xgboost-tuning.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_svmlight_file\n",
    "\n",
    "data_dir=\"/workspace/chipper_data/chipper-0.3.0-data/\"\n",
    "\n",
    "train_data = load_svmlight_file(data_dir + \"training_data.ll\")\n",
    "X_train = train_data[0].toarray()\n",
    "y_train = train_data[1]\n",
    "\n",
    "test_data = load_svmlight_file(data_dir + \"testing_data.ll\")\n",
    "X_test = test_data[0].toarray()\n",
    "y_test = test_data[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import xgboost as xgb\n",
    "from xgboost import XGBClassifier\n",
    "from sklearn.model_selection import GridSearchCV, StratifiedKFold\n",
    "from sklearn.metrics import classification_report, confusion_matrix, accuracy_score, matthews_corrcoef\n",
    "\n",
    "model = XGBClassifier(objective=\"binary:logistic\")\n",
    "# Number of boosted trees to fit.\n",
    "n_estimators = [500] # range(50, 500, 50)\n",
    "# Maximum tree depth for base learners.\n",
    "max_depth = [10] #range(1, 11, 2)\n",
    "param_grid = dict(max_depth=max_depth, n_estimators=n_estimators)\n",
    "kfold = StratifiedKFold(n_splits=10, shuffle=True, random_state=7)\n",
    "grid_search = GridSearchCV(model, param_grid, scoring=\"neg_log_loss\", \n",
    "                           n_jobs=1, cv=kfold, verbose=1)\n",
    "grid_result = grid_search.fit(X_train, y_train)\n",
    "\n",
    "# summarize results\n",
    "print(\"Best: %f using %s\" % (grid_result.best_score_, grid_result.best_params_))\n",
    "means = grid_result.cv_results_['mean_test_score']\n",
    "stds = grid_result.cv_results_['std_test_score']\n",
    "params = grid_result.cv_results_['params']\n",
    "for mean, stdev, param in zip(means, stds, params):\n",
    "    print(\"%f (%f) with: %r\" % (mean, stdev, param))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x112783950>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkAAAAF5CAYAAAB3B5XOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112738ad0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "from matplotlib import pyplot\n",
    "\n",
    "# plot results\n",
    "%matplotlib inline\n",
    "\n",
    "\n",
    "scores = np.array(means).reshape(len(max_depth), len(n_estimators))\n",
    "for i, value in enumerate(max_depth):\n",
    "    pyplot.plot(n_estimators, scores[i], label='depth: ' + str(value))\n",
    "pyplot.legend()\n",
    "pyplot.xlabel('n_estimators')\n",
    "pyplot.ylabel('Log Loss')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# save model to file\n",
    "import pickle\n",
    "pickle.dump(grid_result, open(\"/tmp/grid_result.txt\", \"wb\"))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# load model from file\n",
    "#loaded_model = pickle.load(open(\"pima.pickle.dat\", \"rb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#xgboost.plot_importance(grid_result.best_estimator_.booster)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy was 80.29%\n",
      "\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "        0.0       0.79      0.82      0.81       208\n",
      "        1.0       0.81      0.79      0.80       208\n",
      "\n",
      "avg / total       0.80      0.80      0.80       416\n",
      "\n",
      "sensitivity(recall)=78.8, specificity=81.7, precision=81.2\n"
     ]
    }
   ],
   "source": [
    "# make predictions\n",
    "predicted = grid_result.predict(X_test)\n",
    "\n",
    "print \"Accuracy was %.2f%%\\n\" % (100 * accuracy_score(y_test, predicted))\n",
    "print classification_report(y_test, predicted)\n",
    "((tn, fp), (fn, tp)) = confusion_matrix(y_test, predicted)\n",
    "sensitivity = 100.0 * tp / (tp + fn)\n",
    "specificity = 100.0 * tn / (tn + fp)\n",
    "precision = 100.0 * tp / (tp + fp)\n",
    "print \"sensitivity(recall)=%.1f, specificity=%.1f, precision=%.1f\" % (sensitivity, specificity, precision)"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}