org.apache.ignite.ml.composition.boosting.convergence.median

Class MedianOfMedianConvergenceChecker<K,V>

java.lang.Object

org.apache.ignite.ml.composition.boosting.convergence.ConvergenceChecker<K,V>

org.apache.ignite.ml.composition.boosting.convergence.median.MedianOfMedianConvergenceChecker<K,V>

Type Parameters:

K - Type of a key in upstream data.

V - Type of a value in upstream data.

All Implemented Interfaces:

Serializable

public class MedianOfMedianConvergenceChecker<K,V>
extends ConvergenceChecker<K,V>

Use median of median on partitions value of errors for estimating error on dataset. This algorithm may be less sensitive to

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
MedianOfMedianConvergenceChecker(long sampleSize, IgniteFunction<Double,Double> lblMapping, Loss loss, DatasetBuilder<K,V> datasetBuilder, Preprocessor<K,V> preprocessor, double precision) Creates an instance of MedianOfMedianConvergenceChecker.

Method Summary

Modifier and Type	Method and Description
Double	computeMeanErrorOnDataset(Dataset<EmptyContext,? extends FeatureMatrixWithLabelsOnHeapData> dataset, ModelsComposition mdl) Compute error for given model on learning dataset.

Methods inherited from class org.apache.ignite.ml.composition.boosting.convergence.ConvergenceChecker

computeError, isConverged, isConverged

Methods inherited from class java.lang.Object

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

Constructor Detail

MedianOfMedianConvergenceChecker

public MedianOfMedianConvergenceChecker(long sampleSize,
                                        IgniteFunction<Double,Double> lblMapping,
                                        Loss loss,
                                        DatasetBuilder<K,V> datasetBuilder,
                                        Preprocessor<K,V> preprocessor,
                                        double precision)

Creates an instance of MedianOfMedianConvergenceChecker.

Parameters:

sampleSize - Sample size.

lblMapping - External label to internal mapping.

loss - Loss function.

datasetBuilder - Dataset builder.

preprocessor - Upstream preprocessor.

precision - Precision.

Method Detail

computeMeanErrorOnDataset

public Double computeMeanErrorOnDataset(Dataset<EmptyContext,? extends FeatureMatrixWithLabelsOnHeapData> dataset,
                                        ModelsComposition mdl)

Compute error for given model on learning dataset.

Specified by:

computeMeanErrorOnDataset in class ConvergenceChecker<K,V>

Parameters:

dataset - Learning dataset.

mdl - Model.

Returns:

Error mean value.