1st Program - Test and Pickle XGBoost.ipynb

{
 "cells": [
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "November 2022\n",
    "Author: Evan Shlom\n",
    "\n",
    "Model Detail: XGBoost: Approximation tree method, lower Learnng Rt, Lambda regularization\n",
    "\n",
    "Anonymous Data\n",
    "10 categories with 2 types per category\n",
    "Target variable: A1-J1, A2-J2 (20 Targets)\n",
    "\n",
    "Tested on June 2022 to August 2022\n",
    "Also Blind Forecasting September 2022 and October 2022"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import pickle as pkl\n",
    "from xgboost import XGBRegressor\n",
    "import warnings\n",
    "warnings.simplefilter(action='ignore', category=pd.errors.PerformanceWarning)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "file = './Data/dataset.csv'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Actuals:'"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>Timestep</th>\n",
       "      <th>A1</th>\n",
       "      <th>B1</th>\n",
       "      <th>C1</th>\n",
       "      <th>D1</th>\n",
       "      <th>E1</th>\n",
       "      <th>F1</th>\n",
       "      <th>G1</th>\n",
       "      <th>H1</th>\n",
       "      <th>I1</th>\n",
       "      <th>...</th>\n",
       "      <th>A2</th>\n",
       "      <th>B2</th>\n",
       "      <th>C2</th>\n",
       "      <th>D2</th>\n",
       "      <th>E2</th>\n",
       "      <th>F2</th>\n",
       "      <th>G2</th>\n",
       "      <th>H2</th>\n",
       "      <th>I2</th>\n",
       "      <th>J2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>78</td>\n",
       "      <td>2142.5</td>\n",
       "      <td>616.4</td>\n",
       "      <td>2146.6</td>\n",
       "      <td>498.9</td>\n",
       "      <td>416.2</td>\n",
       "      <td>111.5</td>\n",
       "      <td>20994.0</td>\n",
       "      <td>17207.0</td>\n",
       "      <td>7805.0</td>\n",
       "      <td>...</td>\n",
       "      <td>1777</td>\n",
       "      <td>135</td>\n",
       "      <td>3359</td>\n",
       "      <td>183</td>\n",
       "      <td>534</td>\n",
       "      <td>19.0</td>\n",
       "      <td>10730</td>\n",
       "      <td>2687</td>\n",
       "      <td>838</td>\n",
       "      <td>501</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>79</td>\n",
       "      <td>2120.3</td>\n",
       "      <td>520.0</td>\n",
       "      <td>2013.8</td>\n",
       "      <td>370.0</td>\n",
       "      <td>394.9</td>\n",
       "      <td>432.3</td>\n",
       "      <td>20465.0</td>\n",
       "      <td>19480.0</td>\n",
       "      <td>9316.0</td>\n",
       "      <td>...</td>\n",
       "      <td>1592</td>\n",
       "      <td>122</td>\n",
       "      <td>3110</td>\n",
       "      <td>132</td>\n",
       "      <td>466</td>\n",
       "      <td>51.0</td>\n",
       "      <td>10525</td>\n",
       "      <td>2633</td>\n",
       "      <td>969</td>\n",
       "      <td>474</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>80</td>\n",
       "      <td>2187.6</td>\n",
       "      <td>852.1</td>\n",
       "      <td>2160.2</td>\n",
       "      <td>400.5</td>\n",
       "      <td>504.7</td>\n",
       "      <td>200.4</td>\n",
       "      <td>29227.0</td>\n",
       "      <td>18565.0</td>\n",
       "      <td>8922.0</td>\n",
       "      <td>...</td>\n",
       "      <td>1851</td>\n",
       "      <td>152</td>\n",
       "      <td>3384</td>\n",
       "      <td>171</td>\n",
       "      <td>706</td>\n",
       "      <td>142.0</td>\n",
       "      <td>12963</td>\n",
       "      <td>2833</td>\n",
       "      <td>1429</td>\n",
       "      <td>469</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>81</td>\n",
       "      <td>2043.3</td>\n",
       "      <td>797.4</td>\n",
