feature_importance.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import ppscore as pps\n",
    "\n",
    "path = r\"C:\\Users\\Roman\\Documents\\CleanData\\NormByWell\"\n",
    "\n",
    "df = pd.read_csv(r\"C:\\Users\\Roman\\Documents\\CleanData\\NormByWell\\tubesProfiles.csv\").drop(columns=['Unnamed: 0', 'Row','Column','PlateName','Nucleus Width [µm] (2)'])\n",
    "df.rename(columns={'Compound':'y'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>x</th>\n",
       "      <th>y</th>\n",
       "      <th>ppscore</th>\n",
       "      <th>case</th>\n",
       "      <th>is_valid_score</th>\n",
       "      <th>metric</th>\n",
       "      <th>baseline_score</th>\n",
       "      <th>model_score</th>\n",
       "      <th>model</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Cell 488 Threshold Compactness 50% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.197921</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.238388</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Nucleus Width [µm]</td>\n",
       "      <td>y</td>\n",
       "      <td>0.196459</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.236999</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Cell 488 Threshold Compactness 40% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.181294</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.222600</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Cell Mito Threshold Compactness 50% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.179839</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.221218</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Cell 488 Threshold Compactness 30% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.175212</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.216824</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Cell Mito Threshold Compactness 60% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.174113</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.215781</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Nucleus 33342 Threshold Compactness 40% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.171281</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.213092</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Cell 568 Threshold Compactness 50% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.169570</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.211467</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Nucleus 33342 Threshold Compactness 30% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.168557</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.210505</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Cell 488 Threshold Compactness 60% SER-Valley</td>\n",
       "      <td>y</td>\n",
       "      <td>0.167489</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.209491</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Cell 488 Threshold Compactness 60% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.167012</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.209038</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>Cell 488 Threshold Compactness 60% SER-Saddle</td>\n",
       "      <td>y</td>\n",
       "      <td>0.160832</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.203170</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>Cell 568 Threshold Compactness 40% SER-Spot</td>\n",
       "      <td>y</td>\n",
       "      <td>0.160227</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.202595</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>Cell 488 Threshold Compactness 60% SER-Dark</td>\n",
       "      <td>y</td>\n",
       "      <td>0.159956</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.202338</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Cell 568 Threshold Compactness 40% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.159593</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.201993</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>Cell 568 Threshold Compactness 60% SER-Dark</td>\n",
