LENCE_and_R3_Curve.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "LENCE and R3 Curve.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/kumudlakara/Shifts_evaluation_metrics_regression/blob/main/LENCE_and_R3_Curve.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c4dq6YsttUy2"
      },
      "source": [
        "```\n",
        "Credit: Initial part of the code including data preprocessing and basic uncertainty calculation is based on:\n",
        "```\n",
        "https://github.com/yandex-research/shifts/tree/main/weather\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q5UyHGPWwTKH"
      },
      "source": [
        "import pandas as pd\n",
        "import seaborn as sns\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import sys"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DBJDp_P9wOsL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "0f0a0073-f762-420d-8272-2480f2b97bc8"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ojKOkLGtweTS"
      },
      "source": [
        "# **Clone repository for base code**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zYv7HrOZe6YZ",
        "outputId": "07f322fd-02b8-4fb4-e527-e11ab850de0c"
      },
      "source": [
        "! git clone https://github.com/yandex-research/shifts"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Cloning into 'shifts'...\n",
            "remote: Enumerating objects: 2401, done.\u001b[K\n",
            "remote: Counting objects: 100% (649/649), done.\u001b[K\n",
            "remote: Compressing objects: 100% (224/224), done.\u001b[K\n",
            "remote: Total 2401 (delta 435), reused 628 (delta 425), pack-reused 1752\u001b[K\n",
            "Receiving objects: 100% (2401/2401), 10.52 MiB | 29.36 MiB/s, done.\n",
            "Resolving deltas: 100% (1211/1211), done.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6dPoV9RSfVf5"
      },
      "source": [
        "import sys\n",
        "sys.path.append('/content/drive/MyDrive/shifts-main/weather/')"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pa0b5hPtgo7U",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 308
        },
        "outputId": "d37fe505-7ff2-4fde-a919-428b4fe83de4"
      },
      "source": [
        "import pandas as pd\n",
        "\n",
        "# Load each data file as a pandas data frame\n",
        "df_train = pd.read_csv('/content/drive/MyDrive/dataset/data/train.csv')\n",
        "df_train.head()"
      ],
      "execution_count": 5,
      "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>fact_time</th>\n",
              "      <th>fact_latitude</th>\n",
              "      <th>fact_longitude</th>\n",
              "      <th>fact_temperature</th>\n",
              "      <th>fact_cwsm_class</th>\n",
              "      <th>climate</th>\n",
              "      <th>topography_bathymetry</th>\n",
              "      <th>sun_elevation</th>\n",
              "      <th>climate_temperature</th>\n",
              "      <th>climate_pressure</th>\n",
              "      <th>cmc_0_0_0_1000</th>\n",
              "      <th>cmc_0_0_0_2_grad</th>\n",
              "      <th>cmc_0_0_0_2_interpolated</th>\n",
              "      <th>cmc_0_0_0_2_next</th>\n",
              "      <th>cmc_0_0_0_2</th>\n",
              "      <th>cmc_0_0_0_500</th>\n",
              "      <th>cmc_0_0_0_700</th>\n",
              "      <th>cmc_0_0_0_850</th>\n",
              "      <th>cmc_0_0_0_925</th>\n",
              "      <th>cmc_0_0_6_2</th>\n",
              "      <th>cmc_0_0_7_1000</th>\n",
              "      <th>cmc_0_0_7_2</th>\n",
              "      <th>cmc_0_0_7_500</th>\n",
              "      <th>cmc_0_0_7_700</th>\n",
              "      <th>cmc_0_0_7_850</th>\n",
              "      <th>cmc_0_0_7_925</th>\n",
              "      <th>cmc_0_1_0_0</th>\n",
              "      <th>cmc_0_1_11_0</th>\n",
              "      <th>cmc_0_1_65_0</th>\n",
              "      <th>cmc_0_1_66_0</th>\n",
              "      <th>cmc_0_1_67_0</th>\n",
              "      <th>cmc_0_1_68_0</th>\n",
              "      <th>cmc_0_1_7_0</th>\n",
              "      <th>cmc_0_2_2_10</th>\n",
              "      <th>cmc_0_2_2_1000</th>\n",
              "      <th>cmc_0_2_2_500</th>\n",
              "      <th>cmc_0_2_2_700</th>\n",
              "      <th>cmc_0_2_2_850</th>\n",
              "      <th>cmc_0_2_2_925</th>\n",
              "      <th>cmc_0_2_3_10</th>\n",
              "      <th>...</th>\n",
              "      <th>gfs_temperature_90000</th>\n",
              "      <th>gfs_temperature_92500</th>\n",
              "      <th>gfs_temperature_95000</th>\n",
              "      <th>gfs_temperature_97500</th>\n",
              "      <th>gfs_temperature_sea</th>\n",
              "      <th>gfs_temperature_sea_grad</th>\n",
              "      <th>gfs_temperature_sea_interpolated</th>\n",
              "      <th>gfs_temperature_sea_next</th>\n",
              "      <th>gfs_timedelta_s</th>\n",
              "      <th>gfs_total_clouds_cover_high</th>\n",
              "      <th>gfs_total_clouds_cover_low</th>\n",
              "      <th>gfs_total_clouds_cover_middle</th>\n",
              "      <th>gfs_u_wind</th>\n",
              "      <th>gfs_v_wind</th>\n",
              "      <th>gfs_wind_speed</th>\n",
              "      <th>wrf_available</th>\n",
              "      <th>wrf_t2</th>\n",
              "      <th>wrf_t2_next</th>\n",
              "      <th>wrf_psfc</th>\n",
              "      <th>wrf_rh2</th>\n",
              "      <th>wrf_wind_u</th>\n",
              "      <th>wrf_wind_v</th>\n",
              "      <th>wrf_rain</th>\n",
              "      <th>wrf_snow</th>\n",
              "      <th>wrf_graupel</th>\n",
