb2ZWAE72ZmYF4GQ/RvlErpnlOdmPcaNpHLCNHZWOuWaOxUZuR9wuvdhmpznZj2GV7vltZsXjZG9mVgBO9mZmBeBkb2ZWAE72ZmYF4GRvZlYATvYF5BE5/aEfP4dOxFR6W+ROjxJrZF3N3IK5XLzVhprm33cvP3MnezOzAnCyNzMrgDGV7Hv9NWk0ye8n7zOzsW9MJXszMyuv1R8ct1HMLXqz4nDL3sysANyyN8A/W2jd1exdMesd8thonXbJb6vS31NpPBsXnl41xnb9Xbplb8+rdoLbJ7/NRjcne6vJI3fMRj9341hZzVyF6G4gs/7VsZa9pFMlPSBpvaT5ndrOCLc4u6uo3TrdPq7N2qUjyV7SvsDXgdOAGcBZkmZ0YlvWfs3cL6QIfFzbaNaplv3xwPqI2BAR/w/4NjCrExsqaguzHxXgc+jacW3Wbp3qs58KPJJ7vQl4fTMrqtYfXIDk0vcaGQpXb59+uc+81nHQpfMFbTuuzbpNEdH+lUrvAk6NiPel1+cAr4+IC3J15gHz0stXAg9UWN0k4LG2B9m4fokDHEs51eI4LCIGWt1APcd1Kq/32O6Efvk8KnF8rSmNr+5ju1Mt+83AtNzrQ1LZ8yJiEbCo1ookDUfEUHvDa1y/xAGOpYdx1Dyuof5juxP65fOoxPG1ppX4OtVn/1NguqTDJb0ImA0s7dC2zLrFx7WNWh1p2UfEbkkXAP8b2Be4MiLWdmJbZt3i49pGs45dVBURy4HlbVhVT74Ol9EvcYBjKacrcbTxuO6Ufvk8KnF8rWk6vo6coDUzs/7ie+OYmRVAz5J9rcvOJe0v6fo0/3ZJg7l5n0zlD0g6pQuxfEzSfZLWSLpJ0mG5eXskrU6Plk/W1RHLuZK257b5vty8OZLWpcecDsdxWS6GX0jakZvX7n1ypaRtku6tMF+SvppiXSPpdbl5bdsn/aSVY7Yf4svV+1NJIamrI2DqiU/SmWkfrpX0z/0Un6RDJa2U9LP0Gc+sudKI6PqD7OTWg8ARwIuAu4EZJXU+CPx9mp4NXJ+mZ6T6+wOHp/Xs2+FY3gq8JE1/YCSW9HpXl/fLucDXyiw7EdiQniek6QmdiqOk/ofITla2fZ+k9b0ZeB1wb4X5M4HvAwJOAG5v9z7pp0erx2w/xJfqHQTcCqwChvopPmA68LOR4wV4RZ/Ftwj4QJqeAWystd5etezruex8FrA4Td8AnCRJqfzbEfFMRDwErE/r61gsEbEyIn6TXq4iG1/dCa1cjn8KsCIinoiIJ4EVwKldiuMs4Lomt1VTRNwKPFGlyizg6sisAsZLmkJ790k/6adjtqn4kouBLwK/62JsUF98fw58PR03RMS2PosvgD9I0y8DflVrpb1K9uUuO59aqU5E7AaeAl5e57LtjiVvLlkrcsQBkoYlrZJ0RgtxNBLLn6avbjdIGrnIp537pe51pe6Bw4Gbc8Xt3Cf1qBRvu4+VftHqMdtpNeNLXW3TIqIX9zypZ/8dDRwt6d/TcdzNRkI98S0Azpa0iWx02IdqrdT3s2+ApLOBIeAtueLDImKzpCOAmyXdExEPdjCMfwWui4hnJJ1H9u3nbR3cXi2zgRsiYk+urNv7xCqocMz2lKR9gEvJuiT71TiyrpwTyb4V3SrpP0TEjqpLdc9ZwFUR8WVJbwCukfSaiHiu0gK9atnXc9n583UkjSP7qvJ4ncu2OxYknQx8CviTiHhmpDwiNqfnDcAtwHGdjCUiHs9t/x+B/9jI+2hXHDmzKenCafM+qUeleNt9rPSLlo7ZLqgV30HAa4BbJG0kO8+ytIsnaevZf5uApRHxbOou/gVZ8u+X+OYCSwAi4jbgALL75lTWrZMOJScXxpGdLDuc35+AOKakzvm88ATtkjR9DC88QbuB1k7Q1hPLcWQnTKaXlE8A9k/Tk4B1VDmR2aZYpuSm3wmsStMTgYdSTBPS9MROxZHqvQrYSLpeoxP7JLfeQSqfoD2dF56gvaPd+6SfHq0cs/0SX0n9W+juCdp69t+pwOI0PYmsW+XlfRTf94Fz0/SryfrsVXW93T4QcsHOJPtv+SDwqVT2ObJWCGT/qf6F7ATsHcARuWU/lZZ7ADitC7H8CNgKrE6Ppan8PwP3pA/jHmBuF2L5G2Bt2uZK4FW5Zd+b9td64M86GUd6vQBYWLJcJ/bJdcAW4FmyFtdc4P3A+9N8kf2oyINpm0O5Zdu2T/rp0ewx2y/xldS9hS4m+zr3n8i6mu5Lx9TsPotvBvDv6e9sNfBHtdbpK2jNzArAV9CamRWAk72ZWQE42ZuZFYCTvZlZATjZm5kVgJO9mVkBONmbmRWAk72ZWQH8f0x3BiaJYrZkAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yOtg4Bb3lVjN"
      },
      "source": [
        "# Batching Logic \n",
        "\n",
        "This section defines the necessary utilities for producing batches of variable length sequences."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tWXCuJmTXgYp"
      },
      "source": [
        "def GetInitialBatchIndex(X, input_len, prediction_window_len=48):\n",
        "  \"\"\" \n",
        "      Returns a random index from which to sample `input_len` inputs.\n",
        "      This was written separately so that we can test that it produces an\n",
        "      even distribution among potential starting indices.\n",
        "  \"\"\"\n",
        "  left_idx = np.random.randint(\n",
        "      low=0,\n",
        "      high=max(len(X) - prediction_window_len * 2 - input_len - 1, 1))\n",
        "  right_idx = left_idx + input_len - 1\n",
        "  return left_idx"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BaOJOfx0bhHs",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "outputId": "0f89b0ff-c11b-4232-8cf2-f936a21cd339"
      },
      "source": [
        "# Confirm that starting indices for bathches are evenly distributed.\n",
