Package elki.datasource.filter.normalization.columnwise

Class AttributeWiseCDFNormalization<V extends NumberVector>

java.lang.Object

elki.datasource.filter.normalization.columnwise.AttributeWiseCDFNormalization<V>

Type Parameters:

V - vector type

All Implemented Interfaces:

Normalization<V>, ObjectFilter

Direct Known Subclasses:

AttributeWiseBetaNormalization

public class AttributeWiseCDFNormalization<V extends NumberVector>
extends java.lang.Object
implements Normalization<V>

Class to perform and undo a normalization on real vectors by estimating the distribution of values along each dimension independently, then rescaling objects to the cumulative density function (CDF) value at the original coordinate.

This process is for example also mentioned in section 3.4 of

Effects of Feature Normalization on Image Retrieval
S. Aksoy, R. M. Haralick

but they do not detail how to obtain an appropriate function `F`.

Since:

0.6.0

Author:

Erich Schubert

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	AttributeWiseCDFNormalization.Par<V extends NumberVector>	Parameterization class.

Field Summary

Fields

Modifier and Type	Field	Description
protected java.util.List<Distribution>	dists	Stores the estimated distributions
protected java.util.List<? extends DistributionEstimator<?>>	estimators	Stores the distribution estimators
protected NumberVector.Factory<V>	factory	Number vector factory.
private static Logging	LOG	Class logger.

Constructor Summary

Constructors

Constructor	Description
AttributeWiseCDFNormalization(java.util.List<? extends DistributionEstimator<?>> estimators)	Constructor.

Method Summary

Modifier and Type	Method	Description
protected boolean	constantZero(java.util.List<V> column, AttributeWiseCDFNormalization.Adapter adapter)	Test if an attribute is constant zero.
MultipleObjectsBundle	filter(MultipleObjectsBundle objects)	Filter a set of object packages.
protected Distribution	findBestFit(java.util.List<V> col, AttributeWiseCDFNormalization.Adapter adapter, int d, double[] test)	Find the best fitting distribution.
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Methods inherited from interface elki.datasource.filter.normalization.Normalization

restore, transform

Field Detail

LOG

private static final Logging LOG

Class logger.

estimators

protected java.util.List<? extends DistributionEstimator<?>> estimators

Stores the distribution estimators

dists

protected java.util.List<Distribution> dists

Stores the estimated distributions

factory

protected NumberVector.Factory<V extends NumberVector> factory

Number vector factory.

Constructor Detail

AttributeWiseCDFNormalization

public AttributeWiseCDFNormalization(java.util.List<? extends DistributionEstimator<?>> estimators)

Constructor.

Parameters:

estimators - Distribution estimators

Method Detail

filter

public MultipleObjectsBundle filter(MultipleObjectsBundle objects)

Description copied from interface: ObjectFilter

Filter a set of object packages.

Specified by:

filter in interface ObjectFilter

Parameters:

objects - Object to filter

Returns:

Filtered objects

findBestFit

protected Distribution findBestFit(java.util.List<V> col,
                                   AttributeWiseCDFNormalization.Adapter adapter,
                                   int d,
                                   double[] test)

Find the best fitting distribution.

Parameters:

col - Column of table

adapter - Adapter for accessing the data

d - Dimension

test - Scatch space for testing goodness of fit

Returns:

Best fit distribution

constantZero

protected boolean constantZero(java.util.List<V> column,
                               AttributeWiseCDFNormalization.Adapter adapter)

Test if an attribute is constant zero.

Parameters:

column - Column

adapter - Data accessor.

Returns:

true if all values are zero

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object