{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import h2o\n",
    "from h2o.estimators.gbm import H2OGradientBoostingEstimator\n",
    "\n",
    "h2o.init()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "gbm Model Build progress: |███████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "from h2o.utils.shared_utils import _locate # private function. used to find files within h2o git project directory.\n",
    "\n",
    "prostate_train = h2o.import_file(path=_locate(\"smalldata/logreg/prostate.csv\"))\n",
    "prostate_train[\"CAPSULE\"] = prostate_train[\"CAPSULE\"].asfactor()\n",
    "\n",
    "ntrees = 100\n",
    "learning_rate = 0.1\n",
    "depth = 5\n",
    "min_rows = 10\n",
    "# Build H2O GBM classification model:\n",
    "gbm = H2OGradientBoostingEstimator(ntrees=ntrees, learn_rate=learning_rate,\n",
    "                                       max_depth=depth,\n",
    "                                       min_rows=min_rows,\n",
    "                                       distribution=\"bernoulli\")\n",
    "gbm.train(x=list(range(1, prostate_train.ncol)), y=\"CAPSULE\", training_frame=prostate_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Variable importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>variable</th>\n",
       "      <th>relative_importance</th>\n",
       "      <th>scaled_importance</th>\n",
       "      <th>percentage</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GLEASON</td>\n",
       "      <td>113.942337</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.314205</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PSA</td>\n",
       "      <td>88.346039</td>\n",
       "      <td>0.775357</td>\n",
       "      <td>0.243621</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>VOL</td>\n",
       "      <td>60.350849</td>\n",
       "      <td>0.529661</td>\n",
       "      <td>0.166422</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>AGE</td>\n",
       "      <td>50.993736</td>\n",
       "      <td>0.447540</td>\n",
       "      <td>0.140619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>DPROS</td>\n",
       "      <td>42.008102</td>\n",
       "      <td>0.368679</td>\n",
       "      <td>0.115841</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>RACE</td>\n",
       "      <td>4.168409</td>\n",
       "      <td>0.036583</td>\n",
       "      <td>0.011495</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>DCAPS</td>\n",
       "      <td>2.827885</td>\n",
       "      <td>0.024819</td>\n",
       "      <td>0.007798</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  variable  relative_importance  scaled_importance  percentage\n",
       "0  GLEASON           113.942337           1.000000    0.314205\n",
       "1      PSA            88.346039           0.775357    0.243621\n",
       "2      VOL            60.350849           0.529661    0.166422\n",
       "3      AGE            50.993736           0.447540    0.140619\n",
       "4    DPROS            42.008102           0.368679    0.115841\n",
       "5     RACE             4.168409           0.036583    0.011495\n",
       "6    DCAPS             2.827885           0.024819    0.007798"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gbm.varimp_plot()\n",
    "gbm.varimp(use_pandas=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Permutation Variable Importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Variable Importances: \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>Variable</th>\n",
       "      <th>Relative Importance</th>\n",
       "      <th>Scaled Importance</th>\n",
       "      <th>Percentage</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>PSA</td>\n",
       "      <td>0.133771</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.292864</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>GLEASON</td>\n",
       "      <td>0.127062</td>\n",
       "      <td>0.949849</td>\n",
       "      <td>0.278177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>VOL</td>\n",
       "      <td>0.072039</td>\n",
       "      <td>0.538528</td>\n",
       "      <td>0.157716</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>DPROS</td>\n",
       "      <td>0.061919</td>\n",
       "      <td>0.462871</td>\n",
       "      <td>0.135558</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>AGE</td>\n",
