Scikit-learn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "# Machine Learning with Scikit-Learn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "# Global imports and settings\n",
    "\n",
    "# Matplotlib\n",
    "%matplotlib inline\n",
    "from matplotlib import pyplot as plt\n",
    "plt.rcParams[\"figure.figsize\"] = (8, 8)\n",
    "plt.rcParams[\"figure.max_open_warning\"] = -1\n",
    "\n",
    "# Print options\n",
    "import numpy as np\n",
    "np.set_printoptions(precision=3)\n",
    "\n",
    "# Slideshow\n",
    "from notebook.services.config import ConfigManager\n",
    "cm = ConfigManager()\n",
    "cm.update('livereveal', {'width': 1440, 'height': 768, 'scroll': True, 'theme': 'simple'})\n",
    "\n",
    "# Silence warnings\n",
    "import warnings\n",
    "warnings.simplefilter(action=\"ignore\", category=FutureWarning)\n",
    "warnings.simplefilter(action=\"ignore\", category=UserWarning)\n",
    "warnings.simplefilter(action=\"ignore\", category=RuntimeWarning)\n",
    "\n",
    "# Helper functions\n",
    "def plot_surface(clf, X, y, \n",
    "                 xlim=(-10, 10), ylim=(-10, 10), n_steps=250, \n",
    "                 subplot=None, show=True):\n",
    "    if subplot is None:\n",
    "        fig = plt.figure()\n",
    "    else:\n",
    "        plt.subplot(*subplot)\n",
    "        \n",
    "    xx, yy = np.meshgrid(np.linspace(xlim[0], xlim[1], n_steps), \n",
    "                         np.linspace(ylim[0], ylim[1], n_steps))\n",
    "    \n",
    "    if hasattr(clf, \"decision_function\"):\n",
    "        z = clf.decision_function(np.c_[xx.ravel(), yy.ravel()])\n",
    "    else:\n",
    "        z = clf.predict_proba(np.c_[xx.ravel(), yy.ravel()])[:, 1]\n",
    "        \n",
    "    z = z.reshape(xx.shape)\n",
    "    plt.contourf(xx, yy, z, alpha=0.8, cmap=plt.cm.RdBu_r)\n",
    "    plt.scatter(X[:, 0], X[:, 1], c=y)\n",
    "    plt.xlim(*xlim)\n",
    "    plt.ylim(*ylim)\n",
    "    \n",
    "    if show:\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "# %%javascript\n",
    "# Reveal.addEventListener(\"slidechanged\", function(event){ window.location.hash = \"header\"; });"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Scikit-Learn\n",
    "\n",
    "* Machine learning library written in __Python__\n",
    "* __Simple and efficient__, for both experts and non-experts\n",
    "* Classical, __well-established machine learning algorithms__\n",
    "* Shipped with <a href=\"http://scikit-learn.org/dev/documentation.html\">documentation</a> and <a href=\"http://scikit-learn.org/dev/auto_examples/index.html\">examples</a>\n",
    "* Community-driven"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Algorithms\n",
    "See the [Reference](http://scikit-learn.org/dev/modules/classes.html)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "__Supervised learning:__\n",
    "\n",
    "* Linear models (Ridge, Lasso, Elastic Net, ...)\n",
    "* Support Vector Machines\n",
    "* Tree-based methods (Classification/Regression Trees, Random Forests,...)\n",
    "* Nearest neighbors\n",
    "* Neural networks \n",
    "* Gaussian Processes\n",
    "* Feature selection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "__Unsupervised learning:__\n",
    "\n",
    "* Clustering (KMeans, ...)\n",
    "* Matric Decomposition (PCA, ...)\n",
    "* Manifold Learning (Embeddings)\n",
    "* Density estimation\n",
    "* Outlier detection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "__Model selection and evaluation:__\n",
    "\n",
    "* Cross-validation\n",
    "* Grid-search\n",
    "* Lots of metrics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "# Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Data \n",
    "\n",
    "- Input data = Numpy arrays or Scipy sparse matrices ;\n",
    "- Algorithms are expressed using high-level operations defined on matrices or vectors (similar to MATLAB) ;\n",
    "    - Leverage efficient low-leverage implementations ;\n",
    "    - Keep code short and readable. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[-6.453 -8.764]\n",
      " [ 0.29   0.147]\n",
      " [-5.184 -1.253]\n",
      " ..., \n",
      " [-0.231 -1.608]\n",
      " [-0.603  6.873]\n",
      " [ 2.284  4.874]]\n",
      "['r' 'r' 'b' 'r' 'b']\n"
     ]
    }
   ],
   "source": [
    "# Generate data\n",
    "from sklearn.datasets import make_blobs\n",
    "X, y = make_blobs(n_samples=1000, centers=20, random_state=123)\n",
    "labels = [\"b\", \"r\"]\n",
    "y = np.take(labels, (y < 10)) # Relabels numeric values to b,r\n",
    "print(X)\n",
    "print(y[:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1000, 2)\n",
