gold_recovery/gold_recovery_project.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "beff1f08-db1e-4ed7-8936-b682ddf08374",
   "metadata": {},
   "source": [
    "# Data preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f7b8b047-fac0-4b47-b344-731760e8b802",
   "metadata": {},
   "source": [
    "## Uploading data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "113b133e-3a6c-4f61-b165-9f721a9c63dd",
   "metadata": {},
   "source": [
    "In order to achieve the objective that the company has set for us, we need to first take a look at the data that we have been provided with. Thus, let's import all libraries, functions and data structures which will be particularly useful for completing the project."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "a240825e-906b-40d0-bda0-172e61d86c96",
   "metadata": {},
   "outputs": [],
   "source": [
    "import functools\n",
    "import itertools\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "from IPython.display import display\n",
    "from sklearn.base import clone\n",
    "from sklearn.dummy import DummyRegressor\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.metrics import make_scorer, mean_absolute_error\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "from tqdm import notebook\n",
    "\n",
    "RANDOM_STATE = 12345\n",
    "CV_FOLDS = 3"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "171c1ab3-99bc-4b8d-a62a-6388a009a3e5",
   "metadata": {},
   "source": [
    "Now, let's load the data contained in each *csv*-file into a separate DataFrame. As stated in the project description, the data is to be indexed by the date and time of obtaining information so we will use `date` feature as an index, thus specifying `index_col=\"date\"` and simultaneously converting it to *datetime* using `parse_dates=True`.\n",
    "\n",
    "Notably, in order not to repeat writing data loading using `read_csv()` with all these parameters three times, we can just clone this function with the required parameters using `partial()` function from `functools` library."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "34f88b8a-e8e8-4381-9bb5-37fab6d910f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Cloning 'read_csv' function\n",
    "read_csv_dates = functools.partial(pd.read_csv, index_col=\"date\", parse_dates=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "38a7c119-4eb1-4721-a4a6-bb52e24a8f95",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Loading the data to DataFrames\n",
    "gold_recovery_train = read_csv_dates(\"datasets/gold_recovery_train_new.csv\")\n",
    "gold_recovery_test = read_csv_dates(\"datasets/gold_recovery_test_new.csv\")\n",
    "gold_recovery_full = read_csv_dates(\"datasets/gold_recovery_full_new.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fd3b891b-46db-4274-8551-f9b01bff3fa1",
   "metadata": {},
   "source": [
    "We see that we have now three DataFrames for each of the files provided:\n",
    "\n",
    "* `gold_recovery_train` => training set;\n",
    "* `gold_recovery_test` => test set;\n",
    "* `gold_recovery_full` => full data."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8c3f730-8a81-4e9d-aaaa-e9a5fadf2094",
   "metadata": {},
   "source": [
    "## Data overview"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5d4c092c-a742-4962-aca4-66a056b205e1",
   "metadata": {},
   "source": [
    "Let's look at the data contained in these DataFrames and print the first five rows of each one."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "e4c31087-ba4c-4c32-8148-8c98c8abb207",
   "metadata": {},
   "outputs": [
    {
     "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>final.output.concentrate_ag</th>\n",
       "      <th>final.output.concentrate_pb</th>\n",
       "      <th>final.output.concentrate_sol</th>\n",
       "      <th>final.output.concentrate_au</th>\n",
       "      <th>final.output.recovery</th>\n",
       "      <th>final.output.tail_ag</th>\n",
       "      <th>final.output.tail_pb</th>\n",
       "      <th>final.output.tail_sol</th>\n",
       "      <th>final.output.tail_au</th>\n",
       "      <th>primary_cleaner.input.sulfate</th>\n",
       "      <th>...</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_a_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_a_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_b_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_b_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_a_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_a_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_b_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_b_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank6_a_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank6_a_level</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2016-01-15 00:00:00</th>\n",
       "      <td>6.055403</td>\n",
       "      <td>9.889648</td>\n",
       "      <td>5.507324</td>\n",
       "      <td>42.192020</td>\n",
       "      <td>70.541216</td>\n",
       "      <td>10.411962</td>\n",
       "      <td>0.895447</td>\n",
       "      <td>16.904297</td>\n",
       "      <td>2.143149</td>\n",
       "      <td>127.092003</td>\n",
       "      <td>...</td>\n",
       "      <td>14.016835</td>\n",
       "      <td>-502.488007</td>\n",
       "      <td>12.099931</td>\n",
       "      <td>-504.715942</td>\n",
       "      <td>9.925633</td>\n",
       "      <td>-498.310211</td>\n",
       "      <td>8.079666</td>\n",
       "      <td>-500.470978</td>\n",
       "      <td>14.151341</td>\n",
       "      <td>-605.841980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-15 01:00:00</th>\n",
       "      <td>6.029369</td>\n",
       "      <td>9.968944</td>\n",
       "      <td>5.257781</td>\n",
       "      <td>42.701629</td>\n",
       "      <td>69.266198</td>\n",
       "      <td>10.462676</td>\n",
       "      <td>0.927452</td>\n",
       "      <td>16.634514</td>\n",
       "      <td>2.224930</td>\n",
       "      <td>125.629232</td>\n",
       "      <td>...</td>\n",
       "      <td>13.992281</td>\n",
       "      <td>-505.503262</td>\n",
       "      <td>11.950531</td>\n",
       "      <td>-501.331529</td>\n",
       "      <td>10.039245</td>\n",
       "      <td>-500.169983</td>\n",
       "      <td>7.984757</td>\n",
       "      <td>-500.582168</td>\n",
       "      <td>13.998353</td>\n",
       "      <td>-599.787184</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-15 02:00:00</th>\n",
       "      <td>6.055926</td>\n",
       "      <td>10.213995</td>\n",
       "      <td>5.383759</td>\n",
       "      <td>42.657501</td>\n",
       "      <td>68.116445</td>\n",
       "      <td>10.507046</td>\n",
       "      <td>0.953716</td>\n",
       "      <td>16.208849</td>\n",
       "      <td>2.257889</td>\n",
       "      <td>123.819808</td>\n",
       "      <td>...</td>\n",
       "      <td>14.015015</td>\n",
       "      <td>-502.520901</td>\n",
       "      <td>11.912783</td>\n",
       "      <td>-501.133383</td>\n",
       "      <td>10.070913</td>\n",
       "      <td>-500.129135</td>\n",
       "      <td>8.013877</td>\n",
       "      <td>-500.517572</td>\n",
       "      <td>14.028663</td>\n",
       "      <td>-601.427363</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-15 03:00:00</th>\n",
       "      <td>6.047977</td>\n",
       "      <td>9.977019</td>\n",
       "      <td>4.858634</td>\n",
       "      <td>42.689819</td>\n",
       "      <td>68.347543</td>\n",
       "      <td>10.422762</td>\n",
       "      <td>0.883763</td>\n",
       "      <td>16.532835</td>\n",
       "      <td>2.146849</td>\n",
       "      <td>122.270188</td>\n",
       "      <td>...</td>\n",
       "      <td>14.036510</td>\n",
       "      <td>-500.857308</td>\n",
       "      <td>11.999550</td>\n",
       "      <td>-501.193686</td>\n",
       "      <td>9.970366</td>\n",
       "      <td>-499.201640</td>\n",
       "      <td>7.977324</td>\n",
       "      <td>-500.255908</td>\n",
       "      <td>14.005551</td>\n",
       "      <td>-599.996129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-15 04:00:00</th>\n",
       "      <td>6.148599</td>\n",
       "      <td>10.142511</td>\n",
       "      <td>4.939416</td>\n",
       "      <td>42.774141</td>\n",
       "      <td>66.927016</td>\n",
       "      <td>10.360302</td>\n",
       "      <td>0.792826</td>\n",
       "      <td>16.525686</td>\n",
       "      <td>2.055292</td>\n",
       "      <td>117.988169</td>\n",
       "      <td>...</td>\n",
       "      <td>14.027298</td>\n",
       "      <td>-499.838632</td>\n",
       "      <td>11.953070</td>\n",
       "      <td>-501.053894</td>\n",
       "      <td>9.925709</td>\n",
       "      <td>-501.686727</td>\n",
       "      <td>7.894242</td>\n",
       "      <td>-500.356035</td>\n",
       "      <td>13.996647</td>\n",
       "      <td>-601.496691</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 86 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                     final.output.concentrate_ag  final.output.concentrate_pb  \\\n",
       "date                                                                            \n",
       "2016-01-15 00:00:00                     6.055403                     9.889648   \n",
       "2016-01-15 01:00:00                     6.029369                     9.968944   \n",
       "2016-01-15 02:00:00                     6.055926                    10.213995   \n",
       "2016-01-15 03:00:00                     6.047977                     9.977019   \n",
       "2016-01-15 04:00:00                     6.148599                    10.142511   \n",
       "\n",
       "                     final.output.concentrate_sol  \\\n",
       "date                                                \n",
       "2016-01-15 00:00:00                      5.507324   \n",
       "2016-01-15 01:00:00                      5.257781   \n",
       "2016-01-15 02:00:00                      5.383759   \n",
       "2016-01-15 03:00:00                      4.858634   \n",
       "2016-01-15 04:00:00                      4.939416   \n",
       "\n",
       "                     final.output.concentrate_au  final.output.recovery  \\\n",
       "date                                                                      \n",
       "2016-01-15 00:00:00                    42.192020              70.541216   \n",
       "2016-01-15 01:00:00                    42.701629              69.266198   \n",
       "2016-01-15 02:00:00                    42.657501              68.116445   \n",
       "2016-01-15 03:00:00                    42.689819              68.347543   \n",
       "2016-01-15 04:00:00                    42.774141              66.927016   \n",
       "\n",
       "                     final.output.tail_ag  final.output.tail_pb  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00             10.411962              0.895447   \n",
       "2016-01-15 01:00:00             10.462676              0.927452   \n",
       "2016-01-15 02:00:00             10.507046              0.953716   \n",
       "2016-01-15 03:00:00             10.422762              0.883763   \n",
       "2016-01-15 04:00:00             10.360302              0.792826   \n",
       "\n",
       "                     final.output.tail_sol  final.output.tail_au  \\\n",
       "date                                                               \n",
       "2016-01-15 00:00:00              16.904297              2.143149   \n",
       "2016-01-15 01:00:00              16.634514              2.224930   \n",
       "2016-01-15 02:00:00              16.208849              2.257889   \n",
       "2016-01-15 03:00:00              16.532835              2.146849   \n",
       "2016-01-15 04:00:00              16.525686              2.055292   \n",
       "\n",
       "                     primary_cleaner.input.sulfate  ...  \\\n",
       "date                                                ...   \n",
       "2016-01-15 00:00:00                     127.092003  ...   \n",
       "2016-01-15 01:00:00                     125.629232  ...   \n",
       "2016-01-15 02:00:00                     123.819808  ...   \n",
       "2016-01-15 03:00:00                     122.270188  ...   \n",
       "2016-01-15 04:00:00                     117.988169  ...   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_a_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                 14.016835   \n",
       "2016-01-15 01:00:00                                 13.992281   \n",
       "2016-01-15 02:00:00                                 14.015015   \n",
       "2016-01-15 03:00:00                                 14.036510   \n",
       "2016-01-15 04:00:00                                 14.027298   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_a_level  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00                                 -502.488007   \n",
       "2016-01-15 01:00:00                                 -505.503262   \n",
       "2016-01-15 02:00:00                                 -502.520901   \n",
       "2016-01-15 03:00:00                                 -500.857308   \n",
       "2016-01-15 04:00:00                                 -499.838632   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_b_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                 12.099931   \n",
       "2016-01-15 01:00:00                                 11.950531   \n",
       "2016-01-15 02:00:00                                 11.912783   \n",
       "2016-01-15 03:00:00                                 11.999550   \n",
       "2016-01-15 04:00:00                                 11.953070   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_b_level  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00                                 -504.715942   \n",
       "2016-01-15 01:00:00                                 -501.331529   \n",
       "2016-01-15 02:00:00                                 -501.133383   \n",
       "2016-01-15 03:00:00                                 -501.193686   \n",
       "2016-01-15 04:00:00                                 -501.053894   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_a_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                  9.925633   \n",
       "2016-01-15 01:00:00                                 10.039245   \n",
       "2016-01-15 02:00:00                                 10.070913   \n",
       "2016-01-15 03:00:00                                  9.970366   \n",
       "2016-01-15 04:00:00                                  9.925709   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_a_level  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00                                 -498.310211   \n",
       "2016-01-15 01:00:00                                 -500.169983   \n",
       "2016-01-15 02:00:00                                 -500.129135   \n",
       "2016-01-15 03:00:00                                 -499.201640   \n",
       "2016-01-15 04:00:00                                 -501.686727   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_b_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                  8.079666   \n",
       "2016-01-15 01:00:00                                  7.984757   \n",
       "2016-01-15 02:00:00                                  8.013877   \n",
       "2016-01-15 03:00:00                                  7.977324   \n",
       "2016-01-15 04:00:00                                  7.894242   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_b_level  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00                                 -500.470978   \n",
       "2016-01-15 01:00:00                                 -500.582168   \n",
       "2016-01-15 02:00:00                                 -500.517572   \n",
       "2016-01-15 03:00:00                                 -500.255908   \n",
       "2016-01-15 04:00:00                                 -500.356035   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank6_a_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                 14.151341   \n",
       "2016-01-15 01:00:00                                 13.998353   \n",
       "2016-01-15 02:00:00                                 14.028663   \n",
       "2016-01-15 03:00:00                                 14.005551   \n",
       "2016-01-15 04:00:00                                 13.996647   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank6_a_level  \n",
       "date                                                             \n",
       "2016-01-15 00:00:00                                 -605.841980  \n",
       "2016-01-15 01:00:00                                 -599.787184  \n",
       "2016-01-15 02:00:00                                 -601.427363  \n",
       "2016-01-15 03:00:00                                 -599.996129  \n",
       "2016-01-15 04:00:00                                 -601.496691  \n",
       "\n",
       "[5 rows x 86 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gold_recovery_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "cb596745-f0db-4dff-89ca-4cad12db1a58",
   "metadata": {},
   "outputs": [
    {
     "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>primary_cleaner.input.sulfate</th>\n",
       "      <th>primary_cleaner.input.depressant</th>\n",
       "      <th>primary_cleaner.input.feed_size</th>\n",
       "      <th>primary_cleaner.input.xanthate</th>\n",
       "      <th>primary_cleaner.state.floatbank8_a_air</th>\n",
       "      <th>primary_cleaner.state.floatbank8_a_level</th>\n",
       "      <th>primary_cleaner.state.floatbank8_b_air</th>\n",
       "      <th>primary_cleaner.state.floatbank8_b_level</th>\n",
       "      <th>primary_cleaner.state.floatbank8_c_air</th>\n",
       "      <th>primary_cleaner.state.floatbank8_c_level</th>\n",
       "      <th>...</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_a_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_a_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_b_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_b_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_a_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_a_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_b_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_b_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank6_a_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank6_a_level</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2016-09-01 00:59:59</th>\n",
       "      <td>210.800909</td>\n",
       "      <td>14.993118</td>\n",
       "      <td>8.080000</td>\n",
       "      <td>1.005021</td>\n",
       "      <td>1398.981301</td>\n",
       "      <td>-500.225577</td>\n",
       "      <td>1399.144926</td>\n",