              "      <th>wrf_hail</th>\n",
              "      <th>wrf_t2_interpolated</th>\n",
              "      <th>wrf_t2_grad</th>\n",
              "      <th>cmc_0_1_65_0_grad</th>\n",
              "      <th>cmc_0_1_65_0_next</th>\n",
              "      <th>cmc_0_1_66_0_grad</th>\n",
              "      <th>cmc_0_1_66_0_next</th>\n",
              "      <th>cmc_0_1_67_0_grad</th>\n",
              "      <th>cmc_0_1_67_0_next</th>\n",
              "      <th>cmc_0_1_68_0_grad</th>\n",
              "      <th>cmc_0_1_68_0_next</th>\n",
              "      <th>gfs_2m_dewpoint_grad</th>\n",
              "      <th>gfs_2m_dewpoint_next</th>\n",
              "      <th>gfs_total_clouds_cover_low_grad</th>\n",
              "      <th>gfs_total_clouds_cover_low_next</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1.543321e+09</td>\n",
              "      <td>26.968800</td>\n",
              "      <td>-99.248901</td>\n",
              "      <td>2.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>dry</td>\n",
              "      <td>127.0</td>\n",
              "      <td>-17.526443</td>\n",
              "      <td>14.613571</td>\n",
              "      <td>754.263405</td>\n",
              "      <td>282.702881</td>\n",
              "      <td>6.684790</td>\n",
              "      <td>276.300439</td>\n",
              "      <td>282.985229</td>\n",
              "      <td>276.300439</td>\n",
              "      <td>262.333789</td>\n",
              "      <td>276.962134</td>\n",
              "      <td>280.083057</td>\n",
              "      <td>281.141064</td>\n",
              "      <td>272.500946</td>\n",
              "      <td>14.250</td>\n",
              "      <td>3.750</td>\n",
              "      <td>29.905515</td>\n",
              "      <td>15.500</td>\n",
              "      <td>29.750</td>\n",
              "      <td>15.750</td>\n",
              "      <td>0.0034</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>-1.327548</td>\n",
              "      <td>-2.914091</td>\n",
              "      <td>18.357608</td>\n",
              "      <td>8.480661</td>\n",
              "      <td>-1.510544</td>\n",
              "      <td>-0.465033</td>\n",
              "      <td>-0.126199</td>\n",
              "      <td>...</td>\n",
              "      <td>7.345331</td>\n",
              "      <td>7.926080</td>\n",
              "      <td>8.841516</td>\n",
              "      <td>9.350854</td>\n",
              "      <td>9.077112</td>\n",
              "      <td>1.966156</td>\n",
              "      <td>9.077112</td>\n",
              "      <td>11.043268</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>-2.328149</td>\n",
              "      <td>0.509014</td>\n",
              "      <td>2.383144</td>\n",
              "      <td>1.0</td>\n",
              "      <td>280.888000</td>\n",
              "      <td>280.420593</td>\n",
              "      <td>100927.414062</td>\n",
              "      <td>0.623658</td>\n",
              "      <td>-2.533346</td>\n",
              "      <td>-3.091468</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>280.888000</td>\n",
              "      <td>-0.467407</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>-2.600006</td>\n",
              "      <td>-2.750006</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1.538776e+09</td>\n",
              "      <td>29.374201</td>\n",
              "      <td>-100.927002</td>\n",
              "      <td>31.0</td>\n",
              "      <td>20.0</td>\n",
              "      <td>mild temperate</td>\n",
              "      <td>297.0</td>\n",
              "      <td>41.531032</td>\n",
              "      <td>26.992143</td>\n",
              "      <td>733.117168</td>\n",
              "      <td>305.860669</td>\n",
              "      <td>-1.361548</td>\n",
              "      <td>303.629541</td>\n",
              "      <td>302.267993</td>\n",
              "      <td>303.629541</td>\n",
              "      <td>267.178882</td>\n",
              "      <td>286.211719</td>\n",
              "      <td>291.642969</td>\n",
              "      <td>298.615479</td>\n",
              "      <td>293.701019</td>\n",
              "      <td>11.500</td>\n",
              "      <td>9.875</td>\n",
              "      <td>22.468424</td>\n",
              "      <td>26.500</td>\n",
              "      <td>1.375</td>\n",
              "      <td>6.250</td>\n",
              "      <td>0.0172</td>\n",
              "      <td>0.0</td>\n",
              "      <td>5.67340</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>-3.568011</td>\n",
              "      <td>-3.632005</td>\n",
              "      <td>4.370947</td>\n",
              "      <td>0.653320</td>\n",
              "      <td>-3.227249</td>\n",
              "      <td>-4.499048</td>\n",
              "      <td>7.287793</td>\n",
              "      <td>...</td>\n",
              "      <td>24.699792</td>\n",
              "      <td>27.050012</td>\n",
              "      <td>29.526605</td>\n",
              "      <td>31.950006</td>\n",
              "      <td>32.831812</td>\n",
              "      <td>-2.257446</td>\n",
              "      <td>32.831812</td>\n",
              "      <td>30.574365</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>23.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>-3.851841</td>\n",
              "      <td>7.031523</td>\n",
              "      <td>8.017418</td>\n",
              "      <td>1.0</td>\n",
              "      <td>302.893005</td>\n",
              "      <td>303.356781</td>\n",
              "      <td>97645.507812</td>\n",
              "      <td>0.651412</td>\n",