      "(1000,)\n"
     ]
    }
   ],
   "source": [
    "# X is a 2 dimensional array, with 1000 rows and 2 columns\n",
    "print(X.shape)\n",
    " \n",
    "# y is a vector of 1000 elements\n",
    "print(y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.29   0.147]\n",
      " [-5.184 -1.253]\n",
      " [-4.714  3.674]]\n",
      "[[-6.453 -8.764]\n",
      " [ 0.29   0.147]\n",
      " [-5.184 -1.253]\n",
      " [-4.714  3.674]\n",
      " [ 4.516 -2.881]]\n",
      "[-4.438 -2.46   4.331 -7.921  1.57   0.565  4.996  4.758 -1.604  1.101]\n",
      "[[-5.184 -1.253]\n",
      " [ 4.516 -2.881]\n",
      " [ 1.708  2.624]\n",
      " [-0.526  8.96 ]\n",
      " [-1.076  9.787]]\n"
     ]
    }
   ],
   "source": [
    "# Rows and columns can be accessed with lists, slices or masks\n",
    "print(X[[1, 2, 3]])     # rows 1, 2 and 3\n",
    "print(X[:5])            # 5 first rows\n",
    "print(X[500:510, 0])    # values from row 500 to row 510 at column 0\n",
    "print(X[y == \"b\"][:5])  # 5 first rows for which y is \"b\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAewAAAHfCAYAAACSxOtjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEUlEQVRCSAygYBNCCCEx4P8BNGvZLBnHwekAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1232404e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot\n",
    "plt.figure()\n",
    "for label in labels:\n",
    "    mask = (y == label)\n",
    "    plt.scatter(X[mask, 0], X[mask, 1], c=label)\n",
    "plt.xlim(-10, 10)\n",
    "plt.ylim(-10, 10)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Loading external data\n",
    "\n",
    "- Numpy provides some [simple tools](https://docs.scipy.org/doc/numpy/reference/routines.io.html) for loading data from files (CSV, binary, etc);\n",
    "\n",
    "- For structured data, Pandas provides more [advanced tools](http://pandas.pydata.org/pandas-docs/stable/io.html) (CSV, JSON, Excel, HDF5, SQL, etc);\n",
    "\n",
    "- For ROOT files, ``root_numpy`` provides loaders and converters to Numpy arrays."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Loading data from OpenML\n",
    "* OpenML: An open machine learning collaboration platform with many datasets, models, experiments\n",
    "* Register on [openml.org](http://www.openml.org), go to your profile to find API your key\n",
    "    - Store it in a file (e.g. `.openml/apikey.txt`)\n",
    "* Browse openml.org for interesting datasets, download by their ID"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "from openml import datasets, tasks, runs\n",
    "import pandas as pd\n",
    "import os"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Authentication\n",
    "To talk to the OpenML server you need an API key that can be found in your profile on openml.org (after you are logged in)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Set your API key in scripts\n",
    "#openml.config.apikey = FILL_IN_YOUR_KEY\n",
    "\n",
    "# To avoid showing your API key in scripts, create a file in your home dir '~/.openml/config' with the line:\n",
    "#apikey = FILL_IN_YOUR_KEY"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### List ALL the datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "First 10 of 19488 datasets...\n",
      "                name NumberOfInstances NumberOfFeatures\n",
      "did                                                    \n",
      "1             anneal               898               39\n",
      "2             anneal               898               39\n",
      "3           kr-vs-kp              3196               37\n",
      "4              labor                57               17\n",
      "5         arrhythmia               452              280\n",
      "6             letter             20000               17\n",
      "7          audiology               226               70\n",
      "8    liver-disorders               345                7\n",
      "9              autos               205               26\n",
      "10             lymph               148               19\n"
     ]
    }
   ],
   "source": [
    "# Ask OpenML for all datasets\n",
    "datalist = datasets.list_datasets()\n",
    "\n",
    "# Convert to pandas dataframe\n",
    "data_sets = pd.DataFrame(datalist).T\n",
    "data_sets.set_index('did', inplace=True)\n",
    "print(\"First 10 of %s datasets...\" % len(datalist))\n",
    "print(data_sets[:10][['name','NumberOfInstances','NumberOfFeatures']])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "OpenML returns a lot of meta-data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MajorityClassSize                        684\n",
      "MaxNominalAttDistinctValues               10\n",