       "      <td>-499.919735</td>\n",
       "      <td>1400.102998</td>\n",
       "      <td>-500.704369</td>\n",
       "      <td>...</td>\n",
       "      <td>12.023554</td>\n",
       "      <td>-497.795834</td>\n",
       "      <td>8.016656</td>\n",
       "      <td>-501.289139</td>\n",
       "      <td>7.946562</td>\n",
       "      <td>-432.317850</td>\n",
       "      <td>4.872511</td>\n",
       "      <td>-500.037437</td>\n",
       "      <td>26.705889</td>\n",
       "      <td>-499.709414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-09-01 01:59:59</th>\n",
       "      <td>215.392455</td>\n",
       "      <td>14.987471</td>\n",
       "      <td>8.080000</td>\n",
       "      <td>0.990469</td>\n",
       "      <td>1398.777912</td>\n",
       "      <td>-500.057435</td>\n",
       "      <td>1398.055362</td>\n",
       "      <td>-499.778182</td>\n",
       "      <td>1396.151033</td>\n",
       "      <td>-499.240168</td>\n",
       "      <td>...</td>\n",
       "      <td>12.058140</td>\n",
       "      <td>-498.695773</td>\n",
       "      <td>8.130979</td>\n",
       "      <td>-499.634209</td>\n",
       "      <td>7.958270</td>\n",
       "      <td>-525.839648</td>\n",
       "      <td>4.878850</td>\n",
       "      <td>-500.162375</td>\n",
       "      <td>25.019940</td>\n",
       "      <td>-499.819438</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-09-01 02:59:59</th>\n",
       "      <td>215.259946</td>\n",
       "      <td>12.884934</td>\n",
       "      <td>7.786667</td>\n",
       "      <td>0.996043</td>\n",
       "      <td>1398.493666</td>\n",
       "      <td>-500.868360</td>\n",
       "      <td>1398.860436</td>\n",
       "      <td>-499.764529</td>\n",
       "      <td>1398.075709</td>\n",
       "      <td>-502.151509</td>\n",
       "      <td>...</td>\n",
       "      <td>11.962366</td>\n",
       "      <td>-498.767484</td>\n",
       "      <td>8.096893</td>\n",
       "      <td>-500.827423</td>\n",
       "      <td>8.071056</td>\n",
       "      <td>-500.801673</td>\n",
       "      <td>4.905125</td>\n",
       "      <td>-499.828510</td>\n",
       "      <td>24.994862</td>\n",
       "      <td>-500.622559</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-09-01 03:59:59</th>\n",
       "      <td>215.336236</td>\n",
       "      <td>12.006805</td>\n",
       "      <td>7.640000</td>\n",
       "      <td>0.863514</td>\n",
       "      <td>1399.618111</td>\n",
       "      <td>-498.863574</td>\n",
       "      <td>1397.440120</td>\n",
       "      <td>-499.211024</td>\n",
       "      <td>1400.129303</td>\n",
       "      <td>-498.355873</td>\n",
       "      <td>...</td>\n",
       "      <td>12.033091</td>\n",
       "      <td>-498.350935</td>\n",
       "      <td>8.074946</td>\n",
       "      <td>-499.474407</td>\n",
       "      <td>7.897085</td>\n",
       "      <td>-500.868509</td>\n",
       "      <td>4.931400</td>\n",
       "      <td>-499.963623</td>\n",
       "      <td>24.948919</td>\n",
       "      <td>-498.709987</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-09-01 04:59:59</th>\n",
       "      <td>199.099327</td>\n",
       "      <td>10.682530</td>\n",
       "      <td>7.530000</td>\n",
       "      <td>0.805575</td>\n",
       "      <td>1401.268123</td>\n",
       "      <td>-500.808305</td>\n",
       "      <td>1398.128818</td>\n",
       "      <td>-499.504543</td>\n",
       "      <td>1402.172226</td>\n",
       "      <td>-500.810606</td>\n",
       "      <td>...</td>\n",
       "      <td>12.025367</td>\n",
       "      <td>-500.786497</td>\n",
       "      <td>8.054678</td>\n",
       "      <td>-500.397500</td>\n",
       "      <td>8.107890</td>\n",
       "      <td>-509.526725</td>\n",
       "      <td>4.957674</td>\n",
       "      <td>-500.360026</td>\n",
       "      <td>25.003331</td>\n",
       "      <td>-500.856333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 52 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                     primary_cleaner.input.sulfate  \\\n",
       "date                                                 \n",
       "2016-09-01 00:59:59                     210.800909   \n",
       "2016-09-01 01:59:59                     215.392455   \n",
       "2016-09-01 02:59:59                     215.259946   \n",
       "2016-09-01 03:59:59                     215.336236   \n",
       "2016-09-01 04:59:59                     199.099327   \n",
       "\n",
       "                     primary_cleaner.input.depressant  \\\n",
       "date                                                    \n",
       "2016-09-01 00:59:59                         14.993118   \n",
       "2016-09-01 01:59:59                         14.987471   \n",
       "2016-09-01 02:59:59                         12.884934   \n",
       "2016-09-01 03:59:59                         12.006805   \n",
       "2016-09-01 04:59:59                         10.682530   \n",
       "\n",
       "                     primary_cleaner.input.feed_size  \\\n",
       "date                                                   \n",
       "2016-09-01 00:59:59                         8.080000   \n",
       "2016-09-01 01:59:59                         8.080000   \n",
       "2016-09-01 02:59:59                         7.786667   \n",
       "2016-09-01 03:59:59                         7.640000   \n",
       "2016-09-01 04:59:59                         7.530000   \n",
       "\n",
       "                     primary_cleaner.input.xanthate  \\\n",
       "date                                                  \n",
       "2016-09-01 00:59:59                        1.005021   \n",
       "2016-09-01 01:59:59                        0.990469   \n",
       "2016-09-01 02:59:59                        0.996043   \n",
       "2016-09-01 03:59:59                        0.863514   \n",
       "2016-09-01 04:59:59                        0.805575   \n",
       "\n",
       "                     primary_cleaner.state.floatbank8_a_air  \\\n",
       "date                                                          \n",
       "2016-09-01 00:59:59                             1398.981301   \n",
       "2016-09-01 01:59:59                             1398.777912   \n",
       "2016-09-01 02:59:59                             1398.493666   \n",
       "2016-09-01 03:59:59                             1399.618111   \n",
       "2016-09-01 04:59:59                             1401.268123   \n",
       "\n",
       "                     primary_cleaner.state.floatbank8_a_level  \\\n",
       "date                                                            \n",
       "2016-09-01 00:59:59                               -500.225577   \n",
       "2016-09-01 01:59:59                               -500.057435   \n",
       "2016-09-01 02:59:59                               -500.868360   \n",
       "2016-09-01 03:59:59                               -498.863574   \n",
       "2016-09-01 04:59:59                               -500.808305   \n",
       "\n",
       "                     primary_cleaner.state.floatbank8_b_air  \\\n",
       "date                                                          \n",
       "2016-09-01 00:59:59                             1399.144926   \n",
       "2016-09-01 01:59:59                             1398.055362   \n",
       "2016-09-01 02:59:59                             1398.860436   \n",
       "2016-09-01 03:59:59                             1397.440120   \n",
       "2016-09-01 04:59:59                             1398.128818   \n",
       "\n",
       "                     primary_cleaner.state.floatbank8_b_level  \\\n",
       "date                                                            \n",
       "2016-09-01 00:59:59                               -499.919735   \n",
       "2016-09-01 01:59:59                               -499.778182   \n",
       "2016-09-01 02:59:59                               -499.764529   \n",
       "2016-09-01 03:59:59                               -499.211024   \n",
       "2016-09-01 04:59:59                               -499.504543   \n",
       "\n",
       "                     primary_cleaner.state.floatbank8_c_air  \\\n",
       "date                                                          \n",
       "2016-09-01 00:59:59                             1400.102998   \n",
       "2016-09-01 01:59:59                             1396.151033   \n",
       "2016-09-01 02:59:59                             1398.075709   \n",
       "2016-09-01 03:59:59                             1400.129303   \n",
       "2016-09-01 04:59:59                             1402.172226   \n",
       "\n",
       "                     primary_cleaner.state.floatbank8_c_level  ...  \\\n",
       "date                                                           ...   \n",
       "2016-09-01 00:59:59                               -500.704369  ...   \n",
       "2016-09-01 01:59:59                               -499.240168  ...   \n",
       "2016-09-01 02:59:59                               -502.151509  ...   \n",
       "2016-09-01 03:59:59                               -498.355873  ...   \n",
       "2016-09-01 04:59:59                               -500.810606  ...   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_a_air  \\\n",
       "date                                                            \n",
       "2016-09-01 00:59:59                                 12.023554   \n",
       "2016-09-01 01:59:59                                 12.058140   \n",
       "2016-09-01 02:59:59                                 11.962366   \n",
       "2016-09-01 03:59:59                                 12.033091   \n",
       "2016-09-01 04:59:59                                 12.025367   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_a_level  \\\n",
       "date                                                              \n",
       "2016-09-01 00:59:59                                 -497.795834   \n",
       "2016-09-01 01:59:59                                 -498.695773   \n",
       "2016-09-01 02:59:59                                 -498.767484   \n",
       "2016-09-01 03:59:59                                 -498.350935   \n",
       "2016-09-01 04:59:59                                 -500.786497   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_b_air  \\\n",
       "date                                                            \n",
       "2016-09-01 00:59:59                                  8.016656   \n",
       "2016-09-01 01:59:59                                  8.130979   \n",
       "2016-09-01 02:59:59                                  8.096893   \n",
       "2016-09-01 03:59:59                                  8.074946   \n",
       "2016-09-01 04:59:59                                  8.054678   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_b_level  \\\n",
       "date                                                              \n",
       "2016-09-01 00:59:59                                 -501.289139   \n",
       "2016-09-01 01:59:59                                 -499.634209   \n",
       "2016-09-01 02:59:59                                 -500.827423   \n",
       "2016-09-01 03:59:59                                 -499.474407   \n",
       "2016-09-01 04:59:59                                 -500.397500   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_a_air  \\\n",
       "date                                                            \n",
       "2016-09-01 00:59:59                                  7.946562   \n",
       "2016-09-01 01:59:59                                  7.958270   \n",
       "2016-09-01 02:59:59                                  8.071056   \n",
       "2016-09-01 03:59:59                                  7.897085   \n",
       "2016-09-01 04:59:59                                  8.107890   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_a_level  \\\n",
       "date                                                              \n",
       "2016-09-01 00:59:59                                 -432.317850   \n",
       "2016-09-01 01:59:59                                 -525.839648   \n",
       "2016-09-01 02:59:59                                 -500.801673   \n",
       "2016-09-01 03:59:59                                 -500.868509   \n",
       "2016-09-01 04:59:59                                 -509.526725   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_b_air  \\\n",
       "date                                                            \n",
       "2016-09-01 00:59:59                                  4.872511   \n",
       "2016-09-01 01:59:59                                  4.878850   \n",
       "2016-09-01 02:59:59                                  4.905125   \n",
       "2016-09-01 03:59:59                                  4.931400   \n",
       "2016-09-01 04:59:59                                  4.957674   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_b_level  \\\n",
       "date                                                              \n",
       "2016-09-01 00:59:59                                 -500.037437   \n",
       "2016-09-01 01:59:59                                 -500.162375   \n",
       "2016-09-01 02:59:59                                 -499.828510   \n",
       "2016-09-01 03:59:59                                 -499.963623   \n",
       "2016-09-01 04:59:59                                 -500.360026   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank6_a_air  \\\n",
       "date                                                            \n",
       "2016-09-01 00:59:59                                 26.705889   \n",
       "2016-09-01 01:59:59                                 25.019940   \n",
       "2016-09-01 02:59:59                                 24.994862   \n",
       "2016-09-01 03:59:59                                 24.948919   \n",
       "2016-09-01 04:59:59                                 25.003331   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank6_a_level  \n",
       "date                                                             \n",
       "2016-09-01 00:59:59                                 -499.709414  \n",
       "2016-09-01 01:59:59                                 -499.819438  \n",
       "2016-09-01 02:59:59                                 -500.622559  \n",
       "2016-09-01 03:59:59                                 -498.709987  \n",
       "2016-09-01 04:59:59                                 -500.856333  \n",
       "\n",
       "[5 rows x 52 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gold_recovery_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "d11102c0-28fc-4fed-8733-4d0878a4d4b4",
   "metadata": {},
   "outputs": [
    {
     "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>final.output.concentrate_ag</th>\n",
       "      <th>final.output.concentrate_pb</th>\n",
       "      <th>final.output.concentrate_sol</th>\n",
       "      <th>final.output.concentrate_au</th>\n",
       "      <th>final.output.recovery</th>\n",
       "      <th>final.output.tail_ag</th>\n",
       "      <th>final.output.tail_pb</th>\n",
       "      <th>final.output.tail_sol</th>\n",
       "      <th>final.output.tail_au</th>\n",
       "      <th>primary_cleaner.input.sulfate</th>\n",
       "      <th>...</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_a_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_a_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_b_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank4_b_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_a_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_a_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_b_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank5_b_level</th>\n",
       "      <th>secondary_cleaner.state.floatbank6_a_air</th>\n",
       "      <th>secondary_cleaner.state.floatbank6_a_level</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2016-01-15 00:00:00</th>\n",
       "      <td>6.055403</td>\n",
       "      <td>9.889648</td>\n",
       "      <td>5.507324</td>\n",
       "      <td>42.192020</td>\n",
       "      <td>70.541216</td>\n",
       "      <td>10.411962</td>\n",
       "      <td>0.895447</td>\n",
       "      <td>16.904297</td>\n",
       "      <td>2.143149</td>\n",
       "      <td>127.092003</td>\n",
       "      <td>...</td>\n",
       "      <td>14.016835</td>\n",
       "      <td>-502.488007</td>\n",
       "      <td>12.099931</td>\n",
       "      <td>-504.715942</td>\n",
       "      <td>9.925633</td>\n",
       "      <td>-498.310211</td>\n",
       "      <td>8.079666</td>\n",
       "      <td>-500.470978</td>\n",
       "      <td>14.151341</td>\n",
       "      <td>-605.841980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-15 01:00:00</th>\n",
       "      <td>6.029369</td>\n",
       "      <td>9.968944</td>\n",
       "      <td>5.257781</td>\n",
       "      <td>42.701629</td>\n",
       "      <td>69.266198</td>\n",
       "      <td>10.462676</td>\n",
       "      <td>0.927452</td>\n",
       "      <td>16.634514</td>\n",
       "      <td>2.224930</td>\n",
       "      <td>125.629232</td>\n",
       "      <td>...</td>\n",
       "      <td>13.992281</td>\n",
       "      <td>-505.503262</td>\n",
       "      <td>11.950531</td>\n",
       "      <td>-501.331529</td>\n",
       "      <td>10.039245</td>\n",
       "      <td>-500.169983</td>\n",
       "      <td>7.984757</td>\n",
       "      <td>-500.582168</td>\n",
       "      <td>13.998353</td>\n",
       "      <td>-599.787184</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-15 02:00:00</th>\n",
       "      <td>6.055926</td>\n",
       "      <td>10.213995</td>\n",
       "      <td>5.383759</td>\n",
       "      <td>42.657501</td>\n",
       "      <td>68.116445</td>\n",
       "      <td>10.507046</td>\n",
       "      <td>0.953716</td>\n",
       "      <td>16.208849</td>\n",
       "      <td>2.257889</td>\n",
       "      <td>123.819808</td>\n",
       "      <td>...</td>\n",
       "      <td>14.015015</td>\n",
       "      <td>-502.520901</td>\n",
       "      <td>11.912783</td>\n",
       "      <td>-501.133383</td>\n",
       "      <td>10.070913</td>\n",
       "      <td>-500.129135</td>\n",
       "      <td>8.013877</td>\n",
       "      <td>-500.517572</td>\n",
       "      <td>14.028663</td>\n",
       "      <td>-601.427363</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-15 03:00:00</th>\n",
       "      <td>6.047977</td>\n",
       "      <td>9.977019</td>\n",
       "      <td>4.858634</td>\n",
       "      <td>42.689819</td>\n",
       "      <td>68.347543</td>\n",
       "      <td>10.422762</td>\n",
       "      <td>0.883763</td>\n",
       "      <td>16.532835</td>\n",
       "      <td>2.146849</td>\n",
       "      <td>122.270188</td>\n",
