org.apache.ignite.ml.selection.scoring.metric

Class AbstractMetrics<M extends MetricValues>

java.lang.Object

org.apache.ignite.ml.selection.scoring.metric.AbstractMetrics<M>

All Implemented Interfaces:

Metric<Double>

Direct Known Subclasses:

BinaryClassificationMetrics, RegressionMetrics

public abstract class AbstractMetrics<M extends MetricValues>
extends Object
implements Metric<Double>

Abstract metrics calculator. It could be used in two ways: to caculate all regression metrics or custom regression metric.

Field Summary

Fields

Modifier and Type	Field and Description
protected Function<M,Double>	metric The main metric to get individual score.

Constructor Summary

Constructors

Constructor and Description
AbstractMetrics()

Method Summary

Modifier and Type	Method and Description
double	score(Iterator<LabelPair<Double>> iter) Calculates score.
abstract M	scoreAll(Iterator<LabelPair<Double>> iter) Calculates metrics values.
AbstractMetrics	withMetric(Function<M,Double> metric)

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.apache.ignite.ml.selection.scoring.metric.Metric

name

Field Detail

metric

protected Function<M extends MetricValues,Double> metric

The main metric to get individual score.

Constructor Detail

AbstractMetrics

public AbstractMetrics()

Method Detail

scoreAll

public abstract M scoreAll(Iterator<LabelPair<Double>> iter)

Calculates metrics values.

Parameters:

iter - Iterator that supplies pairs of truth values and predicated.

Returns:

Scores for all regression metrics.

withMetric

public AbstractMetrics withMetric(Function<M,Double> metric)

score

public double score(Iterator<LabelPair<Double>> iter)

Calculates score.

Specified by:

score in interface Metric<Double>

Parameters:

iter - Iterator that supplies pairs of truth values and predicated.

Returns:

Score.