      "MinorityClassSize                          0\n",
      "NumberOfClasses                            6\n",
      "NumberOfFeatures                          39\n",
      "NumberOfInstances                        898\n",
      "NumberOfInstancesWithMissingValues         0\n",
      "NumberOfMissingValues                      0\n",
      "NumberOfNumericFeatures                    6\n",
      "NumberOfSymbolicFeatures                  33\n",
      "format                                  ARFF\n",
      "name                                  anneal\n",
      "status                                active\n",
      "Name: 1, dtype: object\n"
     ]
    }
   ],
   "source": [
    "# Dataset 1 (anneal)\n",
    "print(data_sets.loc[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Filter datasets based on any property"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "First 10 of 226 datasets...\n",
      "                               name  NumberOfInstances\n",
      "did                                                   \n",
      "1588                            w8a              64700\n",
      "4533   KEGGMetabolicReactionNetwork              65554\n",
      "1591                      connect-4              67557\n",
      "554                       mnist_784              70000\n",
      "1578                       real-sim              72309\n",
      "1213                        BNG(mv)              78732\n",
      "23396               COMET_MC_SAMPLE              89640\n",
      "23395               COMET_MC_SAMPLE              89640\n",
      "23517                   numerai28.6              96320\n",
      "4532                          higgs              98050\n"
     ]
    }
   ],
   "source": [
    "filter = (data_sets.NumberOfInstances > 60000) & (data_sets.NumberOfFeatures > 5)\n",
    "big_data = data_sets.loc[filter].sort_values(by='NumberOfInstances', ascending=True)\n",
    "print(\"First 10 of %s datasets...\" % len(big_data))\n",
    "print(big_data[:10][['name', 'NumberOfInstances']])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "Download a specific dataset. This is done based on the dataset ID (called 'did' in the table above)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "This is dataset 'iris', the target feature is called 'class'\n",
      "URL: http://www.openml.org/data/download/61/dataset_61_iris.arff\n",
      "**Author**: R.A. Fisher  \n",
      "**Source**: [UCI](https://archive.ics.uci.edu/ml/datasets/Iris) - 1936 - Donated by Michael Marshall  \n",
      "**Please cite**:   \n",
      "\n",
      "**Iris Plants Database**  \n",
      "This is perhaps the best known database to be found in the pattern recognition literature.  Fisher's paper is a classic in the field and is referenced frequently to this day.  (See Duda & Hart, for example.)  The data set contains 3 classes of 50 instances each, where each class refers to a type of iris plant.  One class \n"
     ]
    }
   ],
   "source": [
    "dataset = datasets.get_dataset(61)\n",
    "\n",
    "print(\"This is dataset '%s', the target feature is called '%s'\" % (dataset.name, dataset.default_target_attribute))\n",
    "print(\"URL: %s\" % dataset.url)\n",
    "print(dataset.description[:500])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "Get the actual data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 5.1,  3.5,  1.4,  0.2,  0. ],\n",
       "       [ 4.9,  3. ,  1.4,  0.2,  0. ],\n",
       "       [ 4.7,  3.2,  1.3,  0.2,  0. ],\n",
       "       [ 4.6,  3.1,  1.5,  0.2,  0. ],\n",
       "       [ 5. ,  3.6,  1.4,  0.2,  0. ],\n",
       "       [ 5.4,  3.9,  1.7,  0.4,  0. ],\n",
       "       [ 4.6,  3.4,  1.4,  0.3,  0. ],\n",
       "       [ 5. ,  3.4,  1.5,  0.2,  0. ],\n",
       "       [ 4.4,  2.9,  1.4,  0.2,  0. ],\n",
       "       [ 4.9,  3.1,  1.5,  0.1,  0. ],\n",
       "       [ 5.4,  3.7,  1.5,  0.2,  0. ],\n",
       "       [ 4.8,  3.4,  1.6,  0.2,  0. ],\n",
       "       [ 4.8,  3. ,  1.4,  0.1,  0. ],\n",
       "       [ 4.3,  3. ,  1.1,  0.1,  0. ],\n",
       "       [ 5.8,  4. ,  1.2,  0.2,  0. ],\n",
       "       [ 5.7,  4.4,  1.5,  0.4,  0. ],\n",
       "       [ 5.4,  3.9,  1.3,  0.4,  0. ],\n",
       "       [ 5.1,  3.5,  1.4,  0.3,  0. ],\n",
       "       [ 5.7,  3.8,  1.7,  0.3,  0. ],\n",
       "       [ 5.1,  3.8,  1.5,  0.3,  0. ],\n",
       "       [ 5.4,  3.4,  1.7,  0.2,  0. ],\n",
       "       [ 5.1,  3.7,  1.5,  0.4,  0. ],\n",
       "       [ 4.6,  3.6,  1. ,  0.2,  0. ],\n",
       "       [ 5.1,  3.3,  1.7,  0.5,  0. ],\n",
       "       [ 4.8,  3.4,  1.9,  0.2,  0. ],\n",
       "       [ 5. ,  3. ,  1.6,  0.2,  0. ],\n",
