Package se.hb.jcp.bindings.jlibsvm

Class SVMClassifier

java.lang.Object

se.hb.jcp.ml.ClassifierBase

se.hb.jcp.bindings.jlibsvm.SVMClassifier

All Implemented Interfaces:

java.io.Serializable, IClassifier, IClassifierInformation, IClassProbabilityClassifier, ISVMClassifier

public class SVMClassifier
extends ClassifierBase
implements ISVMClassifier, IClassProbabilityClassifier, java.io.Serializable

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
protected libsvm.svm_model	_model
protected libsvm.svm_parameter	_parameters

Constructor Summary

Constructors

Constructor	Description
SVMClassifier()
SVMClassifier(libsvm.svm_parameter parameters)
SVMClassifier(org.json.JSONObject parameters)

Method Summary

Modifier and Type	Method	Description
double	distanceFromSeparatingPlane(cern.colt.matrix.DoubleMatrix1D instance)	Returns the signed distance between the separating hyperplane and the instance.
IClassifier	fitNew(cern.colt.matrix.DoubleMatrix2D x, double[] y)	Trains and returns a copy of this classifier using the supplied data.
protected void	internalFit(cern.colt.matrix.DoubleMatrix2D x, double[] y)
cern.colt.matrix.DoubleMatrix1D	nativeStorageTemplate()	Returns a value of the DoubleMatrix1D derived class that is the native storage format for the classifier.
double	predict(cern.colt.matrix.DoubleMatrix1D instance)	Predicts the target for the supplied instance.
double	predict(cern.colt.matrix.DoubleMatrix1D instance, double[] probabilityEstimates)	Predicts the target probabilities for the supplied instance.

Methods inherited from class se.hb.jcp.ml.ClassifierBase

fit, getAttributeCount, getLabels, isTrained

Methods inherited from class java.lang.Object

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

Methods inherited from interface se.hb.jcp.ml.IClassifier

fit

Methods inherited from interface se.hb.jcp.ml.IClassifierInformation

getAttributeCount, getLabels, isTrained

Field Detail

_parameters

protected libsvm.svm_parameter _parameters

_model

protected libsvm.svm_model _model

Constructor Detail

SVMClassifier

public SVMClassifier()

SVMClassifier

public SVMClassifier(org.json.JSONObject parameters)

SVMClassifier

public SVMClassifier(libsvm.svm_parameter parameters)

Method Detail

internalFit

protected void internalFit(cern.colt.matrix.DoubleMatrix2D x,
                           double[] y)

Specified by:

internalFit in class ClassifierBase

fitNew

public IClassifier fitNew(cern.colt.matrix.DoubleMatrix2D x,
                          double[] y)

Description copied from interface: IClassifier

Trains and returns a copy of this classifier using the supplied data.

Specified by:

fitNew in interface IClassifier

Parameters:

x - the attributes of the instances.

y - the targets of the instances.

Returns:

a new IClassifier instance trained with the supplied data and using the same algorithm and parameter settings as the parent instance.

predict

public double predict(cern.colt.matrix.DoubleMatrix1D instance)

Description copied from interface: IClassifier

Predicts the target for the supplied instance.

Specified by:

predict in interface IClassifier

Parameters:

instance - the instance

Returns:

the predicted target of the instance.

predict

public double predict(cern.colt.matrix.DoubleMatrix1D instance,
                      double[] probabilityEstimates)

Description copied from interface: IClassProbabilityClassifier

Predicts the target probabilities for the supplied instance.

Specified by:

predict in interface IClassProbabilityClassifier

Parameters:

instance - the instance

probabilityEstimates - a double[] array for storing the predicted probabilities for each of target values in the order assumed by JCP.

Returns:

the predicted target of the instance.

distanceFromSeparatingPlane

public double distanceFromSeparatingPlane(cern.colt.matrix.DoubleMatrix1D instance)

Returns the signed distance between the separating hyperplane and the instance.

Specified by:

distanceFromSeparatingPlane in interface ISVMClassifier

Parameters:

instance - the instance.

Returns:

the signed distance between the separating hyperplane and the instance.

nativeStorageTemplate

public cern.colt.matrix.DoubleMatrix1D nativeStorageTemplate()

Description copied from interface: IClassifierInformation

Returns a value of the DoubleMatrix1D derived class that is the native storage format for the classifier.

Specified by:

nativeStorageTemplate in interface IClassifierInformation

Returns:

a value of the DoubleMatrix1D derived class of the native storage format for the classifier.