06 - Grid Searches for Hyper Parameters.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Grid Searches\n",
    "================="
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"figures/grid_search_cross_validation.svg\" width=100%>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Grid-Search with build-in cross validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.grid_search import GridSearchCV\n",
    "from sklearn.svm import SVC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_digits\n",
    "from sklearn.cross_validation import train_test_split\n",
    "digits = load_digits()\n",
    "X_train, X_test, y_train, y_test = train_test_split(digits.data,\n",
    "                                                    digits.target)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define parameter grid:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "param_grid = {'C': 10. ** np.arange(-3, 3),\n",
    "              'gamma' : 10. ** np.arange(-5, 0)}\n",
    "\n",
    "np.set_printoptions(suppress=True)\n",
    "print(param_grid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grid_search = GridSearchCV(SVC(), param_grid, verbose=3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A GridSearchCV object behaves just like a normal classifier."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "grid_search.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grid_search.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grid_search.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grid_search.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# We extract just the scores\n",
    "\n",
    "scores = [x.mean_validation_score for x in grid_search.grid_scores_]\n",
    "scores = np.array(scores).reshape(6, 5)\n",
    "\n",
    "plt.matshow(scores)\n",
    "plt.xlabel('gamma')\n",
    "plt.ylabel('C')\n",
    "plt.colorbar()\n",
    "plt.xticks(np.arange(5), param_grid['gamma'])\n",
    "plt.yticks(np.arange(6), param_grid['C']);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Nested Cross-validation in scikit-learn:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.cross_validation import cross_val_score\n",
    "cross_val_score(GridSearchCV(SVC(), param_grid),\n",
    "                digits.data, digits.target)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Exercises\n",
    "Use GridSearchCV to adjust max_depth and max_features of a RandomForestClassifier (from ``sklearn.ensemble``) on the digits dataset.\n",
    "\n",
    "Visualize the results as a heat map.\n",
    "\n",
    "Should you also adjust ``n_estimators``?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# %load solutions/grid_search_forest.py"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Randomized Search"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"figures/randomized_search.png\" width=100%>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_iris\n",
    "iris = load_iris()\n",
    "X_train, X_test, y_train, y_test = train_test_split(iris.data, iris.target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from scipy.stats import expon\n",
    "plt.hist([expon.rvs(scale=0.001) for x in range(10000)], bins=100, normed=True);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.grid_search import RandomizedSearchCV\n",
    "\n",
    "param_distributions = {'C': expon(), 'gamma': expon()}\n",
    "rs = RandomizedSearchCV(SVC(), param_distributions=param_distributions,\n",
    "                        n_iter=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "rs.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "rs.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "rs.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "scores, Cs, gammas = zip(*[(score.mean_validation_score, score.parameters['C'], score.parameters['gamma'])\n",
    "                           for score in rs.grid_scores_])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "plt.scatter(Cs, gammas, s=50, c=scores, linewidths=0)\n",
    "plt.xlabel(\"C\")\n",
    "plt.ylabel(\"gamma\")\n",
    "plt.xscale(\"log\")\n",
    "plt.yscale(\"log\")\n",
    "plt.colorbar()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Exercise\n",
    "Add parameters that are not relevant (like ``coef0``) to param_distribution. Observe that it doesn't change the runtime and be happy."
   ]
  }
 ],
 "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
}