[1-1]_RF_model_build_density.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Random forest regression with hyperparameter tuning by GPyOpt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import library and dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import sys\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.metrics import mean_squared_error, r2_score\n",
    "\n",
    "import GPy, GPyOpt\n",
    "from GPyOpt.methods import BayesianOptimization\n",
    "\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(594, 12)\n"
     ]
    },
    {
     "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>Kod</th>\n",
       "      <th>GlasNo</th>\n",
       "      <th>SiO2</th>\n",
       "      <th>B2O3</th>\n",
       "      <th>Al2O3</th>\n",
       "      <th>MgO</th>\n",
       "      <th>CaO</th>\n",
       "      <th>Na2O</th>\n",
       "      <th>K2O</th>\n",
       "      <th>E</th>\n",
       "      <th>density</th>\n",
       "      <th>Tg</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2498</td>\n",
       "      <td>34459</td>\n",
       "      <td>71.01</td>\n",
       "      <td>0.45</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.0</td>\n",
       "      <td>13.01</td>\n",
       "      <td>12.13</td>\n",
       "      <td>3.08</td>\n",
       "      <td>71.8</td>\n",
       "      <td>2.51</td>\n",
       "      <td>540.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2498</td>\n",
       "      <td>34460</td>\n",
       "      <td>71.77</td>\n",
       "      <td>11.09</td>\n",
       "      <td>0.25</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>9.76</td>\n",
       "      <td>6.29</td>\n",
       "      <td>82.5</td>\n",
       "      <td>2.52</td>\n",
       "      <td>538.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3714</td>\n",
       "      <td>10815</td>\n",
       "      <td>75.01</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>10.69</td>\n",
       "      <td>14.30</td>\n",
       "      <td>0.00</td>\n",
       "      <td>77.5</td>\n",
       "      <td>2.46</td>\n",
       "      <td>548.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3714</td>\n",
       "      <td>10817</td>\n",
       "      <td>69.93</td>\n",
       "      <td>5.08</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>10.71</td>\n",
       "      <td>14.29</td>\n",
       "      <td>0.00</td>\n",
       "      <td>80.4</td>\n",
       "      <td>2.51</td>\n",
       "      <td>563.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3724</td>\n",
       "      <td>10860</td>\n",
       "      <td>52.50</td>\n",
       "      <td>24.30</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>23.20</td>\n",
       "      <td>0.00</td>\n",
       "      <td>80.4</td>\n",
       "      <td>2.56</td>\n",
       "      <td>529.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Kod  GlasNo   SiO2   B2O3  Al2O3  MgO    CaO   Na2O   K2O     E  density  \\\n",
       "0  2498   34459  71.01   0.45   0.31  0.0  13.01  12.13  3.08  71.8     2.51   \n",
       "1  2498   34460  71.77  11.09   0.25  0.0   0.00   9.76  6.29  82.5     2.52   \n",
       "2  3714   10815  75.01   0.00   0.00  0.0  10.69  14.30  0.00  77.5     2.46   \n",
       "3  3714   10817  69.93   5.08   0.00  0.0  10.71  14.29  0.00  80.4     2.51   \n",
       "4  3724   10860  52.50  24.30   0.00  0.0   0.00  23.20  0.00  80.4     2.56   \n",
       "\n",
       "      Tg  \n",
       "0  540.0  \n",
       "1  538.0  \n",
       "2  548.0  \n",
       "3  563.0  \n",
       "4  529.0  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load dataset\n",
    "df = pd.read_csv('./data_common/df_source.csv')\n",
    "print(df.shape)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(594, 12)\n"
     ]
    },
    {
     "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>Kod</th>\n",
       "      <th>GlasNo</th>\n",
       "      <th>SiO2</th>\n",
       "      <th>B2O3</th>\n",
       "      <th>Al2O3</th>\n",
       "      <th>MgO</th>\n",
       "      <th>CaO</th>\n",
       "      <th>Na2O</th>\n",
       "      <th>K2O</th>\n",
       "      <th>E</th>\n",
       "      <th>density</th>\n",
       "      <th>Tg</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>240</th>\n",
       "      <td>39756</td>\n",
       "      <td>465909</td>\n",
       "      <td>40.00</td>\n",
       "      <td>40.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>20.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>75.1</td>\n",
       "      <td>2.42</td>\n",
       "      <td>545.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>132</th>\n",
       "      <td>27734</td>\n",
       "      <td>359612</td>\n",
       "      <td>72.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>14.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>14.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>70.2</td>\n",
       "      <td>2.45</td>\n",
       "      <td>551.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424</th>\n",
       "      <td>42246</td>\n",