              "      <td>-3.482976</td>\n",
              "      <td>5.238701</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>302.893005</td>\n",
              "      <td>0.463776</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>5.673400</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>-0.600006</td>\n",
              "      <td>17.950006</td>\n",
              "      <td>-12.0</td>\n",
              "      <td>11.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1.552115e+09</td>\n",
              "      <td>22.149599</td>\n",
              "      <td>113.592003</td>\n",
              "      <td>17.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>mild temperate</td>\n",
              "      <td>-1.0</td>\n",
              "      <td>43.916531</td>\n",
              "      <td>18.842143</td>\n",
              "      <td>761.571076</td>\n",
              "      <td>294.469702</td>\n",
              "      <td>0.282837</td>\n",
              "      <td>295.153361</td>\n",
              "      <td>295.341919</td>\n",
              "      <td>295.059082</td>\n",
              "      <td>268.584375</td>\n",
              "      <td>280.319653</td>\n",
              "      <td>288.922583</td>\n",
              "      <td>292.259375</td>\n",
              "      <td>293.914505</td>\n",
              "      <td>0.750</td>\n",
              "      <td>1.000</td>\n",
              "      <td>25.403430</td>\n",
              "      <td>1.625</td>\n",
              "      <td>2.375</td>\n",
              "      <td>0.125</td>\n",
              "      <td>0.0161</td>\n",
              "      <td>0.0</td>\n",
              "      <td>82.66500</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.674461</td>\n",
              "      <td>0.812148</td>\n",
              "      <td>28.046518</td>\n",
              "      <td>20.135425</td>\n",
              "      <td>11.603281</td>\n",
              "      <td>5.047708</td>\n",
              "      <td>1.483844</td>\n",
              "      <td>...</td>\n",
              "      <td>17.325555</td>\n",
              "      <td>18.692010</td>\n",
              "      <td>20.252496</td>\n",
              "      <td>21.868280</td>\n",
              "      <td>23.101434</td>\n",
              "      <td>0.287476</td>\n",
              "      <td>23.197260</td>\n",
              "      <td>23.388910</td>\n",
              "      <td>3600.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>7.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.512065</td>\n",
              "      <td>4.074417</td>\n",
              "      <td>4.345942</td>\n",
              "      <td>1.0</td>\n",
              "      <td>295.782043</td>\n",
              "      <td>295.726929</td>\n",
              "      <td>100751.773438</td>\n",
              "      <td>0.989764</td>\n",
              "      <td>3.433286</td>\n",
              "      <td>5.181113</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>295.782043</td>\n",
              "      <td>-0.055115</td>\n",
              "      <td>0.771400</td>\n",
              "      <td>83.436400</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>-0.233978</td>\n",
              "      <td>21.450006</td>\n",
              "      <td>1.0</td>\n",
              "      <td>8.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1.549566e+09</td>\n",
              "      <td>34.678699</td>\n",
              "      <td>-86.684799</td>\n",
              "      <td>24.0</td>\n",
              "      <td>20.0</td>\n",
              "      <td>mild temperate</td>\n",
              "      <td>193.0</td>\n",
              "      <td>40.240955</td>\n",
              "      <td>8.303571</td>\n",
              "      <td>747.524910</td>\n",
              "      <td>295.118774</td>\n",
              "      <td>1.091162</td>\n",
              "      <td>296.376611</td>\n",
              "      <td>297.467773</td>\n",
              "      <td>296.376611</td>\n",
              "      <td>262.566675</td>\n",
              "      <td>278.622852</td>\n",
              "      <td>285.923389</td>\n",
              "      <td>290.274927</td>\n",
              "      <td>288.752838</td>\n",
              "      <td>6.125</td>\n",
              "      <td>7.500</td>\n",
              "      <td>5.767445</td>\n",
              "      <td>14.000</td>\n",
              "      <td>4.875</td>\n",
              "      <td>4.125</td>\n",
              "      <td>0.0120</td>\n",
              "      <td>0.0</td>\n",
              "      <td>4.08775</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>2.791129</td>\n",
              "      <td>2.760229</td>\n",
              "      <td>29.527142</td>\n",
              "      <td>25.678796</td>\n",
              "      <td>14.091809</td>\n",
              "      <td>6.503445</td>\n",
              "      <td>6.488293</td>\n",
              "      <td>...</td>\n",
              "      <td>12.791437</td>\n",
              "      <td>13.875513</td>\n",
              "      <td>15.751978</td>\n",
              "      <td>17.743524</td>\n",
              "      <td>19.321313</td>\n",
              "      <td>2.133026</td>\n",
              "      <td>19.321313</td>\n",
              "      <td>21.454340</td>\n",
              "      <td>0.0</td>\n",
              "      <td>21.0</td>\n",
              "      <td>99.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2.277510</td>\n",
              "      <td>7.151445</td>\n",
              "      <td>7.505346</td>\n",
              "      <td>1.0</td>\n",
              "      <td>295.235687</td>\n",
              "      <td>296.000824</td>\n",
              "      <td>99450.273438</td>\n",
              "      <td>0.749565</td>\n",
