practical 4.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Practical 4\n",
    "### Cross validation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This practical examines applying cross validation. It is straight forward to apply using sklearn, however be sure you understand also what cross validation means: review the text book, lecture 5 and the manual for sklearn cross validation and model selection  (http://scikit-learn.org/stable/modules/cross_validation.html). Also do a search online (e.g. see https://blog.csdn.net/zjsghww/article/details/51603814)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Pandas \n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "# matplotlib\n",
    "import matplotlib.pylab as plt\n",
    "%matplotlib inline\n",
    "\n",
    "# Decision trees and logistic regression\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "# Train, test, and evaluation functions\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.model_selection  import cross_val_score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data\n",
    "We're going to use a dataset from Kaggle that contains information on some passengers from the sinking of the Titanic. This data set has 10 columns.\n",
    "\n",
    "![title](images/titanic.jpg)\n",
    "\n",
    "The columns (features) are:\n",
    "\n",
    "```\n",
    "Col.  Name        Description\n",
    "----- ----------- ----------------------------------------------------------------\n",
    "1     Survived    Survival (0 = No; 1 = Yes)\n",
    "2     Pclass      Passenger Class (1 = 1st; 2 = 2nd; 3 = 3rd)\n",
    "3     Sex         Sex (0 = Female, 1 = Male)\n",
    "4     Age         Age\n",
    "5     SibSp       Number of Siblings/Spouses Aboard\n",
    "6     Parch       Number of Parents/Children Aboard\n",
    "7     Fare        Passenger Fare\n",
    "8     Embarked_S  Embarked from Southampton\n",
    "9     Embarked_C  Embarked from Cherbourg\n",
    "10    Embarked_Q  Embarked from Queenstown\n",
    "```\n",
    "\n",
    "Our goal is to build a model to predict if people survived the sinking (the first column, called `\"Survived\"`."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Grid Searching for a Decision Tree\n",
    "Below is some simple code to use 10-fold cross validation to get the performance for different decision trees with varying levels of complexity."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEKCAYAAADjDHn2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x225701da588>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Read data using pandas\n",
    "data = pd.read_csv(\"data/titanic.csv\")\n",
    "\n",
    "# Split into X and Y\n",
    "X = data.drop(['Survived'], 1)\n",
    "Y = data['Survived']\n",
    "\n",
    "# I will try maximum depths from 1 to 16\n",
    "complexity_values = range(1,16)\n",
    "\n",
    "# I need an empty list to hold my accuracies\n",
    "accuracies = []\n",
    "\n",
    "# Go through each complexity_value one at a time.\n",
    "for complexity_value in complexity_values:\n",
    "    # This will create an empty tree with a specific maximum depth (complexity)\n",
    "    tree = DecisionTreeClassifier(criterion=\"entropy\", max_depth=complexity_value)\n",
    "    \n",
    "    # This will get us 10-fold cross validation accuracy with our tree and our data\n",
    "    # We can do this in one line!\n",
    "    cross_fold_accuracies = cross_val_score(tree, X, Y, scoring=\"accuracy\", cv=10)\n",
    "    \n",
    "    # Average accuracy for all tests\n",
    "    average_cross_fold_accuracy = np.mean(cross_fold_accuracies)\n",
    "    \n",
    "    # Append this accuracy to a list of accuracies\n",
    "    accuracies.append(average_cross_fold_accuracy)\n",
    "\n",
    "# Plot results for different complexity levels\n",
    "plt.plot(complexity_values, accuracies)\n",
    "plt.xlabel(\"Maximum Depth\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### Team work\n",
    "Get together with your teams and modify this code block to try different logistic regressions. \n",
    "\n",
    "Remember, here complexity if called $C$. $C$ is used like `max_depth` for instance is used to control complexity in a decision tree.\n",
    "\n",
    "This value is similar to the $\\lambda$ value discussed in class --- see slide 41 in lecture 5, and p 137 in the text  --- however note that here smaller values of $C$ correspond to a stronger regularization penalty, i.e. $C = \\frac{1}{\\lambda}$). \n",
    "\n",
    "Normal ranges for $C$ are powers of 10. For example: ..., 0.01, 0.1, 1, 1.0, 10.0, ... \n",
    "\n",
    "**Try different values of $C$ to try to find the best model.**\n",
    "\n",
    "**What are your thoughts in comparing decision trees and logistic regression?**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAELCAYAAAAoUKpTAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYWFmZiU5LMzMrCSHhZmZleSwMDOzkv4/s8Mp3PXG+JkAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x22570aa2ba8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Read data using pandas\n",
    "data = pd.read_csv(\"data/titanic.csv\")\n",
    "\n",
    "# Split into X and Y\n",
    "X = data.drop(['Survived'], 1)\n",
    "Y = data['Survived']\n",
    "\n",
    "# Try different regularizations (powers of 10)\n",
    "complexity_values = [0.0001,0.001,0.01,0.1,1,10]\n",
    "\n",
    "# We need an empty list to hold the accuracies\n",
    "accuracies = []\n",
    "\n",
    "# I will go through each complexity_value one at a time.\n",
    "for complexity_value in complexity_values:\n",
    "    # This will create an empty logistic regression with a particular regularization (C)\n",
    "    logistic = LogisticRegression(C = complexity_value)\n",
    "    \n",
    "    # This will get us 10-fold cross validation accuracy with our logistic reg. and our data\n",
    "    # We can do this in one line!\n",
    "    cross_fold_accuracies = cross_val_score(logistic, X, Y, scoring=\"accuracy\", cv = 10)\n",
    "    \n",
    "    # Average accuracy\n",
    "    average_cross_fold_accuracy = np.mean(cross_fold_accuracies)\n",
    "    \n",
    "    # Append this accuracy to a list of accuracies\n",
    "    accuracies.append(average_cross_fold_accuracy)\n",
    "\n",
    "# We want to plot our results\n",
    "plt.plot(complexity_values, accuracies)\n",
    "plt.xlabel(\"complexity_values\")\n",
    "plt.ylabel(\"accuracies\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}