Package com.tibco.patterns.learn.training

Class VectorExample

java.lang.Object

com.tibco.patterns.learn.training.VectorExample

Direct Known Subclasses:

RecPairExample

public class VectorExample
extends java.lang.Object

Stores a feature vector and a label that are used for training the model. Also stores the current and previous predictions from the model.

Constructor Summary

Constructors

Constructor	Description
VectorExample(double[] featureValues, boolean label)	Creates example with given parameters and no prediction.

Method Summary

Modifier and Type	Method	Description
boolean	correctPrediction()	Checks if the RLink prediction is correct.
double[]	getFeatureValues()
boolean	getLabel()
int	getNFeatures()
RLinkOut	getPrediction()
RLinkOut	getPrevPrediction()	Used from TrainingObservers that need to compare previous and current iterations.
void	predict(int modelId)	Calculates prediction using default options.
void	predict(int modelId, PredictOptions predictOpts)	Calculates prediction using the given model and options, stores it in this example.
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Constructor Detail

VectorExample

public VectorExample(double[] featureValues,
                     boolean label)

Creates example with given parameters and no prediction.

Parameters:

featureValues - - the feature vector.

label - - actual label of this example.

Throws:

java.lang.IllegalArgumentException - if featureValues is null, empty, or contains an invalid value. All values must be between 0 and 1, or -1 for an empty feature.

Method Detail

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

getFeatureValues

public double[] getFeatureValues()

Returns:

a copy of the feature values array.

getNFeatures

public int getNFeatures()

Returns:

the number of features in the feature vector.

getLabel

public boolean getLabel()

Returns:

actual label of this example.

correctPrediction

public boolean correctPrediction()

Checks if the RLink prediction is correct.

Returns:

true if the model prediction for this example is correct.

Throws:

java.lang.IllegalStateException - if this example does not contain an RLink prediction.

predict

public void predict(int modelId,
                    PredictOptions predictOpts)

Calculates prediction using the given model and options, stores it in this example. Saves the previously calculated prediction, if any.

Parameters:

modelId - - ID of an existing RLink model.

predictOpts - - the options to use for this prediction. If null, uses default uptions.

predict

public void predict(int modelId)

Calculates prediction using default options. See predict(int, PredictOptions).

getPrediction

public RLinkOut getPrediction()

Returns:

internal reference to the model prediction for this example. Null if any predict() method was never called, or predictions have been cleared.

getPrevPrediction

public RLinkOut getPrevPrediction()

Used from TrainingObservers that need to compare previous and current iterations.

Returns:

internal reference to the previous model prediction. Null if any predict() method was called less than 2 times, or predictions have been cleared.