Developer API

org.apache.spark.ml

Class Transformer

Object

org.apache.spark.ml.PipelineStage

org.apache.spark.ml.Transformer

All Implemented Interfaces:

java.io.Serializable, Params, Identifiable

Direct Known Subclasses:

Binarizer, ColumnPruner, HashingTF, IndexToString, Interaction, Model, OneHotEncoder, SQLTransformer, StopWordsRemover, UnaryTransformer, VectorAssembler, VectorAttributeRewriter, VectorSlicer

public abstract class Transformer
extends PipelineStage

Abstract class for transformers that transform one dataset into another.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
Transformer()

Method Summary

Methods

Modifier and Type	Method and Description
abstract Transformer	copy(ParamMap extra) Creates a copy of this instance with the same UID and some extra params.
abstract Dataset<Row>	transform(Dataset<?> dataset) Transforms the input dataset.
Dataset<Row>	transform(Dataset<?> dataset, ParamMap paramMap) Transforms the dataset with provided parameter map as additional parameters.
Dataset<Row>	transform(Dataset<?> dataset, ParamPair<?> firstParamPair, ParamPair<?>... otherParamPairs) Transforms the dataset with optional parameters
Dataset<Row>	transform(Dataset<?> dataset, ParamPair<?> firstParamPair, scala.collection.Seq<ParamPair<?>> otherParamPairs) Transforms the dataset with optional parameters

Methods inherited from class org.apache.spark.ml.PipelineStage

transformSchema

Methods inherited from class Object

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

Methods inherited from interface org.apache.spark.ml.param.Params

clear, copyValues, defaultCopy, defaultParamMap, explainParam, explainParams, extractParamMap, extractParamMap, get, getDefault, getOrDefault, getParam, hasDefault, hasParam, isDefined, isSet, paramMap, params, set, set, set, setDefault, setDefault, shouldOwn, validateParams

Methods inherited from interface org.apache.spark.ml.util.Identifiable

toString, uid

Constructor Detail

Transformer

public Transformer()

Method Detail

transform

public Dataset<Row> transform(Dataset<?> dataset,
                     ParamPair<?> firstParamPair,
                     ParamPair<?>... otherParamPairs)

Transforms the dataset with optional parameters

Parameters:

dataset - input dataset

firstParamPair - the first param pair, overwrite embedded params

otherParamPairs - other param pairs, overwrite embedded params

Returns:

transformed dataset

transform

public Dataset<Row> transform(Dataset<?> dataset,
                     ParamPair<?> firstParamPair,
                     scala.collection.Seq<ParamPair<?>> otherParamPairs)

Transforms the dataset with optional parameters

Parameters:

dataset - input dataset

firstParamPair - the first param pair, overwrite embedded params

otherParamPairs - other param pairs, overwrite embedded params

Returns:

transformed dataset

transform

public Dataset<Row> transform(Dataset<?> dataset,
                     ParamMap paramMap)

Transforms the dataset with provided parameter map as additional parameters.

Parameters:

dataset - input dataset

paramMap - additional parameters, overwrite embedded params

Returns:

transformed dataset

transform

public abstract Dataset<Row> transform(Dataset<?> dataset)

Transforms the input dataset.

Parameters:

dataset - (undocumented)

Returns:

(undocumented)

copy

public abstract Transformer copy(ParamMap extra)

Description copied from interface: Params

Creates a copy of this instance with the same UID and some extra params. Subclasses should implement this method and set the return type properly. See defaultCopy().

Specified by:

copy in interface Params

Specified by:

copy in class PipelineStage

Parameters:

extra - (undocumented)

Returns:

(undocumented)