       "      <td>y</td>\n",
       "      <td>0.159446</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.201854</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>Cell 568 Threshold Compactness 60% SER-Hole</td>\n",
       "      <td>y</td>\n",
       "      <td>0.158359</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.200822</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>Cell 568 Threshold Compactness 60% SER-Bright</td>\n",
       "      <td>y</td>\n",
       "      <td>0.156604</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.199155</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>Nucleus 33342 Threshold Compactness 60% SER-Ridge</td>\n",
       "      <td>y</td>\n",
       "      <td>0.154596</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.197249</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>Cell 568 Threshold Compactness 50% SER-Spot</td>\n",
       "      <td>y</td>\n",
       "      <td>0.151984</td>\n",
       "      <td>classification</td>\n",
       "      <td>True</td>\n",
       "      <td>weighted F1</td>\n",
       "      <td>0.050452</td>\n",
       "      <td>0.194768</td>\n",
       "      <td>DecisionTreeClassifier()</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                    x  y   ppscore  \\\n",
       "0         Cell 488 Threshold Compactness 50% SER-Hole  y  0.197921   \n",
       "1                                  Nucleus Width [µm]  y  0.196459   \n",
       "2         Cell 488 Threshold Compactness 40% SER-Hole  y  0.181294   \n",
       "3        Cell Mito Threshold Compactness 50% SER-Hole  y  0.179839   \n",
       "4         Cell 488 Threshold Compactness 30% SER-Hole  y  0.175212   \n",
       "5        Cell Mito Threshold Compactness 60% SER-Hole  y  0.174113   \n",
       "6    Nucleus 33342 Threshold Compactness 40% SER-Hole  y  0.171281   \n",
       "7         Cell 568 Threshold Compactness 50% SER-Hole  y  0.169570   \n",
       "8    Nucleus 33342 Threshold Compactness 30% SER-Hole  y  0.168557   \n",
       "9       Cell 488 Threshold Compactness 60% SER-Valley  y  0.167489   \n",
       "10        Cell 488 Threshold Compactness 60% SER-Hole  y  0.167012   \n",
       "11      Cell 488 Threshold Compactness 60% SER-Saddle  y  0.160832   \n",
       "12        Cell 568 Threshold Compactness 40% SER-Spot  y  0.160227   \n",
       "13        Cell 488 Threshold Compactness 60% SER-Dark  y  0.159956   \n",
       "14        Cell 568 Threshold Compactness 40% SER-Hole  y  0.159593   \n",
       "15        Cell 568 Threshold Compactness 60% SER-Dark  y  0.159446   \n",
       "16        Cell 568 Threshold Compactness 60% SER-Hole  y  0.158359   \n",
       "17      Cell 568 Threshold Compactness 60% SER-Bright  y  0.156604   \n",
       "18  Nucleus 33342 Threshold Compactness 60% SER-Ridge  y  0.154596   \n",
       "19        Cell 568 Threshold Compactness 50% SER-Spot  y  0.151984   \n",
       "\n",
       "              case  is_valid_score       metric  baseline_score  model_score  \\\n",
       "0   classification            True  weighted F1        0.050452     0.238388   \n",
       "1   classification            True  weighted F1        0.050452     0.236999   \n",
       "2   classification            True  weighted F1        0.050452     0.222600   \n",
       "3   classification            True  weighted F1        0.050452     0.221218   \n",
       "4   classification            True  weighted F1        0.050452     0.216824   \n",
       "5   classification            True  weighted F1        0.050452     0.215781   \n",
       "6   classification            True  weighted F1        0.050452     0.213092   \n",
       "7   classification            True  weighted F1        0.050452     0.211467   \n",