       "      <td>483234</td>\n",
       "      <td>52.60</td>\n",
       "      <td>20.60</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>26.80</td>\n",
       "      <td>0.00</td>\n",
       "      <td>80.3</td>\n",
       "      <td>2.54</td>\n",
       "      <td>535.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>16195</td>\n",
       "      <td>134085</td>\n",
       "      <td>79.40</td>\n",
       "      <td>10.44</td>\n",
       "      <td>2.48</td>\n",
       "      <td>0.78</td>\n",
       "      <td>1.13</td>\n",
       "      <td>4.08</td>\n",
       "      <td>0.67</td>\n",
       "      <td>66.0</td>\n",
       "      <td>2.34</td>\n",
       "      <td>550.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>170</th>\n",
       "      <td>30511</td>\n",
       "      <td>401108</td>\n",
       "      <td>67.35</td>\n",
       "      <td>0.00</td>\n",
       "      <td>10.49</td>\n",
       "      <td>5.29</td>\n",
       "      <td>0.35</td>\n",
       "      <td>13.45</td>\n",
       "      <td>2.40</td>\n",
       "      <td>72.0</td>\n",
       "      <td>2.43</td>\n",
       "      <td>623.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Kod  GlasNo   SiO2   B2O3  Al2O3    MgO   CaO   Na2O   K2O     E  \\\n",
       "240  39756  465909  40.00  40.00   0.00   0.00  0.00  20.00  0.00  75.1   \n",
       "132  27734  359612  72.00   0.00   0.00  14.00  0.00  14.00  0.00  70.2   \n",
       "424  42246  483234  52.60  20.60   0.00   0.00  0.00  26.80  0.00  80.3   \n",
       "92   16195  134085  79.40  10.44   2.48   0.78  1.13   4.08  0.67  66.0   \n",
       "170  30511  401108  67.35   0.00  10.49   5.29  0.35  13.45  2.40  72.0   \n",
       "\n",
       "     density     Tg  \n",
       "240     2.42  545.0  \n",
       "132     2.45  551.0  \n",
       "424     2.54  535.0  \n",
       "92      2.34  550.0  \n",
       "170     2.43  623.0  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Shuffle the dataset just in case\n",
    "\n",
    "df_s = df.sample(frac = 1, random_state = 0)\n",
    "print(df_s.shape)\n",
    "df_s.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Column the composition starts =  2\n",
      "Column the composition ends   =  9\n"
     ]
    }
   ],
   "source": [
    "# Get column number for computation\n",
    "\n",
    "start = df_s.columns.get_loc('SiO2')\n",
    "end   = df_s.columns.get_loc('K2O') + 1\n",
    "print('Column the composition starts = ', start)\n",
    "print('Column the composition ends   = ', end)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define X variables\n",
    "\n",
    "X = df_s.iloc[:, start:end]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(594, 7)\n"
     ]
    },
    {
     "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>SiO2</th>\n",
       "      <th>B2O3</th>\n",
       "      <th>Al2O3</th>\n",
       "      <th>MgO</th>\n",
       "      <th>CaO</th>\n",
       "      <th>Na2O</th>\n",
       "      <th>K2O</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>240</th>\n",
       "      <td>40.00</td>\n",
       "      <td>40.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>20.00</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>132</th>\n",
       "      <td>72.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>14.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>14.00</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424</th>\n",
       "      <td>52.60</td>\n",
       "      <td>20.60</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>26.80</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>79.40</td>\n",
       "      <td>10.44</td>\n",
       "      <td>2.48</td>\n",
       "      <td>0.78</td>\n",
       "      <td>1.13</td>\n",
       "      <td>4.08</td>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>170</th>\n",
       "      <td>67.35</td>\n",
       "      <td>0.00</td>\n",
       "      <td>10.49</td>\n",
       "      <td>5.29</td>\n",
       "      <td>0.35</td>\n",
       "      <td>13.45</td>\n",
       "      <td>2.40</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      SiO2   B2O3  Al2O3    MgO   CaO   Na2O   K2O\n",
       "240  40.00  40.00   0.00   0.00  0.00  20.00  0.00\n",
       "132  72.00   0.00   0.00  14.00  0.00  14.00  0.00\n",
       "424  52.60  20.60   0.00   0.00  0.00  26.80  0.00\n",
       "92   79.40  10.44   2.48   0.78  1.13   4.08  0.67\n",
       "170  67.35   0.00  10.49   5.29  0.35  13.45  2.40"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(X.shape)\n",
    "X.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define target y value\n",
    "\n",
    "def choose_y(df, property):\n",
    "    y_value  = df[property]\n",
    "    return y_value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "property_name = 'density'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(594,)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "240    2.42\n",
       "132    2.45\n",
       "424    2.54\n",
       "92     2.34\n",
       "170    2.43\n",
