org.tribuo

Class Prediction<T extends Output<T>>

java.lang.Object

org.tribuo.Prediction<T>

All Implemented Interfaces:

Serializable

public class Prediction<T extends Output<T>>
extends Object
implements Serializable

A prediction made by a Model. Contains the output, and optionally and a map of scores over the possible outputs. If hasProbabilities() == true then it has a probability distribution over outputs otherwise it is unnormalized scores over outputs.

If possible it also contains the number of features that were used to make a prediction, and how many features originally existed in the Example.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
Prediction(Prediction<T> other, int numUsed, Example<T> example) Constructs a prediction from the supplied arguments.
Prediction(T output, int numUsed, Example<T> example) Constructs a prediction from the supplied arguments.
Prediction(T output, Map<String,T> outputScores, int numUsed, Example<T> example, boolean probability) Constructs a prediction from the supplied arguments.

Method Summary

Modifier and Type	Method and Description
boolean	distributionEquals(Prediction<T> other) Checks that the other prediction has the same distribution as this prediction, using the Output.fullEquals(T) method.
Example<T>	getExample() Returns the example itself.
int	getExampleSize() Returns the number of features in the example.
int	getNumActiveFeatures() Returns the number of features used in the prediction.
T	getOutput() Returns the predicted output.
Map<String,T>	getOutputScores() Gets the output scores for each output.
boolean	hasProbabilities() Are the scores probabilities?
String	toString()

Methods inherited from class java.lang.Object

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

Constructor Detail

Prediction

public Prediction(T output,
                  Map<String,T> outputScores,
                  int numUsed,
                  Example<T> example,
                  boolean probability)

Constructs a prediction from the supplied arguments.

Parameters:

output - The predicted output (i.e., the one with the maximum score).

outputScores - The output score distribution.

numUsed - The number of features used to make the prediction.

example - The input example.

probability - Are the scores probabilities?

Prediction

public Prediction(T output,
                  int numUsed,
                  Example<T> example)

Constructs a prediction from the supplied arguments.

Parameters:

output - The predicted output.

numUsed - The number of features used to make the prediction.

example - The input example.

Prediction

public Prediction(Prediction<T> other,
                  int numUsed,
                  Example<T> example)

Constructs a prediction from the supplied arguments.

Parameters:

other - The prediction to copy.

numUsed - The number of features used to make the prediction.

example - The input example.

Method Detail

getOutput

public T getOutput()

Returns the predicted output.

Returns:

The predicted output.

getNumActiveFeatures

public int getNumActiveFeatures()

Returns the number of features used in the prediction.

Returns:

The number of features used.

getExampleSize

public int getExampleSize()

Returns the number of features in the example.

Returns:

The number of features in the example.

getExample

public Example<T> getExample()

Returns the example itself.

Returns:

The example.

getOutputScores

public Map<String,T> getOutputScores()

Gets the output scores for each output.

May be an empty map if it did not generate scores.

Returns:

A Map.

hasProbabilities

public boolean hasProbabilities()

Are the scores probabilities?

Returns:

True if the scores are probabilities.

toString

public String toString()

Overrides:

toString in class Object

distributionEquals

public boolean distributionEquals(Prediction<T> other)

Checks that the other prediction has the same distribution as this prediction, using the Output.fullEquals(T) method.

Parameters:

other - The prediction to compare.

Returns:

True if they have the same distributions.