       "8   classification            True  weighted F1        0.050452     0.210505   \n",
       "9   classification            True  weighted F1        0.050452     0.209491   \n",
       "10  classification            True  weighted F1        0.050452     0.209038   \n",
       "11  classification            True  weighted F1        0.050452     0.203170   \n",
       "12  classification            True  weighted F1        0.050452     0.202595   \n",
       "13  classification            True  weighted F1        0.050452     0.202338   \n",
       "14  classification            True  weighted F1        0.050452     0.201993   \n",
       "15  classification            True  weighted F1        0.050452     0.201854   \n",
       "16  classification            True  weighted F1        0.050452     0.200822   \n",
       "17  classification            True  weighted F1        0.050452     0.199155   \n",
       "18  classification            True  weighted F1        0.050452     0.197249   \n",
       "19  classification            True  weighted F1        0.050452     0.194768   \n",
       "\n",
       "                       model  \n",
       "0   DecisionTreeClassifier()  \n",
       "1   DecisionTreeClassifier()  \n",
       "2   DecisionTreeClassifier()  \n",
       "3   DecisionTreeClassifier()  \n",
       "4   DecisionTreeClassifier()  \n",
       "5   DecisionTreeClassifier()  \n",
       "6   DecisionTreeClassifier()  \n",
       "7   DecisionTreeClassifier()  \n",
       "8   DecisionTreeClassifier()  \n",
       "9   DecisionTreeClassifier()  \n",
       "10  DecisionTreeClassifier()  \n",
       "11  DecisionTreeClassifier()  \n",
       "12  DecisionTreeClassifier()  \n",
       "13  DecisionTreeClassifier()  \n",
       "14  DecisionTreeClassifier()  \n",
       "15  DecisionTreeClassifier()  \n",
       "16  DecisionTreeClassifier()  \n",
       "17  DecisionTreeClassifier()  \n",
       "18  DecisionTreeClassifier()  \n",
       "19  DecisionTreeClassifier()  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ppscores = pps.predictors(df, \"y\")\n",
    "ppscores.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Roman\\AppData\\Local\\Temp\\ipykernel_31168\\355047523.py:7: FutureWarning: The series.append method is deprecated and will be removed from pandas in a future version. Use pandas.concat instead.\n",
      "  features_to_keep = ppscores.loc[ppscores['ppscore'] > 0.05, 'x'].append(pd.Series('y'))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of random forest classifier on test set: 0.37\n",
      "ROC AUC score of random forest classifier on test set: 0.88\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>feature</th>\n",
       "      <th>importance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Cytoplasm 488 Profile 4/5</td>\n",
       "      <td>0.045289</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Intensity Nucleus Alexa 555 Mean</td>\n",
       "      <td>0.043754</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Nucleus Alexa 488 SER Hole 0 px</td>\n",
       "      <td>0.036987</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Cytoplasm 488 Profile 5/5</td>\n",
       "      <td>0.032426</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Intensity Ring Region Alexa 488 Mean</td>\n",
       "      <td>0.032341</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124</th>\n",
       "      <td>Cell 568 Threshold Compactness 60% SER-Edge</td>\n",
       "      <td>0.001261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>125</th>\n",
       "      <td>Cell Mito Threshold Compactness 50% SER-Saddle</td>\n",
       "      <td>0.001241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126</th>\n",
       "      <td>Nucleus 33342 Threshold Compactness 40% SER-Dark</td>\n",
       "      <td>0.001127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127</th>\n",
