ir.classifiers
Class NaiveBayes

java.lang.Object
  ir.classifiers.Classifier
      ir.classifiers.NaiveBayes

public class NaiveBayes
extends Classifier

Implements the NaiveBayes Classifier with Laplace smoothing. Stores probabilities internally as logs to prevent underflow problems.

Field Summary

static java.lang.String

name Name of classifier

Fields inherited from class ir.classifiers.Classifier

categories, random

Constructor Summary

NaiveBayes(java.lang.String[] categories, boolean debug)
Create a naive Bayes classifier with these attributes

Method Summary

protected double[]	calculatePriors(java.util.List<Example> trainExamples) Calculates the class priors
protected double[]	calculateProbs(Example testExample) Calculates the prob of the testExample being generated by each category
protected java.util.Hashtable<java.lang.String,double[]>	conditionalProbs(java.util.List<Example> trainExamples) Calculates the conditional probs of each feature in the different categories
protected void	displayProbs(double[] classPriors, java.util.Hashtable<java.lang.String,double[]> featureHash) Displays the probs for each feature in the different categories
double	getEpsilon() Returns value of EPSILON
boolean	getIsLaplace() Returns value of isLaplace
java.lang.String	getName() Returns the name
BayesResult	getTrainResult() Returns training result
void	setDebug(boolean bool) Sets the debug flag
void	setEpsilon(double ep) Sets the value of EPSILON (default 1e-6)
void	setLaplace(boolean bool) Sets the Laplace smoothing flag
boolean	test(Example testExample) Categorizes the test example using the trained Naive Bayes classifier, returning true if the predicted category is same as the actual category
void	train(java.util.List<Example> trainExamples) Trains the Naive Bayes classifier - estimates the prior probs and calculates the counts for each feature in different categories

Methods inherited from class ir.classifiers.Classifier

argMax, getCategories

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

name

public static final java.lang.String name

Name of classifier

See Also:

Constant Field Values

Constructor Detail

NaiveBayes

public NaiveBayes(java.lang.String[] categories,
                  boolean debug)

Create a naive Bayes classifier with these attributes

Parameters:

categories - The array of Strings containing the category names

debug - Flag to turn on detailed output

Method Detail

setDebug

public void setDebug(boolean bool)

Sets the debug flag

setLaplace

public void setLaplace(boolean bool)

Sets the Laplace smoothing flag

setEpsilon

public void setEpsilon(double ep)

Sets the value of EPSILON (default 1e-6)

getName

public java.lang.String getName()

Returns the name

Specified by:

getName in class Classifier

Returns:

the name of a particular classifier

getEpsilon

public double getEpsilon()

Returns value of EPSILON

getTrainResult

public BayesResult getTrainResult()

Returns training result

getIsLaplace

public boolean getIsLaplace()

Returns value of isLaplace

train

public void train(java.util.List<Example> trainExamples)

Trains the Naive Bayes classifier - estimates the prior probs and calculates the counts for each feature in different categories

Specified by:

train in class Classifier

Parameters:

trainExamples - The vector of training examples

test

public boolean test(Example testExample)

Categorizes the test example using the trained Naive Bayes classifier, returning true if the predicted category is same as the actual category

Specified by:

test in class Classifier

Parameters:

testExample - The test example to be categorized

calculatePriors

protected double[] calculatePriors(java.util.List<Example> trainExamples)

Calculates the class priors

Parameters:

trainExamples - The training examples from which class priors will be estimated

conditionalProbs

protected java.util.Hashtable<java.lang.String,double[]> conditionalProbs(java.util.List<Example> trainExamples)

Calculates the conditional probs of each feature in the different categories

Parameters:

trainExamples - The training examples from which counts will be estimated

calculateProbs

protected double[] calculateProbs(Example testExample)

Calculates the prob of the testExample being generated by each category

Parameters:

testExample - The test example to be categorized

displayProbs

protected void displayProbs(double[] classPriors,
                            java.util.Hashtable<java.lang.String,double[]> featureHash)

Displays the probs for each feature in the different categories

Parameters:

classPriors - Prior probs

featureHash - Feature hashtable after training