org.apache.ignite.ml.composition

Class BaggingModelTrainer

java.lang.Object

org.apache.ignite.ml.trainers.DatasetTrainer<ModelsComposition,Double>

org.apache.ignite.ml.composition.BaggingModelTrainer

public abstract class BaggingModelTrainer
extends DatasetTrainer<ModelsComposition,Double>

Abstract trainer implementing bagging logic. In each learning iteration the algorithm trains one model on subset of learning sample and subspace of features space. Each model is produced from same model-class [e.g. Decision Trees].

Nested Class Summary

Nested classes/interfaces inherited from class org.apache.ignite.ml.trainers.DatasetTrainer

DatasetTrainer.EmptyDatasetException

Field Summary

Fields inherited from class org.apache.ignite.ml.trainers.DatasetTrainer

environment

Constructor Summary

Constructors

Constructor and Description
BaggingModelTrainer(PredictionsAggregator predictionsAggregator, int featureVectorSize, int maximumFeaturesCntPerMdl, int ensembleSize, double samplePartSizePerMdl) Constructs new instance of BaggingModelTrainer.

Method Summary

Modifier and Type	Method and Description
protected abstract DatasetTrainer<? extends Model<Vector,Double>,Double>	buildDatasetTrainerForModel() Creates trainer specific to ensemble.
<K,V> ModelsComposition	fit(DatasetBuilder<K,V> datasetBuilder, IgniteBiFunction<K,V,Vector> featureExtractor, IgniteBiFunction<K,V,Double> lbExtractor) Trains model based on the specified data.
<K,V> ModelsComposition	updateModel(ModelsComposition mdl, DatasetBuilder<K,V> datasetBuilder, IgniteBiFunction<K,V,Vector> featureExtractor, IgniteBiFunction<K,V,Double> lbExtractor) Learn new models on dataset and create new Compositions over them and already learned models.

Methods inherited from class org.apache.ignite.ml.trainers.DatasetTrainer

checkState, fit, fit, fit, fit, getLastTrainedModelOrThrowEmptyDatasetException, setEnvironment, update, update, update, update, update

Methods inherited from class java.lang.Object

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

Constructor Detail

BaggingModelTrainer

public BaggingModelTrainer(PredictionsAggregator predictionsAggregator,
                           int featureVectorSize,
                           int maximumFeaturesCntPerMdl,
                           int ensembleSize,
                           double samplePartSizePerMdl)

Constructs new instance of BaggingModelTrainer.

Parameters:

predictionsAggregator - Predictions aggregator.

featureVectorSize - Feature vector size.

maximumFeaturesCntPerMdl - Number of features to draw from original features vector to train each model.

ensembleSize - Ensemble size.

samplePartSizePerMdl - Size of sample part in percent to train one model.

Method Detail

fit

public <K,V> ModelsComposition fit(DatasetBuilder<K,V> datasetBuilder,
                                   IgniteBiFunction<K,V,Vector> featureExtractor,
                                   IgniteBiFunction<K,V,Double> lbExtractor)

Trains model based on the specified data.

Specified by:

fit in class DatasetTrainer<ModelsComposition,Double>

Type Parameters:

K - Type of a key in upstream data.

V - Type of a value in upstream data.

Parameters:

datasetBuilder - Dataset builder.

featureExtractor - Feature extractor.

lbExtractor - Label extractor.

Returns:

Model.

buildDatasetTrainerForModel

protected abstract DatasetTrainer<? extends Model<Vector,Double>,Double> buildDatasetTrainerForModel()

Creates trainer specific to ensemble.

updateModel

public <K,V> ModelsComposition updateModel(ModelsComposition mdl,
                                           DatasetBuilder<K,V> datasetBuilder,
                                           IgniteBiFunction<K,V,Vector> featureExtractor,
                                           IgniteBiFunction<K,V,Double> lbExtractor)

Learn new models on dataset and create new Compositions over them and already learned models.

Specified by:

updateModel in class DatasetTrainer<ModelsComposition,Double>

Type Parameters:

K - Type of a key in upstream data.

V - Type of a value in upstream data.

Parameters:

mdl - Learned model.

datasetBuilder - Dataset builder.

featureExtractor - Feature extractor.

lbExtractor - Label extractor.

Returns:

New models composition.