de.lmu.ifi.dbs.elki.evaluation.outlier

Class OutlierThresholdClustering

java.lang.Object

de.lmu.ifi.dbs.elki.evaluation.outlier.OutlierThresholdClustering

All Implemented Interfaces:

Evaluator, ResultProcessor

public class OutlierThresholdClustering
extends java.lang.Object
implements Evaluator

Pseudo clustering algorithm that builds clusters based on their outlier score. Useful for transforming a numeric outlier score into a 2-class dataset.

Since:

0.5.0

Author:

Erich Schubert

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	OutlierThresholdClustering.Parameterizer Parameterization helper

Field Summary

Fields

Modifier and Type	Field and Description
(package private) ScalingFunction	scaling Scaling function to use
(package private) double[]	threshold Thresholds to use

Constructor Summary

Constructors

Constructor and Description
OutlierThresholdClustering(ScalingFunction scaling, double[] threshold) Constructor.

Method Summary

Modifier and Type	Method and Description
void	processNewResult(ResultHierarchy hier, Result newResult) Process a result.
private Clustering<Model>	split(OutlierResult or)

Methods inherited from class java.lang.Object

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

Field Detail

scaling

ScalingFunction scaling

Scaling function to use

threshold

double[] threshold

Thresholds to use

Constructor Detail

OutlierThresholdClustering

public OutlierThresholdClustering(ScalingFunction scaling,
                                  double[] threshold)

Constructor.

Parameters:

scaling - Scaling function

threshold - Threshold

Method Detail

processNewResult

public void processNewResult(ResultHierarchy hier,
                             Result newResult)

Description copied from interface: ResultProcessor

Process a result.

Specified by:

processNewResult in interface ResultProcessor

Parameters:

hier - The base of the result tree.

newResult - Newly added result subtree.

split

private Clustering<Model> split(OutlierResult or)