       "      <td>2528.1</td>\n",
       "      <td>526.1</td>\n",
       "      <td>484.7</td>\n",
       "      <td>463.1</td>\n",
       "      <td>16531.9</td>\n",
       "      <td>17664.1</td>\n",
       "      <td>6577.7</td>\n",
       "      <td>...</td>\n",
       "      <td>1783</td>\n",
       "      <td>170</td>\n",
       "      <td>3199</td>\n",
       "      <td>159</td>\n",
       "      <td>952</td>\n",
       "      <td>85.0</td>\n",
       "      <td>7150</td>\n",
       "      <td>2713</td>\n",
       "      <td>545</td>\n",
       "      <td>399</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81</th>\n",
       "      <td>82</td>\n",
       "      <td>1822.3</td>\n",
       "      <td>1080.6</td>\n",
       "      <td>2192.8</td>\n",
       "      <td>413.7</td>\n",
       "      <td>736.8</td>\n",
       "      <td>299.0</td>\n",
       "      <td>14620.4</td>\n",
       "      <td>14705.8</td>\n",
       "      <td>5832.7</td>\n",
       "      <td>...</td>\n",
       "      <td>1580</td>\n",
       "      <td>197</td>\n",
       "      <td>3129</td>\n",
       "      <td>159</td>\n",
       "      <td>1143</td>\n",
       "      <td>152.0</td>\n",
       "      <td>6800</td>\n",
       "      <td>2635</td>\n",
       "      <td>467</td>\n",
       "      <td>398</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    Timestep      A1      B1      C1     D1     E1     F1       G1       H1  \\\n",
       "77        78  2142.5   616.4  2146.6  498.9  416.2  111.5  20994.0  17207.0   \n",
       "78        79  2120.3   520.0  2013.8  370.0  394.9  432.3  20465.0  19480.0   \n",
       "79        80  2187.6   852.1  2160.2  400.5  504.7  200.4  29227.0  18565.0   \n",
       "80        81  2043.3   797.4  2528.1  526.1  484.7  463.1  16531.9  17664.1   \n",
       "81        82  1822.3  1080.6  2192.8  413.7  736.8  299.0  14620.4  14705.8   \n",
       "\n",
       "        I1  ...    A2   B2    C2   D2    E2     F2     G2    H2    I2   J2  \n",
       "77  7805.0  ...  1777  135  3359  183   534   19.0  10730  2687   838  501  \n",
       "78  9316.0  ...  1592  122  3110  132   466   51.0  10525  2633   969  474  \n",
       "79  8922.0  ...  1851  152  3384  171   706  142.0  12963  2833  1429  469  \n",
       "80  6577.7  ...  1783  170  3199  159   952   85.0   7150  2713   545  399  \n",
       "81  5832.7  ...  1580  197  3129  159  1143  152.0   6800  2635   467  398  \n",
       "\n",
       "[5 rows x 21 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "'Predictions:'"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>Timestep</th>\n",
       "      <th>A1</th>\n",
       "      <th>B1</th>\n",
       "      <th>C1</th>\n",
       "      <th>D1</th>\n",
       "      <th>E1</th>\n",
       "      <th>F1</th>\n",
       "      <th>G1</th>\n",
       "      <th>H1</th>\n",
       "      <th>I1</th>\n",
       "      <th>...</th>\n",
       "      <th>A2</th>\n",
       "      <th>B2</th>\n",
       "      <th>C2</th>\n",
       "      <th>D2</th>\n",
       "      <th>E2</th>\n",
       "      <th>F2</th>\n",
       "      <th>G2</th>\n",
       "      <th>H2</th>\n",
       "      <th>I2</th>\n",
       "      <th>J2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>78</td>\n",
       "      <td>1704.774170</td>\n",
       "      <td>804.572510</td>\n",
       "      <td>2130.046631</td>\n",
       "      <td>390.480164</td>\n",
       "      <td>392.226868</td>\n",
       "      <td>100.267166</td>\n",
       "      <td>21770.644531</td>\n",
       "      <td>12341.163086</td>\n",
       "      <td>6251.287598</td>\n",
       "      <td>...</td>\n",
       "      <td>1711.361938</td>\n",
       "      <td>207.697327</td>\n",
       "      <td>2850.224365</td>\n",
       "      <td>139.724777</td>\n",
       "      <td>700.687378</td>\n",
