Package com.tibco.patterns.learn.rlink

Class RLinkOut

java.lang.Object

com.tibco.patterns.learn.rlink.RLinkOut

public final class RLinkOut
extends java.lang.Object

Stores the result of RLink model prediction. Immutable class.

Field Summary

Fields

Modifier and Type	Field	Description
static double	RLINK_THRESHOLD	The threshold between "true" and "false" labels used to train models

Constructor Summary

Constructors

Constructor	Description
RLinkOut(double score, double confidence)	Creates object that stores the given score and confidence.
RLinkOut(double score, double confidence, double[] significance)	Creates object that stores the given score, confidence and feature significance values.

Method Summary

Modifier and Type	Method	Description
boolean	correctPrediction(boolean actualLabel)
boolean	correctPrediction(boolean actualLabel, double threshold)	Do not use custom thresholds for model training!
double	getConfidence()	Gets the prediction confidence.
boolean	getLabel()
static boolean	getLabel(double rlinkScore)
double	getScore()
double[]	getSignificance()	Gets the significance of each model feature value.
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

RLINK_THRESHOLD

public static final double RLINK_THRESHOLD

The threshold between "true" and "false" labels used to train models

See Also:

Constant Field Values

Constructor Detail

RLinkOut

public RLinkOut(double score,
                double confidence)

Creates object that stores the given score and confidence.

Parameters:

score - - prediction score.

confidence - - prediction confidence.

RLinkOut

public RLinkOut(double score,
                double confidence,
                double[] significance)

Creates object that stores the given score, confidence and feature significance values.

Parameters:

score - - prediction score.

confidence - - prediction confidence.

significance - - significance of each feature. May be empty or null if significance values have not been calculated.

Method Detail

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

getScore

public double getScore()

Returns:

prediction score

getLabel

public static boolean getLabel(double rlinkScore)

Returns:

the predicted label (according to natural RLink threshold)

getLabel

public boolean getLabel()

Returns:

the predicted label (according to natural RLink threshold)

correctPrediction

public boolean correctPrediction(boolean actualLabel)

Returns:

if the predicted label matches the actualLabel

correctPrediction

public boolean correctPrediction(boolean actualLabel,
                                 double threshold)

Do not use custom thresholds for model training!

Returns:

if the predicted label (determined according to specified threshold) matches the actualLabel.

getConfidence

public double getConfidence()

Gets the prediction confidence. It indicates how well the model was trained for the given feature values.

Returns:

prediction confidence. The value is in the range of [0; 1]. If no confidence was requested, or if the requested confidence measure is not supported by the model, the value is RLink.NO_CONFIDENCE_MEASURE.

getSignificance

public double[] getSignificance()

Gets the significance of each model feature value. Each significance value (between 0.0 and 1.0) is the relative importance of the model feature value in determining the prediction score and label. Feature values with high significance were considered relevant in making the decision, querylets with low scores were not relevant. The sum of all significance values is 1.0.

Returns:

the significance of each feature value in making the prediction. The returned array is empty if significance values have not been calculated (the predicted label is false).