org.apache.ignite.ml.selection.scoring.evaluator.aggregator

Class RegressionMetricStatsAggregator

java.lang.Object

org.apache.ignite.ml.selection.scoring.evaluator.aggregator.RegressionMetricStatsAggregator

All Implemented Interfaces:

Serializable, MetricStatsAggregator<Double,EmptyContext<Double>,RegressionMetricStatsAggregator>

public class RegressionMetricStatsAggregator
extends Object
implements MetricStatsAggregator<Double,EmptyContext<Double>,RegressionMetricStatsAggregator>

Class represents statistics aggregator for regression estimation.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
RegressionMetricStatsAggregator() Creates an instance of RegressionMetricStatsAggregator.
RegressionMetricStatsAggregator(long n, double absoluteError, double rss, double sumOfYs, double sumOfSquaredYs) Creates an instance of RegressionMetricStatsAggregator.

Method Summary

Modifier and Type	Method and Description
void	aggregate(IgniteModel<Vector,Double> model, LabeledVector<Double> vector) Aggregates statistics for metric computation given model and vector with answer.
EmptyContext	createInitializedContext() Returns initialized context.
double	getMAE() Returns mean absolute error.
double	getMSE() Returns mean squared error.
double	getRss() Returns RSS.
void	initByContext(EmptyContext context) Inits this aggtegator by evaluation context.
RegressionMetricStatsAggregator	mergeWith(RegressionMetricStatsAggregator other) Merges statistics of two aggregators to new aggreagator.
double	ysRss() Returns sum of squared errors.
double	ysVariance() Returns label variance.

Methods inherited from class java.lang.Object

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

Constructor Detail

RegressionMetricStatsAggregator

public RegressionMetricStatsAggregator()

Creates an instance of RegressionMetricStatsAggregator.

RegressionMetricStatsAggregator

public RegressionMetricStatsAggregator(long n,
                                       double absoluteError,
                                       double rss,
                                       double sumOfYs,
                                       double sumOfSquaredYs)

Creates an instance of RegressionMetricStatsAggregator.

Parameters:

n - Number of examples in dataset.

absoluteError - Absolute error.

rss - Rss.

sumOfYs - Sum of labels.

sumOfSquaredYs - Sum of squared labels.

Method Detail

aggregate

public void aggregate(IgniteModel<Vector,Double> model,
                      LabeledVector<Double> vector)

Aggregates statistics for metric computation given model and vector with answer.

Specified by:

aggregate in interface MetricStatsAggregator<Double,EmptyContext<Double>,RegressionMetricStatsAggregator>

Parameters:

model - Model.

vector - Vector.

mergeWith

public RegressionMetricStatsAggregator mergeWith(RegressionMetricStatsAggregator other)

Merges statistics of two aggregators to new aggreagator.

Specified by:

mergeWith in interface MetricStatsAggregator<Double,EmptyContext<Double>,RegressionMetricStatsAggregator>

Parameters:

other - Other aggregator.

Returns:

New aggregator.

createInitializedContext

public EmptyContext createInitializedContext()

Returns initialized context.

Specified by:

createInitializedContext in interface MetricStatsAggregator<Double,EmptyContext<Double>,RegressionMetricStatsAggregator>

Returns:

initialized evaluation context.

initByContext

public void initByContext(EmptyContext context)

Inits this aggtegator by evaluation context.

Specified by:

initByContext in interface MetricStatsAggregator<Double,EmptyContext<Double>,RegressionMetricStatsAggregator>

Parameters:

context - Evaluation context.

getMAE

public double getMAE()

Returns mean absolute error.

Returns:

Mean absolute error.

getMSE

public double getMSE()

Returns mean squared error.

Returns:

Mean squared error.

ysRss

public double ysRss()

Returns sum of squared errors.

Returns:

Sum of squared errors

ysVariance

public double ysVariance()

Returns label variance.

Returns:

Label variance.

getRss

public double getRss()

Returns RSS.

Returns:

RSS.