       "      <td>0.058550</td>\n",
       "      <td>0.437688</td>\n",
       "      <td>0.128183</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>RACE</td>\n",
       "      <td>0.002966</td>\n",
       "      <td>0.022170</td>\n",
       "      <td>0.006493</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>DCAPS</td>\n",
       "      <td>0.000461</td>\n",
       "      <td>0.003444</td>\n",
       "      <td>0.001009</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Variable  Relative Importance  Scaled Importance  Percentage\n",
       "0      PSA             0.133771           1.000000    0.292864\n",
       "1  GLEASON             0.127062           0.949849    0.278177\n",
       "2      VOL             0.072039           0.538528    0.157716\n",
       "3    DPROS             0.061919           0.462871    0.135558\n",
       "4      AGE             0.058550           0.437688    0.128183\n",
       "5     RACE             0.002966           0.022170    0.006493\n",
       "6    DCAPS             0.000461           0.003444    0.001009"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "permutation_varimp = gbm.permutation_importance_plot(prostate_train)\n",
    "permutation_varimp"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### How consistent is the permutation variable importance?\n",
    "Let's look at 15 evaluations using the `n_repeats` parameter."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "repeated_varimp = gbm.permutation_importance_plot(prostate_train, n_repeats=15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Variable Importances: \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>Variable</th>\n",
       "      <th>Run 1</th>\n",
       "      <th>Run 2</th>\n",
       "      <th>Run 3</th>\n",
       "      <th>Run 4</th>\n",
       "      <th>Run 5</th>\n",
       "      <th>Run 6</th>\n",
       "      <th>Run 7</th>\n",
       "      <th>Run 8</th>\n",
       "      <th>Run 9</th>\n",
       "      <th>Run 10</th>\n",
       "      <th>Run 11</th>\n",
       "      <th>Run 12</th>\n",
       "      <th>Run 13</th>\n",
       "      <th>Run 14</th>\n",
       "      <th>Run 15</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GLEASON</td>\n",
       "      <td>0.157179</td>\n",
       "      <td>0.153451</td>\n",
       "      <td>0.166163</td>\n",
       "      <td>0.154991</td>\n",
       "      <td>0.168250</td>\n",
       "      <td>0.165947</td>\n",
       "      <td>0.139213</td>\n",
       "      <td>0.150212</td>\n",
       "      <td>0.160836</td>\n",
       "      <td>0.173778</td>\n",
       "      <td>0.127782</td>\n",
       "      <td>0.162362</td>\n",
       "      <td>0.144755</td>\n",
       "      <td>0.148686</td>\n",
       "      <td>0.137456</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PSA</td>\n",
       "      <td>0.113918</td>\n",
       "      <td>0.102675</td>\n",
       "      <td>0.130719</td>\n",
       "      <td>0.102157</td>\n",
       "      <td>0.110938</td>\n",
       "      <td>0.120814</td>\n",
       "      <td>0.127667</td>\n",
       "      <td>0.122686</td>\n",
       "      <td>0.102099</td>\n",
       "      <td>0.112335</td>\n",
       "      <td>0.130661</td>\n",
       "      <td>0.127149</td>\n",
       "      <td>0.129711</td>\n",
       "      <td>0.127278</td>\n",
       "      <td>0.135959</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>VOL</td>\n",
       "      <td>0.068469</td>\n",
       "      <td>0.071046</td>\n",
       "      <td>0.078143</td>\n",
       "      <td>0.070398</td>\n",
       "      <td>0.067346</td>\n",
       "      <td>0.065532</td>\n",
       "      <td>0.062221</td>\n",
       "      <td>0.073882</td>\n",
       "      <td>0.073047</td>\n",
       "      <td>0.052719</td>\n",
       "      <td>0.051078</td>\n",
       "      <td>0.081800</td>\n",
       "      <td>0.051798</td>\n",
       "      <td>0.076358</td>\n",
       "      <td>0.068815</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>AGE</td>\n",
       "      <td>0.064740</td>\n",
       "      <td>0.062394</td>\n",
       "      <td>0.064755</td>\n",
       "      <td>0.053396</td>\n",
       "      <td>0.059227</td>\n",
       "      <td>0.059774</td>\n",
       "      <td>0.063661</td>\n",
       "      <td>0.070312</td>\n",
       "      <td>0.047537</td>\n",
       "      <td>0.053670</td>\n",
