Package com.tibco.patterns.learn.training

Class VectorExperiment

java.lang.Object

com.tibco.patterns.learn.training.RLinkExperiment<VectorExample>

com.tibco.patterns.learn.training.VectorExperiment

public final class VectorExperiment
extends RLinkExperiment<VectorExample>

Manages a training dataset and a validation dataset. The datasets consist of VectorExample objects that store feature vectors.

Field Summary

Fields inherited from class com.tibco.patterns.learn.training.RLinkExperiment

trainSet, vldSet

Constructor Summary

Constructors

Constructor	Description
VectorExperiment(int nFeatures, double vldProportion)	Creates an experiment with two empty datasets and a new random number generator.
VectorExperiment(int nFeatures, double vldProportion, java.util.Random rnd)	Creates an experiment with two empty datasets.

Method Summary

Modifier and Type	Method	Description
boolean	add(VectorExample e)	Randomly adds the example to one of the datasets, preserving dataset proportions.

Methods inherited from class com.tibco.patterns.learn.training.RLinkExperiment

clearPredictions, getAllExamples, getNFeatures, getTrainSet, getVldSet

Methods inherited from class java.lang.Object

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

Constructor Detail

VectorExperiment

public VectorExperiment(int nFeatures,
                        double vldProportion,
                        java.util.Random rnd)

Creates an experiment with two empty datasets.

Parameters:

nFeatures - - number of model features in each example of this experiment

vldProportion - - the approximate percentage of examples in validation dataset

rnd - - the random number generator to use when choosing a dataset. Not null.

Throws:

java.lang.IllegalArgumentException - if vldProportion is not in interval [0.0, 1.0)

VectorExperiment

public VectorExperiment(int nFeatures,
                        double vldProportion)

Creates an experiment with two empty datasets and a new random number generator.

Parameters:

nFeatures - - number of model features in each example of this experiment

vldProportion - - the approximate percentage of examples in validation dataset

Throws:

java.lang.IllegalArgumentException - if vldProportion is not in interval [0.0, 1.0)

Method Detail

add

public boolean add(VectorExample e)

Randomly adds the example to one of the datasets, preserving dataset proportions. Not needed for RecPairExperiment, as record pairs are added to the XML structure instead.

Parameters:

e - - the example to be added.

Returns:

true if the example was added to the training dataset.

Throws:

java.lang.IllegalArgumentException - if number of features in the given example does not match the number of features for this experiment.