com.sjm.machlearn.classifiers
Class Classifier

java.lang.Object
  |
  +--com.sjm.machlearn.util.MainClass
        |
        +--com.sjm.machlearn.classifiers.Classifier

Direct Known Subclasses:

BPNeuralNet, C5_0, KNN, LVQ, NaiveBayes, SvmLite

public abstract class Classifier

extends MainClass

Classifier.java : defines the functions needed for a classifier and operations that are contigent on those functions.

Version:

0.1a April 2003

Author:

Sean McIlwain

See Also:

homepage

Field Summary

static int	Fold10Validation CONSTANTS
static int	JackKnifeValidation
static int	Random10Validation

Fields inherited from class com.sjm.machlearn.util.MainClass

debug, debug_level, debug_listeners

Constructor Summary

Classifier()

Method Summary

abstract Feature	classify(Example example)
abstract Classifier	cloneClassifier()
void	doTune(DataSet trainingData, java.lang.Object[] parameters, int tuneMethod)
void	generateROCCurve(DataSet trainset, DataSet testset, java.lang.String filename) generateROCCurve() creates an ROC curve for the trainset and testset.
double	getAccuracy(DataSet testData) getAccuracy() : get the accuracy on the dataset
Example[][]	getBothCat(DataSet testdata) getBothCat() get the correctly predicted examples and the misclassified predicted examples for the returned matrix : [0][n] is the misclassified examples.
double	getExampleWeight(Example e) getExampleWeight() : Returns the weight of an example for the class 0.
Example[]	getMisCat(DataSet testdata) getMisCat() : gets the misclassified examples
abstract java.lang.String	printClassifier()
abstract void	setParameter(int i, java.lang.Object par)
abstract void	train(DataSet trainingData)

Methods inherited from class com.sjm.machlearn.util.MainClass

_internalError, _internalError, addDebugListener, debugMesg, debugMesg, debugMesg, debugMesg, debugMesg, debugMesg, debugOff, debugOn, internalError, internalError, setDebug, setDebugLevel

Methods inherited from class java.lang.Object

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

Field Detail

Fold10Validation

public static final int Fold10Validation

CONSTANTS

Random10Validation

public static final int Random10Validation

JackKnifeValidation

public static final int JackKnifeValidation

Constructor Detail

Classifier

public Classifier()

Method Detail

train

public abstract void train(DataSet trainingData)
                    throws java.lang.Exception

classify

public abstract Feature classify(Example example)

printClassifier

public abstract java.lang.String printClassifier()

cloneClassifier

public abstract Classifier cloneClassifier()

setParameter

public abstract void setParameter(int i,
                                  java.lang.Object par)

doTune

public void doTune(DataSet trainingData,
                   java.lang.Object[] parameters,
                   int tuneMethod)
            throws java.lang.Exception

getAccuracy

public double getAccuracy(DataSet testData)

getAccuracy() : get the accuracy on the dataset

getBothCat

public Example[][] getBothCat(DataSet testdata)

getBothCat() get the correctly predicted examples and the misclassified predicted examples for the returned matrix : [0][n] is the misclassified examples. [1][n] is the correctly classified examples.

Parameters:

testdata - dataset to be tested by the classifier.

Returns:

[0][n] - miscat [1][n] - corcat.

getMisCat

public Example[] getMisCat(DataSet testdata)

getMisCat() : gets the misclassified examples

Parameters:

testdata - dataset to be tested

Returns:

examples that are misclassified.

getExampleWeight

public double getExampleWeight(Example e)

getExampleWeight() : Returns the weight of an example for the class 0. Used for ROC curves. Will be overidden.

generateROCCurve

public void generateROCCurve(DataSet trainset,
                             DataSet testset,
                             java.lang.String filename)
                      throws java.lang.Exception

generateROCCurve() creates an ROC curve for the trainset and testset.