smile.feature

Class FeatureTransform

java.lang.Object

smile.feature.FeatureTransform

Direct Known Subclasses:

MaxAbsScaler, Normalizer, Scaler, Standardizer

public abstract class FeatureTransform
extends java.lang.Object

Feature transformation. In general, learning algorithms benefit from standardization of the data set. If some outliers are present in the set, robust transformers are more appropriate.

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean	copy If false, try to avoid a copy and do inplace transformation instead.

Constructor Summary

Constructors

Constructor and Description
FeatureTransform() Constructor.
FeatureTransform(boolean copy) Constructor.

Method Summary

Modifier and Type	Method and Description
abstract void	learn(smile.data.Attribute[] attributes, double[][] data) Learns transformation parameters from a dataset.
void	learn(double[][] data) Learns transformation parameters from a dataset.
abstract double[]	transform(double[] x) Transform a feature vector.
double[][]	transform(double[][] x) Transform an array of feature vectors.

Methods inherited from class java.lang.Object

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

Field Detail

copy

protected boolean copy

If false, try to avoid a copy and do inplace transformation instead.

Constructor Detail

FeatureTransform

public FeatureTransform()

Constructor. Inplace transformation.

FeatureTransform

public FeatureTransform(boolean copy)

Constructor.

Parameters:

copy - If false, try to avoid a copy and do inplace scaling instead.

Method Detail

learn

public void learn(double[][] data)

Learns transformation parameters from a dataset. All features are assumed numeric.

Parameters:

data - The training data.

learn

public abstract void learn(smile.data.Attribute[] attributes,
                           double[][] data)

Learns transformation parameters from a dataset.

Parameters:

attributes - The variable attributes. Of which, numeric variables will be standardized.

data - The training data to learn scaling parameters. The data will not be modified.

transform

public abstract double[] transform(double[] x)

Transform a feature vector.

Parameters:

x - a feature vector.

Returns:

the transformed feature value.

transform

public double[][] transform(double[][] x)

Transform an array of feature vectors.

Parameters:

x - an array of feature vectors. The feature vectors may be modified on output if copy is false.

Returns:

the transformed feature vectors.