              "      <td>-0.158933</td>\n",
              "      <td>8.494842</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>295.235687</td>\n",
              "      <td>0.765137</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>4.087750</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.059448</td>\n",
              "      <td>16.150018</td>\n",
              "      <td>-58.0</td>\n",
              "      <td>41.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1.552910e+09</td>\n",
              "      <td>46.066667</td>\n",
              "      <td>41.966667</td>\n",
              "      <td>9.0</td>\n",
              "      <td>20.0</td>\n",
              "      <td>dry</td>\n",
              "      <td>90.0</td>\n",
              "      <td>30.394660</td>\n",
              "      <td>6.451429</td>\n",
              "      <td>753.168113</td>\n",
              "      <td>281.233203</td>\n",
              "      <td>-1.461133</td>\n",
              "      <td>282.810181</td>\n",
              "      <td>281.349048</td>\n",
              "      <td>282.810181</td>\n",
              "      <td>242.905981</td>\n",
              "      <td>258.822461</td>\n",
              "      <td>269.483154</td>\n",
              "      <td>274.965063</td>\n",
              "      <td>275.815735</td>\n",
              "      <td>7.000</td>\n",
              "      <td>6.875</td>\n",
              "      <td>19.727903</td>\n",
              "      <td>4.125</td>\n",
              "      <td>1.000</td>\n",
              "      <td>2.250</td>\n",
              "      <td>0.0057</td>\n",
              "      <td>0.0</td>\n",
              "      <td>4.04475</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>2.635770</td>\n",
              "      <td>3.278284</td>\n",
              "      <td>8.752283</td>\n",
              "      <td>6.223914</td>\n",
              "      <td>3.826746</td>\n",
              "      <td>3.674579</td>\n",
              "      <td>-1.011131</td>\n",
              "      <td>...</td>\n",
              "      <td>0.783167</td>\n",
              "      <td>2.854852</td>\n",
              "      <td>5.040155</td>\n",
              "      <td>7.147913</td>\n",
              "      <td>10.949976</td>\n",
              "      <td>-1.080627</td>\n",
              "      <td>10.949976</td>\n",
              "      <td>9.869348</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>74.0</td>\n",
              "      <td>9.0</td>\n",
              "      <td>2.866836</td>\n",
              "      <td>-1.849202</td>\n",
              "      <td>3.411494</td>\n",
              "      <td>1.0</td>\n",
              "      <td>281.861969</td>\n",
              "      <td>282.492462</td>\n",
              "      <td>100790.546875</td>\n",
              "      <td>0.707056</td>\n",
              "      <td>1.406434</td>\n",
              "      <td>1.084141</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>281.861969</td>\n",
              "      <td>0.630493</td>\n",
              "      <td>0.142125</td>\n",
              "      <td>4.186875</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.400024</td>\n",
              "      <td>3.150018</td>\n",
              "      <td>18.0</td>\n",
              "      <td>92.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5 rows × 129 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "      fact_time  ...  gfs_total_clouds_cover_low_next\n",
              "0  1.543321e+09  ...                              0.0\n",
              "1  1.538776e+09  ...                             11.0\n",
              "2  1.552115e+09  ...                              8.0\n",
              "3  1.549566e+09  ...                             41.0\n",
              "4  1.552910e+09  ...                             92.0\n",
              "\n",
              "[5 rows x 129 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ck8D9GwGiVrg"
      },
      "source": [
        "df_dev_in = pd.read_csv('/content/drive/MyDrive/dataset/data/dev_in.csv')\n",
        "df_dev_out = pd.read_csv('/content/drive/MyDrive/dataset/data/dev_out.csv')"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tuLJMohgku0k"
      },
      "source": [
        "def visualize_temperatures(df_list, df_names, plot_type='histogram'):\n",
        "    '''\n",
        "    Compare the distribution of temperature labels in datasets\n",
        "\n",
        "    Supported visualization formats:\n",
        "     - histogram\n",
        "     - violin\n",
        "    '''\n",
        "    temperatures = np.concatenate([df['fact_temperature'] for df in df_list])\n",
        "    names = np.concatenate([[name]*len(df) for name, df in zip(df_names, df_list)])\n",
        "\n",
        "    df_to_plot = pd.DataFrame(data={'Temperature': temperatures, 'Dataset Name':names})\n",
        "\n",
        "    if plot_type == 'histogram':\n",
        "        sns.histplot(\n",
        "            data=df_to_plot, x=\"Temperature\", hue=\"Dataset Name\",\n",
        "            element=\"step\", stat=\"density\", common_norm=False)\n",
        "    elif plot_type == 'violin':\n",
        "        sns.violinplot(x='Temperature', y='Dataset Name',\n",
        "                              data=df_to_plot, scale='width', palette='Set3')\n",
        "    else:\n",
        "        raise ValueError(\"Unsupported plot_type\")\n",
        "    \n",
        "    plt.show()\n",
        "    plt.clf()"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FqKf93cSi72Q"
      },
      "source": [
        "df_names = ['train', 'dev_in', 'dev_out']\n",
        "df_list = [df_train, df_dev_in, df_dev_out]"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ptz1HnfSj03-",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "outputId": "b05a10d9-623e-456b-8963-1d65eece3c1e"
