notebooks/Gain calculations and user study for AAAI TreeSHAP paper.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Path fever gain % for tree A = 0.3333333333333333\n",
      "Path cough gain % for tree A = 0.6666666666666666\n"
     ]
    }
   ],
   "source": [
    "root_mse = 0.25*(80-20)**2 + 0.75*(0-20)**2\n",
    "fever_mse = 0.25*(80-40)**2 + 0.25*(0-40)**2\n",
    "fever_and_cough_mse = 0\n",
    "print(\"Path fever gain % for tree A =\", (root_mse-fever_mse)/root_mse)\n",
    "print(\"Path cough gain % for tree A =\", (fever_mse-fever_and_cough_mse)/root_mse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Path cough gain % for tree B = 0.43859649122807015\n",
      "Path fever gain % for tree B = 0.5614035087719298\n"
     ]
    }
   ],
   "source": [
    "root_mse = 0.25*(90-25)**2 + 0.25*(10-25)**2 + 0.5*(0-25)**2\n",
    "cough_mse = 0.25*(90-50)**2 + 0.25*(10-50)**2\n",
    "fever_and_cough_mse = 0\n",
    "print(\"Path cough gain % for tree B =\", (root_mse-cough_mse)/root_mse)\n",
    "print(\"Path fever gain % for tree B =\", (fever_mse-fever_and_cough_mse)/root_mse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SHAP cough gain % for tree A = 0.5\n",
      "SHAP fever gain % for tree A = 0.5\n"
     ]
    }
   ],
   "source": [
    "import math\n",
    "def shapley_weight(N, S):\n",
    "    return math.factorial(S)*math.factorial(N-S-1)/math.factorial(N)\n",
    "\n",
    "# 11 => cough,fever\n",
    "root_mse = 0.25*(80-20)**2 + 0.75*(0-20)**2\n",
    "s00 = 0\n",
    "s01 = root_mse - (0.25*(80-40)**2 + 0.25*(0-40)**2)\n",
    "s10 = root_mse - (0.25*(0-0)**2 + 0.25*(0-0)**2 + 0.25*(0-40)**2 + 0.25*(80-40)**2)\n",
    "s11 = root_mse - 0\n",
    "\n",
    "print(\"SHAP cough gain % for tree A =\", (shapley_weight(2,0)*(s10-s00) + shapley_weight(2,1)*(s11-s01))/root_mse)\n",
    "print(\"SHAP fever gain % for tree A =\", (shapley_weight(2,0)*(s01-s00) + shapley_weight(2,1)*(s11-s10))/root_mse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SHAP fever gain % for tree B = 0.42105263157894735\n",
      "SHAP cough gain % for tree B = 0.5789473684210527\n"
     ]
    }
   ],
   "source": [
    "# 11 => fever,cough\n",
    "root_mse = 0.25*(90-25)**2 + 0.25*(10-25)**2 + 0.5*(0-25)**2\n",
    "s00 = 0\n",
    "s01 = root_mse - (0.25*(90-50)**2 + 0.25*(10-50)**2)\n",
    "s10 = root_mse - (0.25*(0-5)**2 + 0.25*(10-5)**2 + 0.25*(0-45)**2 + 0.25*(90-45)**2)\n",
    "s11 = root_mse - 0\n",
    "\n",
    "print(\"SHAP fever gain % for tree B =\", (shapley_weight(2,0)*(s10-s00) + shapley_weight(2,1)*(s11-s01))/root_mse)\n",
    "print(\"SHAP cough gain % for tree B =\", (shapley_weight(2,0)*(s01-s00) + shapley_weight(2,1)*(s11-s10))/root_mse)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## User experiment results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "user_results = [{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 20, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 45, \"fever_credit\": 15},\n",
    "{\"cough_credit\": 20, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 60, \"fever_credit\": 20},\n",
    "{\"cough_credit\": 80, \"fever_credit\": 20},\n",
    "{\"cough_credit\": 40, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 20, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 20, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 10, \"fever_credit\": 50},\n",
    "{\"cough_credit\": 80, \"fever_credit\": 0},\n",
    "{\"cough_credit\": 40, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 40, \"fever_credit\": 20},\n",
    "{\"cough_credit\": 35, \"fever_credit\": 25},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 20, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 40, \"fever_credit\": 20},\n",
    "{\"cough_credit\": 20, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 20, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 20, \"fever_credit\": 40},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 40, \"fever_credit\": 20},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30},\n",
    "{\"cough_credit\": 35, \"fever_credit\": 25},\n",
    "{\"cough_credit\": 40, \"fever_credit\": 20},\n",
    "{\"cough_credit\": 60, \"fever_credit\": 20},\n",
    "{\"cough_credit\": 30, \"fever_credit\": 30}]\n",
    "\n",
    "vals = [str(r[\"fever_credit\"])+\"_\"+str(r[\"cough_credit\"]) for r in user_results]\n",
    "\n",
    "from collections import Counter\n",
    "c = Counter( vals )\n",
    "\n",
    "counts = list(c.items())\n",
    "counts.sort(key=lambda x: -x[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaEAAAFZCAYAAAAxaZC2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1109741d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(4.5,3.2))\n",
    "\n",
    "bar_list = plt.bar(np.array(range(len(counts)))*5+2.5, [c[1] for c in counts], width=3)\n",
    "[x.set_color('#bbbbbb') for x in bar_list]\n",
    "bar_list[0].set_color('#1E88E5')\n",
    "bar_list[2].set_color('#F52757')\n",
    "\n",
    "# just stuff for\n",
    "# leg1 = plt.bar(np.array([0]), [1])\n",