       "      <td>...</td>\n",
       "      <td>14.036510</td>\n",
       "      <td>-500.857308</td>\n",
       "      <td>11.999550</td>\n",
       "      <td>-501.193686</td>\n",
       "      <td>9.970366</td>\n",
       "      <td>-499.201640</td>\n",
       "      <td>7.977324</td>\n",
       "      <td>-500.255908</td>\n",
       "      <td>14.005551</td>\n",
       "      <td>-599.996129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-15 04:00:00</th>\n",
       "      <td>6.148599</td>\n",
       "      <td>10.142511</td>\n",
       "      <td>4.939416</td>\n",
       "      <td>42.774141</td>\n",
       "      <td>66.927016</td>\n",
       "      <td>10.360302</td>\n",
       "      <td>0.792826</td>\n",
       "      <td>16.525686</td>\n",
       "      <td>2.055292</td>\n",
       "      <td>117.988169</td>\n",
       "      <td>...</td>\n",
       "      <td>14.027298</td>\n",
       "      <td>-499.838632</td>\n",
       "      <td>11.953070</td>\n",
       "      <td>-501.053894</td>\n",
       "      <td>9.925709</td>\n",
       "      <td>-501.686727</td>\n",
       "      <td>7.894242</td>\n",
       "      <td>-500.356035</td>\n",
       "      <td>13.996647</td>\n",
       "      <td>-601.496691</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 86 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                     final.output.concentrate_ag  final.output.concentrate_pb  \\\n",
       "date                                                                            \n",
       "2016-01-15 00:00:00                     6.055403                     9.889648   \n",
       "2016-01-15 01:00:00                     6.029369                     9.968944   \n",
       "2016-01-15 02:00:00                     6.055926                    10.213995   \n",
       "2016-01-15 03:00:00                     6.047977                     9.977019   \n",
       "2016-01-15 04:00:00                     6.148599                    10.142511   \n",
       "\n",
       "                     final.output.concentrate_sol  \\\n",
       "date                                                \n",
       "2016-01-15 00:00:00                      5.507324   \n",
       "2016-01-15 01:00:00                      5.257781   \n",
       "2016-01-15 02:00:00                      5.383759   \n",
       "2016-01-15 03:00:00                      4.858634   \n",
       "2016-01-15 04:00:00                      4.939416   \n",
       "\n",
       "                     final.output.concentrate_au  final.output.recovery  \\\n",
       "date                                                                      \n",
       "2016-01-15 00:00:00                    42.192020              70.541216   \n",
       "2016-01-15 01:00:00                    42.701629              69.266198   \n",
       "2016-01-15 02:00:00                    42.657501              68.116445   \n",
       "2016-01-15 03:00:00                    42.689819              68.347543   \n",
       "2016-01-15 04:00:00                    42.774141              66.927016   \n",
       "\n",
       "                     final.output.tail_ag  final.output.tail_pb  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00             10.411962              0.895447   \n",
       "2016-01-15 01:00:00             10.462676              0.927452   \n",
       "2016-01-15 02:00:00             10.507046              0.953716   \n",
       "2016-01-15 03:00:00             10.422762              0.883763   \n",
       "2016-01-15 04:00:00             10.360302              0.792826   \n",
       "\n",
       "                     final.output.tail_sol  final.output.tail_au  \\\n",
       "date                                                               \n",
       "2016-01-15 00:00:00              16.904297              2.143149   \n",
       "2016-01-15 01:00:00              16.634514              2.224930   \n",
       "2016-01-15 02:00:00              16.208849              2.257889   \n",
       "2016-01-15 03:00:00              16.532835              2.146849   \n",
       "2016-01-15 04:00:00              16.525686              2.055292   \n",
       "\n",
       "                     primary_cleaner.input.sulfate  ...  \\\n",
       "date                                                ...   \n",
       "2016-01-15 00:00:00                     127.092003  ...   \n",
       "2016-01-15 01:00:00                     125.629232  ...   \n",
       "2016-01-15 02:00:00                     123.819808  ...   \n",
       "2016-01-15 03:00:00                     122.270188  ...   \n",
       "2016-01-15 04:00:00                     117.988169  ...   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_a_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                 14.016835   \n",
       "2016-01-15 01:00:00                                 13.992281   \n",
       "2016-01-15 02:00:00                                 14.015015   \n",
       "2016-01-15 03:00:00                                 14.036510   \n",
       "2016-01-15 04:00:00                                 14.027298   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_a_level  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00                                 -502.488007   \n",
       "2016-01-15 01:00:00                                 -505.503262   \n",
       "2016-01-15 02:00:00                                 -502.520901   \n",
       "2016-01-15 03:00:00                                 -500.857308   \n",
       "2016-01-15 04:00:00                                 -499.838632   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_b_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                 12.099931   \n",
       "2016-01-15 01:00:00                                 11.950531   \n",
       "2016-01-15 02:00:00                                 11.912783   \n",
       "2016-01-15 03:00:00                                 11.999550   \n",
       "2016-01-15 04:00:00                                 11.953070   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank4_b_level  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00                                 -504.715942   \n",
       "2016-01-15 01:00:00                                 -501.331529   \n",
       "2016-01-15 02:00:00                                 -501.133383   \n",
       "2016-01-15 03:00:00                                 -501.193686   \n",
       "2016-01-15 04:00:00                                 -501.053894   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_a_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                  9.925633   \n",
       "2016-01-15 01:00:00                                 10.039245   \n",
       "2016-01-15 02:00:00                                 10.070913   \n",
       "2016-01-15 03:00:00                                  9.970366   \n",
       "2016-01-15 04:00:00                                  9.925709   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_a_level  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00                                 -498.310211   \n",
       "2016-01-15 01:00:00                                 -500.169983   \n",
       "2016-01-15 02:00:00                                 -500.129135   \n",
       "2016-01-15 03:00:00                                 -499.201640   \n",
       "2016-01-15 04:00:00                                 -501.686727   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_b_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                  8.079666   \n",
       "2016-01-15 01:00:00                                  7.984757   \n",
       "2016-01-15 02:00:00                                  8.013877   \n",
       "2016-01-15 03:00:00                                  7.977324   \n",
       "2016-01-15 04:00:00                                  7.894242   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank5_b_level  \\\n",
       "date                                                              \n",
       "2016-01-15 00:00:00                                 -500.470978   \n",
       "2016-01-15 01:00:00                                 -500.582168   \n",
       "2016-01-15 02:00:00                                 -500.517572   \n",
       "2016-01-15 03:00:00                                 -500.255908   \n",
       "2016-01-15 04:00:00                                 -500.356035   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank6_a_air  \\\n",
       "date                                                            \n",
       "2016-01-15 00:00:00                                 14.151341   \n",
       "2016-01-15 01:00:00                                 13.998353   \n",
       "2016-01-15 02:00:00                                 14.028663   \n",
       "2016-01-15 03:00:00                                 14.005551   \n",
       "2016-01-15 04:00:00                                 13.996647   \n",
       "\n",
       "                     secondary_cleaner.state.floatbank6_a_level  \n",
       "date                                                             \n",
       "2016-01-15 00:00:00                                 -605.841980  \n",
       "2016-01-15 01:00:00                                 -599.787184  \n",
       "2016-01-15 02:00:00                                 -601.427363  \n",
       "2016-01-15 03:00:00                                 -599.996129  \n",
       "2016-01-15 04:00:00                                 -601.496691  \n",
       "\n",
       "[5 rows x 86 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gold_recovery_full.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a4d2090b-ad1a-4757-83b9-6ac649ab27f0",
   "metadata": {},
   "source": [
    "We notice a fairly large number of columns in all DataFrames and looking at these small samples of 5 observations in each, it is difficult to say anything specific about the data at hand. Let's use `info()` method to get a concise summary of the data, namely, information on column names, data types and the number of `non-null` values in each of the columns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "57a3bbb6-ace2-440a-92e9-32f88425fd0a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "DatetimeIndex: 14149 entries, 2016-01-15 00:00:00 to 2018-08-18 10:59:59\n",
      "Data columns (total 86 columns):\n",
      " #   Column                                              Non-Null Count  Dtype  \n",
      "---  ------                                              --------------  -----  \n",
      " 0   final.output.concentrate_ag                         14148 non-null  float64\n",
      " 1   final.output.concentrate_pb                         14148 non-null  float64\n",
      " 2   final.output.concentrate_sol                        13938 non-null  float64\n",
      " 3   final.output.concentrate_au                         14149 non-null  float64\n",
      " 4   final.output.recovery                               14149 non-null  float64\n",
      " 5   final.output.tail_ag                                14149 non-null  float64\n",
      " 6   final.output.tail_pb                                14049 non-null  float64\n",
      " 7   final.output.tail_sol                               14144 non-null  float64\n",
      " 8   final.output.tail_au                                14149 non-null  float64\n",
      " 9   primary_cleaner.input.sulfate                       14129 non-null  float64\n",
      " 10  primary_cleaner.input.depressant                    14117 non-null  float64\n",
      " 11  primary_cleaner.input.feed_size                     14149 non-null  float64\n",
      " 12  primary_cleaner.input.xanthate                      14049 non-null  float64\n",
      " 13  primary_cleaner.output.concentrate_ag               14149 non-null  float64\n",
      " 14  primary_cleaner.output.concentrate_pb               14063 non-null  float64\n",
      " 15  primary_cleaner.output.concentrate_sol              13863 non-null  float64\n",
      " 16  primary_cleaner.output.concentrate_au               14149 non-null  float64\n",
      " 17  primary_cleaner.output.tail_ag                      14148 non-null  float64\n",
      " 18  primary_cleaner.output.tail_pb                      14134 non-null  float64\n",
      " 19  primary_cleaner.output.tail_sol                     14103 non-null  float64\n",
      " 20  primary_cleaner.output.tail_au                      14149 non-null  float64\n",
      " 21  primary_cleaner.state.floatbank8_a_air              14145 non-null  float64\n",
      " 22  primary_cleaner.state.floatbank8_a_level            14148 non-null  float64\n",
      " 23  primary_cleaner.state.floatbank8_b_air              14145 non-null  float64\n",
      " 24  primary_cleaner.state.floatbank8_b_level            14148 non-null  float64\n",
      " 25  primary_cleaner.state.floatbank8_c_air              14147 non-null  float64\n",
      " 26  primary_cleaner.state.floatbank8_c_level            14148 non-null  float64\n",
      " 27  primary_cleaner.state.floatbank8_d_air              14146 non-null  float64\n",
      " 28  primary_cleaner.state.floatbank8_d_level            14148 non-null  float64\n",
      " 29  rougher.calculation.sulfate_to_au_concentrate       14148 non-null  float64\n",
      " 30  rougher.calculation.floatbank10_sulfate_to_au_feed  14148 non-null  float64\n",
      " 31  rougher.calculation.floatbank11_sulfate_to_au_feed  14148 non-null  float64\n",
      " 32  rougher.calculation.au_pb_ratio                     14149 non-null  float64\n",
      " 33  rougher.input.feed_ag                               14149 non-null  float64\n",
      " 34  rougher.input.feed_pb                               14049 non-null  float64\n",
      " 35  rougher.input.feed_rate                             14141 non-null  float64\n",
      " 36  rougher.input.feed_size                             14005 non-null  float64\n",
      " 37  rougher.input.feed_sol                              14071 non-null  float64\n",
      " 38  rougher.input.feed_au                               14149 non-null  float64\n",
      " 39  rougher.input.floatbank10_sulfate                   14120 non-null  float64\n",
      " 40  rougher.input.floatbank10_xanthate                  14141 non-null  float64\n",
      " 41  rougher.input.floatbank11_sulfate                   14113 non-null  float64\n",
      " 42  rougher.input.floatbank11_xanthate                  13721 non-null  float64\n",
      " 43  rougher.output.concentrate_ag                       14149 non-null  float64\n",
      " 44  rougher.output.concentrate_pb                       14149 non-null  float64\n",
      " 45  rougher.output.concentrate_sol                      14127 non-null  float64\n",
      " 46  rougher.output.concentrate_au                       14149 non-null  float64\n",
      " 47  rougher.output.recovery                             14149 non-null  float64\n",
      " 48  rougher.output.tail_ag                              14148 non-null  float64\n",
      " 49  rougher.output.tail_pb                              14149 non-null  float64\n",
      " 50  rougher.output.tail_sol                             14149 non-null  float64\n",
      " 51  rougher.output.tail_au                              14149 non-null  float64\n",
      " 52  rougher.state.floatbank10_a_air                     14148 non-null  float64\n",
      " 53  rougher.state.floatbank10_a_level                   14148 non-null  float64\n",
      " 54  rougher.state.floatbank10_b_air                     14148 non-null  float64\n",
      " 55  rougher.state.floatbank10_b_level                   14148 non-null  float64\n",
      " 56  rougher.state.floatbank10_c_air                     14148 non-null  float64\n",
      " 57  rougher.state.floatbank10_c_level                   14148 non-null  float64\n",
      " 58  rougher.state.floatbank10_d_air                     14149 non-null  float64\n",
      " 59  rougher.state.floatbank10_d_level                   14149 non-null  float64\n",
      " 60  rougher.state.floatbank10_e_air                     13713 non-null  float64\n",
      " 61  rougher.state.floatbank10_e_level                   14149 non-null  float64\n",
      " 62  rougher.state.floatbank10_f_air                     14149 non-null  float64\n",
      " 63  rougher.state.floatbank10_f_level                   14149 non-null  float64\n",
      " 64  secondary_cleaner.output.tail_ag                    14147 non-null  float64\n",
      " 65  secondary_cleaner.output.tail_pb                    14139 non-null  float64\n",
      " 66  secondary_cleaner.output.tail_sol                   12544 non-null  float64\n",
      " 67  secondary_cleaner.output.tail_au                    14149 non-null  float64\n",
      " 68  secondary_cleaner.state.floatbank2_a_air            13932 non-null  float64\n",
      " 69  secondary_cleaner.state.floatbank2_a_level          14148 non-null  float64\n",
      " 70  secondary_cleaner.state.floatbank2_b_air            14128 non-null  float64\n",
      " 71  secondary_cleaner.state.floatbank2_b_level          14148 non-null  float64\n",
      " 72  secondary_cleaner.state.floatbank3_a_air            14145 non-null  float64\n",
      " 73  secondary_cleaner.state.floatbank3_a_level          14148 non-null  float64\n",
      " 74  secondary_cleaner.state.floatbank3_b_air            14148 non-null  float64\n",
      " 75  secondary_cleaner.state.floatbank3_b_level          14148 non-null  float64\n",
      " 76  secondary_cleaner.state.floatbank4_a_air            14143 non-null  float64\n",
      " 77  secondary_cleaner.state.floatbank4_a_level          14148 non-null  float64\n",
      " 78  secondary_cleaner.state.floatbank4_b_air            14148 non-null  float64\n",
      " 79  secondary_cleaner.state.floatbank4_b_level          14148 non-null  float64\n",
      " 80  secondary_cleaner.state.floatbank5_a_air            14148 non-null  float64\n",
      " 81  secondary_cleaner.state.floatbank5_a_level          14148 non-null  float64\n",
      " 82  secondary_cleaner.state.floatbank5_b_air            14148 non-null  float64\n",
      " 83  secondary_cleaner.state.floatbank5_b_level          14148 non-null  float64\n",
      " 84  secondary_cleaner.state.floatbank6_a_air            14147 non-null  float64\n",
      " 85  secondary_cleaner.state.floatbank6_a_level          14148 non-null  float64\n",
      "dtypes: float64(86)\n",
      "memory usage: 9.4 MB\n"
     ]
    }
   ],
   "source": [
    "gold_recovery_train.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "a65a2690-995c-42e2-8342-0c00c6f63a6b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "DatetimeIndex: 5290 entries, 2016-09-01 00:59:59 to 2017-12-31 23:59:59\n",
      "Data columns (total 52 columns):\n",
      " #   Column                                      Non-Null Count  Dtype  \n",