       "       [ 5. ,  3.4,  1.6,  0.4,  0. ],\n",
       "       [ 5.2,  3.5,  1.5,  0.2,  0. ],\n",
       "       [ 5.2,  3.4,  1.4,  0.2,  0. ],\n",
       "       [ 4.7,  3.2,  1.6,  0.2,  0. ],\n",
       "       [ 4.8,  3.1,  1.6,  0.2,  0. ],\n",
       "       [ 5.4,  3.4,  1.5,  0.4,  0. ],\n",
       "       [ 5.2,  4.1,  1.5,  0.1,  0. ],\n",
       "       [ 5.5,  4.2,  1.4,  0.2,  0. ],\n",
       "       [ 4.9,  3.1,  1.5,  0.1,  0. ],\n",
       "       [ 5. ,  3.2,  1.2,  0.2,  0. ],\n",
       "       [ 5.5,  3.5,  1.3,  0.2,  0. ],\n",
       "       [ 4.9,  3.1,  1.5,  0.1,  0. ],\n",
       "       [ 4.4,  3. ,  1.3,  0.2,  0. ],\n",
       "       [ 5.1,  3.4,  1.5,  0.2,  0. ],\n",
       "       [ 5. ,  3.5,  1.3,  0.3,  0. ],\n",
       "       [ 4.5,  2.3,  1.3,  0.3,  0. ],\n",
       "       [ 4.4,  3.2,  1.3,  0.2,  0. ],\n",
       "       [ 5. ,  3.5,  1.6,  0.6,  0. ],\n",
       "       [ 5.1,  3.8,  1.9,  0.4,  0. ],\n",
       "       [ 4.8,  3. ,  1.4,  0.3,  0. ],\n",
       "       [ 5.1,  3.8,  1.6,  0.2,  0. ],\n",
       "       [ 4.6,  3.2,  1.4,  0.2,  0. ],\n",
       "       [ 5.3,  3.7,  1.5,  0.2,  0. ],\n",
       "       [ 5. ,  3.3,  1.4,  0.2,  0. ],\n",
       "       [ 7. ,  3.2,  4.7,  1.4,  1. ],\n",
       "       [ 6.4,  3.2,  4.5,  1.5,  1. ],\n",
       "       [ 6.9,  3.1,  4.9,  1.5,  1. ],\n",
       "       [ 5.5,  2.3,  4. ,  1.3,  1. ],\n",
       "       [ 6.5,  2.8,  4.6,  1.5,  1. ],\n",
       "       [ 5.7,  2.8,  4.5,  1.3,  1. ],\n",
       "       [ 6.3,  3.3,  4.7,  1.6,  1. ],\n",
       "       [ 4.9,  2.4,  3.3,  1. ,  1. ],\n",
       "       [ 6.6,  2.9,  4.6,  1.3,  1. ],\n",
       "       [ 5.2,  2.7,  3.9,  1.4,  1. ],\n",
       "       [ 5. ,  2. ,  3.5,  1. ,  1. ],\n",
       "       [ 5.9,  3. ,  4.2,  1.5,  1. ],\n",
       "       [ 6. ,  2.2,  4. ,  1. ,  1. ],\n",
       "       [ 6.1,  2.9,  4.7,  1.4,  1. ],\n",
       "       [ 5.6,  2.9,  3.6,  1.3,  1. ],\n",
       "       [ 6.7,  3.1,  4.4,  1.4,  1. ],\n",
       "       [ 5.6,  3. ,  4.5,  1.5,  1. ],\n",
       "       [ 5.8,  2.7,  4.1,  1. ,  1. ],\n",
       "       [ 6.2,  2.2,  4.5,  1.5,  1. ],\n",
       "       [ 5.6,  2.5,  3.9,  1.1,  1. ],\n",
       "       [ 5.9,  3.2,  4.8,  1.8,  1. ],\n",
       "       [ 6.1,  2.8,  4. ,  1.3,  1. ],\n",
       "       [ 6.3,  2.5,  4.9,  1.5,  1. ],\n",
       "       [ 6.1,  2.8,  4.7,  1.2,  1. ],\n",
       "       [ 6.4,  2.9,  4.3,  1.3,  1. ],\n",
       "       [ 6.6,  3. ,  4.4,  1.4,  1. ],\n",
       "       [ 6.8,  2.8,  4.8,  1.4,  1. ],\n",
       "       [ 6.7,  3. ,  5. ,  1.7,  1. ],\n",
       "       [ 6. ,  2.9,  4.5,  1.5,  1. ],\n",
       "       [ 5.7,  2.6,  3.5,  1. ,  1. ],\n",
       "       [ 5.5,  2.4,  3.8,  1.1,  1. ],\n",
       "       [ 5.5,  2.4,  3.7,  1. ,  1. ],\n",
       "       [ 5.8,  2.7,  3.9,  1.2,  1. ],\n",
       "       [ 6. ,  2.7,  5.1,  1.6,  1. ],\n",
       "       [ 5.4,  3. ,  4.5,  1.5,  1. ],\n",
       "       [ 6. ,  3.4,  4.5,  1.6,  1. ],\n",
       "       [ 6.7,  3.1,  4.7,  1.5,  1. ],\n",
       "       [ 6.3,  2.3,  4.4,  1.3,  1. ],\n",
       "       [ 5.6,  3. ,  4.1,  1.3,  1. ],\n",
       "       [ 5.5,  2.5,  4. ,  1.3,  1. ],\n",
       "       [ 5.5,  2.6,  4.4,  1.2,  1. ],\n",
       "       [ 6.1,  3. ,  4.6,  1.4,  1. ],\n",
       "       [ 5.8,  2.6,  4. ,  1.2,  1. ],\n",
       "       [ 5. ,  2.3,  3.3,  1. ,  1. ],\n",
       "       [ 5.6,  2.7,  4.2,  1.3,  1. ],\n",
       "       [ 5.7,  3. ,  4.2,  1.2,  1. ],\n",
       "       [ 5.7,  2.9,  4.2,  1.3,  1. ],\n",
       "       [ 6.2,  2.9,  4.3,  1.3,  1. ],\n",
       "       [ 5.1,  2.5,  3. ,  1.1,  1. ],\n",
       "       [ 5.7,  2.8,  4.1,  1.3,  1. ],\n",
       "       [ 6.3,  3.3,  6. ,  2.5,  2. ],\n",
       "       [ 5.8,  2.7,  5.1,  1.9,  2. ],\n",
       "       [ 7.1,  3. ,  5.9,  2.1,  2. ],\n",
       "       [ 6.3,  2.9,  5.6,  1.8,  2. ],\n",
       "       [ 6.5,  3. ,  5.8,  2.2,  2. ],\n",
       "       [ 7.6,  3. ,  6.6,  2.1,  2. ],\n",
       "       [ 4.9,  2.5,  4.5,  1.7,  2. ],\n",
       "       [ 7.3,  2.9,  6.3,  1.8,  2. ],\n",
       "       [ 6.7,  2.5,  5.8,  1.8,  2. ],\n",
       "       [ 7.2,  3.6,  6.1,  2.5,  2. ],\n",
       "       [ 6.5,  3.2,  5.1,  2. ,  2. ],\n",
       "       [ 6.4,  2.7,  5.3,  1.9,  2. ],\n",
       "       [ 6.8,  3. ,  5.5,  2.1,  2. ],\n",
       "       [ 5.7,  2.5,  5. ,  2. ,  2. ],\n",
       "       [ 5.8,  2.8,  5.1,  2.4,  2. ],\n",
       "       [ 6.4,  3.2,  5.3,  2.3,  2. ],\n",
       "       [ 6.5,  3. ,  5.5,  1.8,  2. ],\n",
       "       [ 7.7,  3.8,  6.7,  2.2,  2. ],\n",