        "no_trials = 100000\n",
        "start_indices_samples = []\n",
        "\n",
        "for _ in range(no_trials):\n",
        "  start_indices_samples.append(GetInitialBatchIndex(X_train, 15))\n",
        "\n",
        "plt.hist(start_indices_samples)\n",
        "plt.show()"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jCUO69OREbVh"
      },
      "source": [
        "def GetSample(y, input_len, prediction_window_len=48, use_noise_reg=False):\n",
        "  \"\"\"Return a single input sequence and its respective targets.\n",
        "  \n",
        "  Args:\n",
        "    y: np.array from which to extract the regression targets corresponding\n",
        "       to the selected inputs.\n",
        "    prediction_window_len: length of the prediction window, which describes\n",
        "      how many predictions we have to do for a given input sequence.\n",
        "    use_noise_reg: bool indication whether to inject the labels with noise.\n",
        "\n",
        "  Return:\n",
        "    y_input_tensor: tensorized inputs.\n",
        "    y_target_tensor: tensorized targets.\n",
        "\n",
        "  Note, since we can use forecasts, we also include future data in our\n",
        "  sequences.\n",
        "  \"\"\"\n",
        "  left_idx = GetInitialBatchIndex(y, input_len, prediction_window_len)\n",
        "  right_idx = left_idx + input_len - 1\n",
        "\n",
        "  # Add `prediction_window_len` to our right index, since we can use future\n",
        "  # values as well, given that we always expect to have forecasts availabel.\n",
        "  y_input_tensor = torch.Tensor(\n",
        "      y[left_idx : right_idx + 1])\n",
        "  y_target_tensor = torch.Tensor(\n",
        "      y[right_idx + 1 : prediction_window_len + right_idx + 1])\n",
        "\n",
        "  return (y_input_tensor,\n",
        "          InjectGaussianNoise(y_target_tensor) \n",
        "          if use_noise_reg else y_target_tensor)"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2swVcYlWkhRW"
      },
      "source": [
        "def GetBatch(y, batch_size, input_len, device, prediction_window_len=48,\n",
        "             use_noise_reg=False):\n",
        "  y_inputs, y_targets = [], []\n",
        "  for _ in range(batch_size):\n",
        "    y_input_tensor, y_target_tensor = GetSample(y, \n",
        "                                   input_len,\n",
        "                                   prediction_window_len,\n",
        "                                   use_noise_reg)\n",
        "    y_inputs.append(y_input_tensor)\n",
        "    y_targets.append(y_target_tensor)\n",
        "  return (torch.stack(y_inputs).to(device),\n",
        "          torch.stack(y_targets).to(device))"
      ],
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0nLPp6uKT9eS",
        "outputId": "22693a31-f1c6-40d2-cc27-0a4aef0561c4"
      },
      "source": [
        "y1, y2 = GetBatch(y_train, 12, 5, 'cpu')\n",
        "y1.shape, y2.shape"
      ],
      "execution_count": 108,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(torch.Size([12, 5]), torch.Size([12, 48]))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 108
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ic5KL4tghLas"
      },
      "source": [
        "# Neural Models Definitions\n",
        "\n",
        "This section covers defining the classes needed for our regression models:\n",
        "\n",