      "---  ------                                      --------------  -----  \n",
      " 0   primary_cleaner.input.sulfate               5286 non-null   float64\n",
      " 1   primary_cleaner.input.depressant            5285 non-null   float64\n",
      " 2   primary_cleaner.input.feed_size             5290 non-null   float64\n",
      " 3   primary_cleaner.input.xanthate              5286 non-null   float64\n",
      " 4   primary_cleaner.state.floatbank8_a_air      5290 non-null   float64\n",
      " 5   primary_cleaner.state.floatbank8_a_level    5290 non-null   float64\n",
      " 6   primary_cleaner.state.floatbank8_b_air      5290 non-null   float64\n",
      " 7   primary_cleaner.state.floatbank8_b_level    5290 non-null   float64\n",
      " 8   primary_cleaner.state.floatbank8_c_air      5290 non-null   float64\n",
      " 9   primary_cleaner.state.floatbank8_c_level    5290 non-null   float64\n",
      " 10  primary_cleaner.state.floatbank8_d_air      5290 non-null   float64\n",
      " 11  primary_cleaner.state.floatbank8_d_level    5290 non-null   float64\n",
      " 12  rougher.input.feed_ag                       5290 non-null   float64\n",
      " 13  rougher.input.feed_pb                       5290 non-null   float64\n",
      " 14  rougher.input.feed_rate                     5287 non-null   float64\n",
      " 15  rougher.input.feed_size                     5289 non-null   float64\n",
      " 16  rougher.input.feed_sol                      5269 non-null   float64\n",
      " 17  rougher.input.feed_au                       5290 non-null   float64\n",
      " 18  rougher.input.floatbank10_sulfate           5285 non-null   float64\n",
      " 19  rougher.input.floatbank10_xanthate          5290 non-null   float64\n",
      " 20  rougher.input.floatbank11_sulfate           5282 non-null   float64\n",
      " 21  rougher.input.floatbank11_xanthate          5265 non-null   float64\n",
      " 22  rougher.state.floatbank10_a_air             5290 non-null   float64\n",
      " 23  rougher.state.floatbank10_a_level           5290 non-null   float64\n",
      " 24  rougher.state.floatbank10_b_air             5290 non-null   float64\n",
      " 25  rougher.state.floatbank10_b_level           5290 non-null   float64\n",
      " 26  rougher.state.floatbank10_c_air             5290 non-null   float64\n",
      " 27  rougher.state.floatbank10_c_level           5290 non-null   float64\n",
      " 28  rougher.state.floatbank10_d_air             5290 non-null   float64\n",
      " 29  rougher.state.floatbank10_d_level           5290 non-null   float64\n",
      " 30  rougher.state.floatbank10_e_air             5290 non-null   float64\n",
      " 31  rougher.state.floatbank10_e_level           5290 non-null   float64\n",
      " 32  rougher.state.floatbank10_f_air             5290 non-null   float64\n",
      " 33  rougher.state.floatbank10_f_level           5290 non-null   float64\n",
      " 34  secondary_cleaner.state.floatbank2_a_air    5287 non-null   float64\n",
      " 35  secondary_cleaner.state.floatbank2_a_level  5290 non-null   float64\n",
      " 36  secondary_cleaner.state.floatbank2_b_air    5288 non-null   float64\n",
      " 37  secondary_cleaner.state.floatbank2_b_level  5290 non-null   float64\n",
      " 38  secondary_cleaner.state.floatbank3_a_air    5281 non-null   float64\n",
      " 39  secondary_cleaner.state.floatbank3_a_level  5290 non-null   float64\n",
      " 40  secondary_cleaner.state.floatbank3_b_air    5290 non-null   float64\n",
      " 41  secondary_cleaner.state.floatbank3_b_level  5290 non-null   float64\n",
      " 42  secondary_cleaner.state.floatbank4_a_air    5290 non-null   float64\n",
      " 43  secondary_cleaner.state.floatbank4_a_level  5290 non-null   float64\n",
      " 44  secondary_cleaner.state.floatbank4_b_air    5290 non-null   float64\n",
      " 45  secondary_cleaner.state.floatbank4_b_level  5290 non-null   float64\n",
      " 46  secondary_cleaner.state.floatbank5_a_air    5290 non-null   float64\n",
      " 47  secondary_cleaner.state.floatbank5_a_level  5290 non-null   float64\n",
      " 48  secondary_cleaner.state.floatbank5_b_air    5290 non-null   float64\n",
      " 49  secondary_cleaner.state.floatbank5_b_level  5290 non-null   float64\n",
      " 50  secondary_cleaner.state.floatbank6_a_air    5290 non-null   float64\n",
      " 51  secondary_cleaner.state.floatbank6_a_level  5290 non-null   float64\n",
      "dtypes: float64(52)\n",
      "memory usage: 2.1 MB\n"
     ]
    }
   ],
   "source": [
    "gold_recovery_test.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "7aa542b5-0a80-43f1-be49-adc2e29b6391",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "DatetimeIndex: 19439 entries, 2016-01-15 00:00:00 to 2018-08-18 10:59:59\n",
      "Data columns (total 86 columns):\n",
      " #   Column                                              Non-Null Count  Dtype  \n",
      "---  ------                                              --------------  -----  \n",
      " 0   final.output.concentrate_ag                         19438 non-null  float64\n",
      " 1   final.output.concentrate_pb                         19438 non-null  float64\n",
      " 2   final.output.concentrate_sol                        19228 non-null  float64\n",
      " 3   final.output.concentrate_au                         19439 non-null  float64\n",
      " 4   final.output.recovery                               19439 non-null  float64\n",
      " 5   final.output.tail_ag                                19438 non-null  float64\n",
      " 6   final.output.tail_pb                                19338 non-null  float64\n",
      " 7   final.output.tail_sol                               19433 non-null  float64\n",
      " 8   final.output.tail_au                                19439 non-null  float64\n",
      " 9   primary_cleaner.input.sulfate                       19415 non-null  float64\n",
      " 10  primary_cleaner.input.depressant                    19402 non-null  float64\n",
      " 11  primary_cleaner.input.feed_size                     19439 non-null  float64\n",
      " 12  primary_cleaner.input.xanthate                      19335 non-null  float64\n",
      " 13  primary_cleaner.output.concentrate_ag               19439 non-null  float64\n",
      " 14  primary_cleaner.output.concentrate_pb               19323 non-null  float64\n",
      " 15  primary_cleaner.output.concentrate_sol              19069 non-null  float64\n",
      " 16  primary_cleaner.output.concentrate_au               19439 non-null  float64\n",
      " 17  primary_cleaner.output.tail_ag                      19435 non-null  float64\n",
      " 18  primary_cleaner.output.tail_pb                      19418 non-null  float64\n",
      " 19  primary_cleaner.output.tail_sol                     19377 non-null  float64\n",
      " 20  primary_cleaner.output.tail_au                      19439 non-null  float64\n",
      " 21  primary_cleaner.state.floatbank8_a_air              19435 non-null  float64\n",
      " 22  primary_cleaner.state.floatbank8_a_level            19438 non-null  float64\n",
      " 23  primary_cleaner.state.floatbank8_b_air              19435 non-null  float64\n",
      " 24  primary_cleaner.state.floatbank8_b_level            19438 non-null  float64\n",
      " 25  primary_cleaner.state.floatbank8_c_air              19437 non-null  float64\n",
      " 26  primary_cleaner.state.floatbank8_c_level            19438 non-null  float64\n",
      " 27  primary_cleaner.state.floatbank8_d_air              19436 non-null  float64\n",
      " 28  primary_cleaner.state.floatbank8_d_level            19438 non-null  float64\n",
      " 29  rougher.calculation.sulfate_to_au_concentrate       19437 non-null  float64\n",
      " 30  rougher.calculation.floatbank10_sulfate_to_au_feed  19437 non-null  float64\n",
      " 31  rougher.calculation.floatbank11_sulfate_to_au_feed  19437 non-null  float64\n",
      " 32  rougher.calculation.au_pb_ratio                     19439 non-null  float64\n",
      " 33  rougher.input.feed_ag                               19439 non-null  float64\n",
      " 34  rougher.input.feed_pb                               19339 non-null  float64\n",
      " 35  rougher.input.feed_rate                             19428 non-null  float64\n",
      " 36  rougher.input.feed_size                             19294 non-null  float64\n",
      " 37  rougher.input.feed_sol                              19340 non-null  float64\n",
      " 38  rougher.input.feed_au                               19439 non-null  float64\n",
      " 39  rougher.input.floatbank10_sulfate                   19405 non-null  float64\n",
      " 40  rougher.input.floatbank10_xanthate                  19431 non-null  float64\n",
      " 41  rougher.input.floatbank11_sulfate                   19395 non-null  float64\n",
      " 42  rougher.input.floatbank11_xanthate                  18986 non-null  float64\n",
      " 43  rougher.output.concentrate_ag                       19439 non-null  float64\n",
      " 44  rougher.output.concentrate_pb                       19439 non-null  float64\n",
      " 45  rougher.output.concentrate_sol                      19416 non-null  float64\n",
      " 46  rougher.output.concentrate_au                       19439 non-null  float64\n",
      " 47  rougher.output.recovery                             19439 non-null  float64\n",
      " 48  rougher.output.tail_ag                              19438 non-null  float64\n",
      " 49  rougher.output.tail_pb                              19439 non-null  float64\n",
      " 50  rougher.output.tail_sol                             19439 non-null  float64\n",
      " 51  rougher.output.tail_au                              19439 non-null  float64\n",
      " 52  rougher.state.floatbank10_a_air                     19438 non-null  float64\n",
      " 53  rougher.state.floatbank10_a_level                   19438 non-null  float64\n",
      " 54  rougher.state.floatbank10_b_air                     19438 non-null  float64\n",
      " 55  rougher.state.floatbank10_b_level                   19438 non-null  float64\n",
      " 56  rougher.state.floatbank10_c_air                     19438 non-null  float64\n",
      " 57  rougher.state.floatbank10_c_level                   19438 non-null  float64\n",
      " 58  rougher.state.floatbank10_d_air                     19439 non-null  float64\n",
      " 59  rougher.state.floatbank10_d_level                   19439 non-null  float64\n",
      " 60  rougher.state.floatbank10_e_air                     19003 non-null  float64\n",
      " 61  rougher.state.floatbank10_e_level                   19439 non-null  float64\n",
      " 62  rougher.state.floatbank10_f_air                     19439 non-null  float64\n",
      " 63  rougher.state.floatbank10_f_level                   19439 non-null  float64\n",
      " 64  secondary_cleaner.output.tail_ag                    19437 non-null  float64\n",
      " 65  secondary_cleaner.output.tail_pb                    19427 non-null  float64\n",
      " 66  secondary_cleaner.output.tail_sol                   17691 non-null  float64\n",
      " 67  secondary_cleaner.output.tail_au                    19439 non-null  float64\n",
      " 68  secondary_cleaner.state.floatbank2_a_air            19219 non-null  float64\n",
      " 69  secondary_cleaner.state.floatbank2_a_level          19438 non-null  float64\n",
      " 70  secondary_cleaner.state.floatbank2_b_air            19416 non-null  float64\n",
      " 71  secondary_cleaner.state.floatbank2_b_level          19438 non-null  float64\n",
      " 72  secondary_cleaner.state.floatbank3_a_air            19426 non-null  float64\n",
      " 73  secondary_cleaner.state.floatbank3_a_level          19438 non-null  float64\n",
      " 74  secondary_cleaner.state.floatbank3_b_air            19438 non-null  float64\n",
      " 75  secondary_cleaner.state.floatbank3_b_level          19438 non-null  float64\n",
      " 76  secondary_cleaner.state.floatbank4_a_air            19433 non-null  float64\n",
      " 77  secondary_cleaner.state.floatbank4_a_level          19438 non-null  float64\n",
      " 78  secondary_cleaner.state.floatbank4_b_air            19438 non-null  float64\n",
      " 79  secondary_cleaner.state.floatbank4_b_level          19438 non-null  float64\n",
      " 80  secondary_cleaner.state.floatbank5_a_air            19438 non-null  float64\n",
      " 81  secondary_cleaner.state.floatbank5_a_level          19438 non-null  float64\n",
      " 82  secondary_cleaner.state.floatbank5_b_air            19438 non-null  float64\n",
      " 83  secondary_cleaner.state.floatbank5_b_level          19438 non-null  float64\n",
      " 84  secondary_cleaner.state.floatbank6_a_air            19437 non-null  float64\n",
      " 85  secondary_cleaner.state.floatbank6_a_level          19438 non-null  float64\n",
      "dtypes: float64(86)\n",
      "memory usage: 12.9 MB\n"
     ]
    }
   ],
   "source": [
    "gold_recovery_full.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "e7477903-e484-4a6c-9377-57c99d3d5db2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Verifying the correctness of data split\n",
    "assert (\n",
    "    gold_recovery_train.shape[0] + gold_recovery_test.shape[0]\n",
    "    == gold_recovery_full.shape[0]\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e9385810-e1db-41d8-84ca-46d095e00a38",
   "metadata": {},
   "source": [
    "We can see that the original data from `gold_recovery_full` DataFrame has been divided into training and test sets correctly. Furthermore, the columns for all DataFrames are characterized by the same `float64` data type so there are no problems with data types. We can also notice that many columns have a certain number of missing values, which we will have to deal with at the data preprocessing stage."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9c911078-6340-4279-9de5-f494f350dbf4",
   "metadata": {},
   "source": [
    "Now let's take a closer look at the column names. It can be seen that these represent three certain words (blocks) that are separated by dots. In fact the names have a specific structure as well as meaning. Take, for instance, a column called `final.output.concentrate_au` and consider what its name is made up of: \n",
    "\n",
    "$$\n",
    "\\underbrace{final}_\\text{[stage]}.\\underbrace{output}_\\text{[parameter_type]}.\\underbrace{concentrate\\_au}_\\text{[parameter_name]}\n",
    "$$\n",
    "\n",
    "It can be seen that the name starts with a stage of the gold recovery process (`final`). Second, we see the type of the parameter (`output`). Lastly, there is a description of a certain parameter used in the recovery process (`concentrate_au`). \n",
    "\n",
    "These three blocks can take multiple different values as seen from the column names printed out above so let's consider all columns and find out what values each of these blocks can take. To do this, we can simply separate the column names by dots and display all possible unique values in each of these three blocks. This can be done using the `print_separated_column_names()` function, which we will write below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "88f1353e-17d1-4415-9079-bf3c351c146a",
   "metadata": {},
   "outputs": [],
   "source": [
    "def print_separated_column_names(\n",
    "    data,\n",
    "    separator=\".\",\n",
    "    include_unique_values=True,\n",
    "    show_stage_names=False,\n",
    "    show_parameter_types=False,\n",
    "    show_parameter_names=False,\n",
    "):\n",
    "    \"\"\"Splits column names of a DataFrame by a separator and prints the results.\n",
    "\n",
    "    Column names are split into three words assigned to distinct\n",
    "    three blocks according to the separator passed, and their unique\n",
    "    values (by default) are then printed out by blocks.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    data : DataFrame\n",
    "        DataFrame with gold-bearing ore extraction\n",
    "        and purification parameters.\n",
    "    separator : str, default=\".\"\n",
    "        Column names separator.\n",
    "    include_unique_values : bool, default=True\n",
    "        Boolean value indicating including only unique\n",
    "        values in each block.\n",
    "    show_stage_names : bool, default=False\n",
    "        Boolean value indicating showing possible names\n",
    "        of stages of gold purification process.\n",
    "    show_parameter_types : bool, default=False\n",
    "        Boolean value indicating showing possible types\n",
    "        of parameters of gold purification process.\n",
    "    show_parameter_names : bool, default=False\n",
    "        Boolean value indicating showing possible names of\n",
    "        parameters of gold purification process.\n",
    "    \"\"\"\n",
    "    # Defining blocks to store column name separation results\n",
    "    stage_names = []\n",
    "    parameter_types = []\n",
    "    parameter_names = []\n",
    "\n",
    "    # Splitting column names by separator\n",
    "    for column in data.columns:\n",
    "        column_names_splitted = column.split(separator)\n",
    "\n",
    "        stage_names.append(column_names_splitted[0])\n",
    "        parameter_types.append(column_names_splitted[1])\n",
    "        parameter_names.append(column_names_splitted[2])\n",
    "\n",
    "    # Including only unique values in each block\n",
    "    if include_unique_values:\n",
    "        stage_names = set(stage_names)\n",
    "        parameter_types = set(parameter_types)\n",
    "        parameter_names = set(parameter_names)\n",
    "\n",
    "    # Printing out separated column names\n",
    "    if show_stage_names:\n",
    "        print(\"Possible values in [stage] block:\")\n",
    "        for i, sep_value in enumerate(sorted(stage_names)):\n",
    "            print(f\"{i+1}. {sep_value}\")\n",
    "    if show_parameter_types:\n",
    "        print(\"Possible values in [parameter_type] block:\")\n",
    "        for i, sep_value in enumerate(sorted(parameter_types)):\n",
    "            print(f\"{i+1}. {sep_value}\")\n",
    "    if show_parameter_names:\n",
    "        print(\"Possible values in [parameter_name] block:\")\n",
    "        for i, sep_value in enumerate(sorted(parameter_names)):\n",
    "            print(f\"{i+1}. {sep_value}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4cbec0e4-b1b6-4768-9a28-dd29338fae44",
   "metadata": {},
   "source": [
    "Let's apply this function to the DataFrame with full information (`gold_recovery_full`) and consider each block."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "cfd50f13-88b2-4439-98dd-40e202bc001b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Possible values in [stage] block:\n",
      "1. final\n",
      "2. primary_cleaner\n",
      "3. rougher\n",
      "4. secondary_cleaner\n"
     ]
