Package edu.cmu.tetrad.classify

Class ClassifierBayesUpdaterDiscrete

java.lang.Object

edu.cmu.tetrad.classify.ClassifierBayesUpdaterDiscrete

All Implemented Interfaces:

ClassifierDiscrete, TetradSerializable, Serializable

public final class ClassifierBayesUpdaterDiscrete
extends Object
implements ClassifierDiscrete, TetradSerializable

This class contains a method classify which uses an instantiated Bayes net (BayesIm) provided in the constructor. For each case (record) in the DataSet it uses the values of all variables but the target variable to update the distributions of all the variables. It then computes an estimated value for the target variable by selecting the value with the greatest probability in the updated distribution. The method returns a crosstabulation table in the form of a two-dimensional integer array in which coefs of observed versus estimated values of the target variable are stored. Note that the variables must be the same in the dataset and the Bayes net.

Version:

$Id: $Id

Author:

Frank Wimberly based on a specification by Clark Glymour

See Also:

• Serialized Form

Constructor Summary

Constructors

Constructor	Description
ClassifierBayesUpdaterDiscrete(BayesIm bayesIm, DataSet testData)	The constructor sets the values of the private member variables.

Method Summary

Modifier and Type	Method	Description
int[]	classify()	Computes and returns the cross-tabulation of observed versus estimated values of the target variable as described above.
int[][]	crossTabulation()	Computes the "confusion matrix" of coefs of the number of cases associated with each combination of estimated and observed values in the test dataset.
BayesIm	getBayesIm()	Returns the Bayes IM being used.
List<Node>	getBayesImVars()	Returns the variables of the BayesIM.
int[]	getClassifications()	Returns the classification for the target variable.
double[][]	getMarginals()	Returns the marginals.
int	getNumCases()	Returns the number of cases for the data.
double	getPercentCorrect()	Getter for the field percentCorrect.
DiscreteVariable	getTargetVariable()	Getter for the field targetVariable.
DataSet	getTestData()	Returns the test data being used.
int	getTotalUsableCases()	Returns the number of usable cases for the data.
static ClassifierBayesUpdaterDiscrete	serializableInstance()	Generates a simple exemplar of this class to test serialization.
void	setTarget(String target)	Sets the target variable.

Methods inherited from class java.lang.Object

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

Constructor Details

ClassifierBayesUpdaterDiscrete

public ClassifierBayesUpdaterDiscrete(BayesIm bayesIm,
 DataSet testData)

The constructor sets the values of the private member variables. The BayesIm instance is used to create an updater. The dataset to be classified is the test data. The variables in the dataset to be classified are the same as the variables in the Bayes net. The target variable is not set.

Parameters:

bayesIm - the BayesIm instance used to create an updater.

testData - the dataset to be classified.

Method Details

serializableInstance

public static ClassifierBayesUpdaterDiscrete serializableInstance()

Generates a simple exemplar of this class to test serialization.

Returns:

a simple exemplar of this class to test serialization.

setTarget

public void setTarget(String target)

Sets the target variable.

Parameters:

target - This variable.

classify

public int[] classify()

Computes and returns the cross-tabulation of observed versus estimated values of the target variable as described above.

Specified by:

classify in interface ClassifierDiscrete

Returns:

this cross-tabulation.

crossTabulation

public int[][] crossTabulation()

Computes the "confusion matrix" of coefs of the number of cases associated with each combination of estimated and observed values in the test dataset. Each row, column i,j corresponds to the ith and jth categories of the target variable.

Specified by:

crossTabulation in interface ClassifierDiscrete

Returns:

an int[][] array containing the coefs, or null if the target variable is not in the test data.

getPercentCorrect

public double getPercentCorrect()

Getter for the field percentCorrect.

Specified by:

getPercentCorrect in interface ClassifierDiscrete

Returns:

the percentage of cases in which the target variable is correctly classified.

getTargetVariable

public DiscreteVariable getTargetVariable()

Getter for the field targetVariable.

Returns:

the discrete variables which is the target variable.

See Also:

• DiscreteVariable

getBayesIm

public BayesIm getBayesIm()

Returns the Bayes IM being used.

Returns:

This IM

See Also:

• BayesIm

getTestData

public DataSet getTestData()

Returns the test data being used.

Returns:

This data.

getClassifications

public int[] getClassifications()

Returns the classification for the target variable.

Returns:

Thsi classification.

getMarginals

public double[][] getMarginals()

Returns the marginals.

Returns:

This.

getNumCases

public int getNumCases()

Returns the number of cases for the data.

Returns:

This number.

getTotalUsableCases

public int getTotalUsableCases()

Returns the number of usable cases for the data.

Returns:

Thsi number.

getBayesImVars

public List<Node> getBayesImVars()

Returns the variables of the BayesIM.

Returns:

These variables as a list.

See Also:

• BayesIm