       "      <td>17.893074</td>\n",
       "      <td>10088.442383</td>\n",
       "      <td>2405.543213</td>\n",
       "      <td>791.522156</td>\n",
       "      <td>352.084808</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>79</td>\n",
       "      <td>1808.200684</td>\n",
       "      <td>738.215332</td>\n",
       "      <td>2130.046631</td>\n",
       "      <td>365.784088</td>\n",
       "      <td>340.679199</td>\n",
       "      <td>109.836723</td>\n",
       "      <td>25146.480469</td>\n",
       "      <td>12036.032227</td>\n",
       "      <td>6071.300293</td>\n",
       "      <td>...</td>\n",
       "      <td>1554.525513</td>\n",
       "      <td>238.618225</td>\n",
       "      <td>2761.366943</td>\n",
       "      <td>135.115173</td>\n",
       "      <td>605.188232</td>\n",
       "      <td>30.010580</td>\n",
       "      <td>7115.916992</td>\n",
       "      <td>2238.628418</td>\n",
       "      <td>612.454407</td>\n",
       "      <td>345.645111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>80</td>\n",
       "      <td>1436.302246</td>\n",
       "      <td>725.188660</td>\n",
       "      <td>2117.963867</td>\n",
       "      <td>387.438293</td>\n",
       "      <td>386.191864</td>\n",
       "      <td>100.936295</td>\n",
       "      <td>18697.460938</td>\n",
       "      <td>11697.129883</td>\n",
       "      <td>4935.188965</td>\n",
       "      <td>...</td>\n",
       "      <td>1314.789307</td>\n",
       "      <td>167.058701</td>\n",
       "      <td>2535.916016</td>\n",
       "      <td>143.572571</td>\n",
       "      <td>754.067444</td>\n",
       "      <td>17.291162</td>\n",
       "      <td>9683.092773</td>\n",
       "      <td>2389.021484</td>\n",
       "      <td>494.133972</td>\n",
       "      <td>251.128845</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>81</td>\n",
       "      <td>1530.706299</td>\n",
       "      <td>716.571045</td>\n",
       "      <td>2130.046631</td>\n",
       "      <td>390.130981</td>\n",
       "      <td>431.680939</td>\n",
       "      <td>95.107376</td>\n",
       "      <td>17967.669922</td>\n",
       "      <td>8631.333008</td>\n",
       "      <td>4780.489746</td>\n",
       "      <td>...</td>\n",
       "      <td>1321.965576</td>\n",
       "      <td>188.630051</td>\n",
       "      <td>2821.939209</td>\n",
       "      <td>145.365372</td>\n",
       "      <td>859.664246</td>\n",
       "      <td>17.291162</td>\n",
       "      <td>7489.082031</td>\n",
       "      <td>2473.566406</td>\n",
       "      <td>312.242523</td>\n",
       "      <td>352.319916</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>82</td>\n",
       "      <td>1588.990479</td>\n",
       "      <td>737.624268</td>\n",
       "      <td>2130.046631</td>\n",
       "      <td>381.330994</td>\n",
       "      <td>658.211792</td>\n",
       "      <td>87.053566</td>\n",
       "      <td>27199.398438</td>\n",
       "      <td>8389.782227</td>\n",
       "      <td>5621.973633</td>\n",
       "      <td>...</td>\n",
       "      <td>1211.456909</td>\n",
       "      <td>163.557022</td>\n",
       "      <td>2831.188721</td>\n",
       "      <td>139.737091</td>\n",
       "      <td>801.965515</td>\n",
       "      <td>15.463931</td>\n",
       "      <td>7772.947754</td>\n",
       "      <td>2146.727539</td>\n",
       "      <td>599.530151</td>\n",
       "      <td>347.738770</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Timestep           A1          B1           C1          D1          E1  \\\n",
       "0        78  1704.774170  804.572510  2130.046631  390.480164  392.226868   \n",
       "1        79  1808.200684  738.215332  2130.046631  365.784088  340.679199   \n",
