Package elki.clustering.subspace

Class DiSH.Instance

java.lang.Object

elki.clustering.optics.GeneralizedOPTICS.Instance<DiSH.DiSHClusterOrder>

elki.clustering.subspace.DiSH.Instance

All Implemented Interfaces:

java.util.Comparator<DBIDRef>

Enclosing class:

DiSH

private class DiSH.Instance
extends GeneralizedOPTICS.Instance<DiSH.DiSHClusterOrder>

OPTICS variant used by DiSH internally.

Author:

Erich Schubert

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
private class	DiSH.Instance.Sorter	Sort new candidates by their distance, for determining the core size.

Field Summary

Fields

Modifier and Type	Field	Description
private ArrayModifiableDBIDs	clusterOrder	Cluster order.
private WritableDataStore<long[]>	commonPreferenceVectors	Shared preference vectors.
private WritableIntegerDataStore	correlationValue	Correlation value.
protected WritableDataStore<long[]>	preferenceVectors	The precomputed preference vectors.
private Relation<? extends NumberVector>	relation	Data relation.
(package private) java.util.Comparator<DBIDRef>	tmpcomp	Sort object by the temporary fields.
private WritableIntegerDataStore	tmpCorrelation	Temporary storage of correlation values.
private WritableDoubleDataStore	tmpDistance	Temporary storage of distances.
private ArrayModifiableDBIDs	tmpIds	Temporary ids.
private WritableDataStore<long[]>	tmpPreferenceVectors	Temporary storage for new preference vectors.

Fields inherited from class elki.clustering.optics.GeneralizedOPTICS.Instance

candidates, predecessor, processedIDs, reachability

Constructor Summary

Constructors

Constructor	Description
Instance(Relation<? extends NumberVector> relation)	Constructor.

Method Summary

Modifier and Type	Method	Description
protected DiSH.DiSHClusterOrder	buildResult()	Build the final result.
int	compare(DBIDRef o1, DBIDRef o2)
private long[]	determinePreferenceVector(ModifiableDBIDs[] neighborIDs, java.lang.StringBuilder msg)	Determines the preference vector according to the specified neighbor ids.
private long[]	determinePreferenceVectorByApriori(ModifiableDBIDs[] neighborIDs, java.lang.StringBuilder msg)	Determines the preference vector with the apriori strategy.
private long[]	determinePreferenceVectorByMaxIntersection(ModifiableDBIDs[] neighborIDs, java.lang.StringBuilder msg)	Determines the preference vector with the max intersection strategy.
protected void	expandDBID(DBIDRef id)	Add the current DBID to the cluster order, and expand its neighbors if minPts and similar conditions are satisfied.
protected Logging	getLogger()	Get the class logger.
protected void	initialDBID(DBIDRef id)	Initialize for a new DBID.
private int	max(java.util.Map<java.lang.Integer,ModifiableDBIDs> candidates)	Returns the set with the maximum size contained in the specified map.
private int	maxIntersection(java.util.Map<java.lang.Integer,ModifiableDBIDs> candidates, ModifiableDBIDs set)	Returns the index of the set having the maximum intersection set with the specified set contained in the specified map.
DiSH.DiSHClusterOrder	run()	Process the data set.

Methods inherited from class java.lang.Object

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

Methods inherited from interface java.util.Comparator

equals, reversed, thenComparing, thenComparing, thenComparing, thenComparingDouble, thenComparingInt, thenComparingLong

Field Detail

relation

private Relation<? extends NumberVector> relation

Data relation.

clusterOrder

private ArrayModifiableDBIDs clusterOrder

Cluster order.

correlationValue

private WritableIntegerDataStore correlationValue

Correlation value.

commonPreferenceVectors

private WritableDataStore<long[]> commonPreferenceVectors

Shared preference vectors.

tmpIds

private ArrayModifiableDBIDs tmpIds

Temporary ids.

tmpCorrelation

private WritableIntegerDataStore tmpCorrelation

Temporary storage of correlation values.

tmpDistance

private WritableDoubleDataStore tmpDistance

Temporary storage of distances.

tmpcomp

java.util.Comparator<DBIDRef> tmpcomp

Sort object by the temporary fields.

preferenceVectors

protected WritableDataStore<long[]> preferenceVectors

The precomputed preference vectors.

tmpPreferenceVectors

private WritableDataStore<long[]> tmpPreferenceVectors

Temporary storage for new preference vectors.

