org.apache.ignite.ml.preprocessing.encoding

Class EncoderTrainer<K,V>

java.lang.Object

org.apache.ignite.ml.preprocessing.encoding.EncoderTrainer<K,V>

Type Parameters:

K - Type of a key in upstream data.

V - Type of a value in upstream data.

All Implemented Interfaces:

PreprocessingTrainer<K,V>

public class EncoderTrainer<K,V>
extends Object
implements PreprocessingTrainer<K,V>

Trainer of the String Encoder and One-Hot Encoder preprocessors.

Constructor Summary

Constructors

Constructor and Description
EncoderTrainer()

Method Summary

Modifier and Type	Method and Description
EncoderPreprocessor<K,V>	fit(LearningEnvironmentBuilder envBuilder, DatasetBuilder<K,V> datasetBuilder, Preprocessor<K,V> basePreprocessor) Fits preprocessor.
EncoderTrainer<K,V>	labeled(Integer targetLabelIndex) Sets the target label index.
EncoderTrainer<K,V>	minCategorySize(Long minCategorySize) Sets the min category size for category target encoding.
EncoderTrainer<K,V>	minSamplesLeaf(Integer minSamplesLeaf) Sets the minSamplesLeaf for target encoding.
EncoderTrainer<K,V>	smoothing(Double smoothing) Sets the smoothing for target encoding.
EncoderTrainer<K,V>	withEncodedFeature(int idx) Add the index of encoded feature.
EncoderTrainer<K,V>	withEncodedFeatures(Set<Integer> handledIndices) Sets the indices of features which should be encoded.
EncoderTrainer<K,V>	withEncoderIndexingStrategy(EncoderSortingStrategy encoderSortingStgy) Sets the encoder indexing strategy.
EncoderTrainer<K,V>	withEncoderType(EncoderType type) Sets the encoder preprocessor type.

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.preprocessing.PreprocessingTrainer

fit, fit, fit, fit, fit, learningEnvironment

Constructor Detail

EncoderTrainer

public EncoderTrainer()

Method Detail

fit

public EncoderPreprocessor<K,V> fit(LearningEnvironmentBuilder envBuilder,
                                    DatasetBuilder<K,V> datasetBuilder,
                                    Preprocessor<K,V> basePreprocessor)

Fits preprocessor.

Specified by:

fit in interface PreprocessingTrainer<K,V>

Parameters:

envBuilder - Learning environment builder.

datasetBuilder - Dataset builder.

basePreprocessor - Base preprocessor.

Returns:

Preprocessor.

withEncodedFeature

public EncoderTrainer<K,V> withEncodedFeature(int idx)

Add the index of encoded feature.

Parameters:

idx - The index of encoded feature.

Returns:

The changed trainer.

withEncoderIndexingStrategy

public EncoderTrainer<K,V> withEncoderIndexingStrategy(EncoderSortingStrategy encoderSortingStgy)

Sets the encoder indexing strategy.

Parameters:

encoderSortingStgy - The encoder indexing strategy.

Returns:

The changed trainer.

withEncoderType

public EncoderTrainer<K,V> withEncoderType(EncoderType type)

Sets the encoder preprocessor type.

Parameters:

type - The encoder preprocessor type.

Returns:

The changed trainer.

withEncodedFeatures

public EncoderTrainer<K,V> withEncodedFeatures(Set<Integer> handledIndices)

Sets the indices of features which should be encoded.

Parameters:

handledIndices - Indices of features which should be encoded.

Returns:

The changed trainer.

labeled

public EncoderTrainer<K,V> labeled(Integer targetLabelIndex)

Sets the target label index.

Parameters:

targetLabelIndex - Index of target label.

Returns:

The changed trainer.

smoothing

public EncoderTrainer<K,V> smoothing(Double smoothing)

Sets the smoothing for target encoding.

Parameters:

smoothing - smoothing value.

Returns:

The changed trainer.

minSamplesLeaf

public EncoderTrainer<K,V> minSamplesLeaf(Integer minSamplesLeaf)

Sets the minSamplesLeaf for target encoding.

Parameters:

minSamplesLeaf - min samples leaf.

Returns:

The changed trainer.

minCategorySize

public EncoderTrainer<K,V> minCategorySize(Long minCategorySize)

Sets the min category size for category target encoding. Category less then minCategorySize will be encoded with avarage target value.

Parameters:

minCategorySize - min samples leaf.

Returns:

The changed trainer.