      },
      "source": [
        "visualize_temperatures(df_list, df_names, plot_type='violin')"
      ],
      "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": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 0 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a_8-lv5nlCdX",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "outputId": "00548d3b-07e1-4772-cc80-d4c15f6ea79e"
      },
      "source": [
        "visualize_temperatures(df_list, df_names, plot_type='histogram')"
      ],
      "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": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 0 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m--CDVcHlPGM"
      },
      "source": [
        "X_train = df_train.iloc[:,6:]\n",
        "X_dev_in = df_dev_in.iloc[:,6:]\n",
        "y_train = df_train['fact_temperature']\n",
        "y_dev_in = df_dev_in['fact_temperature']"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UkvrtcCIlaQF"
      },
      "source": [
        "ensemble_size = 3\n",
        "depth = 2\n",
        "iterations = 200\n",
        "learning_rate = 0.03"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "q57EAonqlzF2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f4463164-0b73-4a1b-c369-f1143af550e9"
      },
      "source": [
        "!pip install catboost"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Collecting catboost\n",
            "  Downloading catboost-1.0.0-cp37-none-manylinux1_x86_64.whl (76.4 MB)\n",
            "\u001b[K     |████████████████████████████████| 76.4 MB 1.3 MB/s \n",
            "\u001b[?25hRequirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from catboost) (1.15.0)\n",
            "Requirement already satisfied: plotly in /usr/local/lib/python3.7/dist-packages (from catboost) (4.4.1)\n",
            "Requirement already satisfied: pandas>=0.24.0 in /usr/local/lib/python3.7/dist-packages (from catboost) (1.1.5)\n",
            "Requirement already satisfied: numpy>=1.16.0 in /usr/local/lib/python3.7/dist-packages (from catboost) (1.19.5)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.7/dist-packages (from catboost) (1.4.1)\n",
            "Requirement already satisfied: graphviz in /usr/local/lib/python3.7/dist-packages (from catboost) (0.10.1)\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.7/dist-packages (from catboost) (3.2.2)\n",
            "Requirement already satisfied: python-dateutil>=2.7.3 in /usr/local/lib/python3.7/dist-packages (from pandas>=0.24.0->catboost) (2.8.2)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.7/dist-packages (from pandas>=0.24.0->catboost) (2018.9)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->catboost) (2.4.7)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->catboost) (1.3.2)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.7/dist-packages (from matplotlib->catboost) (0.10.0)\n",
            "Requirement already satisfied: retrying>=1.3.3 in /usr/local/lib/python3.7/dist-packages (from plotly->catboost) (1.3.3)\n",
            "Installing collected packages: catboost\n",
            "Successfully installed catboost-1.0.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rfQT2yObw4Fb"
      },
      "source": [
        "# **Training**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "utqh_2hWlg77"
      },
      "source": [
        "# Train ensemble of models\n",
        "\n",
        "import catboost\n",
        "\n",
        "trained_models = []\n",
        "for seed in range(ensemble_size):\n",
        "    \n",
        "    model = catboost.CatBoostRegressor(\n",
        "        learning_rate = learning_rate,\n",
        "        iterations = iterations,\n",
        "        depth = depth,\n",
        "        loss_function = 'RMSEWithUncertainty',\n",
        "        eval_metric = 'RMSE',\n",
        "        random_seed = seed)\n",
        "    \n",
        "    print(f'\\n Model index: {seed}\\n')\n",
        "    \n",
        "    model.fit(\n",
        "        X_train,\n",
        "        y_train,\n",
        "        verbose = 100,\n",
        "        eval_set = (X_dev_in, y_dev_in))\n",
        "    \n",
        "    trained_models.append(model)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y8svU7cWxErf"
      },
      "source": [
        "# Save models"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZyHpusoelro6"
      },
      "source": [
        "dir_path = '/content/drive/MyDrive/saved-models'\n",
        "for i, model in enumerate(trained_models):\n",
        "    model.save_model(f'{dir_path}/model{i+1}.cbm')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wB1CXBlFxIig"
      },
      "source": [
        "# Load saved models"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2Lzt7VPGo6H0"
      },
      "source": [
        "import catboost"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gP0jWFxJmHqC"
      },
      "source": [
        "dir_path = '/content/drive/MyDrive/saved-models'\n",
        "trained_models = []\n",
        "\n",
        "# 3 models trained above in this tutorial\n",
        "ensemble_size=10\n",
        "\n",
        "for i in range(1, ensemble_size+1):\n",
        "    model = catboost.CatBoostRegressor()\n",
        "    model.load_model(f'{dir_path}/model{i}.cbm')\n",
        "    trained_models.append(model)"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "owhZOfYTmPdP"
      },
      "source": [
        "dir_path = '/content/drive/MyDrive/baseline/baseline-models'\n",
        "baseline_models = []\n",
        "\n",
        "# 10 models provided\n",
        "ensemble_size=10\n",
        "\n",
        "for ind in range(1, ensemble_size+1):\n",
