de.lmu.ifi.dbs.elki.index.preprocessed.subspaceproj
Class AbstractSubspaceProjectionIndex<NV extends NumberVector<?,?>,D extends Distance<D>,P extends ProjectionResult>

java.lang.Object
  de.lmu.ifi.dbs.elki.index.AbstractIndex<O>
      de.lmu.ifi.dbs.elki.index.preprocessed.AbstractPreprocessorIndex<NV,P>
          de.lmu.ifi.dbs.elki.index.preprocessed.subspaceproj.AbstractSubspaceProjectionIndex<NV,D,P>

Type Parameters:

NV - Vector type

All Implemented Interfaces:

Index, LocalProjectionIndex<NV,P>, SubspaceProjectionIndex<NV,P>, Result

Direct Known Subclasses:

FourCSubspaceIndex, PreDeConSubspaceIndex

@Title(value="Local PCA Preprocessor")
@Description(value="Materializes the local PCA and the locally weighted matrix of objects of a database.")
public abstract class AbstractSubspaceProjectionIndex<NV extends NumberVector<?,?>,D extends Distance<D>,P extends ProjectionResult>
extends AbstractPreprocessorIndex<NV,P>
implements SubspaceProjectionIndex<NV,P>

Abstract base class for a local PCA based index.

Nested Class Summary

static class

AbstractSubspaceProjectionIndex.Factory<NV extends NumberVector<?,?>,D extends Distance<D>,I extends AbstractSubspaceProjectionIndex<NV,D,?>> Factory class

Field Summary

protected D	epsilon Contains the value of parameter epsilon;
protected int	minpts Holds the value of parameter minpts.
protected DistanceFunction<NV,D>	rangeQueryDistanceFunction The distance function for the variance analysis.

Fields inherited from class de.lmu.ifi.dbs.elki.index.preprocessed.AbstractPreprocessorIndex

storage

Fields inherited from class de.lmu.ifi.dbs.elki.index.AbstractIndex

relation

Constructor Summary

AbstractSubspaceProjectionIndex(Relation<NV> relation, D epsilon, DistanceFunction<NV,D> rangeQueryDistanceFunction, int minpts)
Constructor.

Method Summary

protected abstract P	computeProjection(DBID id, List<DistanceResultPair<D>> neighbors, Relation<NV> relation) This method implements the type of variance analysis to be computed for a given point.
P	getLocalProjection(DBID objid) Get the precomputed local subspace for a particular object ID.
protected void	preprocess() Preprocessing step.

Methods inherited from class de.lmu.ifi.dbs.elki.index.preprocessed.AbstractPreprocessorIndex

getLogger

Methods inherited from class de.lmu.ifi.dbs.elki.index.AbstractIndex

delete, deleteAll, getLongName, getPageFileStatistics, getShortName, insert, insertAll

Methods inherited from class java.lang.Object

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

Methods inherited from interface de.lmu.ifi.dbs.elki.index.Index

delete, deleteAll, getPageFileStatistics, insert, insertAll

Methods inherited from interface de.lmu.ifi.dbs.elki.result.Result

getLongName, getShortName

Field Detail

epsilon

protected D extends Distance<D> epsilon

Contains the value of parameter epsilon;

rangeQueryDistanceFunction

protected DistanceFunction<NV extends NumberVector<?,?>,D extends Distance<D>> rangeQueryDistanceFunction

The distance function for the variance analysis.

minpts

protected int minpts

Holds the value of parameter minpts.

Constructor Detail

AbstractSubspaceProjectionIndex

public AbstractSubspaceProjectionIndex(Relation<NV> relation,
                                       D epsilon,
                                       DistanceFunction<NV,D> rangeQueryDistanceFunction,
                                       int minpts)

Constructor.

Parameters:

relation - Relation

epsilon - Maximum Epsilon

rangeQueryDistanceFunction - range query

minpts - Minpts

Method Detail

preprocess

protected void preprocess()

Preprocessing step.

getLocalProjection

public P getLocalProjection(DBID objid)

Description copied from interface: SubspaceProjectionIndex

Get the precomputed local subspace for a particular object ID.

Specified by:

getLocalProjection in interface LocalProjectionIndex<NV extends NumberVector<?,?>,P extends ProjectionResult>

Specified by:

getLocalProjection in interface SubspaceProjectionIndex<NV extends NumberVector<?,?>,P extends ProjectionResult>

Parameters:

objid - Object ID

Returns:

Matrix

computeProjection

protected abstract P computeProjection(DBID id,
                                       List<DistanceResultPair<D>> neighbors,
                                       Relation<NV> relation)

This method implements the type of variance analysis to be computed for a given point.

Example1: for 4C, this method should implement a PCA for the given point. Example2: for PreDeCon, this method should implement a simple axis-parallel variance analysis.

Parameters:

id - the given point

neighbors - the neighbors as query results of the given point

relation - the database for which the preprocessing is performed

Returns:

local subspace projection