de.lmu.ifi.dbs.elki.datasource.filter.transform

Class AbstractSupervisedProjectionVectorFilter<V extends NumberVector>

java.lang.Object

de.lmu.ifi.dbs.elki.datasource.filter.transform.AbstractSupervisedProjectionVectorFilter<V>

Type Parameters:

V - Vector type

All Implemented Interfaces:

ObjectFilter

Direct Known Subclasses:

LinearDiscriminantAnalysisFilter

public abstract class AbstractSupervisedProjectionVectorFilter<V extends NumberVector>
extends Object
implements ObjectFilter

Base class for supervised projection methods. TODO: re-add sampling.

Since:

0.6.0

Author:

Angela Peng, Erich Schubert

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	AbstractSupervisedProjectionVectorFilter.Parameterizer<V extends NumberVector> Parameterization class.

Field Summary

Fields

Modifier and Type	Field and Description
protected int	tdim The dimensionality to which the data should be reduced.

Constructor Summary

Constructors

Constructor and Description
AbstractSupervisedProjectionVectorFilter(int projdimension) Constructor.

Method Summary

Methods

Modifier and Type	Method and Description
protected abstract Matrix	computeProjectionMatrix(List<V> vectorcolumn, List<? extends ClassLabel> classcolumn, int dim) computes the projection matrix
protected SimpleTypeInformation<?>	convertedType(SimpleTypeInformation<?> in, NumberVector.Factory<V> factory) Get the output type from the input type after conversion.
MultipleObjectsBundle	filter(MultipleObjectsBundle objects) Filter a set of object packages.
protected abstract Logging	getLogger() Class logger.
protected <O> Map<O,TIntList>	partition(List<? extends O> classcolumn) Partition the bundle based on the class label.

Methods inherited from class java.lang.Object

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

Field Detail

tdim

protected int tdim

The dimensionality to which the data should be reduced.

Constructor Detail

AbstractSupervisedProjectionVectorFilter

public AbstractSupervisedProjectionVectorFilter(int projdimension)

Constructor.

Parameters:

projdimension - Projection dimensionality

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

convertedType

protected SimpleTypeInformation<?> convertedType(SimpleTypeInformation<?> in,
                                     NumberVector.Factory<V> factory)

Get the output type from the input type after conversion.

Parameters:

in - input type restriction

factory - Vector factory

Returns:

output type restriction

getLogger

protected abstract Logging getLogger()

Class logger.

Returns:

Logger

computeProjectionMatrix

protected abstract Matrix computeProjectionMatrix(List<V> vectorcolumn,
                             List<? extends ClassLabel> classcolumn,
                             int dim)

computes the projection matrix

Parameters:

vectorcolumn - Vectors

classcolumn - Class information

dim - Dimensionality Dimensionality

Returns:

Projection matrix

partition

protected <O> Map<O,TIntList> partition(List<? extends O> classcolumn)

Partition the bundle based on the class label.

Parameters:

classcolumn -

Returns:

Partitioned data set.