        "We shall employ a seq2seq approach, where we unroll a Luong attention based decoder step by step, instead of predicting everything at a time.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ACG1HJkTms5n"
      },
      "source": [
        "def InjectGaussianNoise(X, isotropic_std=0.1):\n",
        "  return torch.normal(mean=X, std=isotropic_std)"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G0yn9_oW2d4y"
      },
      "source": [
        "class EncoderRNN(torch.nn.Module):\n",
        "    def __init__(self, input_size, hidden_size, n_layers=1):\n",
        "        super(EncoderRNN, self).__init__()\n",
        "        \n",
        "        self.input_size = input_size\n",
        "        self.hidden_size = hidden_size\n",
        "        self.n_layers = n_layers\n",
        "        \n",
        "        self.embedding = torch.nn.Linear(input_size, hidden_size)\n",
        "        self.gru = torch.nn.GRU(hidden_size, hidden_size, n_layers, True, \n",
        "                                batch_first=True)\n",
        "        \n",
        "    def forward(self, inputs, hidden=None):\n",
        "        embedded = self.embedding(inputs)\n",
        "        output, hidden = self.gru(embedded)\n",
        "        return output, hidden"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qxN44W3ftd1W"
      },
      "source": [
        "# Training/Testing Setup\n",
        "\n",
        "This section covers the training procedure, records training metrics, plots train/test loss curves, and shows how to do inference with a trained model."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wlK3_EmBs2Jm"
      },
      "source": [
        "def _GetDevice():\n",
        "  return torch.device('cuda') if torch.cuda.is_available() else 'cpu'"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MbW1w3_Nr0zN"
      },
      "source": [
        "def train(device=_GetDevice(),\n",
        "          epochs=30,\n",
        "          batch_size=32,\n",
        "          prediction_window_len=24,\n",
        "          input_size=1,\n",
        "          hidden_size=24):\n",
        "\n",
        "  # Encoder, optimizer instantiations.\n",
        "  encoder = EncoderRNN(input_size=input_size,\n",
        "                       hidden_size=hidden_size).to(device)\n",
        "\n",
        "  enc_optimizer = torch.optim.Adam(encoder.parameters(), lr=0.001)\n",
        "  # Regression loss.\n",
        "  mse_loss = torch.nn.MSELoss() \n",
        "\n",
        "  iterations_per_epoch = len(df) // batch_size\n",
        "  train_losses, val_losses = [], []\n",
        "  print(epochs, iterations_per_epoch)\n",
        "\n",
        "  for epoch in range(epochs):\n",
        "    for iter in range(iterations_per_epoch):\n",
        "      enc_optimizer.zero_grad()\n",
        "\n",
        "      # Sample varying sequence lengths to use for prediction.\n",
        "      input_len = np.random.randint(low=2, high=12)\n",
        "      inputs, targets = GetBatch(y_train,\n",
        "                                batch_size,\n",
        "                                input_len,\n",
        "                                device,\n",
        "                                prediction_window_len)\n",
        "      \n",
        "      predictions = torch.mean(encoder(inputs[..., None])[0], dim=1)\n",
        "\n",
        "      # Compute and record MSE loss.\n",
        "      train_loss = mse_loss(predictions, targets)\n",
        "      train_losses.append(train_loss)\n",
        "\n",
        "      # Backprop and update parameters.\n",