    }
   ],
   "source": [
    "# Printing possible stage names\n",
    "print_separated_column_names(data=gold_recovery_full, show_stage_names=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "0f41b413-a5ea-476e-a88f-9ba6924b2c12",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Possible values in [parameter_type] block:\n",
      "1. calculation\n",
      "2. input\n",
      "3. output\n",
      "4. state\n"
     ]
    }
   ],
   "source": [
    "# Printing possible parameter types\n",
    "print_separated_column_names(data=gold_recovery_full, show_parameter_types=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "176a4452-873a-4b4d-abb8-cc8c7c151919",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Possible values in [parameter_name] block:\n",
      "1. au_pb_ratio\n",
      "2. concentrate_ag\n",
      "3. concentrate_au\n",
      "4. concentrate_pb\n",
      "5. concentrate_sol\n",
      "6. depressant\n",
      "7. feed_ag\n",
      "8. feed_au\n",
      "9. feed_pb\n",
      "10. feed_rate\n",
      "11. feed_size\n",
      "12. feed_sol\n",
      "13. floatbank10_a_air\n",
      "14. floatbank10_a_level\n",
      "15. floatbank10_b_air\n",
      "16. floatbank10_b_level\n",
      "17. floatbank10_c_air\n",
      "18. floatbank10_c_level\n",
      "19. floatbank10_d_air\n",
      "20. floatbank10_d_level\n",
      "21. floatbank10_e_air\n",
      "22. floatbank10_e_level\n",
      "23. floatbank10_f_air\n",
      "24. floatbank10_f_level\n",
      "25. floatbank10_sulfate\n",
      "26. floatbank10_sulfate_to_au_feed\n",
      "27. floatbank10_xanthate\n",
      "28. floatbank11_sulfate\n",
      "29. floatbank11_sulfate_to_au_feed\n",
      "30. floatbank11_xanthate\n",
      "31. floatbank2_a_air\n",
      "32. floatbank2_a_level\n",
      "33. floatbank2_b_air\n",
      "34. floatbank2_b_level\n",
      "35. floatbank3_a_air\n",
      "36. floatbank3_a_level\n",
      "37. floatbank3_b_air\n",
      "38. floatbank3_b_level\n",
      "39. floatbank4_a_air\n",
      "40. floatbank4_a_level\n",
      "41. floatbank4_b_air\n",
      "42. floatbank4_b_level\n",
      "43. floatbank5_a_air\n",
      "44. floatbank5_a_level\n",
      "45. floatbank5_b_air\n",
      "46. floatbank5_b_level\n",
      "47. floatbank6_a_air\n",
      "48. floatbank6_a_level\n",
      "49. floatbank8_a_air\n",
      "50. floatbank8_a_level\n",
      "51. floatbank8_b_air\n",
      "52. floatbank8_b_level\n",
      "53. floatbank8_c_air\n",
      "54. floatbank8_c_level\n",
      "55. floatbank8_d_air\n",
      "56. floatbank8_d_level\n",
      "57. recovery\n",
      "58. sulfate\n",
      "59. sulfate_to_au_concentrate\n",
      "60. tail_ag\n",
      "61. tail_au\n",
      "62. tail_pb\n",
      "63. tail_sol\n",
      "64. xanthate\n"
     ]
    }
   ],
   "source": [
    "# Printing possible parameter names\n",
    "print_separated_column_names(data=gold_recovery_full, show_parameter_names=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "edab661e-be99-4958-8044-67f3d6c44110",
   "metadata": {},
   "source": [
    "We see that the first block containing names of the stages of gold purification process has four unique values:\n",
    "\n",
    "* `final`: final characteristics;\n",
    "* `primary_cleaner`: primary purification;\n",
    "* `rougher`: flotation;\n",
    "* `secondary_cleaner`: secondary purification."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fd1ae4c1-17a3-4288-a6fa-513d6a016372",
   "metadata": {},
   "source": [
    "Let's now consider what parameter types there are in the process:\n",
    "\n",
    "* `calculation`: calculated characteristics;\n",
    "* `input`: input parameters;\n",
    "* `output`: output parameters;\n",
    "* `state`: current stage state characteristics."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1a65dd64-ee83-4075-b432-7ac6ec87278d",
   "metadata": {},
   "source": [
    "The last block with parameter names contains a wide range of possible values. We will not describe each of them, since not all of them will be particularly relevant to us during completing this project. Nevertheless, we will later pay a particular attention to some of them.  "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da86863e-7237-4a55-84f9-118d913cd691",
   "metadata": {},
   "source": [
    "## Computing gold recovery efficiency"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "23fc91dd-b03e-4382-861e-15ce562c2d2a",
   "metadata": {},
   "source": [
    "Now, let's compute gold recovery efficiency and verify if it has been calculated correctly. It is important, since this will be one of the targets that we need to predict. Gold recovery efficiency can be computed as follows:\n",
    "\n",
    "$$\n",
    "Recovery = \\frac{C \\times (F - T)}{F \\times (C - T)} \\times 100\\%,\n",
    "$$\n",
    "where $C$ - proportion of gold in concentrate after flotation/purification, $F$ - proportion of gold in input/concentrate before flotation/purification, а $T$ - proportion of gold in tails after flotation/purification. "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2015e3d5-4b75-4393-b34b-f621611f8e69",
   "metadata": {},
   "source": [
    "Basically, all we need to understand to compute this indicator is which columns of `gold_recovery_full` represent the parameters ($C$, $F$, $T$) of the above formula. We can then check whether gold recovery was calculated correctly, which is contained in `rougher.output.recovery` column. In this case, the information we need is contained in the following columns:\n",
    "\n",
    "* `rougher.output.concentrate_au` => $C$ parameter;\n",
    "\n",
    "* `rougher.input.feed_au` => $F$ parameter;\n",
    "\n",
    "* `rougher.output.tail_au` => $T$ parameter.\n",
    "\n",
    "This allows us to write a function called `compute_gold_recovery_efficiency()` that we will define below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "552a5c11-784c-4ee9-922c-82bba1801c02",
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_gold_recovery_efficiency(data):\n",
    "    \"\"\"Computes efficiency of gold recovery as a percentage.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    data : DataFrame\n",
    "        DataFrame with gold-bearing ore extraction\n",
    "        and purification parameters.\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    float\n",
    "        Gold recovery efficiency ratio.\n",
    "\n",
    "    Notes\n",
    "    -----\n",
    "    It is assumed that a function will be applied to the entire\n",
    "    DataFrame using `apply()` method, where, given data for each row\n",
    "    of the columns chosen, indicator will be computed.\n",
    "    \"\"\"\n",
    "    # Selecting data for computing gold recovery\n",
    "    C = data[\"rougher.output.concentrate_au\"]\n",
    "    F = data[\"rougher.input.feed_au\"]\n",
    "    T = data[\"rougher.output.tail_au\"]\n",
    "\n",
    "    # Computing gold recovery efficiency\n",
    "    recovery = (C * (F - T)) / (F * (C - T)) * 100\n",
    "\n",
    "    return recovery"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "731518e7-a8a3-4bca-a6cb-d67be9cde19a",
   "metadata": {},
   "source": [
    "Now, we can compute gold recovery efficiency using `apply()` method, the result of which we will store in a new column called `rougher.output.recovery_computed`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "6a5f8044-5645-4bd4-81e9-e5cc488070e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "gold_recovery_full[\"rougher.output.recovery_computed\"] = gold_recovery_full.apply(\n",
    "    compute_gold_recovery_efficiency, axis=1\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3cbcc93f-8abc-4475-8fcc-42981562ec19",
   "metadata": {},
   "source": [
    "We can verify whether we have correctly computed this indicator by comparing the results with the values of gold recovery already computed beforehand and available in `rougher.output.recovery` column. This can be done by calculating *Mean Absolute Error (MAE)* metric with its arguments being values in `rougher.output.recovery_computed` and `rougher.output.recovery` columns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "b2fd70ef-b9fe-4c93-9acd-672417a05d9b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE: 9.874045668302637e-15\n"
     ]
    }
   ],
   "source": [
    "# Computing MAE\n",
    "mae_gold_recovery = mean_absolute_error(\n",
    "    gold_recovery_full[\"rougher.output.recovery_computed\"],\n",
    "    gold_recovery_full[\"rougher.output.recovery\"],\n",
    ")\n",
    "\n",
    "# Verifying the correctness of computations\n",
    "assert (\n",
    "    np.round(mae_gold_recovery) == 0\n",
    "), \"Incorrect gold recovery efficiency calculation\"\n",
    "\n",
    "# Printing out MAE value\n",
    "print(\"MAE: {}\".format(mae_gold_recovery))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "93d9e1f0-4369-4544-92e5-c0f35fb70d9b",
   "metadata": {},
   "source": [
    "We see that the value of the metric proved to be extremely small, thus showcasing the fact that we have computed gold recovery correctly and proving that the values in `rougher.output.recovery` have been computed correctly. Hence, let's drop a column with the computed gold recovery from our data and move on. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "6793d989-50ec-45fc-84c1-d5aee4066ff0",
   "metadata": {},
   "outputs": [],
   "source": [
    "gold_recovery_full.drop(\"rougher.output.recovery_computed\", axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "caef2610-93ee-4454-b839-2eb5c84a02e6",
   "metadata": {},
   "source": [
    "## Features analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a032de13-ec09-4c4b-8e10-45831a9ae5da",
   "metadata": {},
   "source": [
    "As written in the project description, there are some features that are present in full data (`gold_recovery_full`) but absent in test set (`gold_recovery_test`). Thus, let's take a look at what features are not accessible to the test set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "0b5433e3-88c6-46dc-98b8-0b22bf44209a",
   "metadata": {},
   "outputs": [],
   "source": [
    "gold_recovery_full_columns = gold_recovery_full.columns\n",
    "gold_recovery_test_columns = gold_recovery_test.columns\n",
    "\n",
    "# Identifying columns inaccessible to test set\n",
    "columns_test_inaccessible = []\n",
    "for column in gold_recovery_full_columns:\n",
    "    if column not in gold_recovery_test_columns:\n",
    "        columns_test_inaccessible.append(column)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9f10a7ed-4ea3-47a9-9b44-ee31e8b475f9",
   "metadata": {},
   "source": [
    "Let's choose such columns from our full data in order to apply previously written `print_separated_column_names()` function so as to study what peculiarities inaccessible columns have."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "b89c146a-8396-41e7-8904-b239e90198ff",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Possible values in [stage] block:\n",
      "1. final\n",
      "2. primary_cleaner\n",
      "3. rougher\n",
      "4. secondary_cleaner\n"
     ]
    }
   ],
   "source": [
    "print_separated_column_names(\n",
    "    data=gold_recovery_full[columns_test_inaccessible], show_stage_names=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "3234583e-d3c7-41ec-a562-7380ff202336",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Possible values in [parameter_type] block:\n",
      "1. calculation\n",
      "2. output\n"
     ]
    }
   ],
   "source": [
    "print_separated_column_names(\n",
    "    data=gold_recovery_full[columns_test_inaccessible], show_parameter_types=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "d8d8213a-39dc-482a-ae91-c4453d80861b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Possible values in [parameter_name] block:\n",
      "1. au_pb_ratio\n",
      "2. concentrate_ag\n",
      "3. concentrate_au\n",
      "4. concentrate_pb\n",
      "5. concentrate_sol\n",
      "6. floatbank10_sulfate_to_au_feed\n",
      "7. floatbank11_sulfate_to_au_feed\n",
      "8. recovery\n",
      "9. sulfate_to_au_concentrate\n",
      "10. tail_ag\n",
      "11. tail_au\n",
      "12. tail_pb\n",
      "13. tail_sol\n"
     ]
    }
   ],
   "source": [
    "print_separated_column_names(\n",
    "    data=gold_recovery_full[columns_test_inaccessible], show_parameter_names=True\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f30c7e23-54e8-41cb-a695-aab208c28683",
   "metadata": {},
   "source": [
    "Now, we can see the structure of the column names for those columns that are absent in the test set. It is quite noticeable that calculation (`calculation`) and output (`output`) characteristics of the process are not available, because they cannot be obtained or measured during the execution of the technological process. Conversely, such parameters are computed and measured much later. Moreover, we notice that parameter names contain `recovery` meaning that the test set does not contain our target that will have to be dealt with later."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a76b489-86dc-4a31-93a6-0f037b6f691e",
   "metadata": {},
   "source": [
    "## Data preprocessing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bf8716e2-8dca-48e7-98dd-d55644d6673b",
   "metadata": {},
   "source": [
    "### Missing values"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8e8470f3-4fda-4797-8f53-c09b103263df",
   "metadata": {},
   "source": [
    "Let's look at the missing values in each of the three DataFrames. To do this, we can write `identify_missing_values()` function, which will output the names of columns in which there are missing values, as well as the number and proportion of missing values in them."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e00a73a4-19fa-4e7f-8fc5-49e3b134640e",
   "metadata": {},
   "source": [
    "Let's apply this function to each DataFrame in order to see how serious missing values problem is."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "ce5356a2-b223-4073-ae5b-4ceadd73f558",
   "metadata": {},
   "outputs": [],
   "source": [
    "def identify_missing_values(data):\n",
    "    \"\"\"Performs missing values computation.\n",
    "\n",
    "    Function computes a number and share of missing values\n",
    "    in DataFrame columns which have NaN-values present\n",
    "    and displays data types of such columns.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    data : DataFrame\n",
    "        DataFrame which needs to be checked for missing values.\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    DataFrame or None\n",
    "        DataFrame with column names/their data types, number of\n",
    "        missing values and shares of NaN-values in such columns,\n",
    "        or None if no missing values have been found.\n",
    "    \"\"\"\n",
    "    # Verifying the presence of missing values\n",
    "    miss_vals_num = data.isnull().sum()[data.isnull().sum() > 0]\n",
    "    if miss_vals_num.empty:\n",
    "        return None\n",
    "\n",
    "    # Creating a table with numbers of missing values\n",
    "    cols = {\"missing_count\": miss_vals_num.values}\n",
    "    nans_df = pd.DataFrame(data=cols, index=miss_vals_num.index).sort_values(\n",
    "        by=\"missing_count\", ascending=False\n",
    "    )\n",
    "\n",
    "    # Adding shares of missing values\n",
    "    nans_df[\"missing_fraction\"] = nans_df[\"missing_count\"] / data.shape[0]\n",
    "    nans_df[\"missing_fraction\"] = nans_df[\"missing_fraction\"].round(4)\n",
    "\n",
    "    # Adding data types\n",
    "    nans_df[\"dtype\"] = data[nans_df.index].dtypes\n",
    "    nans_df = nans_df[[\"dtype\", \"missing_count\", \"missing_fraction\"]]\n",
    "\n",
    "    return nans_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "74f08bb4-0f20-480f-9059-2bc69309c23f",
   "metadata": {},
   "outputs": [
    {
     "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>dtype</th>\n",
       "      <th>missing_count</th>\n",
       "      <th>missing_fraction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>secondary_cleaner.output.tail_sol</th>\n",
       "      <td>float64</td>\n",
       "      <td>1605</td>\n",
       "      <td>0.1134</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.state.floatbank10_e_air</th>\n",
       "      <td>float64</td>\n",
       "      <td>436</td>\n",
       "      <td>0.0308</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.input.floatbank11_xanthate</th>\n",
       "      <td>float64</td>\n",
       "      <td>428</td>\n",
       "      <td>0.0302</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>primary_cleaner.output.concentrate_sol</th>\n",
       "      <td>float64</td>\n",
       "      <td>286</td>\n",
       "      <td>0.0202</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>secondary_cleaner.state.floatbank2_a_air</th>\n",
       "      <td>float64</td>\n",
       "      <td>217</td>\n",
       "      <td>0.0153</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.state.floatbank10_a_air</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.state.floatbank10_a_level</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.state.floatbank10_b_air</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.state.floatbank10_b_level</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>secondary_cleaner.state.floatbank6_a_level</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>62 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              dtype  missing_count  \\\n",
       "secondary_cleaner.output.tail_sol           float64           1605   \n",
       "rougher.state.floatbank10_e_air             float64            436   \n",
       "rougher.input.floatbank11_xanthate          float64            428   \n",
       "primary_cleaner.output.concentrate_sol      float64            286   \n",
       "secondary_cleaner.state.floatbank2_a_air    float64            217   \n",
       "...                                             ...            ...   \n",
       "rougher.state.floatbank10_a_air             float64              1   \n",
       "rougher.state.floatbank10_a_level           float64              1   \n",
       "rougher.state.floatbank10_b_air             float64              1   \n",
       "rougher.state.floatbank10_b_level           float64              1   \n",
       "secondary_cleaner.state.floatbank6_a_level  float64              1   \n",
       "\n",
       "                                            missing_fraction  \n",
       "secondary_cleaner.output.tail_sol                     0.1134  \n",
       "rougher.state.floatbank10_e_air                       0.0308  \n",
       "rougher.input.floatbank11_xanthate                    0.0302  \n",
       "primary_cleaner.output.concentrate_sol                0.0202  \n",