    "# [x.set_color('#1E88E5') for x in leg1]\n",
    "# leg2 = plt.bar(np.array([0]), [1])\n",
    "# [x.set_color('#F52757') for x in leg2]\n",
    "# leg3 = plt.bar(np.array([0]), [1])\n",
    "# [x.set_color('#bbbbbb') for x in leg3]\n",
    "\n",
    "import matplotlib.patches as mpatches\n",
    "\n",
    "shap_patch = mpatches.Patch(color='#1E88E5', label='SHAP')\n",
    "path_patch = mpatches.Patch(color='#F52757', label='Path')\n",
    "other_patch = mpatches.Patch(color='#bbbbbb', label='Other')\n",
    "plt.legend(handles=[blue_patch,path_patch,other_patch], frameon=False)\n",
    "\n",
    "\n",
    "plt.tick_params(\n",
    "    axis='x',          # changes apply to the x-axis\n",
    "    which='both',      # both major and minor ticks are affected\n",
    "    bottom='off',      # ticks along the bottom edge are off\n",
    "    top='off',         # ticks along the top edge are off\n",
    "    labelbottom='on') # labels along the bottom edge are off\n",
    "plt.tick_params(\n",
    "    axis='y',          # changes apply to the x-axis\n",
    "    which='both',      # both major and minor ticks are affected\n",
    "    left='on',      # ticks along the bottom edge are off\n",
    "    right='off',         # ticks along the top edge are off\n",
    "    labelbottom='on') # labels along the bottom edge are off\n",
    "\n",
    "ax.spines['top'].set_visible(False)\n",
    "ax.spines['right'].set_visible(False)\n",
    "ax.spines['left'].set_visible(True)\n",
    "plt.xticks(np.array(range(len(counts)))*5+4, [c[0].replace(\"_\", \"/\") for c in counts], fontsize=10, rotation='vertical')\n",
    "#plt.legend([\"SHAP\", \"Path\"], frameon=False, loc=\"upper center\")\n",
    "plt.ylabel(\"# of respondents\", fontsize=12)\n",
    "plt.xlabel(\"feature attribution values (fever/cough)\", fontsize=12)\n",
    "plt.xlim(0,53)\n",
    "plt.savefig(\"tree_max.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'30/30'"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"30_30\".replace(\"_\", \"/\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x10d9adba8>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaEAAAEmCAYAAADLHS+iAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10d8cc550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(4.5,3.2))\n",
    "\n",
    "bar_list = plt.bar(np.array([0,1])*5+1, [30, 30])\n",
    "[x.set_color('#bbbbbb') for x in bar_list]\n",
    "\n",
    "bar_list = plt.bar(np.array([0,1])*5+2, [30, 30])\n",
    "[x.set_color('#1E88E5') for x in bar_list]\n",
    "\n",
    "bar_list = plt.bar(np.array([0,1])*5+3, [20, 40])\n",
    "[x.set_color('#18C45D') for x in bar_list]\n",
    "\n",
    "plt.tick_params(\n",
    "    axis='x',          # changes apply to the x-axis\n",
    "    which='both',      # both major and minor ticks are affected\n",
    "    bottom='on',      # ticks along the bottom edge are off\n",
    "    top='off',         # ticks along the top edge are off\n",
    "    labelbottom='on') # labels along the bottom edge are off\n",
    "plt.tick_params(\n",
    "    axis='y',          # changes apply to the x-axis\n",
    "    which='both',      # both major and minor ticks are affected\n",
    "    left='on',      # ticks along the bottom edge are off\n",
    "    right='off',         # ticks along the top edge are off\n",
    "    labelbottom='on') # labels along the bottom edge are off\n",
    "\n",
    "ax.spines['top'].set_visible(False)\n",
    "ax.spines['right'].set_visible(False)\n",
    "ax.spines['left'].set_visible(True)\n",
    "plt.xticks(np.array([0,1])*5+2.4, [\"Fever\", \"Cough\"], fontsize=10)\n",
    "plt.legend([\"Human\", \"SHAP\", \"Path\"], frameon=False, loc=\"upper center\")\n",
    "plt.ylabel(\"feature attribution value\", fontsize=12)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "counts.sort(key=lambda x: -x[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('40_40', 1),\n",
       " ('20_60', 1),\n",
       " ('20_80', 1),\n",
       " ('15_45', 1),\n",
       " ('40_20', 2),\n",
       " ('30_30', 4)]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "counts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAbEAAAEICAYAAADRFcoMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1126e9588>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "brute_force = [0.000547, 0.007596, 0.01731, 0.050687, 0.317647, 0.471027, 1.491240, 3.9685, 12.283, 40.28, 96.3355, 315.67518]\n",
    "fast_method = [0.000094, 0.000108, 0.000080, 0.000121, 0.000222, 0.000382, 0.000910, 0.00124, 0.0015, 0.0034, 0.00335, 0.00577]\n",
    "fig = plt.figure(figsize=(4.5,2.5))\n",
    "plt.semilogy(fast_method, color='#1E88E5', linewidth=2)\n",
    "plt.semilogy(brute_force, color='#7C52FF', linewidth=2)\n",
    "plt.ylabel(\"seconds of runtime\")\n",
    "plt.xlabel(\"number of input features and max tree depth\")\n",
    "plt.legend([\"TreeSHAP\", \"Brute force\"], frameon=False, loc=\"upper left\")\n",
    "plt.savefig(\"runtime.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}