XGBT Feature Expansion.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from pandas.io.parsers import read_csv\n",
    "%matplotlib inline\n",
    "\n",
    "from matplotlib import pyplot as plt\n",
    "import matplotlib as mpl\n",
    "\n",
    "import scipy\n",
    "\n",
    "import xgboost as xgb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dtrain = xgb.DMatrix(\"train.buffer\")\n",
    "dvalidation = xgb.DMatrix(\"validation.buffer\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "param = {'bst:max_depth':7, 'bst:eta':1, 'silent':1, 'objective':'binary:logistic' }\n",
    "param['nthread'] = 4\n",
    "plst = param.items()\n",
    "plst += [('eval_metric', 'logloss')]\n",
    "plst += [('eval_metric', 'auc')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "evallist  = [(dvalidation,'eval'), (dtrain,'train')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[0]\teval-logloss:0.253334\teval-auc:0.780535\ttrain-logloss:0.248384\ttrain-auc:0.781785\n",
      "[1]\teval-logloss:0.207300\teval-auc:0.785617\ttrain-logloss:0.209092\ttrain-auc:0.787094\n",
      "[2]\teval-logloss:0.197886\teval-auc:0.788099\ttrain-logloss:0.195596\ttrain-auc:0.789908\n",
      "[3]\teval-logloss:0.196388\teval-auc:0.789676\ttrain-logloss:0.193340\ttrain-auc:0.792229\n",
      "[4]\teval-logloss:0.195568\teval-auc:0.793677\ttrain-logloss:0.192150\ttrain-auc:0.796082\n",
      "[5]\teval-logloss:0.195593\teval-auc:0.795120\ttrain-logloss:0.191933\ttrain-auc:0.798321\n",
      "[6]\teval-logloss:0.195982\teval-auc:0.796112\ttrain-logloss:0.192192\ttrain-auc:0.799354\n",
      "[7]\teval-logloss:0.195935\teval-auc:0.797171\ttrain-logloss:0.192042\ttrain-auc:0.800681\n",
      "[8]\teval-logloss:0.195762\teval-auc:0.797791\ttrain-logloss:0.191980\ttrain-auc:0.801628\n",
      "[9]\teval-logloss:0.195653\teval-auc:0.799015\ttrain-logloss:0.191947\ttrain-auc:0.802798\n",
      "[10]\teval-logloss:0.195533\teval-auc:0.799382\ttrain-logloss:0.191716\ttrain-auc:0.803155\n",
      "[11]\teval-logloss:0.195838\teval-auc:0.800146\ttrain-logloss:0.191925\ttrain-auc:0.804124\n",
      "[12]\teval-logloss:0.195927\teval-auc:0.800358\ttrain-logloss:0.191971\ttrain-auc:0.804732\n",
      "[13]\teval-logloss:0.195886\teval-auc:0.800860\ttrain-logloss:0.191593\ttrain-auc:0.805456\n",
      "[14]\teval-logloss:0.196272\teval-auc:0.801299\ttrain-logloss:0.191675\ttrain-auc:0.805989\n",
      "[15]\teval-logloss:0.196464\teval-auc:0.801970\ttrain-logloss:0.190474\ttrain-auc:0.806655\n",
      "[16]\teval-logloss:0.196404\teval-auc:0.802252\ttrain-logloss:0.190403\ttrain-auc:0.807044\n",
      "[17]\teval-logloss:0.196262\teval-auc:0.802891\ttrain-logloss:0.190237\ttrain-auc:0.807700\n",
      "[18]\teval-logloss:0.196599\teval-auc:0.803300\ttrain-logloss:0.190356\ttrain-auc:0.808317\n",
      "[19]\teval-logloss:0.196430\teval-auc:0.803652\ttrain-logloss:0.190241\ttrain-auc:0.808611\n",
      "[20]\teval-logloss:0.196436\teval-auc:0.804174\ttrain-logloss:0.190326\ttrain-auc:0.809188\n",
      "[21]\teval-logloss:0.196494\teval-auc:0.804444\ttrain-logloss:0.190349\ttrain-auc:0.809547\n",
      "[22]\teval-logloss:0.196617\teval-auc:0.804801\ttrain-logloss:0.190343\ttrain-auc:0.810022\n",
      "[23]\teval-logloss:0.196617\teval-auc:0.805046\ttrain-logloss:0.190262\ttrain-auc:0.810345\n",
      "[24]\teval-logloss:0.196569\teval-auc:0.805390\ttrain-logloss:0.190094\ttrain-auc:0.810903\n",
      "[25]\teval-logloss:0.196540\teval-auc:0.805762\ttrain-logloss:0.189940\ttrain-auc:0.811338\n",
      "[26]\teval-logloss:0.196496\teval-auc:0.806066\ttrain-logloss:0.189812\ttrain-auc:0.811782\n",
      "[27]\teval-logloss:0.196562\teval-auc:0.806010\ttrain-logloss:0.189837\ttrain-auc:0.811808\n",
      "[28]\teval-logloss:0.196723\teval-auc:0.806168\ttrain-logloss:0.190012\ttrain-auc:0.812114\n",
      "[29]\teval-logloss:0.197192\teval-auc:0.806515\ttrain-logloss:0.190270\ttrain-auc:0.812461\n",
      "[30]\teval-logloss:0.197457\teval-auc:0.806607\ttrain-logloss:0.190326\ttrain-auc:0.812706\n",
      "[31]\teval-logloss:0.197358\teval-auc:0.806797\ttrain-logloss:0.190127\ttrain-auc:0.813055\n",
      "[32]\teval-logloss:0.197435\teval-auc:0.807031\ttrain-logloss:0.190063\ttrain-auc:0.813546\n",
      "[33]\teval-logloss:0.197510\teval-auc:0.807228\ttrain-logloss:0.190088\ttrain-auc:0.813798\n",
      "[34]\teval-logloss:0.197452\teval-auc:0.807467\ttrain-logloss:0.190039\ttrain-auc:0.814018\n",
      "[35]\teval-logloss:0.197579\teval-auc:0.807707\ttrain-logloss:0.190044\ttrain-auc:0.814438\n",
      "[36]\teval-logloss:0.197523\teval-auc:0.807915\ttrain-logloss:0.189938\ttrain-auc:0.814649\n",
      "[37]\teval-logloss:0.197562\teval-auc:0.808144\ttrain-logloss:0.189920\ttrain-auc:0.815010\n",
      "[38]\teval-logloss:0.197583\teval-auc:0.808244\ttrain-logloss:0.189900\ttrain-auc:0.815211\n",
      "[39]\teval-logloss:0.197524\teval-auc:0.808328\ttrain-logloss:0.189815\ttrain-auc:0.815430\n"
     ]
    }
   ],
   "source": [
    "num_round = 40\n",
    "bst = xgb.train( plst, dtrain, num_round, evallist )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "bst.save_model(\"0001.model\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "tr_leafindex = bst.predict(dtrain, pred_leaf=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "va_leafindex = bst.predict(dvalidation, pred_leaf=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "tr_y = dtrain.get_label()\n",
    "va_y = dvalidation.get_label()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.save(\"tr.xgb\", tr_leafindex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.save(\"va.xgb\", va_leafindex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.save(\"tr.label\", tr_y)\n",
    "np.save(\"va.label\", va_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}