       "      <td>0.058104</td>\n",
       "      <td>0.060018</td>\n",
       "      <td>0.061213</td>\n",
       "      <td>0.056535</td>\n",
       "      <td>0.055253</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>DPROS</td>\n",
       "      <td>0.038409</td>\n",
       "      <td>0.044830</td>\n",
       "      <td>0.062653</td>\n",
       "      <td>0.056290</td>\n",
       "      <td>0.058478</td>\n",
       "      <td>0.062711</td>\n",
       "      <td>0.049351</td>\n",
       "      <td>0.065590</td>\n",
       "      <td>0.040051</td>\n",
       "      <td>0.049898</td>\n",
       "      <td>0.037344</td>\n",
       "      <td>0.061386</td>\n",
       "      <td>0.074084</td>\n",
       "      <td>0.055052</td>\n",
       "      <td>0.063574</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>RACE</td>\n",
       "      <td>0.003887</td>\n",
       "      <td>0.002822</td>\n",
       "      <td>0.002332</td>\n",
       "      <td>0.002994</td>\n",
       "      <td>0.004261</td>\n",
       "      <td>0.002159</td>\n",
       "      <td>0.003081</td>\n",
       "      <td>0.003599</td>\n",
       "      <td>0.005384</td>\n",
       "      <td>0.003398</td>\n",
       "      <td>0.001900</td>\n",
       "      <td>0.002505</td>\n",
       "      <td>0.007285</td>\n",
       "      <td>0.004261</td>\n",
       "      <td>0.004578</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>DCAPS</td>\n",
       "      <td>0.001037</td>\n",
       "      <td>0.000288</td>\n",
       "      <td>0.001756</td>\n",
       "      <td>0.000749</td>\n",
       "      <td>0.003455</td>\n",
       "      <td>0.000317</td>\n",
       "      <td>0.000547</td>\n",
       "      <td>0.001238</td>\n",
       "      <td>0.000576</td>\n",
       "      <td>0.000374</td>\n",
       "      <td>0.000230</td>\n",
       "      <td>0.000806</td>\n",
       "      <td>0.001123</td>\n",
       "      <td>0.000547</td>\n",
       "      <td>0.000202</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Variable     Run 1     Run 2     Run 3     Run 4     Run 5     Run 6  \\\n",
       "0  GLEASON  0.157179  0.153451  0.166163  0.154991  0.168250  0.165947   \n",
       "1      PSA  0.113918  0.102675  0.130719  0.102157  0.110938  0.120814   \n",
       "2      VOL  0.068469  0.071046  0.078143  0.070398  0.067346  0.065532   \n",
       "3      AGE  0.064740  0.062394  0.064755  0.053396  0.059227  0.059774   \n",
       "4    DPROS  0.038409  0.044830  0.062653  0.056290  0.058478  0.062711   \n",
       "5     RACE  0.003887  0.002822  0.002332  0.002994  0.004261  0.002159   \n",
       "6    DCAPS  0.001037  0.000288  0.001756  0.000749  0.003455  0.000317   \n",
       "\n",
       "      Run 7     Run 8     Run 9    Run 10    Run 11    Run 12    Run 13  \\\n",
       "0  0.139213  0.150212  0.160836  0.173778  0.127782  0.162362  0.144755   \n",
       "1  0.127667  0.122686  0.102099  0.112335  0.130661  0.127149  0.129711   \n",
       "2  0.062221  0.073882  0.073047  0.052719  0.051078  0.081800  0.051798   \n",
       "3  0.063661  0.070312  0.047537  0.053670  0.058104  0.060018  0.061213   \n",
       "4  0.049351  0.065590  0.040051  0.049898  0.037344  0.061386  0.074084   \n",
       "5  0.003081  0.003599  0.005384  0.003398  0.001900  0.002505  0.007285   \n",
       "6  0.000547  0.001238  0.000576  0.000374  0.000230  0.000806  0.001123   \n",
       "\n",
       "     Run 14    Run 15  \n",
       "0  0.148686  0.137456  \n",
       "1  0.127278  0.135959  \n",
       "2  0.076358  0.068815  \n",
       "3  0.056535  0.055253  \n",
       "4  0.055052  0.063574  \n",
       "5  0.004261  0.004578  \n",
       "6  0.000547  0.000202  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "repeated_varimp"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### How much does it differ between different metrics?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "gbm.permutation_importance_plot(prostate_train, n_repeats=15, metric=\"auc\")  # AUC is default metric for binary classification\n",
    "gbm.permutation_importance_plot(prostate_train, n_repeats=15, metric=\"pr_auc\")\n",
    "gbm.permutation_importance_plot(prostate_train, n_repeats=15, metric=\"logloss\");"
   ]
  }
 ],
 "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": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}