       "      <td>Cell Mito Threshold Compactness 40% SER-Saddle</td>\n",
       "      <td>0.001014</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>128</th>\n",
       "      <td>Cell 568 Threshold Compactness 40% SER-Ridge</td>\n",
       "      <td>0.000945</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>129 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              feature  importance\n",
       "0                          Cytoplasm 488 Profile 4/5     0.045289\n",
       "1                    Intensity Nucleus Alexa 555 Mean    0.043754\n",
       "2                     Nucleus Alexa 488 SER Hole 0 px    0.036987\n",
       "3                          Cytoplasm 488 Profile 5/5     0.032426\n",
       "4                Intensity Ring Region Alexa 488 Mean    0.032341\n",
       "..                                                ...         ...\n",
       "124       Cell 568 Threshold Compactness 60% SER-Edge    0.001261\n",
       "125    Cell Mito Threshold Compactness 50% SER-Saddle    0.001241\n",
       "126  Nucleus 33342 Threshold Compactness 40% SER-Dark    0.001127\n",
       "127    Cell Mito Threshold Compactness 40% SER-Saddle    0.001014\n",
       "128      Cell 568 Threshold Compactness 40% SER-Ridge    0.000945\n",
       "\n",
       "[129 rows x 2 columns]"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "# filter out features with low ppscore but keep 'y' column\n",
    "features_to_keep = ppscores.loc[ppscores['ppscore'] > 0.05, 'x'].append(pd.Series('y'))\n",
    "df_filt = df[features_to_keep]\n",
    "\n",
    "X = df_filt.drop(columns=['y','Concentration'])\n",
    "y = df_filt['y']\n",
    "X_train,X_test,y_train,y_test = train_test_split(X,y,random_state=42, stratify=y, test_size=0.2)\n",
    "\n",
    "rf = RandomForestClassifier(n_estimators=500, random_state=42, max_depth=5)\n",
    "rf.fit(X_train, y_train)\n",
    "\n",
    "y_pred = pd.Series(rf.predict(X_test))\n",
    "y_pred_proba = rf.predict_proba(X_test)\n",
    "\n",
    "\n",
    "print('Accuracy of random forest classifier on test set: {:.2f}'.format(rf.score(X_test, y_test)))\n",
    "print('ROC AUC score of random forest classifier on test set: {:.2f}'.format(roc_auc_score(y_test, y_pred_proba, multi_class='ovr')))\n",
    "\n",
    "conf_matrix = confusion_matrix(y_test, y_pred)\n",
    "\n",
    "\n",
    "# feature importance\n",
    "feature_importance = pd.DataFrame({'feature': X_train.columns, 'importance': rf.feature_importances_})\n",
    "feature_importance.sort_values(by='importance', ascending=False, inplace=True)\n",
    "feature_importance.reset_index(drop=True, inplace=True)\n",
    "feature_importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'importances_mean': array([ 1.49253731e-03,  2.46268657e-03,  1.86567164e-03,  2.42537313e-03,\n",
       "         2.16417910e-03,  6.23134328e-03,  1.06343284e-02,  4.10447761e-03,\n",
       "         2.98507463e-03, -7.46268657e-05,  5.33582090e-03,  5.41044776e-03,\n",
       "         1.71641791e-03,  1.79104478e-03,  1.11940299e-03,  4.85074627e-04,\n",
       "         4.21641791e-03,  1.34328358e-03,  5.22388060e-04,  1.26865672e-03,\n",
       "         2.83582090e-03,  1.28358209e-02,  7.83582090e-03,  1.26865672e-03,\n",
       "         1.60447761e-03,  1.41791045e-03,  1.97761194e-03,  1.75373134e-03,\n",
       "         4.47761194e-04,  1.41791045e-03,  1.64179104e-03,  6.64179104e-03,\n",
       "         7.42537313e-03,  7.01492537e-03,  1.23134328e-03,  1.41791045e-03,\n",
       "         2.91044776e-03,  5.59701493e-03,  3.54477612e-03,  2.91044776e-03,\n",
       "         2.42537313e-03,  1.60447761e-03,  2.94776119e-03,  3.61940299e-03,\n",
       "         2.20149254e-03,  0.00000000e+00,  1.11940299e-03,  1.82835821e-03,\n",
       "         5.97014925e-04,  8.95522388e-04,  4.10447761e-04,  1.64179104e-03,\n",