       "       [ 7.7,  2.6,  6.9,  2.3,  2. ],\n",
       "       [ 6. ,  2.2,  5. ,  1.5,  2. ],\n",
       "       [ 6.9,  3.2,  5.7,  2.3,  2. ],\n",
       "       [ 5.6,  2.8,  4.9,  2. ,  2. ],\n",
       "       [ 7.7,  2.8,  6.7,  2. ,  2. ],\n",
       "       [ 6.3,  2.7,  4.9,  1.8,  2. ],\n",
       "       [ 6.7,  3.3,  5.7,  2.1,  2. ],\n",
       "       [ 7.2,  3.2,  6. ,  1.8,  2. ],\n",
       "       [ 6.2,  2.8,  4.8,  1.8,  2. ],\n",
       "       [ 6.1,  3. ,  4.9,  1.8,  2. ],\n",
       "       [ 6.4,  2.8,  5.6,  2.1,  2. ],\n",
       "       [ 7.2,  3. ,  5.8,  1.6,  2. ],\n",
       "       [ 7.4,  2.8,  6.1,  1.9,  2. ],\n",
       "       [ 7.9,  3.8,  6.4,  2. ,  2. ],\n",
       "       [ 6.4,  2.8,  5.6,  2.2,  2. ],\n",
       "       [ 6.3,  2.8,  5.1,  1.5,  2. ],\n",
       "       [ 6.1,  2.6,  5.6,  1.4,  2. ],\n",
       "       [ 7.7,  3. ,  6.1,  2.3,  2. ],\n",
       "       [ 6.3,  3.4,  5.6,  2.4,  2. ],\n",
       "       [ 6.4,  3.1,  5.5,  1.8,  2. ],\n",
       "       [ 6. ,  3. ,  4.8,  1.8,  2. ],\n",
       "       [ 6.9,  3.1,  5.4,  2.1,  2. ],\n",
       "       [ 6.7,  3.1,  5.6,  2.4,  2. ],\n",
       "       [ 6.9,  3.1,  5.1,  2.3,  2. ],\n",
       "       [ 5.8,  2.7,  5.1,  1.9,  2. ],\n",
       "       [ 6.8,  3.2,  5.9,  2.3,  2. ],\n",
       "       [ 6.7,  3.3,  5.7,  2.5,  2. ],\n",
       "       [ 6.7,  3. ,  5.2,  2.3,  2. ],\n",
       "       [ 6.3,  2.5,  5. ,  1.9,  2. ],\n",
       "       [ 6.5,  3. ,  5.2,  2. ,  2. ],\n",
       "       [ 6.2,  3.4,  5.4,  2.3,  2. ],\n",
       "       [ 5.9,  3. ,  5.1,  1.8,  2. ]], dtype=float32)"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = dataset.get_data()\n",
    "X"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "Or in a nice pandas dataframe:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   sepallength  sepalwidth  petallength  petalwidth  class\n",
      "0          5.1         3.5          1.4         0.2      0\n",
      "1          4.9         3.0          1.4         0.2      0\n",
      "2          4.7         3.2          1.3         0.2      0\n",
      "3          4.6         3.1          1.5         0.2      0\n",
      "4          5.0         3.6          1.4         0.2      0\n",
      "5          5.4         3.9          1.7         0.4      0\n",
      "6          4.6         3.4          1.4         0.3      0\n",
      "7          5.0         3.4          1.5         0.2      0\n",
      "8          4.4         2.9          1.4         0.2      0\n",
      "9          4.9         3.1          1.5         0.1      0\n"
     ]
    }
   ],
   "source": [
    "# Get data, as well as the labels (target) and attribute names\n",
    "X2, y2, attribute_names = dataset.get_data(target=dataset.default_target_attribute, return_attribute_names=True)\n",
    "iris = pd.DataFrame(X2, columns=attribute_names)\n",
    "iris['class'] = y2\n",
    "print(iris[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x122faa860>"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfsAAAHuCAYAAACYkKX6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ncQp7ERERj/v/i5N5E5qDM8UAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1157ab748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "iris.plot(kind='scatter', x='petallength', y='petalwidth', c='class', colormap='gist_rainbow', colorbar=False, s=50)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In case you are working with scikit-learn, you can use this data right away"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=10, n_jobs=1,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False)"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn import preprocessing, ensemble\n",
    "X, y, categorical = dataset.get_data(target=dataset.default_target_attribute, return_categorical_indicator=True)\n",
    "enc = preprocessing.OneHotEncoder(categorical_features=categorical)\n",
    "clf = ensemble.RandomForestClassifier()\n",
    "clf.fit(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## A simple and unified API\n",
    "\n",
    "All learning algorithms in scikit-learn share a uniform and limited API consisting of complementary interfaces:\n",
    "\n",
    "- an `estimator` interface for building and fitting models;\n",
    "- a `predictor` interface for making predictions;\n",
    "- a `transformer` interface for converting data.\n",
    "\n",
    "You can __swap or plug algorithms__ "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### Estimators"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [],
   "source": [
    "class Estimator(object):\n",
    "    def fit(self, X, y=None):\n",