        "    model = catboost.CatBoostRegressor()\n",
        "    model.load_model(f'{dir_path}/seed{ind}.cbm')\n",
        "    baseline_models.append(model)\n"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rwdjYWTQr8e9"
      },
      "source": [
        "# Create a combined evaluation dataset and keep only the features\n",
        "df_dev = pd.concat([df_dev_in, df_dev_out])\n",
        "X_dev = df_dev.iloc[:,6:]"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "olkqpohyRzVt"
      },
      "source": [
        "# Get Predictions"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GTIzy5TPr_QS"
      },
      "source": [
        "# Get ensemble of predictions for each data point\n",
        "\n",
        "def get_predictions(features_df, model):\n",
        "    '''\n",
        "    Calculates predictions on df features for specified model\n",
        "    \n",
        "    Return: array [num_samples x 2],\n",
        "        where\n",
        "            num_samples = number of rows in features_df\n",
        "            2 = [mean, variance]\n",
        "    \n",
        "    '''\n",
        "    return model.predict(features_df)\n",
        "\n",
        "\n",
        "def get_all_predictions(features_df, models_list):\n",
        "    '''\n",
        "    Return: array [ensemble_size x num_samples x 2],\n",
        "        where\n",
        "            ensemble_size = number of models in models_list\n",
        "            num_samples = number of rows in features_df\n",
        "            2 = [mean, variance]\n",
        "    '''\n",
        "    all_preds = []\n",
        "    for model in models_list:\n",
        "        preds = np.asarray(get_predictions(features_df, model))\n",
        "        all_preds.append(preds)\n",
        "    return np.stack(all_preds, axis=0)\n",
        "\n",
        "\n",
        "all_preds = get_all_predictions(X_dev, baseline_models)"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VVpng1mDsFHX"
      },
      "source": [
        "def entropy_of_expected_class(probs, epsilon=1e-10):\n",
        "    \"\"\"\n",
        "    :param probs: array [num_models, num_examples, num_classes]\n",
        "    :return: array [num_examples}\n",
        "    \"\"\"\n",
        "    mean_probs = np.mean(probs, axis=0)\n",
        "    log_probs = -np.log(mean_probs + epsilon)\n",
        "    return np.sum(mean_probs * log_probs, axis=1)\n",
        "\n",
        "\n",
        "def expected_entropy_class(probs, epsilon=1e-10):\n",
        "    \"\"\"\n",
        "    :param probs: array [num_models, num_examples, num_classes]\n",
        "    :return: array [num_examples}\n",
        "    \"\"\"\n",
        "    log_probs = -np.log(probs + epsilon)\n",
        "\n",
        "    return np.mean(np.sum(probs * log_probs, axis=2), axis=0)\n",
        "\n",
        "\n",
        "def ensemble_uncertainties_classification(probs, epsilon=1e-10):\n",
        "    \"\"\"\n",
        "    :param probs: array [num_models, num_examples, num_classes]\n",
        "    :return: Dictionary of uncertaintties\n",
        "    \"\"\"\n",
        "    mean_probs = np.mean(probs, axis=0)\n",
        "    mean_lprobs = np.mean(np.log(probs + epsilon), axis=0)\n",
        "    conf = np.max(mean_probs, axis=1)\n",
        "\n",
        "    eoe = entropy_of_expected_class(probs, epsilon)\n",
        "    exe = expected_entropy_class(probs, epsilon)\n",
        "\n",
        "    mutual_info = eoe - exe\n",
        "\n",
        "    epkl = -np.sum(mean_probs * mean_lprobs, axis=1) - exe\n",
        "\n",
        "    uncertainty = {'confidence': conf,\n",
        "                   'entropy_of_expected': eoe,\n",
        "                   'expected_entropy': exe,\n",
        "                   'mutual_information': mutual_info,\n",
        "                   'epkl': epkl,\n",
        "                   'reverse_mutual_information': epkl - mutual_info\n",
        "                   }\n",
        "\n",
        "    return uncertainty\n",
        "\n",
        "\n",
        "def epkl_reg(preds, epsilon=1e-10):\n",
        "    \"\"\"\n",
        "    preds: array [n_models, n_samples, 2] - mean and var along last axis.\n",
        "    \"\"\"\n",
        "    means = preds[:, :, 0]\n",
        "    vars = preds[:, :, 1] + epsilon\n",
        "    logvars = np.log(vars)\n",
        "\n",
        "    avg_means = np.mean(means, axis=0)\n",
        "    avg_second_moments = np.mean(means * means + vars, axis=0)\n",
        "\n",
        "    inv_vars = 1.0 / vars\n",
        "    avg_inv_vars = np.mean(inv_vars, axis=0)\n",
        "    mean_inv_var = inv_vars * means\n",
        "    avg_mean_inv_var = np.mean(mean_inv_var, axis=0)\n",
        "    avg_mean2_inv_var = np.mean(means * mean_inv_var + logvars, axis=0) + np.log(2 * np.pi)\n",
        "\n",
        "    epkl = 0.5 * (avg_second_moments * avg_inv_vars - 2 * avg_means * avg_mean_inv_var + avg_mean2_inv_var)\n",
        "\n",
        "    return epkl\n",
        "\n",
        "\n",
        "\n",
        "def ensemble_uncertainties_regression(preds, errors):\n",
        "    \"\"\"\n",
        "    preds: array [n_models, n_samples, 2] - last dim is mean, var\n",
        "    \"\"\"\n",
        "    epkl = epkl_reg(preds)\n",
        "    var_mean = np.var(preds[:, :, 0], axis=0)\n",
        "    mean_var = np.mean(preds[:, :, 1], axis=0)\n",
        "\n",
        "    uncertainty = {'tvar': var_mean + mean_var,\n",
        "                   'mvar': mean_var,\n",
        "                   'varm': var_mean,\n",
        "                   'epkl': epkl}\n",
        "\n",
        "    return uncertainty"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0JWp_pPhsbIm"