       "Name: density, dtype: float64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y = choose_y(df_s, property_name)\n",
    "print(y.shape)\n",
    "y.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(475, 7) (119, 7) (475,) (119,)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)\n",
    "print(X_train.shape, X_test.shape, y_train.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Default"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestRegressor()"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rfr = RandomForestRegressor()\n",
    "rfr.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R2 =  0.8009129054449335\n"
     ]
    }
   ],
   "source": [
    "pred_rfr = rfr.predict(X_test)\n",
    "R2_rfr = r2_score(pred_rfr, y_test)\n",
    "print('R2 = ', R2_rfr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Hyperparameter tuning using GPyOpt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# GPyOpt for RF regression\n",
    "\n",
    "bounds_rfr = [\n",
    "    {'name': 'n_estimators', 'type': 'discrete', 'domain': tuple(range(5, 500))},\n",
    "    {'name': 'max_depth', 'type': 'discrete', 'domain': tuple(range(1, 300))},\n",
    "    {'name': 'min_sample_split', 'type': 'discrete', 'domain': tuple(range(2, 20))},\n",
    "    {'name': 'min_samples_leaf', 'type': 'discrete', 'domain': tuple(range(1, 10))}\n",
    "]\n",
    "\n",
    "# Optimization objective function\n",
    "\n",
    "def rmse_rfr(*args):\n",
    "    params = args[0]\n",
    "    \n",
    "    rfr = RandomForestRegressor(n_estimators = int(params[0][0]),\n",
    "                                max_depth = int(params[0][1]),\n",
    "                                min_samples_split = int(params[0][2]),\n",
    "                                min_samples_leaf = int(params[0][3]),\n",
    "                                bootstrap = True,\n",
    "                                random_state = 0,\n",
    "                                criterion = 'mse')\n",
    "                \n",
    "    rfr.fit(X_train, y_train)\n",
    "    \n",
    "    pred_rfr = rfr.predict(X_test)\n",
    "    RMSE = np.sqrt(mean_squared_error(pred_rfr, y_test))\n",
    "    \n",
    "    return RMSE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 2min 17s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "optimizer = BayesianOptimization(f = rmse_rfr, domain = bounds_rfr)\n",
    "optimizer.run_optimization(max_iter = 100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([459., 202.,   2.,   1.])"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "optimizer.x_opt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R2 =  0.8188\n"
     ]
    }
   ],
   "source": [
    "rfr_best = RandomForestRegressor(n_estimators = int(optimizer.x_opt[0]),\n",
    "                                 max_depth = int(optimizer.x_opt[1]),\n",
    "                                 min_samples_split = int(optimizer.x_opt[2]),\n",
    "                                 min_samples_leaf = int(optimizer.x_opt[3]),\n",
    "                                 bootstrap = True,\n",
    "                                 random_state = 0,\n",
    "                                 criterion = 'mse')\n",
    "\n",
    "rfr_best.fit(X_train, y_train)\n",
    "y_pred_rfr = rfr_best.predict(X_test)\n",
    "\n",
    "print('R2 = ', '{:.4}'.format(r2_score(y_pred_rfr, y_test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "optimizer.plot_convergence()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check model performance by CV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R2 =   0.9808\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 288x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Use all data\n",
    "X_train, y_train = X, y\n",
    "\n",
    "rfr_best = RandomForestRegressor(n_estimators = int(optimizer.x_opt[0]),\n",
    "                                 max_depth = int(optimizer.x_opt[1]),\n",
    "                                 min_samples_split = int(optimizer.x_opt[2]),\n",
    "                                 min_samples_leaf = int(optimizer.x_opt[3]),\n",
    "                                 bootstrap = True,\n",
    "                                 random_state = 0,\n",
    "                                 criterion = 'mse')\n",
    "\n",
    "rfr_best.fit(X_train, y_train)\n",
    "y_pred = rfr_best.predict(X_train)\n",
    "\n",
    "print('R2 = ', ' {:.4}'.format(r2_score(y, y_pred)))\n",
    "\n",
    "plt.figure(figsize = [4, 4])\n",
    "slp_1_begin = 0.99 * y.min()\n",
    "slp_1_end   = 1.01 * y.max()\n",
    "\n",
    "plt.scatter(y, y_pred, c = 'r', alpha = 0.5)\n",
    "plt.plot([slp_1_begin, slp_1_end], [slp_1_begin, slp_1_end], c = 'b')\n",
    "plt.title('Fitting check')\n",
    "plt.xlabel('Observed value')\n",
    "plt.ylabel('Predicted value')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save the best model\n",
    "\n",
    "import pickle\n",
    "with open('./data_common/best_RF_density.pickle', mode='wb') as f:\n",
    "    pickle.dump(rfr_best, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}