       "secondary_cleaner.state.floatbank2_a_air              0.0153  \n",
       "...                                                      ...  \n",
       "rougher.state.floatbank10_a_air                       0.0001  \n",
       "rougher.state.floatbank10_a_level                     0.0001  \n",
       "rougher.state.floatbank10_b_air                       0.0001  \n",
       "rougher.state.floatbank10_b_level                     0.0001  \n",
       "secondary_cleaner.state.floatbank6_a_level            0.0001  \n",
       "\n",
       "[62 rows x 3 columns]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "identify_missing_values(data=gold_recovery_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "648800e1-f59f-4b40-8d9f-cb050fa8d3df",
   "metadata": {},
   "outputs": [
    {
     "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>dtype</th>\n",
       "      <th>missing_count</th>\n",
       "      <th>missing_fraction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>rougher.input.floatbank11_xanthate</th>\n",
       "      <td>float64</td>\n",
       "      <td>25</td>\n",
       "      <td>0.0047</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.input.feed_sol</th>\n",
       "      <td>float64</td>\n",
       "      <td>21</td>\n",
       "      <td>0.0040</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>secondary_cleaner.state.floatbank3_a_air</th>\n",
       "      <td>float64</td>\n",
       "      <td>9</td>\n",
       "      <td>0.0017</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.input.floatbank11_sulfate</th>\n",
       "      <td>float64</td>\n",
       "      <td>8</td>\n",
       "      <td>0.0015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>primary_cleaner.input.depressant</th>\n",
       "      <td>float64</td>\n",
       "      <td>5</td>\n",
       "      <td>0.0009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.input.floatbank10_sulfate</th>\n",
       "      <td>float64</td>\n",
       "      <td>5</td>\n",
       "      <td>0.0009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>primary_cleaner.input.sulfate</th>\n",
       "      <td>float64</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>primary_cleaner.input.xanthate</th>\n",
       "      <td>float64</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.input.feed_rate</th>\n",
       "      <td>float64</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>secondary_cleaner.state.floatbank2_a_air</th>\n",
       "      <td>float64</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>secondary_cleaner.state.floatbank2_b_air</th>\n",
       "      <td>float64</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.input.feed_size</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0002</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            dtype  missing_count  \\\n",
       "rougher.input.floatbank11_xanthate        float64             25   \n",
       "rougher.input.feed_sol                    float64             21   \n",
       "secondary_cleaner.state.floatbank3_a_air  float64              9   \n",
       "rougher.input.floatbank11_sulfate         float64              8   \n",
       "primary_cleaner.input.depressant          float64              5   \n",
       "rougher.input.floatbank10_sulfate         float64              5   \n",
       "primary_cleaner.input.sulfate             float64              4   \n",
       "primary_cleaner.input.xanthate            float64              4   \n",
       "rougher.input.feed_rate                   float64              3   \n",
       "secondary_cleaner.state.floatbank2_a_air  float64              3   \n",
       "secondary_cleaner.state.floatbank2_b_air  float64              2   \n",
       "rougher.input.feed_size                   float64              1   \n",
       "\n",
       "                                          missing_fraction  \n",
       "rougher.input.floatbank11_xanthate                  0.0047  \n",
       "rougher.input.feed_sol                              0.0040  \n",
       "secondary_cleaner.state.floatbank3_a_air            0.0017  \n",
       "rougher.input.floatbank11_sulfate                   0.0015  \n",
       "primary_cleaner.input.depressant                    0.0009  \n",
       "rougher.input.floatbank10_sulfate                   0.0009  \n",
       "primary_cleaner.input.sulfate                       0.0008  \n",
       "primary_cleaner.input.xanthate                      0.0008  \n",
       "rougher.input.feed_rate                             0.0006  \n",
       "secondary_cleaner.state.floatbank2_a_air            0.0006  \n",
       "secondary_cleaner.state.floatbank2_b_air            0.0004  \n",
       "rougher.input.feed_size                             0.0002  "
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "identify_missing_values(data=gold_recovery_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "ff30461a-a4ec-4c1d-a09e-d0a852a0382c",
   "metadata": {},
   "outputs": [
    {
     "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>dtype</th>\n",
       "      <th>missing_count</th>\n",
       "      <th>missing_fraction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>secondary_cleaner.output.tail_sol</th>\n",
       "      <td>float64</td>\n",
       "      <td>1748</td>\n",
       "      <td>0.0899</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.input.floatbank11_xanthate</th>\n",
       "      <td>float64</td>\n",
       "      <td>453</td>\n",
       "      <td>0.0233</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rougher.state.floatbank10_e_air</th>\n",
       "      <td>float64</td>\n",
       "      <td>436</td>\n",
       "      <td>0.0224</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>primary_cleaner.output.concentrate_sol</th>\n",
       "      <td>float64</td>\n",
       "      <td>370</td>\n",
       "      <td>0.0190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>secondary_cleaner.state.floatbank2_a_air</th>\n",
       "      <td>float64</td>\n",
       "      <td>220</td>\n",
       "      <td>0.0113</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>primary_cleaner.state.floatbank8_c_level</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>primary_cleaner.state.floatbank8_b_level</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>primary_cleaner.state.floatbank8_a_level</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>final.output.tail_ag</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>secondary_cleaner.state.floatbank6_a_level</th>\n",
       "      <td>float64</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>63 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              dtype  missing_count  \\\n",
       "secondary_cleaner.output.tail_sol           float64           1748   \n",
       "rougher.input.floatbank11_xanthate          float64            453   \n",
       "rougher.state.floatbank10_e_air             float64            436   \n",
       "primary_cleaner.output.concentrate_sol      float64            370   \n",
       "secondary_cleaner.state.floatbank2_a_air    float64            220   \n",
       "...                                             ...            ...   \n",
       "primary_cleaner.state.floatbank8_c_level    float64              1   \n",
       "primary_cleaner.state.floatbank8_b_level    float64              1   \n",
       "primary_cleaner.state.floatbank8_a_level    float64              1   \n",
       "final.output.tail_ag                        float64              1   \n",
       "secondary_cleaner.state.floatbank6_a_level  float64              1   \n",
       "\n",
       "                                            missing_fraction  \n",
       "secondary_cleaner.output.tail_sol                     0.0899  \n",
       "rougher.input.floatbank11_xanthate                    0.0233  \n",
       "rougher.state.floatbank10_e_air                       0.0224  \n",
       "primary_cleaner.output.concentrate_sol                0.0190  \n",
       "secondary_cleaner.state.floatbank2_a_air              0.0113  \n",
       "...                                                      ...  \n",
       "primary_cleaner.state.floatbank8_c_level              0.0001  \n",
       "primary_cleaner.state.floatbank8_b_level              0.0001  \n",
       "primary_cleaner.state.floatbank8_a_level              0.0001  \n",
       "final.output.tail_ag                                  0.0001  \n",
       "secondary_cleaner.state.floatbank6_a_level            0.0001  \n",
       "\n",
       "[63 rows x 3 columns]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "identify_missing_values(data=gold_recovery_full)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7cc3b7d6-066b-4001-8aef-218df5a2f1d8",
   "metadata": {},
   "source": [
    "We see that there are not so many missing values on the scale of the whole DataFrames (the maximum share of omissions is 11.34%). Since the data represent sequential technological processes, we can fill in the gaps in the data with the latest `non-null` values in the data by specifying `method=\"ffill\"` for the `fillna()` method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "0db1e7e9-fa01-4b25-b668-ae17e2d76a95",
   "metadata": {},
   "outputs": [],
   "source": [
    "for gold_recovery_data in (gold_recovery_train, gold_recovery_test, gold_recovery_full):\n",
    "    gold_recovery_data.fillna(method=\"ffill\", inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4da284d8-46c6-40b3-ac0b-4c4f1135f47a",
   "metadata": {},
   "source": [
    "Let's check the DataFrames again for missing values in order to verify whether the changes made have been successful."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "ee2bb857-4012-4ac3-9ea9-a643c89766a6",
   "metadata": {},
   "outputs": [],
   "source": [
    "identify_missing_values(data=gold_recovery_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "3c419a44-9e1b-4bde-af54-ddc9e31cbbf1",
   "metadata": {},
   "outputs": [],
   "source": [
    "identify_missing_values(data=gold_recovery_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "e84c63f9-2c2d-4cac-b416-13453ab5aaa3",
   "metadata": {},
   "outputs": [],
   "source": [
    "identify_missing_values(data=gold_recovery_full)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5673e09a-4b4a-4010-ba9b-6be331629829",
   "metadata": {},
   "source": [
    "### Duplicates"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8292db1e-fcb8-41ca-a454-093af1e7a8db",
   "metadata": {},
   "source": [
    "Next, let's make sure that there are no duplicated rows in our data. If we find such objects, we will leave exactly the last duplicate found, because if duplicates have arisen as a result of some data changes, then we need to have the latest data, which we specify as `keep=\"last\"`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "75c9bf5d-bc96-4298-a690-2574397a4059",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Identical rows are not found.\n",
      "Identical rows are not found.\n",
      "Identical rows are not found.\n"
     ]
    }
   ],
   "source": [
    "for gold_recovery_data in (gold_recovery_train, gold_recovery_test, gold_recovery_full):\n",
    "    duplicated_rows_num = gold_recovery_data.duplicated().sum()\n",
    "    if duplicated_rows_num == 0:\n",
    "        print(\"Identical rows are not found.\")\n",
    "    else:\n",
    "        gold_recovery_data.drop_duplicates(keep=\"last\", inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1424b338-6627-48cb-a39b-9ed53cbbe3e3",
   "metadata": {},
   "source": [
    "### Including targets"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1bfffd80-94f8-4e1e-a205-e8684707908f",
   "metadata": {},
   "source": [
    "Basically, we have two targets that we will need to later predict:\n",
    "\n",
    "1. `rougher.output.recovery`\n",
    "2. `final.output.recovery`\n",
    "\n",
    "As noticed earlier, these columns were not present in our test data so let's verify this observation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "61abdf66-58ce-49e2-81b1-5c235de9659a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Targets 'rougher.output.recovery' and 'final.output.recovery' are not present in test data.\n"
     ]
    }
   ],
   "source": [
    "target_name_rougher = \"rougher.output.recovery\"\n",
    "target_name_final = \"final.output.recovery\"\n",
    "\n",
    "# Checking the presence of targets in test set\n",
    "try:\n",
    "    assert (target_name_rougher in columns_test_inaccessible) & (\n",
    "        target_name_final in columns_test_inaccessible\n",
    "    ) == False\n",
    "except AssertionError:\n",
    "    print(\n",
    "        f\"Targets '{target_name_rougher}' and '{target_name_final}' are not present in test data.\"\n",
    "    )\n",
    "else:\n",
    "    print(\n",
    "        f\"Targets '{target_name_rougher}' and '{target_name_final}' are present in test data.\"\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "536ae197-d855-4a95-a5c0-d97892444dc2",
   "metadata": {},
   "source": [
    "We see that, indeed, these two target features are missing from our test data. We will have to add them to the test set ourselves. This can be easily done by selecting the corresponding indices of `gold_recovery_test` in `gold_recovery_full`, thus considering the values of target features that relate only to objects in the test sample."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "12b5bdc1-5dc8-47dd-80f2-8be6c595a2f6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Adding targets to test set\n",
    "gold_recovery_test[target_name_rougher] = gold_recovery_full[target_name_rougher].loc[\n",
    "    gold_recovery_test.index\n",
    "]\n",
    "gold_recovery_test[target_name_final] = gold_recovery_full[target_name_final].loc[\n",
    "    gold_recovery_test.index\n",
    "]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "34b0bf2f-69d3-4217-866a-176b6d96c69c",
   "metadata": {},
   "source": [
    "Let's check the changes made by printing out the first five rows of the test set for the last two columns that we have created."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "0e21b50b-917d-4dc5-85e5-fd78ef425124",
   "metadata": {},
   "outputs": [
    {
     "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>rougher.output.recovery</th>\n",
       "      <th>final.output.recovery</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2016-09-01 00:59:59</th>\n",
       "      <td>89.993421</td>\n",
       "      <td>70.273583</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-09-01 01:59:59</th>\n",
       "      <td>88.089657</td>\n",
       "      <td>68.910432</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-09-01 02:59:59</th>\n",
       "      <td>88.412756</td>\n",
       "      <td>68.143213</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-09-01 03:59:59</th>\n",
       "      <td>87.360133</td>\n",
       "      <td>67.776393</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-09-01 04:59:59</th>\n",
       "      <td>83.236367</td>\n",
       "      <td>61.467078</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     rougher.output.recovery  final.output.recovery\n",
       "date                                                               \n",
       "2016-09-01 00:59:59                89.993421              70.273583\n",
       "2016-09-01 01:59:59                88.089657              68.910432\n",
       "2016-09-01 02:59:59                88.412756              68.143213\n",
       "2016-09-01 03:59:59                87.360133              67.776393\n",
       "2016-09-01 04:59:59                83.236367              61.467078"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gold_recovery_test[gold_recovery_test.columns[-2:]].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4f1fc0ad-af7e-408b-9017-2c8c84fd16f7",
   "metadata": {},
   "source": [
    "**Conclusion**\n",
    "\n",
    "At this stage, we have managed to upload, examine and prepare the data. In particular, a fairly large number of features were found in each of the DataFrames and, what is more, main data problems were identified (missing values and the absence of target features in the test set). Furthermore, at the data preprocessing stage, the missing values were filled with the latest available values, and the target features were added to the test set. Additionally, we checked the data for duplicates rows."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "607fd543-8533-494a-841b-49abd7a80b40",
   "metadata": {},
   "source": [
    "# Exploratory data analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "20a87238-0639-4ffc-a44d-c9e9819d7460",
   "metadata": {},
   "source": [
    "## Chemical elements concentration dynamics"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6c34751a-cbcf-4021-96fe-cbf983768e3f",
   "metadata": {},
   "source": [
    "In order to look at the changes in the concentration of metals, we will use `plot_concentration_dynamics()` function written below, where for a certain chemical element we will plot a series of histograms reflecting the concentration data for this element, broken down by purification stages."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "d8e7ac2a-b9a5-45b3-81d9-390920545e66",
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_concentration_dynamics(data, chemical_element_name, hist_bins_number=40):\n",
    "    \"\"\"Makes a chemical concentration plot.\n",
    "\n",
    "    Plots a series of histograms with concentration of a\n",
    "    particular chemical element for different gold purification\n",
    "    stages.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    data : DataFrame\n",
    "        DataFrame with gold-bearing ore\n",
    "        extraction and purification parameters.\n",
    "    chemical_element_name : str\n",
    "        Latin two-letter name of a chemical element.\n",
    "    hist_bins_number : int, default=40\n",
    "        Number of total bins to show on a plot.\n",
    "    \"\"\"\n",
    "    chemical_element_name = chemical_element_name.lower()\n",
    "\n",
    "    # Filtering data by chemical element\n",
    "    data_filtered = data.loc[\n",
    "        :,\n",
    "        (data.columns.str.endswith(\"concentrate_\" + chemical_element_name))\n",
    "        | (data.columns.str.endswith(\"feed_\" + chemical_element_name)),\n",
    "    ]\n",
    "\n",
    "    # Verifying the presence of the element in data\n",
    "    if data_filtered.empty == True:\n",
    "        print(f\"'{chemical_element_name.capitalize()}' element is not found.\")\n",
    "        return\n",
    "\n",
    "    # Plotting histograms\n",
    "    element_histplot = sns.histplot(data=data_filtered, bins=hist_bins_number)\n",
    "    element_histplot.set(\n",
    "        title=f\"{chemical_element_name.capitalize()}: Concentration dynamics\",\n",