       "         2.38805970e-03,  1.08208955e-03,  1.08208955e-03,  4.02985075e-03,\n",
       "         1.19402985e-03,  6.34328358e-03,  1.56716418e-03,  3.99253731e-03,\n",
       "         3.99253731e-03,  6.34328358e-04,  2.27611940e-03,  1.67910448e-03,\n",
       "         1.71641791e-03,  1.41791045e-03,  1.56716418e-03,  2.08955224e-03,\n",
       "         5.33582090e-03,  1.04477612e-03,  6.15671642e-03,  6.60447761e-03,\n",
       "         3.35820896e-03,  2.42537313e-03,  1.52985075e-03,  2.27611940e-03,\n",
       "         7.38805970e-03,  3.43283582e-03,  1.19402985e-03,  1.11940299e-03,\n",
       "         1.45522388e-03,  7.46268657e-04,  2.08955224e-03,  5.41044776e-03,\n",
       "         6.71641791e-04,  7.08955224e-04,  1.71641791e-03,  1.86567164e-04,\n",
       "         1.41791045e-03,  3.88059701e-03,  1.19402985e-03,  8.95522388e-04,\n",
       "         2.79850746e-03,  1.23134328e-03, -3.73134328e-05,  1.80597015e-02,\n",
       "         3.50746269e-03,  4.10447761e-04,  1.11940299e-02,  7.12686567e-03,\n",
       "         2.60074627e-02,  1.97761194e-03,  1.45522388e-03, -1.00746269e-03,\n",
       "         6.41791045e-03,  4.25373134e-03,  2.60447761e-02,  1.82835821e-03,\n",
       "         3.52611940e-02,  8.50746269e-03,  5.52238806e-03, -7.46268657e-05,\n",
       "         1.13059701e-02,  1.58582090e-02,  3.13432836e-03,  1.67910448e-03,\n",
       "         2.00000000e-02,  5.59701493e-04,  5.70895522e-03,  1.19402985e-03,\n",
       "         1.35820896e-02,  1.72761194e-02,  4.40298507e-03,  2.98507463e-04,\n",
       "         1.45522388e-03,  3.95522388e-03,  5.18656716e-03,  3.76865672e-03,\n",
       "         3.91791045e-03]),\n",
       " 'importances_std': array([0.00117995, 0.00065058, 0.00091399, 0.00153167, 0.00049502,\n",
       "        0.00131447, 0.00166244, 0.00124875, 0.0008175 , 0.00046604,\n",
       "        0.00134587, 0.00227122, 0.00094691, 0.00046604, 0.00078269,\n",
       "        0.00070896, 0.00177425, 0.00050614, 0.00073119, 0.00065058,\n",
       "        0.00078624, 0.00276824, 0.00178169, 0.00034198, 0.00078358,\n",
       "        0.0006633 , 0.00110752, 0.00072833, 0.00076105, 0.00131182,\n",
       "        0.00088614, 0.0022807 , 0.00157823, 0.00102595, 0.00088379,\n",
       "        0.00068397, 0.00086387, 0.00140608, 0.0004793 , 0.00064196,\n",
       "        0.00091779, 0.00066853, 0.00063433, 0.001493  , 0.00079066,\n",
       "        0.00050061, 0.00055345, 0.00071677, 0.00038052, 0.00050614,\n",
       "        0.0005888 , 0.00055846, 0.00076833, 0.00038956, 0.00045547,\n",
       "        0.00068397, 0.0006199 , 0.00117995, 0.00074253, 0.0009447 ,\n",
       "        0.00176639, 0.00033582, 0.00061199, 0.00044931, 0.00078624,\n",
       "        0.00074253, 0.00076105, 0.00050614, 0.00132502, 0.00086387,\n",
       "        0.00093284, 0.0016183 , 0.00098722, 0.00129527, 0.00141053,\n",
       "        0.00053943, 0.00099844, 0.00059701, 0.00054839, 0.00062437,\n",
       "        0.0020468 , 0.00104211, 0.00113423, 0.00071287, 0.00072353,\n",
       "        0.00045547, 0.00071189, 0.00087109, 0.00074253, 0.00088614,\n",
       "        0.00064196, 0.00029851, 0.00117107, 0.00044307, 0.00035201,\n",
       "        0.00369836, 0.00120563, 0.00026119, 0.00205732, 0.00254416,\n",
       "        0.00367135, 0.00092985, 0.00061199, 0.001601  , 0.00068397,\n",
       "        0.00160231, 0.00251859, 0.00045547, 0.00281834, 0.0026374 ,\n",
       "        0.00156361, 0.00049502, 0.00217605, 0.00194782, 0.00067164,\n",
       "        0.00062992, 0.00269174, 0.00058405, 0.00196773, 0.00049502,\n",
       "        0.00193311, 0.00254197, 0.0006199 , 0.00046604, 0.00053943,\n",
       "        0.0014944 , 0.00121998, 0.00119694, 0.00153167]),\n",
       " 'importances': array([[0.00373134, 0.00149254, 0.00037313, ..., 0.00037313, 0.00149254,\n",
       "         0.00149254],\n",