       "2        80  1436.302246  725.188660  2117.963867  387.438293  386.191864   \n",
       "3        81  1530.706299  716.571045  2130.046631  390.130981  431.680939   \n",
       "4        82  1588.990479  737.624268  2130.046631  381.330994  658.211792   \n",
       "\n",
       "           F1            G1            H1           I1  ...           A2  \\\n",
       "0  100.267166  21770.644531  12341.163086  6251.287598  ...  1711.361938   \n",
       "1  109.836723  25146.480469  12036.032227  6071.300293  ...  1554.525513   \n",
       "2  100.936295  18697.460938  11697.129883  4935.188965  ...  1314.789307   \n",
       "3   95.107376  17967.669922   8631.333008  4780.489746  ...  1321.965576   \n",
       "4   87.053566  27199.398438   8389.782227  5621.973633  ...  1211.456909   \n",
       "\n",
       "           B2           C2          D2          E2         F2            G2  \\\n",
       "0  207.697327  2850.224365  139.724777  700.687378  17.893074  10088.442383   \n",
       "1  238.618225  2761.366943  135.115173  605.188232  30.010580   7115.916992   \n",
       "2  167.058701  2535.916016  143.572571  754.067444  17.291162   9683.092773   \n",
       "3  188.630051  2821.939209  145.365372  859.664246  17.291162   7489.082031   \n",
       "4  163.557022  2831.188721  139.737091  801.965515  15.463931   7772.947754   \n",
       "\n",
       "            H2          I2          J2  \n",
       "0  2405.543213  791.522156  352.084808  \n",
       "1  2238.628418  612.454407  345.645111  \n",
       "2  2389.021484  494.133972  251.128845  \n",
       "3  2473.566406  312.242523  352.319916  \n",
       "4  2146.727539  599.530151  347.738770  \n",
       "\n",
       "[5 rows x 21 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df = pd.read_csv(f'{file}')\\\n",
    "       .fillna(0)\n",
    "\n",
    "# https://xgboost.readthedocs.io/en/stable/parameter.html\n",
    "xgb = XGBRegressor(booster='gbtree',\n",
    "                   n_estimators=25,\n",
    "                   learning_rate=0.15, # default=0.3\n",
    "                   max_depth=3, # default=6\n",
    "                   \n",
    "                   reg_lambda=4, # default=1 # L2 regularization\n",
    "                   \n",
    "                   tree_method='approx', # default='auto' aka 'exact'\n",
    "                   grow_policy='lossguide', # approx only\n",
    "                   max_leaves=5, #default=0, # approx only\n",
    "                   max_bin=512, #default=256, # approx only\n",
    "                   random_state=0)\n",
    "\n",
    "def count_y(df):\n",
    "    num_y = len(df.columns[3:])\n",
    "    return num_y\n",
    "\n",
    "def prepare_dataset(df, num_y, train_interval):\n",
    "    for label, content in df.iloc[:,3:(3+num_y)].items(): # cols 0:2 are datetime cols # can 3+num_y be changed to just -1?    \n",
    "        col_name = label+f\" lag+{train_interval}\"\n",
    "        df[col_name] = df[label].shift(train_interval)\n",
    "    df.drop(index=df.index[[0]], axis=0, inplace=True) # lag cols have null rows\n",
    "    return df\n",
    "\n",
    "def train_test_split(df, num_y, test_rows):\n",
    "    X = pd.concat((df['Timestep'], df.iloc[:,(3+num_y):]), axis=1) # timestep, lag X1, lag X2\n",
    "    y = df.iloc[:, 3:(3+num_y)] # actual X1, actual X2\n",
    "\n",
    "    num_rows = len(df.index)\n",
    "    split_rows = num_rows - test_rows\n",
    "    \n",
    "    X_train = X.iloc[:split_rows]\n",
    "    X_test = X.iloc[split_rows:]\n",
    "    y_train = y.iloc[:split_rows]\n",
    "    y_test = y.iloc[split_rows:]\n",
    "    \n",
    "    return [X, X_train, X_test, y, y_train, y_test]\n",
    "\n",
    "def get_results(model, X, X_train, X_test, y, y_train, y_test):\n",
    "    # Run model\n",
    "    model = model.fit(X_train,y_train)\n",