      },
      "source": [
        "# Get ensemble predictions\n",
        "all_preds_mean = all_preds[0,:,0]\n",
        "avg_preds = all_preds_mean#np.squeeze(np.mean(all_preds_mean, axis=0))#all_preds_mean \n",
        "# Get target values\n",
        "targets = np.asarray(df_dev['fact_temperature'])\n",
        "\n",
        "# squared error\n",
        "errors = (avg_preds - targets) ** 2"
      ],
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "io3QefcosYBg"
      },
      "source": [
        "all_uncertainty = ensemble_uncertainties_regression(all_preds, errors)\n",
        "uncertainties = all_uncertainty['tvar']"
      ],
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Pm8gTuIUAfy0"
      },
      "source": [
        "# **ENCE and Covariance Coeffecient**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "nEraVg7rrp3a",
        "outputId": "82d62f3e-4244-46fb-c4a5-5e7bf9e89fc9"
      },
      "source": [
        "all_preds_var = all_preds[0,:,1]\n",
        "avg_preds_var = all_preds_var#np.squeeze(np.mean(all_preds_var, axis=0))#all_preds_var \n",
        "print(avg_preds.shape, avg_preds_var.shape)\n",
        "print(avg_preds_var.min())\n",
        "print(avg_preds_var.max())\n",
        "print(avg_preds_var[0])"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(100000,) (100000,)\n",
            "1.5482729070319812e-16\n",
            "1.1033528453839036e+37\n",
            "1.963126860320886\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aVOb-pr6rkGf",
        "outputId": "21a38789-fa62-4eff-92e4-258397de5eac"
      },
      "source": [
        "def arrange(apv, ap, errs):\n",
        "  idx = apv.argsort()\n",
        "  apv = apv[idx]\n",
        "  ap = ap[idx]\n",
        "  errs = errs[idx]\n",
        "  return apv, ap, errs\n",
        "\n",
        "apv_arr,ap_arr, err_arr = arrange(avg_preds_var, avg_preds, errors)\n",
        "print(avg_preds[0], ap_arr[0])\n",
        "print(avg_preds_var[0], apv_arr[0])\n",
        "print(err_arr.shape)"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "10.040559583957656 19.445638001618967\n",
            "1.963126860320886 1.5482729070319812e-16\n",
            "(100000,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5BXGe8exrkgP"
      },
      "source": [
        "def create_bins(ap_arr, apv_arr):\n",
        "  N = 500           #number of bins\n",
        "  bins = {}\n",
        "  j = 200\n",
        "  for i in range(0,N):\n",
        "    idx = j*i\n",
        "    bins[i+1] = {'mean':ap_arr[idx:idx+200], 'var':apv_arr[idx:idx+200]}\n",
        "  return bins\n",
        "\n",
        "bins = create_bins(ap_arr, apv_arr)\n",
        "#print(bins)"
      ],
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MVJRbLE_rkp4"
      },
      "source": [
        "def root_mean_var(bins):\n",
        "  rmv_list = []\n",
        "  for i, mv_dict in bins.items():\n",
        "    vars = mv_dict['var']\n",
        "    rms = np.sqrt(np.sum(np.array(vars))/vars.shape[0])\n",
        "    rmv_list.append(rms)\n",
        "  return rmv_list\n",
        "\n",
        "rmv_list = root_mean_var(bins)\n",
        "#print(rmv_list)"
      ],
      "execution_count": 34,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "48euMxaYrksn"
      },
      "source": [
        "def root_mean_squared_error(bins, errs):\n",
        "  rmse_list = []\n",
        "  for i, mv_dict in bins.items():\n",
        "    means = mv_dict['mean']\n",
        "    idx = (i-1)*means.shape[0]\n",
        "    rmse = np.sqrt(np.sum(errs[idx:idx+200])/means.shape[0])\n",
        "    rmse_list.append(rmse)\n",
        "  return rmse_list\n",
        "\n",
        "rmse_list = root_mean_squared_error(bins, err_arr)\n",
        "#print(rmse_list)"
      ],
      "execution_count": 35,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SHB14C0jrkzV",
        "outputId": "36f0b114-e74f-4450-c984-18a9aa1283cb"
      },
      "source": [
        "def get_ence(rmv_list, rmse_list):\n",
        "  rmv_array = np.array(rmv_list)\n",
        "  rmse_array = np.array(rmse_list)\n",
        "  ence = (1/rmv_array.shape[0])*np.sum(((np.abs(rmv_array - rmse_array))/rmv_array))\n",
        "\n",
        "  return ence\n",
        "\n",
        "ence = get_ence(rmv_list, rmse_list)\n",
        "print(ence)"
      ],
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "3.551325869148706\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BhWVz322rk19",
        "outputId": "f933e4f9-e11e-47b5-b314-fe52c55aeb13"
      },
      "source": [
        "def var_coeff_std(avg_preds_var):\n",
        "  stds = np.sqrt(avg_preds_var)\n",
        "  mu = (np.sum(stds))/avg_preds_var.shape[0]\n",
        "  cv = (np.sqrt((np.sum((stds - mu)**2))/(avg_preds_var.shape[0] - 1)))/mu\n",
        "  return cv\n",
        "\n",
        "\n",
        "avg_preds_var = np.delete(avg_preds_var, np.where(avg_preds_var > 10))\n",
        "cv = var_coeff_std(avg_preds_var)\n",
        "print(cv)"
      ],
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.5504970092627393\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SKJ02PdErk4S",
        "outputId": "af274f62-64ac-4951-d49c-746d2042108a"
      },
      "source": [
        "def lence(ence, cv):\n",
        "  return (np.log(ence + (1/cv)))\n",
        "\n",
        "print(lence(ence, cv))"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1.6804304673299435\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y7kbdM1asSnU"
      },
      "source": [
        "# **R3 Curve**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-kI-EbvisvO6"
      },
      "source": [
        "# Ensemble"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3ly8tLn8dbSR"
      },
      "source": [