Constructor Detail

Instance

public Instance(Relation<? extends NumberVector> relation)

Constructor.

Parameters:

relation - Relation

Method Detail

run

public DiSH.DiSHClusterOrder run()

Description copied from class: GeneralizedOPTICS.Instance

Process the data set.

Overrides:

run in class GeneralizedOPTICS.Instance<DiSH.DiSHClusterOrder>

Returns:

Cluster order result.

determinePreferenceVector

private long[] determinePreferenceVector(ModifiableDBIDs[] neighborIDs,
                                         java.lang.StringBuilder msg)

Determines the preference vector according to the specified neighbor ids.

Parameters:

neighborIDs - the list of ids of the neighbors in each dimension

msg - a string buffer for debug messages

Returns:

the preference vector

determinePreferenceVectorByApriori

private long[] determinePreferenceVectorByApriori(ModifiableDBIDs[] neighborIDs,
                                                  java.lang.StringBuilder msg)

Determines the preference vector with the apriori strategy.

Parameters:

neighborIDs - the list of ids of the neighbors in each dimension

msg - a string buffer for debug messages

Returns:

the preference vector

determinePreferenceVectorByMaxIntersection

private long[] determinePreferenceVectorByMaxIntersection(ModifiableDBIDs[] neighborIDs,
                                                          java.lang.StringBuilder msg)

Determines the preference vector with the max intersection strategy.

Parameters:

neighborIDs - the list of ids of the neighbors in each dimension

msg - a string buffer for debug messages

Returns:

the preference vector

max

private int max(java.util.Map<java.lang.Integer,ModifiableDBIDs> candidates)

Returns the set with the maximum size contained in the specified map.

Parameters:

candidates - the map containing the sets

Returns:

the set with the maximum size

maxIntersection

private int maxIntersection(java.util.Map<java.lang.Integer,ModifiableDBIDs> candidates,
                            ModifiableDBIDs set)

Returns the index of the set having the maximum intersection set with the specified set contained in the specified map.

Parameters:

candidates - the map containing the sets

set - the set to intersect with and output the result to

Returns:

the set with the maximum size

buildResult

protected DiSH.DiSHClusterOrder buildResult()

Description copied from class: GeneralizedOPTICS.Instance

Build the final result.

Specified by:

buildResult in class GeneralizedOPTICS.Instance<DiSH.DiSHClusterOrder>

Returns:

Result

initialDBID

protected void initialDBID(DBIDRef id)

Description copied from class: GeneralizedOPTICS.Instance

Initialize for a new DBID.

Specified by:

initialDBID in class GeneralizedOPTICS.Instance<DiSH.DiSHClusterOrder>

Parameters:

id - Current object ID

expandDBID

protected void expandDBID(DBIDRef id)

Description copied from class: GeneralizedOPTICS.Instance

Add the current DBID to the cluster order, and expand its neighbors if minPts and similar conditions are satisfied.

Specified by:

expandDBID in class GeneralizedOPTICS.Instance<DiSH.DiSHClusterOrder>

Parameters:

id - Current object ID

compare

public int compare(DBIDRef o1,
                   DBIDRef o2)

Specified by:

compare in interface java.util.Comparator<DBIDRef>

Overrides:

compare in class GeneralizedOPTICS.Instance<DiSH.DiSHClusterOrder>

getLogger

protected Logging getLogger()

Description copied from class: GeneralizedOPTICS.Instance

Get the class logger.

Specified by:

getLogger in class GeneralizedOPTICS.Instance<DiSH.DiSHClusterOrder>

Returns:

Class logger