    "    predictions = model.predict(X_test)\n",
    "    df_valid = pd.concat([X_test.Timestep, y_test],axis=1)\n",
    "    \n",
    "    X_test.reset_index(drop=True, inplace=True)\n",
    "    df_y_preds = pd.DataFrame(predictions, columns=y.columns)\n",
    "    df_preds = pd.concat([X_test.Timestep, df_y_preds], axis=1)\n",
    "    \n",
    "    return model, df_preds, df_valid\n",
    "\n",
    "num_y = count_y(df)\n",
    "train_interval = 12 # months to shift for lag columns # 12 gives best performance\n",
    "df = prepare_dataset(df, num_y, train_interval)\n",
    "test_rows = 5 # rows in the dataset that're not for training\n",
    "tts_list = train_test_split(df, num_y, test_rows)\n",
    "model, df_preds, df_valid = get_results(xgb, *tts_list)\n",
    "\n",
    "display(\"Actuals:\",df_valid,\n",
    "        \"Predictions:\",df_preds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Percent Variance:'"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>Timestep</th>\n",
       "      <th>A1</th>\n",
       "      <th>B1</th>\n",
       "      <th>C1</th>\n",
       "      <th>D1</th>\n",
       "      <th>E1</th>\n",
       "      <th>F1</th>\n",
       "      <th>G1</th>\n",
       "      <th>H1</th>\n",
       "      <th>I1</th>\n",
       "      <th>...</th>\n",
       "      <th>A2</th>\n",
       "      <th>B2</th>\n",
       "      <th>C2</th>\n",
       "      <th>D2</th>\n",
       "      <th>E2</th>\n",
       "      <th>F2</th>\n",
       "      <th>G2</th>\n",
       "      <th>H2</th>\n",
       "      <th>I2</th>\n",
       "      <th>J2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.256765</td>\n",
       "      <td>0.766121</td>\n",
       "      <td>1.007771</td>\n",
       "      <td>1.277658</td>\n",
       "      <td>1.061121</td>\n",
       "      <td>1.112029</td>\n",
       "      <td>0.964326</td>\n",
       "      <td>1.394277</td>\n",
       "      <td>1.248543</td>\n",
       "      <td>...</td>\n",
       "      <td>1.038354</td>\n",
       "      <td>0.649984</td>\n",
       "      <td>1.178504</td>\n",
       "      <td>1.309718</td>\n",
       "      <td>0.762109</td>\n",
       "      <td>1.061863</td>\n",
       "      <td>1.063593</td>\n",
       "      <td>1.117003</td>\n",
       "      <td>1.058720</td>\n",
       "      <td>1.422953</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.172602</td>\n",
       "      <td>0.704402</td>\n",
       "      <td>0.945425</td>\n",
       "      <td>1.011526</td>\n",
       "      <td>1.159155</td>\n",
       "      <td>3.935842</td>\n",
       "      <td>0.813832</td>\n",
       "      <td>1.618474</td>\n",
       "      <td>1.534432</td>\n",
       "      <td>...</td>\n",
       "      <td>1.024107</td>\n",
       "      <td>0.511277</td>\n",
       "      <td>1.126254</td>\n",
       "      <td>0.976944</td>\n",
       "      <td>0.770008</td>\n",
       "      <td>1.699401</td>\n",
       "      <td>1.479079</td>\n",
       "      <td>1.176167</td>\n",
       "      <td>1.582159</td>\n",
       "      <td>1.371349</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.523078</td>\n",
       "      <td>1.175005</td>\n",
       "      <td>1.019942</td>\n",
       "      <td>1.033713</td>\n",
       "      <td>1.306863</td>\n",
       "      <td>1.985411</td>\n",
       "      <td>1.563153</td>\n",
       "      <td>1.587141</td>\n",
       "      <td>1.807834</td>\n",
       "      <td>...</td>\n",
       "      <td>1.407830</td>\n",
       "      <td>0.909860</td>\n",
       "      <td>1.334429</td>\n",
       "      <td>1.191035</td>\n",
       "      <td>0.936256</td>\n",
       "      <td>8.212288</td>\n",
       "      <td>1.338725</td>\n",
       "      <td>1.185841</td>\n",
       "      <td>2.891928</td>\n",