    "        \"\"\"Fits estimator to data.\"\"\"\n",
    "        # set state of ``self``\n",
    "        return self"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'make_blobs' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-8-6b0847955a85>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# Back to blobs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmake_blobs\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mn_samples\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1000\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcenters\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m20\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrandom_state\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m123\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      3\u001b[0m \u001b[0mlabels\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m\"b\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"r\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtake\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlabels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0my\u001b[0m \u001b[0;34m<\u001b[0m \u001b[0;36m10\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;31m# Relabels numeric values to b,r\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'make_blobs' is not defined"
     ]
    }
   ],
   "source": [
    "# Back to blobs\n",
    "X, y = make_blobs(n_samples=1000, centers=20, random_state=123)\n",
    "labels = [\"b\", \"r\"]\n",
    "y = np.take(labels, (y < 10)) # Relabels numeric values to b,r\n",
    "\n",
    "# Import the decision tree class\n",
    "from sklearn.tree import DecisionTreeClassifier  # Change this to try \n",
    "                                                 # something else \n",
    "\n",
    "# Set hyper-parameters, for controlling algorithm\n",
    "clf = DecisionTreeClassifier()\n",
    "\n",
    "\n",
    "# Learn a model from training data\n",
    "clf.fit(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### Predictors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "# Make predictions  \n",
    "print(clf.predict(X[:5])) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "# Compute (approximate) class probabilities\n",
    "print(clf.predict_proba(X[:5]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "plot_surface(clf, X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "clf = DecisionTreeClassifier(criterion='entropy', max_depth=None) # information gain\n",
    "clf.fit(X, y)\n",
    "plot_surface(clf, X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Classifier zoo"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### K-nearest neighbours\n",
    "\n",
    "Idea: Make prediction based on target class of k nearest neighbors (vote)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "# Set hyper-parameters, for controlling algorithm\n",
    "clf = KNeighborsClassifier(n_neighbors=5)\n",
    "clf.fit(X, y)\n",
    "plot_surface(clf, X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### Random Forests\n",
    "\n",
    "Idea: Build several decision trees with controlled randomness and average their decisions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier \n",
    "clf = RandomForestClassifier(n_estimators=500)\n",
    "clf.fit(X, y)\n",
    "plot_surface(clf, X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### Support vector machines\n",
    "\n",
    "Idea: Find the hyperplane which has the largest distance to the nearest training points of any class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.svm import SVC\n",
    "clf = SVC(kernel=\"rbf\")  # try kernel=\"rbf\" instead\n",
    "clf.fit(X, y)\n",
    "plot_surface(clf, X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "# Model evaluation and selection\n",
    "Meant as demonstration. Theory given in next lecture."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Training error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.metrics import zero_one_loss\n",
    "clf = DecisionTreeClassifier()\n",
    "clf.fit(X, y)\n",
    "print(\"Training error =\", zero_one_loss(y, clf.predict(X)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Test error\n",
    "\n",
    "Issue: the training error is a __biased__ estimate of the generalization error.\n",
    "\n",
    "Solution: Divide data into two disjoint parts called training and test sets (usually using 70% for training and 30% for test).\n",
    "- Use the training set for fitting the model;\n",
    "- Use the test set for evaluation only, thereby yielding an unbiased estimate.\n",
