test_dart.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### This is a Fork of this [Kaggle script](https://www.kaggle.com/yuansun/santander-customer-satisfaction/lb-0-84-for-starters/code)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import xgboost as xgb\n",
    "from sklearn.metrics import roc_auc_score\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('ggplot')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(76020, 371)\n"
     ]
    }
   ],
   "source": [
    "train = pd.read_csv('ORG/train.csv')\n",
    "print(train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# remove constant columns\n",
    "remove = []\n",
    "for col in train.columns:\n",
    "    if train[col].std() == 0:\n",
    "        remove.append(col)\n",
    "train.drop(remove, axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# remove duplicated columns\n",
    "remove = []\n",
    "col    = train.columns\n",
    "for ii in range(len(col)-1):\n",
    "    v = train[col[ii]].values\n",
    "    for jj in range(ii+1, len(col)):\n",
    "        if np.array_equal(v, train[col[jj]].values):\n",
    "            remove.append(col[jj])\n",
    "train.drop(remove, axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[76020, 76020]\n"
     ]
    }
   ],
   "source": [
    "y = train['TARGET']\n",
    "x = train.drop(['ID','TARGET'], axis=1)\n",
    "print([len(y), len(x)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.random.seed(0)\n",
    "tv = np.random.choice([False, True], len(y), replace=True, p=[0.3, 0.7])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "mt = xgb.DMatrix(x[tv].values, label=y[tv].values,\n",
    "                 feature_names=x.columns.values)\n",
    "mv = xgb.DMatrix(x[np.logical_not(tv)].values,\n",
    "                 feature_names=x.columns.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "num_round = 350\n",
    "prm = {'max_depth': 5,\n",
    "       'learning_rate': 0.03,\n",
    "       'subsample': 0.95,\n",
    "       'colsample_bytree': 0.85,\n",
    "       'seed': 4242,\n",
    "       'objective': 'binary:logistic',\n",
    "       'silent': True}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.840068231475\n"
     ]
    }
   ],
   "source": [
    "prm['booster'] = 'gbtree'\n",
    "xgb_gbtree = xgb.train(prm, mt, num_round)\n",
    "auc_gbtree = roc_auc_score(y[np.logical_not(tv)],\n",
    "                           xgb_gbtree.predict(mv))\n",
    "print(auc_gbtree)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "prm['booster']        = 'dart'\n",
    "prm['sample_type']    = 'uniform'\n",
    "prm['normalize_type'] = 'forest'\n",
    "prm['rate_drop']      = 0.1\n",
    "rate     = [0.1, 0.3, 0.5, 0.7, 0.9]\n",
    "auc_dart = []\n",
    "for p in rate:\n",
    "    prm['skip_drop'] = p\n",
    "    xgb_dart = xgb.train(prm, mt, num_round)\n",
    "    auc_dart.append(roc_auc_score(y[np.logical_not(tv)],\n",
    "                                  xgb_dart.predict(mv, num_round)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df = pd.DataFrame(index=rate)\n",
    "df['gbtree'] = auc_gbtree\n",
    "df['dart']   = auc_dart"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x10e1e82e8>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAvEAAAHwCAYAAAAvjDDZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nNvP/ARTL/NVcF613AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10dc9a6d8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = df.plot(style='x-', figsize=(12, 8))\n",
    "ax.set_xlabel(\"skip_drop\")\n",
    "ax.set_ylabel(\"valid AUC\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}