       "      <td>1.867567</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.334874</td>\n",
       "      <td>1.112800</td>\n",
       "      <td>1.186875</td>\n",
       "      <td>1.348521</td>\n",
       "      <td>1.122820</td>\n",
       "      <td>4.869233</td>\n",
       "      <td>0.920091</td>\n",
       "      <td>2.046509</td>\n",
       "      <td>1.375947</td>\n",
       "      <td>...</td>\n",
       "      <td>1.348749</td>\n",
       "      <td>0.901235</td>\n",
       "      <td>1.133618</td>\n",
       "      <td>1.093796</td>\n",
       "      <td>1.107409</td>\n",
       "      <td>4.915806</td>\n",
       "      <td>0.954723</td>\n",
       "      <td>1.096797</td>\n",
       "      <td>1.745438</td>\n",
       "      <td>1.132493</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.146829</td>\n",
       "      <td>1.464973</td>\n",
       "      <td>1.029461</td>\n",
       "      <td>1.084884</td>\n",
       "      <td>1.119397</td>\n",
       "      <td>3.434667</td>\n",
       "      <td>0.537527</td>\n",
       "      <td>1.752823</td>\n",
       "      <td>1.037483</td>\n",
       "      <td>...</td>\n",
       "      <td>1.304215</td>\n",
       "      <td>1.204473</td>\n",
       "      <td>1.105189</td>\n",
       "      <td>1.137851</td>\n",
       "      <td>1.425248</td>\n",
       "      <td>9.829325</td>\n",
       "      <td>0.874829</td>\n",
       "      <td>1.227450</td>\n",
       "      <td>0.778943</td>\n",
       "      <td>1.144537</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    Timestep        A1        B1        C1        D1        E1        F1  \\\n",
       "77       1.0  1.256765  0.766121  1.007771  1.277658  1.061121  1.112029   \n",
       "78       1.0  1.172602  0.704402  0.945425  1.011526  1.159155  3.935842   \n",
       "79       1.0  1.523078  1.175005  1.019942  1.033713  1.306863  1.985411   \n",
       "80       1.0  1.334874  1.112800  1.186875  1.348521  1.122820  4.869233   \n",
       "81       1.0  1.146829  1.464973  1.029461  1.084884  1.119397  3.434667   \n",
       "\n",
       "          G1        H1        I1  ...        A2        B2        C2        D2  \\\n",
       "77  0.964326  1.394277  1.248543  ...  1.038354  0.649984  1.178504  1.309718   \n",
       "78  0.813832  1.618474  1.534432  ...  1.024107  0.511277  1.126254  0.976944   \n",
       "79  1.563153  1.587141  1.807834  ...  1.407830  0.909860  1.334429  1.191035   \n",
       "80  0.920091  2.046509  1.375947  ...  1.348749  0.901235  1.133618  1.093796   \n",
       "81  0.537527  1.752823  1.037483  ...  1.304215  1.204473  1.105189  1.137851   \n",
       "\n",
       "          E2        F2        G2        H2        I2        J2  \n",
       "77  0.762109  1.061863  1.063593  1.117003  1.058720  1.422953  \n",
       "78  0.770008  1.699401  1.479079  1.176167  1.582159  1.371349  \n",
       "79  0.936256  8.212288  1.338725  1.185841  2.891928  1.867567  \n",
       "80  1.107409  4.915806  0.954723  1.096797  1.745438  1.132493  \n",
       "81  1.425248  9.829325  0.874829  1.227450  0.778943  1.144537  \n",
       "\n",
       "[5 rows x 21 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_var = df_valid.div(df_preds.values)\n",
    "display(\"Percent Variance:\",df_var) # standard metric key performance indicator"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "      "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_preds.to_excel('./Outputs/e_em_xgb_lambda4_junoct22_preds_export.xlsx')\n",
    "df_valid.to_excel('./Outputs/e_em_xgb_lambda4_junoct22_valid_export.xlsx')\n",
    "df_var.to_excel('./Outputs/e_em_xgb_lambda4_junoct22_var_export.xlsx')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "pkl.dump(model, open('./Model/model.pkl','wb'))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}