Package elki.clustering.subspace

Class DOC

java.lang.Object

elki.clustering.subspace.DOC

All Implemented Interfaces:

elki.Algorithm, ClusteringAlgorithm<Clustering<SubspaceModel>>, SubspaceClusteringAlgorithm<SubspaceModel>

Direct Known Subclasses:

FastDOC

@Title("DOC: Density-based Optimal projective Clustering")
@Reference(authors="C. M. Procopiuc, M. Jones, P. K. Agarwal, T. M. Murali",
           title="A Monte Carlo algorithm for fast projective clustering",
           booktitle="Proc. ACM SIGMOD Int. Conf. on Management of Data (SIGMOD \'02)",
           url="https://doi.org/10.1145/564691.564739",
           bibkey="DBLP:conf/sigmod/ProcopiucJAM02")
public class DOC
extends java.lang.Object
implements SubspaceClusteringAlgorithm<SubspaceModel>

DOC is a sampling based subspace clustering algorithm.

Reference:

C. M. Procopiuc, M. Jones, P. K. Agarwal, T. M. Murali
A Monte Carlo algorithm for fast projective clustering
In: Proc. ACM SIGMOD Int. Conf. on Management of Data (SIGMOD '02).

Since:

0.6.0

Author:

Florian Nuecke

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	DOC.Par	Parameterization class.

Nested classes/interfaces inherited from interface elki.Algorithm

elki.Algorithm.Utils

Field Summary

Fields

Modifier and Type	Field	Description
protected double	alpha	Relative density threshold parameter alpha.
protected double	beta	Balancing parameter for importance of points vs. dimensions
private static elki.logging.Logging	LOG	The logger for this class.
protected elki.utilities.random.RandomFactory	rnd	Randomizer used internally for sampling points.
protected double	w	Half width parameter.

Constructor Summary

Constructors

Constructor	Description
DOC(double alpha, double beta, double w, elki.utilities.random.RandomFactory random)	Constructor.

Method Summary

Modifier and Type	Method	Description
protected double	computeClusterQuality(int clusterSize, int numRelevantDimensions)	Computes the quality of a cluster based on its size and number of relevant attributes, as described via the μ-function from the paper.
protected boolean	dimensionIsRelevant(int dimension, elki.database.relation.Relation<? extends elki.data.NumberVector> relation, elki.database.ids.DBIDs points)	Utility method to test if a given dimension is relevant as determined via a set of reference points (i.e. if the variance along the attribute is lower than the threshold).
protected elki.database.ids.DBIDs	findNeighbors(elki.database.ids.DBIDRef q, long[] nD, elki.database.ids.ArrayModifiableDBIDs S, elki.database.relation.Relation<? extends elki.data.NumberVector> relation)	Find the neighbors of point q in the given subspace
elki.data.type.TypeInformation[]	getInputTypeRestriction()
protected Cluster<SubspaceModel>	makeCluster(elki.database.relation.Relation<? extends elki.data.NumberVector> relation, elki.database.ids.DBIDs C, long[] D)	Utility method to create a subspace cluster from a list of DBIDs and the relevant attributes.
Clustering<SubspaceModel>	run(elki.database.relation.Relation<? extends elki.data.NumberVector> relation)	Performs the DOC or FastDOC (as configured) algorithm.
protected Cluster<SubspaceModel>	runDOC(elki.database.relation.Relation<? extends elki.data.NumberVector> relation, elki.database.ids.ArrayModifiableDBIDs S, int d, int n, int m, int r, int minClusterSize)	Performs a single run of DOC, finding a single cluster.

Methods inherited from class java.lang.Object

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

Methods inherited from interface elki.clustering.ClusteringAlgorithm

autorun

Field Detail

LOG

private static final elki.logging.Logging LOG

The logger for this class.

alpha

protected double alpha

Relative density threshold parameter alpha.

beta

protected double beta

Balancing parameter for importance of points vs. dimensions

w

protected double w

Half width parameter.

rnd

protected elki.utilities.random.RandomFactory rnd

Randomizer used internally for sampling points.

Constructor Detail

DOC

public DOC(double alpha,
           double beta,
           double w,
           elki.utilities.random.RandomFactory random)

Constructor.

Parameters:

alpha - α relative density threshold.

beta - β balancing parameter for size vs. dimensionality.

w - half width parameter.

random - Random factory

Method Detail

getInputTypeRestriction

public elki.data.type.TypeInformation[] getInputTypeRestriction()

Specified by:

getInputTypeRestriction in interface elki.Algorithm

run

public Clustering<SubspaceModel> run(elki.database.relation.Relation<? extends elki.data.NumberVector> relation)

Performs the DOC or FastDOC (as configured) algorithm.

This will run exhaustively, i.e. run DOC until no clusters are found anymore / the database size has shrunk below the threshold for minimum cluster size.

Parameters:

relation - Data relation

runDOC

protected Cluster<SubspaceModel> runDOC(elki.database.relation.Relation<? extends elki.data.NumberVector> relation,
                                        elki.database.ids.ArrayModifiableDBIDs S,
                                        int d,
                                        int n,
                                        int m,
                                        int r,
                                        int minClusterSize)

Performs a single run of DOC, finding a single cluster.

Parameters:

relation - used to get actual values for DBIDs.

S - The set of points we're working on.

d - Dimensionality of the data set we're currently working on.

r - Size of random samples.

m - Number of inner iterations (per seed point).

n - Number of outer iterations (seed points).

minClusterSize - Minimum size a cluster must have to be accepted.

Returns:

a cluster, if one is found, else null.

findNeighbors

protected elki.database.ids.DBIDs findNeighbors(elki.database.ids.DBIDRef q,
                                                long[] nD,
                                                elki.database.ids.ArrayModifiableDBIDs S,
                                                elki.database.relation.Relation<? extends elki.data.NumberVector> relation)

Find the neighbors of point q in the given subspace

Parameters:

q - Query point

nD - Subspace mask

S - Remaining data points

relation - Data relation

Returns:

Neighbors

dimensionIsRelevant

protected boolean dimensionIsRelevant(int dimension,
                                      elki.database.relation.Relation<? extends elki.data.NumberVector> relation,
                                      elki.database.ids.DBIDs points)

Utility method to test if a given dimension is relevant as determined via a set of reference points (i.e. if the variance along the attribute is lower than the threshold).

Parameters:

dimension - the dimension to test.

relation - used to get actual values for DBIDs.

points - the points to test.

Returns:

true if the dimension is relevant.

makeCluster

protected Cluster<SubspaceModel> makeCluster(elki.database.relation.Relation<? extends elki.data.NumberVector> relation,
                                             elki.database.ids.DBIDs C,
                                             long[] D)

Utility method to create a subspace cluster from a list of DBIDs and the relevant attributes.

Parameters:

relation - to compute a centroid.

C - the cluster points.

D - the relevant dimensions.

Returns:

an object representing the subspace cluster.

computeClusterQuality

protected double computeClusterQuality(int clusterSize,
                                       int numRelevantDimensions)

Computes the quality of a cluster based on its size and number of relevant attributes, as described via the μ-function from the paper.

Parameters:

clusterSize - the size of the cluster.

numRelevantDimensions - the number of dimensions relevant to the cluster.

Returns:

a quality measure (only use this to compare the quality to that other clusters).