       "        [0.00186567, 0.00298507, 0.00298507, ..., 0.00149254, 0.00261194,\n",
       "         0.00335821],\n",
       "        [0.00186567, 0.0011194 , 0.00223881, ..., 0.0011194 , 0.00149254,\n",
       "         0.00186567],\n",
       "        ...,\n",
       "        [0.00485075, 0.00708955, 0.00485075, ..., 0.00447761, 0.00634328,\n",
       "         0.00708955],\n",
       "        [0.00559701, 0.00559701, 0.00373134, ..., 0.00298507, 0.00447761,\n",
       "         0.00149254],\n",
       "        [0.00559701, 0.00485075, 0.00559701, ..., 0.00149254, 0.00447761,\n",
       "         0.00261194]])}"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.inspection import permutation_importance\n",
    "\n",
    "result = permutation_importance(rf, X_train, y_train, n_repeats=10, random_state=42, n_jobs=2)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 1.49253731e-03,  2.46268657e-03,  1.86567164e-03,  2.42537313e-03,\n",
       "        2.16417910e-03,  6.23134328e-03,  1.06343284e-02,  4.10447761e-03,\n",
       "        2.98507463e-03, -7.46268657e-05,  5.33582090e-03,  5.41044776e-03,\n",
       "        1.71641791e-03,  1.79104478e-03,  1.11940299e-03,  4.85074627e-04,\n",
       "        4.21641791e-03,  1.34328358e-03,  5.22388060e-04,  1.26865672e-03,\n",
       "        2.83582090e-03,  1.28358209e-02,  7.83582090e-03,  1.26865672e-03,\n",
       "        1.60447761e-03,  1.41791045e-03,  1.97761194e-03,  1.75373134e-03,\n",
       "        4.47761194e-04,  1.41791045e-03,  1.64179104e-03,  6.64179104e-03,\n",
       "        7.42537313e-03,  7.01492537e-03,  1.23134328e-03,  1.41791045e-03,\n",
       "        2.91044776e-03,  5.59701493e-03,  3.54477612e-03,  2.91044776e-03,\n",
       "        2.42537313e-03,  1.60447761e-03,  2.94776119e-03,  3.61940299e-03,\n",
       "        2.20149254e-03,  0.00000000e+00,  1.11940299e-03,  1.82835821e-03,\n",
       "        5.97014925e-04,  8.95522388e-04,  4.10447761e-04,  1.64179104e-03,\n",
       "        2.38805970e-03,  1.08208955e-03,  1.08208955e-03,  4.02985075e-03,\n",
       "        1.19402985e-03,  6.34328358e-03,  1.56716418e-03,  3.99253731e-03,\n",
       "        3.99253731e-03,  6.34328358e-04,  2.27611940e-03,  1.67910448e-03,\n",
       "        1.71641791e-03,  1.41791045e-03,  1.56716418e-03,  2.08955224e-03,\n",
       "        5.33582090e-03,  1.04477612e-03,  6.15671642e-03,  6.60447761e-03,\n",
       "        3.35820896e-03,  2.42537313e-03,  1.52985075e-03,  2.27611940e-03,\n",
       "        7.38805970e-03,  3.43283582e-03,  1.19402985e-03,  1.11940299e-03,\n",
       "        1.45522388e-03,  7.46268657e-04,  2.08955224e-03,  5.41044776e-03,\n",
       "        6.71641791e-04,  7.08955224e-04,  1.71641791e-03,  1.86567164e-04,\n",
       "        1.41791045e-03,  3.88059701e-03,  1.19402985e-03,  8.95522388e-04,\n",
       "        2.79850746e-03,  1.23134328e-03, -3.73134328e-05,  1.80597015e-02,\n",
       "        3.50746269e-03,  4.10447761e-04,  1.11940299e-02,  7.12686567e-03,\n",
       "        2.60074627e-02,  1.97761194e-03,  1.45522388e-03, -1.00746269e-03,\n",
       "        6.41791045e-03,  4.25373134e-03,  2.60447761e-02,  1.82835821e-03,\n",
       "        3.52611940e-02,  8.50746269e-03,  5.52238806e-03, -7.46268657e-05,\n",
       "        1.13059701e-02,  1.58582090e-02,  3.13432836e-03,  1.67910448e-03,\n",
       "        2.00000000e-02,  5.59701493e-04,  5.70895522e-03,  1.19402985e-03,\n",
       "        1.35820896e-02,  1.72761194e-02,  4.40298507e-03,  2.98507463e-04,\n",
       "        1.45522388e-03,  3.95522388e-03,  5.18656716e-03,  3.76865672e-03,\n",
       "        3.91791045e-03])"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result.importances_mean\n",
    "rf.\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "py310",
   "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": "3.10.6"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}