Package de.jungblut.classification.eval

Class TestSetIterationCallback<T extends Classifier>

java.lang.Object

de.jungblut.classification.eval.TestSetIterationCallback<T>

All Implemented Interfaces:

IterationCompletionListener

public class TestSetIterationCallback<T extends Classifier>
extends java.lang.Object
implements IterationCompletionListener

This callback is used to evaluate the performance on a held-out test set. It saves the best performing parameter inside to be retrieved later.

Author:

thomas.jungblut

Constructor Summary

Constructors

Constructor	Description
TestSetIterationCallback(EvaluationSplit split, WeightMapper<T> mapper, java.util.Comparator<Evaluator.EvaluationResult> resultComparison)	Creates a new test set iteration callback instance.
TestSetIterationCallback(EvaluationSplit split, WeightMapper<T> mapper, java.util.Comparator<Evaluator.EvaluationResult> resultComparison, int evaluationInterval)	Creates a new test set iteration callback instance.

Method Summary

Modifier and Type	Method	Description
Evaluator.EvaluationResult	getBestResult()
de.jungblut.math.DoubleVector	getBestWeights()
void	onIterationFinished(int iteration, double cost, de.jungblut.math.DoubleVector currentWeights)	This callback is called from a AbstractMinimizer when a iteration of a minimization objective is finished.

Methods inherited from class java.lang.Object

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

Constructor Detail

TestSetIterationCallback

public TestSetIterationCallback(EvaluationSplit split,
                                WeightMapper<T> mapper,
                                java.util.Comparator<Evaluator.EvaluationResult> resultComparison,
                                int evaluationInterval)

Creates a new test set iteration callback instance.

Parameters:

split - the split to test on

mapper - the weight mapper to transform a weight to a classifier.

resultComparison - the comparator to compare what result is better and should be kept.

evaluationInterval - how often the result should be evaluated.

TestSetIterationCallback

public TestSetIterationCallback(EvaluationSplit split,
                                WeightMapper<T> mapper,
                                java.util.Comparator<Evaluator.EvaluationResult> resultComparison)

Creates a new test set iteration callback instance. The callback executes every 10 iterations.

Parameters:

split - the split to test on

mapper - the weight mapper to transform a weight to a classifier.

resultComparison - the comparator to compare what result is better and should be kept.

Method Detail

onIterationFinished

public void onIterationFinished(int iteration,
                                double cost,
                                de.jungblut.math.DoubleVector currentWeights)

Description copied from interface: IterationCompletionListener

This callback is called from a AbstractMinimizer when a iteration of a minimization objective is finished.

Specified by:

onIterationFinished in interface IterationCompletionListener

Parameters:

iteration - the number of the current iteration.

cost - the cost at the current iteration.

currentWeights - the current optimal weights.

getBestResult

public Evaluator.EvaluationResult getBestResult()

getBestWeights

public de.jungblut.math.DoubleVector getBestWeights()