    "        xlabel=\"Concentration (percent)\",\n",
    "        ylabel=\"Number of observations\",\n",
    "    )\n",
    "    plt.tight_layout()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d5210a2a-e30a-4853-a5f8-43bd260e5be7",
   "metadata": {},
   "source": [
    "Now, let's apply this function to each of elements which are of interest to us:\n",
    "\n",
    "* Gold (*Au*)\n",
    "* Silver (*Ag*)\n",
    "* Lead (*Pb*)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "eb471807-2b41-4967-8ee8-d2e35c45b8e9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting gold concentration\n",
    "sns.set_theme(style=\"darkgrid\")\n",
    "plot_concentration_dynamics(\n",
    "    data=gold_recovery_train, chemical_element_name=\"Au\", hist_bins_number=100\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "8ac5bcff-119d-43d1-8912-484931600fad",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting silver concentration\n",
    "plot_concentration_dynamics(\n",
    "    data=gold_recovery_train, chemical_element_name=\"Ag\", hist_bins_number=100\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "c695e194-90e8-4c9d-a43b-935e32e41884",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting lead concentration\n",
    "plot_concentration_dynamics(\n",
    "    data=gold_recovery_train, chemical_element_name=\"Pb\", hist_bins_number=100\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b5f7d9eb-62bb-47b6-b0e7-707e586249ff",
   "metadata": {},
   "source": [
    "We can see that during the purification process, the concentration of gold increases, which is the meaning of purifying gold-bearing ore. Next, let's look at silver: its concentration decreases with each purification stage. As far as lead is concerned, it represents a byproduct of chemical reaction and that is precisely why we see its concentration moderately increasing."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5eed7d86-ace6-4eb9-8c8d-5efb2eca3d26",
   "metadata": {},
   "source": [
    "## Comparing feed sizes"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ee9fe3ae-e954-4c69-9f2b-957c5453fd29",
   "metadata": {},
   "source": [
    "Now, let's compare the distributions of the size of input materials granules, the data on which is contained in `rougher.input.feed_size` column. Hence, let's look at the distribution of values on the histograms for training and test sets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "8a0fccc0-52a2-42be-b80d-59294c3bc3b5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Comparing feed sizes distributions\n",
    "feed_size_column_name = \"rougher.input.feed_size\"\n",
    "\n",
    "fig, axes = plt.subplots(1, 2, figsize=(5 * 2, 5 * 1))\n",
    "plt.suptitle(f\"Feed sizes comparison ({feed_size_column_name})\")\n",
    "\n",
    "sns.kdeplot(data=gold_recovery_train, x=feed_size_column_name, ax=axes[0])\n",
    "axes[0].set_title(\"Training set\")\n",
    "axes[0].set_xlabel(\"Feed size\")\n",
    "axes[0].set_ylabel(\"Density\")\n",
    "\n",
    "sns.kdeplot(data=gold_recovery_test, x=feed_size_column_name, ax=axes[1])\n",
    "axes[1].set_title(\"Test set\")\n",
    "axes[1].set_xlabel(\"Feed size\")\n",
    "axes[1].set_ylabel(\"Density\")\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "db6468b4-01ab-4664-87bb-31a10901b4f3",
   "metadata": {},
   "source": [
    "It can be seen that the histograms are similar, which is a good sign, because the estimates of the models will be more correct and accurate. Let's compare the distribution data in more detail and calculate descriptive statistics for each set using the `describe()` method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "c1e632c3-1c59-42b3-87f4-e3ca0b03ee4d",
   "metadata": {},
   "outputs": [
    {
     "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>gold_recovery_train</th>\n",
       "      <th>gold_recovery_test</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>14149.000000</td>\n",
       "      <td>5290.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>60.239856</td>\n",
       "      <td>55.952754</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>23.005310</td>\n",
       "      <td>19.080334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>9.659576</td>\n",
       "      <td>0.046369</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>48.966751</td>\n",
       "      <td>43.906504</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>55.269784</td>\n",
       "      <td>50.846843</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>66.007526</td>\n",
       "      <td>62.451165</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>484.967466</td>\n",
       "      <td>392.494040</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       gold_recovery_train  gold_recovery_test\n",
       "count         14149.000000         5290.000000\n",
       "mean             60.239856           55.952754\n",
       "std              23.005310           19.080334\n",
       "min               9.659576            0.046369\n",
       "25%              48.966751           43.906504\n",
       "50%              55.269784           50.846843\n",
       "75%              66.007526           62.451165\n",
       "max             484.967466          392.494040"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "stats_train = gold_recovery_train[feed_size_column_name].describe()\n",
    "stats_test = gold_recovery_test[feed_size_column_name].describe()\n",
    "\n",
    "stats_df = pd.concat([stats_train, stats_test], axis=1)\n",
    "stats_df.columns = [\"gold_recovery_train\", \"gold_recovery_test\"]\n",
    "stats_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6711cd42-9987-4205-9f1c-96a89bd0cb1e",
   "metadata": {},
   "source": [
    "We see that the values of most descriptive statistics are quite close to each other, which again proves the similarity of these two distributions."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "60642954-5057-49bb-97c6-f987cf6c2b2b",
   "metadata": {},
   "source": [
    "## Total concentration analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "16c13666-ab37-4864-9896-6434615b101d",
   "metadata": {},
   "source": [
    "Now let's look at the total concentration of all elements available on the histograms. To begin with, we will create new columns in the training data, where we will sum up all the available chemical elements and save the results in the appropriate columns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "d05272d9-ad6c-4b3b-bf27-71b25fed187d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Computing total concentration (input)\n",
    "gold_recovery_train[\"rougher.input.feed_total\"] = (\n",
    "    gold_recovery_train[\"rougher.input.feed_au\"]\n",
    "    + gold_recovery_train[\"rougher.input.feed_ag\"]\n",
    "    + gold_recovery_train[\"rougher.input.feed_pb\"]\n",
    "    + gold_recovery_train[\"rougher.input.feed_sol\"]\n",
    ")\n",
    "\n",
    "# Computing total concentration (rougher)\n",
    "gold_recovery_train[\"rougher.output.concentrate_total\"] = (\n",
    "    gold_recovery_train[\"rougher.output.concentrate_au\"]\n",
    "    + gold_recovery_train[\"rougher.output.concentrate_ag\"]\n",
    "    + gold_recovery_train[\"rougher.output.concentrate_pb\"]\n",
    "    + gold_recovery_train[\"rougher.output.concentrate_sol\"]\n",
    ")\n",
    "\n",
    "# Computing total concentration (final)\n",
    "gold_recovery_train[\"final.output.concentrate_total\"] = (\n",
    "    gold_recovery_train[\"final.output.concentrate_au\"]\n",
    "    + gold_recovery_train[\"final.output.concentrate_ag\"]\n",
    "    + gold_recovery_train[\"final.output.concentrate_pb\"]\n",
    "    + gold_recovery_train[\"final.output.concentrate_sol\"]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "0c1efe06-5083-4900-9f2d-2f6c39ee29a9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Filtering data\n",
    "gold_recovery_train_total = gold_recovery_train.loc[\n",
    "    :, (gold_recovery_train.columns.str.endswith(\"total\"))\n",
    "]\n",
    "\n",
    "# Plotting histograms\n",
    "total_histplot = sns.histplot(data=gold_recovery_train_total, bins=100)\n",
    "total_histplot.set(\n",
    "    title=\"Total concentration dynamics\",\n",
    "    xlabel=\"Concentration (percent)\",\n",
    "    ylabel=\"Number of observations\",\n",
    ")\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "be950f0b-58f1-45f1-8904-383afdda9975",
   "metadata": {},
   "source": [
    "The graph shows that as we move through the purification stages, the distribution of the total concentration of substances becomes narrower, which means that the concentration increases and eventually takes higher values in a narrower region."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5be33350-ef21-4af0-8c24-3eeecc9b5c7b",
   "metadata": {},
   "source": [
    "## Anomalies analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7772d202-a298-4d88-b160-28148d4be488",
   "metadata": {},
   "source": [
    "When we plot the histograms on the total concentration, we notice a small bin noticeable in the area of zero values. Such values may represent anomalies, because the total concentration should grow throughout the entire technological process. Let's try to find such values in the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "fd376931-413a-4dad-8353-e6b1a95e1831",
   "metadata": {},
   "outputs": [
    {
     "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>rougher.input.feed_total</th>\n",
       "      <th>rougher.output.concentrate_total</th>\n",
       "      <th>final.output.concentrate_total</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2016-01-16 21:00:00</th>\n",
       "      <td>58.602038</td>\n",
       "      <td>0.0</td>\n",
       "      <td>66.175996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-16 22:00:00</th>\n",
       "      <td>59.921613</td>\n",
       "      <td>0.0</td>\n",
       "      <td>66.794250</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-16 23:00:00</th>\n",
       "      <td>57.269041</td>\n",
       "      <td>0.0</td>\n",
       "      <td>67.572604</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-17 00:00:00</th>\n",
       "      <td>25.352358</td>\n",
       "      <td>0.0</td>\n",
       "      <td>65.598903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016-01-17 02:00:00</th>\n",
       "      <td>40.721375</td>\n",
       "      <td>0.0</td>\n",
       "      <td>63.445359</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-11 23:59:59</th>\n",
       "      <td>58.106496</td>\n",
       "      <td>0.0</td>\n",
       "      <td>70.703344</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-12 00:59:59</th>\n",
       "      <td>59.500740</td>\n",
       "      <td>0.0</td>\n",
       "      <td>70.814703</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-12 01:59:59</th>\n",
       "      <td>60.129638</td>\n",
       "      <td>0.0</td>\n",
       "      <td>71.329178</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-12 02:59:59</th>\n",
       "      <td>59.267147</td>\n",
       "      <td>0.0</td>\n",
       "      <td>71.195925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-12 03:59:59</th>\n",
       "      <td>59.490974</td>\n",
       "      <td>0.0</td>\n",
       "      <td>71.148257</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>354 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                     rougher.input.feed_total  \\\n",
       "date                                            \n",
       "2016-01-16 21:00:00                 58.602038   \n",
       "2016-01-16 22:00:00                 59.921613   \n",
       "2016-01-16 23:00:00                 57.269041   \n",
       "2016-01-17 00:00:00                 25.352358   \n",
       "2016-01-17 02:00:00                 40.721375   \n",
       "...                                       ...   \n",
       "2018-02-11 23:59:59                 58.106496   \n",
       "2018-02-12 00:59:59                 59.500740   \n",
       "2018-02-12 01:59:59                 60.129638   \n",
       "2018-02-12 02:59:59                 59.267147   \n",
       "2018-02-12 03:59:59                 59.490974   \n",
       "\n",
       "                     rougher.output.concentrate_total  \\\n",
       "date                                                    \n",
       "2016-01-16 21:00:00                               0.0   \n",
       "2016-01-16 22:00:00                               0.0   \n",
       "2016-01-16 23:00:00                               0.0   \n",
       "2016-01-17 00:00:00                               0.0   \n",
       "2016-01-17 02:00:00                               0.0   \n",
       "...                                               ...   \n",
       "2018-02-11 23:59:59                               0.0   \n",
       "2018-02-12 00:59:59                               0.0   \n",
       "2018-02-12 01:59:59                               0.0   \n",
       "2018-02-12 02:59:59                               0.0   \n",
       "2018-02-12 03:59:59                               0.0   \n",
       "\n",
       "                     final.output.concentrate_total  \n",
       "date                                                 \n",
       "2016-01-16 21:00:00                       66.175996  \n",
       "2016-01-16 22:00:00                       66.794250  \n",
       "2016-01-16 23:00:00                       67.572604  \n",
       "2016-01-17 00:00:00                       65.598903  \n",
       "2016-01-17 02:00:00                       63.445359  \n",
       "...                                             ...  \n",
       "2018-02-11 23:59:59                       70.703344  \n",
       "2018-02-12 00:59:59                       70.814703  \n",
       "2018-02-12 01:59:59                       71.329178  \n",
       "2018-02-12 02:59:59                       71.195925  \n",
       "2018-02-12 03:59:59                       71.148257  \n",
       "\n",
       "[354 rows x 3 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "anomalies_train = gold_recovery_train_total[\n",
    "    (gold_recovery_train_total[\"rougher.input.feed_total\"] == 0)\n",
    "    | (gold_recovery_train_total[\"rougher.output.concentrate_total\"] == 0)\n",
    "    | (gold_recovery_train_total[\"final.output.concentrate_total\"] == 0)\n",
    "]\n",
    "display(anomalies_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f491a590-fec5-47a4-a2c9-1146fb9672a4",
   "metadata": {},
   "source": [
    "We see that there are indeed anomalies in `rougher.output.concentrate_total` column as well as in `final.output_concentrate_total` column (not visible). In fact, the bins for these two columns were visible on the graph above in the area of zero values. Let's try to delete these rows from the training set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "07a87afa-69f6-43eb-bee6-d4b43bad8032",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training set size (with anomalies): 14,149\n",
      "Training set size (without anomalies): 13,795\n",
      "Anomalies found and deleted: 354\n"
     ]
    }
   ],
   "source": [
    "print(\"Training set size (with anomalies): {:,}\".format(gold_recovery_train.shape[0]))\n",
    "# Deleting anomalies\n",
    "gold_recovery_train.drop(anomalies_train.index, inplace=True)\n",
    "print(\n",
    "    \"Training set size (without anomalies): {:,}\".format(gold_recovery_train.shape[0])\n",
    ")\n",
    "print(\"Anomalies found and deleted: {}\".format(anomalies_train.index.shape[0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e2ebcf18-72c8-4e30-91a1-7760a324c164",
   "metadata": {},
   "source": [
    "We have managed to clean the data from anomalies successfully. Let's delete columns with total concentration from data and move on to modeling."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "770d5ccd-e2a6-409c-a39a-ff9e65ef27dd",
   "metadata": {},
   "outputs": [],
   "source": [
    "gold_recovery_train = gold_recovery_train.drop(\n",
    "    gold_recovery_train.columns[gold_recovery_train.columns.str.endswith(\"total\")],\n",
    "    axis=1,\n",
    ")\n",
    "\n",
    "# Verifying the correctness of columns deletion\n",
    "assert (\n",
    "    gold_recovery_train.columns.shape[0] == gold_recovery_full.columns.shape[0]\n",
    "), \"Number of columns discrepancy\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fd3beda3-79c4-4681-b1cc-5ea492f65ae3",
   "metadata": {},
   "source": [
    "**Conclusion**\n",
    "\n",
    "At this stage, exploratory data analysis of the data was carried out, during which we studied the concentration of the main chemical elements involved in the purification process (gold, silver, lead) at various stages, analyzed the size of granules in both training and test samples, and also considered changes in the stages of the total concentration. The analysis helped to identify a number of abnormally low values, which we managed to get rid of."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5d4771b2-6153-4ae1-8f85-48d724a07e55",
   "metadata": {},
   "source": [
    "# Modelling"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "500840dc-adca-4496-82b3-42d4ba0b9c96",
   "metadata": {},
   "source": [
    "## Final data preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f0f9dd6-8b88-4396-80c0-96044acac4ec",
   "metadata": {},
   "source": [
    "Before proceeding to the training models, we need to make sure that the features in the training and test sets match. Earlier it was noticed that this is not the case, since some features cannot yet be calculated and measured. Therefore, we will filter the training data set so that it has the same columns as the test set does."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "a6ef5b61-45ba-4e84-b2d4-a15d261efbd2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Filtering training set\n",
    "gold_recovery_train = gold_recovery_train[gold_recovery_test.columns]\n",
    "\n",
    "# Verifying the correctness of changes made\n",
    "assert (\n",
    "    np.mean(gold_recovery_train.columns == gold_recovery_test.columns) == 1\n",
    "), \"Columns differ in two sets\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab76c15e-49c3-4bfd-96fc-888b5b4cc921",
   "metadata": {},
   "source": [
    "We did not get any errors so we can be sure that two sets are now homogeneous. Lastly, let's separate features from targets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "688cfd82-0045-4795-ad56-02dfa1629fb8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separating features from targets in the training set\n",
    "features_train = gold_recovery_train.drop(\n",
    "    [target_name_rougher, target_name_final], axis=1\n",
    ").values\n",
    "target_train_rougher = gold_recovery_train[target_name_rougher].values\n",
