Class Projection

java.lang.Object

smile.feature.extraction.Projection

All Implemented Interfaces:

Serializable, Function<Tuple,Tuple>, Transform

Direct Known Subclasses:

GHA, KernelPCA, PCA, ProbabilisticPCA, RandomProjection

public class Projection
extends Object
implements Transform

A projection is a kind of feature extraction technique that transforms data from the input space to a feature space, linearly or non-linearly. Often, projections are used to reduce dimensionality, for example PCA and random projection. However, kernel-based methods, e.g. Kernel PCA, can actually map the data into a much higher dimensional space.

See Also:

• Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
final String[]	columns	The fields of input space.
final DenseMatrix	projection	The projection matrix.
final StructType	schema	The schema of output space.

Constructor Summary

Constructors

Constructor	Description
Projection(DenseMatrix projection, String prefix, String... columns)	Constructor.

Method Summary

Modifier and Type	Method	Description
double[]	apply(double[] x)	Project a data point to the feature space.
double[][]	apply(double[][] x)	Project a set of data to the feature space.
DataFrame	apply(DataFrame data)	Applies this transform to the given argument.
Tuple	apply(Tuple x)
protected double[]	postprocess(double[] x)	Postprocess the output vector after projection.
protected double[]	preprocess(double[] x)	Preprocess the input vector before projection.

Methods inherited from class Object

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

Methods inherited from interface Function

andThen, compose

Methods inherited from interface Transform

andThen, compose

Field Details

projection

public final DenseMatrix projection

The projection matrix. The dimension reduced data can be obtained by y = W * x.

schema

public final StructType schema

The schema of output space.

columns

public final String[] columns

The fields of input space.

Constructor Details

Projection

public Projection(DenseMatrix projection,
 String prefix,
 String... columns)

Constructor.

Parameters:

projection - the projection matrix.

prefix - the output field name prefix.

columns - the input fields.

Method Details

apply

public Tuple apply(Tuple x)

Specified by:

apply in interface Function<Tuple,Tuple>

apply

public DataFrame apply(DataFrame data)

Description copied from interface: Transform

Applies this transform to the given argument.

Specified by:

apply in interface Transform

Parameters:

data - the input data frame.

Returns:

the transformed data frame.

apply

public double[] apply(double[] x)

Project a data point to the feature space.

Parameters:

x - the data point.

Returns:

the projection in the feature space.

apply

public double[][] apply(double[][] x)

Project a set of data to the feature space.

Parameters:

x - the data set.

Returns:

the projection in the feature space.

preprocess

protected double[] preprocess(double[] x)

Preprocess the input vector before projection.

Parameters:

x - the input vector of projection.

Returns:

the preprocessed vector.

postprocess

protected double[] postprocess(double[] x)

Postprocess the output vector after projection.

Parameters:

x - the output vector of projection.

Returns:

the postprocessed vector.