        "      train_loss.backward()\n",
        "      enc_optimizer.step()\n",
        "      \n",
        "      with torch.no_grad():\n",
        "        input, targets = GetBatch(y_test, batch_size,\n",
        "                                  input_len, device, prediction_window_len)\n",
        "        predictions = torch.mean(encoder(inputs[..., None])[0], dim=1)\n",
        "\n",
        "        val_loss = mse_loss(predictions, targets)\n",
        "        val_losses.append(val_loss)\n",
        "\n",
        "      if epoch % 10 == 0 and iter == iterations_per_epoch % 5 == 0:\n",
        "        mae_loss = torch.nn.L1Loss()(predictions, targets) * y_train_max\n",
        "\n",
        "        print('Epoch:', epoch)\n",
        "        print('Train loss:', train_loss.item())\n",
        "        print('Val loss:', val_loss.item())\n",
        "        print('Val loss:', mae_loss.item())\n",
        "\n",
        "        print('\\n')\n",
        "  return encoder, train_losses, val_losses"
      ],
      "execution_count": 62,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "n02BWBTGtMOl",
        "outputId": "0fd7b685-a647-4048-a49e-c34ec8ee8176"
      },
      "source": [
        "# Customize this loop however you prefer.\n",
        "encoder, train_losses, val_losses = train()"
      ],
      "execution_count": 63,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "30 185\n",
            "Epoch: 0\n",
            "Train loss: 0.16321513056755066\n",
            "Val loss: 0.1413751095533371\n",
            "Val loss: 2864.18896484375\n",
            "\n",
            "\n",
            "Epoch: 10\n",
            "Train loss: 0.004984874278306961\n",
            "Val loss: 0.04585590958595276\n",
            "Val loss: 1597.486572265625\n",
            "\n",
            "\n",
            "Epoch: 20\n",
            "Train loss: 0.006965005304664373\n",
            "Val loss: 0.04365427419543266\n",
            "Val loss: 1647.28759765625\n",
            "\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fhQpM7VskUWu",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "cc662fea-b8f8-4f3b-a9a6-5b5982cee763"
      },
      "source": [
        "plt.plot(train_losses)\n",
        "plt.plot(val_losses)\n",
        "\n",
        "plt.title('Model Losses')\n",
        "plt.legend(['Train', 'Val'], loc='upper left')\n",
        "\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('iteration')\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cTWDU4kt5n4L",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "84845a6c-39b3-4aa9-96de-6ca2bae8d3f2"
      },
      "source": [
        "with torch.no_grad():\n",
        "  input_len = np.random.randint(low=2, high=12)\n",
        "  print('Input length:', input_len)\n",
        "  inputs, targets = GetBatch(y_test,\n",
        "                             batch_size=1,\n",
        "                             input_len=input_len,\n",
        "                             device=_GetDevice(),\n",
        "                             prediction_window_len=24)\n",
        "        \n",
        "  predictions = torch.mean(encoder(inputs[..., None])[0], dim=1)\n",
        "  plt.plot(predictions.cpu().numpy()[0], label='Predictions')\n",
        "  plt.plot(targets.cpu().numpy()[0], label='Targets')\n",
        "  plt.legend()\n",
        "  plt.show()"
      ],
      "execution_count": 49,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Input length: 3\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}