3-Validation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Validation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Model validation: How well is your model performing?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the previous notebook, we tested our data on the same set of data that was used for training. That's cheating!!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If we optimize our model this way, we will **over-fit** the data: that is, we learn the patterns that are true for that particular dataset, but don't generalize well to data that the model hasn't seen before."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Holdout method"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create our own test set (different from the Kaggle one!). \n",
    "\n",
    "- That test set is not going to be used to train the model. \n",
    "- Since, you'll know the true value of the holdout set, you'll be able to assess how good your model is performing without cheating\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df = pd.read_csv('trainDataFeatures.tsv', sep='\\t', index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "columns = df.columns[3:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X = np.asarray(df[columns])\n",
    "y = np.asarray(df.sentiment.transpose())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((18750, 9), (6250, 9))"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.cross_validation import train_test_split\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y)\n",
    "X_train.shape, X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MultinomialNB(alpha=1.0, class_prior=None, fit_prior=True)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.naive_bayes import MultinomialNB\n",
    "nb = MultinomialNB()\n",
    "nb.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4169 / 6250 correct\n"
     ]
    }
   ],
   "source": [
    "y_pred = nb.predict(X_test)\n",
    "print(\"{0} / {1} correct\".format(np.sum(y_test == y_pred), len(y_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Accuracy:** the number of correct matches divided by the total number of samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.66703999999999997"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import accuracy_score\n",
    "accuracy_score(y_test, y_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Crossvalidation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.cross_validation import cross_val_score, cross_val_predict\n",
    "cv = cross_val_score(MultinomialNB(), X_train, y_train, cv=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.67136046475959898"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1493, 1656],\n",
       "       [ 425, 2676]])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "confusion_matrix(y_test, y_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ROC curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_auc_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "fpr, tpr, _ = roc_curve(y_test, y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.66853310208684058"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(y_test, y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "lr = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=100, multi_class='ovr',\n",
       "          penalty='l2', random_state=None, solver='liblinear', tol=0.0001,\n",
       "          verbose=0)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "y_pred = lr.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "fpr, tpr, _ = roc_curve(y_test, y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "roc_curve?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_auc_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "fpr, tpr, _ = roc_curve(y_test, y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "roc_auc_score?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.67023790664030469"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(y_test, y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAqIAAAHzCAYAAAD7O388AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAIyiiAAAAMIIiCgAAACP+D28isI1KE4gyAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1128d9b90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<ggplot: (287889785)>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from ggplot import *\n",
    "\n",
    "preds = lr.predict_proba(X_test)[:,1]\n",
    "fpr, tpr, _ = roc_curve(y_test, preds)\n",
    "\n",
    "df = pd.DataFrame(dict(fpr=fpr, tpr=tpr))\n",
    "ggplot(df, aes(x='fpr', y='tpr')) +\\\n",
    "    geom_line() +\\\n",
    "    geom_abline(linetype='dashed')"
   ]
  }
 ],
 "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": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}