    "- __The same data should not be used both for training and evaluation.__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.cross_validation import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y)\n",
    "clf = DecisionTreeClassifier()\n",
    "clf.fit(X_train, y_train)\n",
    "print(\"Training error =\", zero_one_loss(y_train, clf.predict(X_train)))\n",
    "print(\"Test error =\", zero_one_loss(y_test, clf.predict(X_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Cross-validation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Issue: \n",
    "- When data is small, training on 70% of the data may lead to a model that is significantly different from a model that would have been learned on the entire set. \n",
    "- Yet, increasing the size of the training set (resp. decreasing the size of the test set), might lead to an inaccurate estimate of the generalization error. \n",
    "\n",
    "Solution: K-Fold cross-validation. \n",
    "- Split data into K small disjoint folds. \n",
    "- Train on K-1 folds, evaluate the test error one the held-out fold.\n",
    "- Repeat for all combinations and average the K estimates of the generalization error."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.cross_validation import KFold\n",
    "\n",
    "scores = []\n",
    "\n",
    "for train, test in KFold(n=len(X), n_folds=5, random_state=42):\n",
    "    X_train, y_train = X[train], y[train]\n",
    "    X_test, y_test = X[test], y[test]\n",
    "    clf = DecisionTreeClassifier().fit(X_train, y_train)\n",
    "    scores.append(zero_one_loss(y_test, clf.predict(X_test)))\n",
    "\n",
    "print(\"CV error = %f +-%f\" % (np.mean(scores), np.std(scores)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "# Shortcut\n",
    "from sklearn.cross_validation import cross_val_score\n",
    "scores = cross_val_score(DecisionTreeClassifier(), X, y, \n",
    "                         cv=KFold(n=len(X), n_folds=5, random_state=42), \n",
    "                         scoring=\"accuracy\")\n",
    "print(\"CV error = %f +-%f\" % (1. - np.mean(scores), np.std(scores)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Metrics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### Default score\n",
    "\n",
    "Estimators come with a built-in default evaluation score\n",
    "* Accuracy for classification \n",
    "* R2 score for regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "y_train = (y_train == \"r\")\n",
    "y_test = (y_test == \"r\")\n",
    "clf = KNeighborsClassifier(n_neighbors=5)\n",
    "clf.fit(X_train, y_train) \n",
    "print(\"Default score =\", clf.score(X_test, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### Accuracy\n",
    "\n",
    "Definition: The accuracy is the proportion of correct predictions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score\n",
    "print(\"Accuracy =\", accuracy_score(y_test, clf.predict(X_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### Precision, recall and F-measure\n",
    "\n",
    "$$Precision = \\frac{TP}{TP + FP}$$\n",
    "$$Recall = \\frac{TP}{TP + FN}$$\n",
    "$$F = \\frac{2 * Precision * Recall}{Precision + Recall}$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import precision_score\n",
    "from sklearn.metrics import recall_score\n",
    "from sklearn.metrics import fbeta_score\n",
    "print(\"Precision =\", precision_score(y_test, clf.predict(X_test)))\n",
    "print(\"Recall =\", recall_score(y_test, clf.predict(X_test)))\n",
    "print(\"F =\", fbeta_score(y_test, clf.predict(X_test), beta=1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### ROC AUC\n",
    "\n",
    "Definition: Area under the curve of the false positive rate (FPR) against the true positive rate (TPR) as the decision threshold of the classifier is varied."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import get_scorer\n",
    "roc_auc_scorer = get_scorer(\"roc_auc\")\n",
    "print(\"ROC AUC =\", roc_auc_scorer(clf, X_test, y_test))\n",
    "\n",
    "from sklearn.metrics import roc_curve\n",
    "fpr, tpr, thresholds = roc_curve(y_test, clf.predict_proba(X_test)[:, 1])\n",
    "plt.plot(fpr, tpr)\n",
    "plt.xlabel(\"FPR\")\n",
    "plt.ylabel(\"TPR\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "### Confusion matrix\n",
    "\n",