        "def root_mean_squared_error(errs):\n",
        "  rmse = np.sqrt(np.sum(errs)/errs.shape[0])\n",
        "  return rmse\n",
        "\n",
        "def root_mean_var(vars):\n",
        "  rmv = np.sqrt(np.sum(vars)/vars.shape[0])\n",
        "  return rmv"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yhEfniUWcgOC"
      },
      "source": [
        "#retention fraction\n",
        "N = 100000\n",
        "frac_list = []\n",
        "def arrange(apv, ap, errs):\n",
        "  idx = apv.argsort() #ascending order of variances\n",
        "  apv = apv[idx]\n",
        "  ap = ap[idx]\n",
        "  errs = errs[idx]\n",
        "  return apv, ap, errs\n",
        "\n",
        "apv_arr,ap_arr, err_arr = arrange(avg_preds_var, avg_preds, errors)\n",
        "\n",
        "retention_frac = [0.2, 0.4, 0.6, 0.8, 1.0]\n",
        "\n",
        "\n",
        "for i in retention_frac:\n",
        "  rmse = root_mean_squared_error(ap_arr[0:int(i*N)])\n",
        "  rmv = root_mean_var(apv_arr[0:int(i*N)])\n",
        "  frac = rmse/rmv\n",
        "  #frac = rmv/rmse\n",
        "  frac_list.append(frac)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wLwk9kEWfKy8"
      },
      "source": [
        "frac_list.insert(0,0.0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o1IlXldefq-g"
      },
      "source": [
        "retention_frac.insert(0,0.0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "Jb0IKKFVfcDQ",
        "outputId": "8d591e4f-8124-4784-81f2-df76849a7311"
      },
      "source": [
        "plt.plot(retention_frac, frac_list)\n",
        "plt.ylabel('RMSE/RMV')\n",
        "plt.xlabel(\"Retention fraction\")\n",
        "plt.show()"
      ],
      "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": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z0l_tI3xke8g"
      },
      "source": [
        "frac_list_ens = frac_list\n",
        "ret_frac_list_ens = retention_frac"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1QprHhHagkah"
      },
      "source": [
        "# Single model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s7KMgRoLgCK4"
      },
      "source": [
        "avg_preds = all_preds[0,:,0]\n",
        "#avg_preds = np.squeeze(np.mean(all_preds_mean, axis=0))\n",
        "\n",
        "# Get target values\n",
        "targets = np.asarray(df_dev['fact_temperature'])\n",
        "\n",
        "# squared error\n",
        "errors = (avg_preds - targets) ** 2"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SdtjUoxqgGsV",
        "outputId": "1f000cad-b374-41f2-f507-5b5e2b0f9af2"
      },
      "source": [
        "avg_preds_var = all_preds[0,:,1]\n",
        "#avg_preds_var = np.squeeze(np.mean(all_preds_var, axis=0))\n",
        "print(avg_preds.shape, avg_preds_var.shape)\n",
        "print(avg_preds_var.min())\n",
        "print(avg_preds_var.max())\n",
        "print(avg_preds_var[0])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(100000,) (100000,)\n",
            "1.5482729070319812e-16\n",
            "1.1033528453839036e+37\n",
            "1.963126860320886\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9VlaZ5p9gGsc"
      },
      "source": [
        "def root_mean_squared_error(errs):\n",
        "  rmse = np.sqrt(np.sum(errs)/errs.shape[0])\n",
        "  return rmse\n",
        "\n",
        "def root_mean_var(vars):\n",
        "  rmv = np.sqrt(np.sum(vars)/vars.shape[0])\n",
        "  return rmv"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A5zQzOS8gGse"
      },
      "source": [
        "#retention fraction\n",
        "N = 100000\n",
        "frac_list = []\n",
        "def arrange(apv, ap, errs):\n",
        "  idx = apv.argsort() #ascending order of variances\n",
        "  apv = apv[idx]\n",
        "  ap = ap[idx]\n",
        "  errs = errs[idx]\n",
        "  return apv, ap, errs\n",
        "\n",
        "apv_arr,ap_arr, err_arr = arrange(avg_preds_var, avg_preds, errors)\n",
        "\n",
        "retention_frac = [0.2, 0.4, 0.6, 0.8, 1.0]\n",
        "for i in retention_frac:\n",
        "  rmse = root_mean_squared_error(ap_arr[0:int(i*N)])\n",
        "  rmv = root_mean_var(apv_arr[0:int(i*N)])\n",
        "  frac = rmse/rmv\n",
        "  frac_list.append(frac)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jVqzDeIkgMVh"
      },
      "source": [
        "frac_list.insert(0,0.0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PF2Tu6b0gMVk"
      },
      "source": [
        "retention_frac.insert(0,0.0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4oQ0RjbbkTbx"
      },
      "source": [
        "frac_single = frac_list\n",
        "ret_frac_single = retention_frac"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "qsl3WJtQgMVl",
        "outputId": "330a03ec-12ee-40a4-c042-ec2d2d318719"
      },
      "source": [
        "plt.plot(retention_frac, frac_list)\n",
        "plt.ylabel('RMSE/RMV')\n",
        "plt.xlabel(\"Retention fraction\")\n",
        "plt.show()"
      ],
      "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": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JZCaGHbssylQ"
      },
      "source": [
        "## Combined R3 Plot"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "JR3W5SNCgH4V",
        "outputId": "98eace76-267c-470d-afb7-8e99f91bbbb2"
      },
      "source": [
        "plt.plot(ret_frac_single, frac_single)\n",
        "plt.plot(ret_frac_list_ens, frac_list_ens)\n",
        "plt.ylabel('RMSE/RMV')\n",
        "plt.xlabel(\"Retention fraction\")\n",
        "plt.legend(['Single', 'Ensemble'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f3859b85450>"
            ]
          },
          "metadata": {},
          "execution_count": 38
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}