notebooks/census.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load data, train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import shap\n",
    "import pickle\n",
    "import lightgbm as lgb\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load data\n",
    "X, y = shap.datasets.adult()\n",
    "X_display, y_display = shap.datasets.adult(display=True)\n",
    "num_features = X.shape[1]\n",
    "\n",
    "# Split data\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=7)\n",
    "d_train = lgb.Dataset(X_train, label=y_train)\n",
    "d_test = lgb.Dataset(X_test, label=y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training until validation scores don't improve for 50 rounds\n",
      "Early stopping, best iteration is:\n",
      "[644]\tvalid_0's binary_logloss: 0.278029\n"
     ]
    }
   ],
   "source": [
    "# Train model\n",
    "params = {\n",
    "    \"max_bin\": 512,\n",
    "    \"learning_rate\": 0.05,\n",
    "    \"boosting_type\": \"gbdt\",\n",
    "    \"objective\": \"binary\",\n",
    "    \"metric\": \"binary_logloss\",\n",
    "    \"num_leaves\": 10,\n",
    "    \"verbose\": -1,\n",
    "    \"min_data\": 100,\n",
    "    \"boost_from_average\": True\n",
    "}\n",
    "\n",
    "model = lgb.train(params, d_train, 10000, valid_sets=[d_test], early_stopping_rounds=50, verbose_eval=1000)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SHAP cooperative game"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from shapreg import removal, games, shapley"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make model callable\n",
    "model_lam = lambda x: model.predict(x)\n",
    "\n",
    "# Model extension\n",
    "marginal_extension = removal.MarginalExtension(X_test.values[:512], model_lam)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set up game (single prediction)\n",
    "instance = X.values[0]\n",
    "game = games.PredictionGame(marginal_extension, instance)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "957bc7c79da44767a7de3f226f334988",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/1 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Run estimator\n",
    "explanation = shapley.ShapleyRegression(game, batch_size=32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot with 95% confidence intervals\n",
    "feature_names = X.columns.tolist()\n",
    "explanation.plot(feature_names, title='SHAP Values', sort_features=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# LossSHAP cooperative game"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from shapreg.utils import crossentropyloss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set up game (single prediction)\n",
    "instance = X.values[1]\n",
    "label = y[1].astype(int)\n",
    "game = games.PredictionLossGame(marginal_extension, instance, label, crossentropyloss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3446d0cfb03a4f8c912a6c8878678379",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/1 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Run estimator\n",
    "explanation = shapley.ShapleyRegression(game, batch_size=32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot with 95% confidence intervals\n",
    "explanation.plot(feature_names, title='LossSHAP Values', sort_features=False)"
   ]
  },
  {
   "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": "3.6.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}