    "Definition: number of samples of class $i$ predicted as class $j$."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'y_test' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-9-879849b57689>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0msklearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmetrics\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mconfusion_matrix\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mconfusion_matrix\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my_test\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mclf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX_test\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m: name 'y_test' is not defined"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "confusion_matrix(y_test, clf.predict(X_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "# Transformers, pipelines and feature unions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Transformers\n",
    "\n",
    "- Classification (or regression) is often only one or the last step of a long and complicated process;\n",
    "- In most cases, input data needs to be cleaned, massaged or extended before being fed to a learning algorithm;\n",
    "- For this purpose, Scikit-Learn provides the ``transformer`` API."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "class Transformer(object):    \n",
    "    def fit(self, X, y=None):\n",
    "        \"\"\"Fits estimator to data.\"\"\"\n",
    "        # set state of ``self``\n",
    "        return self\n",
    "    \n",
    "    def transform(self, X):\n",
    "        \"\"\"Transform X into Xt.\"\"\"\n",
    "        # transform X in some way to produce Xt\n",
    "        return Xt\n",
    "    \n",
    "    # Shortcut\n",
    "    def fit_transform(self, X, y=None):\n",
    "        self.fit(X, y)\n",
    "        Xt = self.transform(X)\n",
    "        return Xt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Pipelines\n",
    "\n",
    "Transformers can be chained in sequence to form a pipeline."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'train_test_split' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-10-77435598ec88>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0mdataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdatasets\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_dataset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m337\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdataset\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtarget\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdefault_target_attribute\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m \u001b[0mX_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX_test\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_test\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtrain_test_split\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      8\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;31m# Chain transformers + a classifier to build a new classifier\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'train_test_split' is not defined"
     ]
    }
   ],
   "source": [
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.feature_selection import SelectKBest, f_classif\n",
    " \n",
    "# Get more complex data\n",
    "dataset = datasets.get_dataset(337)\n",
    "X, y = dataset.get_data(target=dataset.default_target_attribute)\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y)\n",
    "\n",
    "# Chain transformers + a classifier to build a new classifier\n",
    "clf = make_pipeline(SelectKBest(score_func=f_classif, k=44), \n",
    "                    RandomForestClassifier())\n",
    "clf.fit(X_train, y_train)\n",
    "print(clf.predict_proba(X_test)[:5])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "## Optimizing parameters\n",
    "Search for the best hyperparameter settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "# Hyper-parameters can be accessed using step names\n",
    "print(\"K =\", clf.get_params()[\"selectkbest__k\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.grid_search import GridSearchCV\n",
    "grid = GridSearchCV(clf, \n",
    "                    param_grid={\"selectkbest__k\": [1, 10, 20, 30, 40],\n",
    "                                \"randomforestclassifier__max_features\": [0.1, 0.25, 0.5]})\n",
    "grid.fit(X_train, y_train)\n",
    "\n",
    "print(\"Best params =\", grid.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "# Summary\n",
    "\n",
    "- Scikit-Learn provides essential tools for machine learning. \n",
    "- It is more than training classifiers!\n",
    "- It integrates within a larger Python scientific ecosystem.\n",
    "- Try it for yourself!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "source": [
    "questions?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## Acknowledgements\n",
    "Based on a tutorial by Gilles Loupe and sci-kit learn documentation"
   ]
  }
 ],
 "metadata": {
  "celltoolbar": "Slideshow",
  "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
}