    "target_train_final = gold_recovery_train[target_name_final].values\n",
    "\n",
    "# Separating features from targets in the test set\n",
    "features_test = gold_recovery_test.drop(\n",
    "    [target_name_rougher, target_name_final], axis=1\n",
    ").values\n",
    "target_test_rougher = gold_recovery_test[target_name_rougher].values\n",
    "target_test_final = gold_recovery_test[target_name_final].values"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ca0ebe79-0320-41a0-bd4d-a59ad80449c5",
   "metadata": {},
   "source": [
    "## *sMAPE* metric"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fe49dbe4-d7bd-47d7-bdb1-c10031d4c51b",
   "metadata": {},
   "source": [
    "We need to focus on the target metric in more detail. Since we will consider the *sMAPE* metric, we will write a function that will perform the calculation. The metric is calculated as follows:\n",
    "\n",
    "$$\n",
    "sMAPE =  \\frac{1}{N} \\sum_{i=1}^{N} \\frac{|y_i - \\hat{y}_i|}{0.5 \\times (|y_i| + |\\hat{y}_i|)} \\times 100 \\%,\n",
    "$$\n",
    "\n",
    "where $y_i$ is real target data, $\\hat{y}_i$ - predicted value for a target and $N$ - number of observations. The overall value of  $sMAPE$ will consist of two computed *sMAPE* metrics for `rougher` and `final`:\n",
    "\n",
    "$$\n",
    "sMAPE_{overall} = 0.25 \\times sMAPE_{rougher} + 0.75 \\times sMAPE_{final}\n",
    "$$\n",
    "\n",
    "Thus, we will define a function, which we will call `smape()` and which will calculate the value of this metric for certain real target values and corresponding predictions. Additionally, we will create a second function called `smape_overall()` for computing the value of the final target metric."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "b793a0ab-5144-4e1f-b331-9d7772b30af0",
   "metadata": {},
   "outputs": [],
   "source": [
    "def smape(y_real, y_pred):\n",
    "    \"\"\"Calculates the value of sMAPE metric.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    y_real : MatrixLike or ArrayLike\n",
    "        Real target data.\n",
    "    y_pred : MatrixLike or ArrayLike\n",
    "        Predicted values of target.\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    float\n",
    "        Floating point number being the value of sMAPE metric.\n",
    "    \"\"\"\n",
    "    smape = (\n",
    "        np.mean(np.abs(y_real - y_pred) / ((np.abs(y_real) + np.abs(y_pred)) / 2)) * 100\n",
    "    )\n",
    "\n",
    "    return smape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "2ed2b9ab-967a-4cf9-926e-bc8d91d17d54",
   "metadata": {},
   "outputs": [],
   "source": [
    "def smape_overall(smape_rougher, smape_final, weight_rougher=0.25, weight_final=0.75):\n",
    "    \"\"\"Computes the value of the final sMAPE metric.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    smape_rougher : float\n",
    "        Value of sMAPE computed for rougher target.\n",
    "    smape_final : float\n",
    "        Value of sMAPE computed for final target.\n",
    "    weight_rougher : float, default=0.25\n",
    "        Weight given to sMAPE computed for rougher target.\n",
    "    weight_final : float, default=0.75\n",
    "        Weight given to sMAPE computed for final target.\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    float\n",
    "        Floating point value of the overall sMAPE metric.\n",
    "    \"\"\"\n",
    "    smape_overall = np.sum([weight_rougher * smape_rougher, weight_final * smape_final])\n",
    "\n",
    "    return smape_overall"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b794f623-a2dc-45d3-99c4-241280e4be92",
   "metadata": {},
   "source": [
    "Now let's move on to the actual modeling process. We will consider three regression models:\n",
    "\n",
    "* *Linear Regression*\n",
    "* *Decision Tree Regressor*\n",
    "* *Random Forest Regressor*\n",
    "\n",
    "We will train models based on cross-validation, where our `smape_score()` function will be used as the target metric. However, in order to correctly use the `sklearn` cross-validation, we need to define a scorer function that can be set using the `make_scorer()` function. Moreover, when defining it, we specify `greater_is_better=False`, since *sMAPE* is essentially a loss that should take as low values as possible."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "0987ea8b-dfc6-490c-a5e2-a154b62378b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining a scorer\n",
    "smape_scorer = make_scorer(smape, greater_is_better=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "65d94b14-c2ac-4cc1-9766-04a90b6e8e32",
   "metadata": {},
   "source": [
    "It should be noted that it will be interesting to consider different hyperparameters during cross-validation. We cannot do grid search for `LinearRegression` but we can try it for tree-based models. We will thus write `run_grid_search` function that will help with that without the need to duplicate the code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "638bf3ad-a1a8-4770-87ca-85808d2b56c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "def run_grid_search(model, grid):\n",
    "    \"\"\"Launches grid search.\n",
    "    \n",
    "    The function makes use of a grid search\n",
    "    algorithm with cross-validation for two\n",
    "    target vectors in question.\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    model : BaseEstimator\n",
    "        Estimator to be used in hyperparameter\n",
    "        search.\n",
    "    grid : dict\n",
    "        Grid of hyperparameter names with their\n",
    "        values to be varied in grid search.\n",
    "        \n",
    "    Returns\n",
    "    -------\n",
    "    BaseEstimator, float, dict\n",
    "        Function outputs the best model, best\n",
    "        value of sMAPE metric and an optimal\n",
    "        hyperparameter configuration.\n",
    "    \"\"\"\n",
    "    # Transforming a grid into a list of all hyperparameter values combinations\n",
    "    keys, values = zip(*grid.items())\n",
    "    model_config_list = [dict(zip(keys, v)) for v in itertools.product(*values)]\n",
    "    \n",
    "    # Initializing starting references for model comparison\n",
    "    best_model = None\n",
    "    best_smape = np.inf\n",
    "    best_config = {}\n",
    "    \n",
    "    # Starting the grid search algorithm\n",
    "    for i in notebook.tqdm(range(len(model_config_list))):\n",
    "        \n",
    "        # Reinitializing the model for the next iteration\n",
    "        candidate_model = clone(model)\n",
    "        candidate_model.set_params(**model_config_list[i])\n",
    "\n",
    "        # Cross-validation for \"rougher\" target\n",
    "        smape_scores_rougher = cross_val_score(\n",
    "            estimator=candidate_model,\n",
    "            X=features_train,\n",
    "            y=target_train_rougher,\n",
    "            cv=CV_FOLDS,\n",
    "            scoring=smape_scorer,\n",
    "            n_jobs=-1,\n",
    "        )\n",
    "\n",
    "        # Cross-validation for \"final\" target\n",
    "        smape_scores_final = cross_val_score(\n",
    "            estimator=candidate_model,\n",
    "            X=features_train,\n",
    "            y=target_train_final,\n",
    "            cv=CV_FOLDS,\n",
    "            scoring=smape_scorer,\n",
    "            n_jobs=-1,\n",
    "        )\n",
    "\n",
    "        # Computing average scores\n",
    "        smape_score_rougher = np.mean(smape_scores_rougher) \n",
    "        smape_score_final = np.mean(smape_scores_final)\n",
    "\n",
    "        # Computing overall sMAPE\n",
    "        smape_score_overall = smape_overall(\n",
    "            smape_rougher=smape_score_rougher, smape_final=smape_score_final\n",
    "        )\n",
    "        smape = np.abs(smape_score_overall)\n",
    "        \n",
    "        # Making a decision about the best model\n",
    "        if smape < best_smape:\n",
    "            best_smape = smape\n",
    "            best_model = candidate_model\n",
    "            best_config = model_config_list[i]\n",
    "\n",
    "    return best_model, best_smape, best_config"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85cfde10-f519-4b46-9d56-4676ef151303",
   "metadata": {},
   "source": [
    "## *LinearRegression*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "87f9c1f4-5d87-46bf-b168-d2d7731d06cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining a scaler\n",
    "scaler = StandardScaler()\n",
    "\n",
    "# Defining a model\n",
    "lin_reg_model = LinearRegression()\n",
    "\n",
    "# Chaining estimators together via a pipeline\n",
    "lin_reg_pipeline = make_pipeline(scaler, lin_reg_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "74541281-bce1-412b-90bd-407083d24910",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Cross-validation for \"rougher\" target\n",
    "smape_scores_rougher = cross_val_score(\n",
    "    lin_reg_pipeline,\n",
    "    features_train,\n",
    "    target_train_rougher,\n",
    "    cv=CV_FOLDS,\n",
    "    scoring=smape_scorer,\n",
    ")\n",
    "\n",
    "# Cross-validation for \"final\" target\n",
    "smape_scores_final = cross_val_score(\n",
    "    lin_reg_pipeline,\n",
    "    features_train,\n",
    "    target_train_final,\n",
    "    cv=CV_FOLDS,\n",
    "    scoring=smape_scorer,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "cf276ae7-f182-4fb7-8b2a-0be4e0eb78f7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sMAPE (Linear regression): 12.64%\n"
     ]
    }
   ],
   "source": [
    "# Computing average scores\n",
    "smape_score_rougher, smape_score_final = np.mean(smape_scores_rougher), np.mean(\n",
    "    smape_scores_final\n",
    ")\n",
    "\n",
    "# Computing overall sMAPE\n",
    "smape_score_overall = smape_overall(\n",
    "    smape_rougher=smape_score_rougher, smape_final=smape_score_final\n",
    ")\n",
    "smape_LR = np.abs(smape_score_overall)\n",
    "\n",
    "print(\"sMAPE (Linear regression): {:.2f}%\".format(smape_LR))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6be5eb03-5270-4eee-a7e6-44633e7d29bc",
   "metadata": {},
   "source": [
    "## *DecisionTreeRegressor*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "6efc8662-e65b-422f-8363-c3ece82f494a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining a Decision tree model\n",
    "dt_model = DecisionTreeRegressor(random_state=RANDOM_STATE)\n",
    "\n",
    "# Chaining estimators\n",
    "dt_pipeline = make_pipeline(scaler, dt_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "86a8f10a-affc-4d5e-bcdc-85c76641d053",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "29fa8623929e44fca8e6fd1197e35828",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/10 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sMAPE (Decision tree): 10.04%\n"
     ]
    }
   ],
   "source": [
    "# Defining a hyperparameter grid\n",
    "dt_params_grid = {\"decisiontreeregressor__max_depth\": np.arange(1, 11)}\n",
    "\n",
    "# Running grid search\n",
    "best_model_DTR, best_smape_DTR, best_params_DTR = run_grid_search(\n",
    "    model=dt_pipeline, grid=dt_params_grid\n",
    ")\n",
    "\n",
    "# Displaying the best score\n",
    "print(f\"sMAPE (Decision tree): {best_smape_DTR:.2f}%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "bb6c68fa-f985-4e3a-b331-c35a093925b8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'decisiontreeregressor__max_depth': 1}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Displaying the best hyperparamerer combination\n",
    "display(best_params_DTR)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "66e11658-93c6-451d-93eb-0528d130a41d",
   "metadata": {},
   "source": [
    "## *RandomForestRegressor*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "b931cf46-e77d-4f04-93b8-c6332ffb53d6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining a Random forest model\n",
    "rf_model = RandomForestRegressor(random_state=RANDOM_STATE)\n",
    "\n",
    "# Chaining estimators\n",
    "rf_pipeline = make_pipeline(scaler, rf_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "3a90c4ad-3e14-4b60-8511-f42b17dcc464",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b23e0c6b9878453fa851f49e6d115815",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/9 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sMAPE (Random forest): 9.76%\n"
     ]
    }
   ],
   "source": [
    "# Defining a hyperparameter grid\n",
    "rf_params_grid = {\n",
    "    \"randomforestregressor__max_depth\": np.arange(1, 4),\n",
    "    \"randomforestregressor__n_estimators\": [10, 20, 30],\n",
    "}\n",
    "\n",
    "# Running grid search\n",
    "best_model_RFR, best_smape_RFR, best_params_RFR = run_grid_search(\n",
    "    model=rf_pipeline, grid=rf_params_grid\n",
    ")\n",
    "\n",
    "# Displaying the best score\n",
    "print(f\"sMAPE (Random forest): {best_smape_RFR:.2f}%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "9808f6f5-8155-44c3-839e-9c57adaf2b83",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'randomforestregressor__max_depth': 2,\n",
       " 'randomforestregressor__n_estimators': 30}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Displaying the best hyperparameter combination\n",
    "display(best_params_RFR)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "096ed41f-3e8a-43e1-9998-a19dd7cde5d5",
   "metadata": {},
   "source": [
    "## Model comparison"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0564d7e9-f848-48e7-8b67-eb1c2a9bb1a6",
   "metadata": {},
   "source": [
    "Now, let's compare the results of the modeling and create a table summarizing the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "d2054bc5-b1a0-4eb9-9081-4b5749079c79",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Random forest         9.76\n",
       "Decision tree        10.04\n",
       "Linear regression    12.64\n",
       "dtype: float64"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Combining the results together\n",
    "grid_search_results = dict()\n",
    "grid_search_results[\"Linear regression\"] = smape_LR\n",
    "grid_search_results[\"Decision tree\"] = best_smape_DTR\n",
    "grid_search_results[\"Random forest\"] = best_smape_RFR\n",
    "\n",
    "# Transforming a dict to Series\n",
    "grid_search_results_series = pd.Series(grid_search_results)\n",
    "\n",
    "# Displaying the final results\n",
    "display(grid_search_results_series.sort_values().round(2))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a85a7cf4-6dc4-4b0b-914d-e994eea1b281",
   "metadata": {},
   "source": [
    "The results show that the best model proves to be a Random Forest with 30 trees with depth of 2 each. The model's value of *sMAPE* metric is 9.76%."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "7c5ad2ee-c684-43d6-9f3b-445057f94e97",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Choosing the best model\n",
    "best_model = clone(best_model_RFR)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8cea9ad5-1c1c-4192-a676-c84a6883d579",
   "metadata": {},
   "source": [
    "# Testing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2431ad60-3193-4e65-92ad-c3647277e6e3",
   "metadata": {},
   "source": [
    "Now, let's test the model we chose and compute the target metric on the test set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "464bcf18-e421-4924-8c8a-7d06d1eab89a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test sMAPE: 9.08%\n",
      "CPU times: total: 17.5 s\n",
      "Wall time: 17.5 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "# Computing rougher sMAPE\n",
    "best_model.fit(features_train, target_train_rougher)\n",
    "predictions_test_rougher = best_model.predict(features_test)\n",
    "smape_score_test_rougher = smape(\n",
    "    y_real=target_test_rougher, y_pred=predictions_test_rougher\n",
    ")\n",
    "\n",
    "# Computing final sMAPE\n",
    "best_model.fit(features_train, target_train_final)\n",
    "predictions_test_final = best_model.predict(features_test)\n",
    "smape_score_test_final = smape(y_real=target_test_final, y_pred=predictions_test_final)\n",
    "\n",
    "# Computing overall sMAPE\n",
    "smape_score_test_overall = smape_overall(\n",
    "    smape_rougher=smape_score_test_rougher, smape_final=smape_score_test_final\n",
    ")\n",
    "\n",
    "# Printing testing results\n",
    "print(\"Test sMAPE: {:.2f}%\".format(smape_score_test_overall))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "88c45585-d76d-475b-be97-7d9f2404f6c4",
   "metadata": {},
   "source": [
    "Additionally, let's compare the results of the model with constant model which we can configure using `DummyRegressor()`. In this case, we will compare test predictions with median target values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "e1398475-dcfa-4c5e-aac3-49ee037e62ce",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dummy sMAPE: 9.44%\n"
     ]
    }
   ],
   "source": [
    "# Comparison\n",
    "dummy_reg_model = DummyRegressor(strategy=\"median\")\n",
    "\n",
    "dummy_reg_model.fit(features_train, target_train_rougher)\n",
    "predictions_dummy_rougher = dummy_reg_model.predict(features_test)\n",
    "smape_score_dummy_rougher = smape(\n",
    "    y_real=target_test_rougher, y_pred=predictions_dummy_rougher\n",
    ")\n",
    "\n",
    "dummy_reg_model.fit(features_train, target_train_final)\n",
    "predictions_dummy_final = dummy_reg_model.predict(features_test)\n",
    "smape_score_dummy_final = smape(\n",
    "    y_real=target_test_final, y_pred=predictions_dummy_final\n",
    ")\n",
    "\n",
    "smape_score_dummy_overall = smape_overall(\n",
    "    smape_rougher=smape_score_dummy_rougher, smape_final=smape_score_dummy_final\n",
    ")\n",
    "\n",
    "# Verifying the adequacy of the model\n",
    "assert smape_score_dummy_overall > smape_score_test_overall, \"Model is inadequate\"\n",
    "\n",
    "print(\"Dummy sMAPE: {:.2f}%\".format(smape_score_dummy_overall))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "029ca83e-081a-4839-b1bc-f319a8d5356a",
   "metadata": {},
   "source": [
    "# General conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da345eda-9c6b-4824-a30f-83333765bb79",
   "metadata": {},
   "source": [
    "Within the framework of this project, data on the purification of gold-bearing ore were analyzed, as a result of which a data overview, preprocessing, as well as exploratory data analysis were carried out. Furthermore, we have built several models in order to predict the recovery rate of gold. According to the modelling results, the best model showing the smallest values of the target metric *sMAPE* is a Random Forest model with 30 trees with a depth of 2."
   ]
  }
 ],
 "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": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}