de.lmu.ifi.dbs.elki.algorithm.timeseries

Class OfflineChangePointDetectionAlgorithm.Instance

java.lang.Object

de.lmu.ifi.dbs.elki.algorithm.timeseries.OfflineChangePointDetectionAlgorithm.Instance

Enclosing class:

OfflineChangePointDetectionAlgorithm

class OfflineChangePointDetectionAlgorithm.Instance
extends java.lang.Object

Instance for a single data set.

Author:

Erich Schubert

Field Summary

Fields

Modifier and Type	Field and Description
(package private) double[]	bstrap Temporary storage for bootstrap testing.
(package private) double[]	column Raw data column.
(package private) int	columnnr Current column number.
(package private) DBIDArrayIter	iter Iterator to reference data positions.
(package private) ChangePoints	result Result to output to.
(package private) java.util.Random	rnd Random generator.
(package private) double[]	sums Cumulative sum.

Constructor Summary

Constructors

Constructor and Description
Instance(java.util.Random rnd) Constructor.

Method Summary

Modifier and Type	Method and Description
private double	bootstrapConfidence(int begin, int end, double thresh) Calculates the confidence for the most probable change point of the given timer series.
private int	multipleChangepointsWithConfidence(int begin, int end) Performs multiple change point detection for a given time series.
ChangePoints	run(Relation<DoubleVector> relation) Run the change point detection algorithm on a data relation.

Methods inherited from class java.lang.Object

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

Field Detail

column

double[] column

Raw data column.

sums

double[] sums

Cumulative sum.

bstrap

double[] bstrap

Temporary storage for bootstrap testing.

iter

DBIDArrayIter iter

Iterator to reference data positions.

result

ChangePoints result

Result to output to.

columnnr

int columnnr

Current column number.

rnd

java.util.Random rnd

Random generator.

Constructor Detail

Instance

public Instance(java.util.Random rnd)

Constructor.

Parameters:

rnd - Random generator

Method Detail

run

public ChangePoints run(Relation<DoubleVector> relation)

Run the change point detection algorithm on a data relation.

Parameters:

relation - Data relation.

Returns:

Change points

multipleChangepointsWithConfidence

private int multipleChangepointsWithConfidence(int begin,
                                               int end)

Performs multiple change point detection for a given time series. This method uses a kind of divide and conquer approach

Parameters:

begin - Interval begin

end - Interval end

Returns:

Last change point position, or begin

bootstrapConfidence

private double bootstrapConfidence(int begin,
                                   int end,
                                   double thresh)

Calculates the confidence for the most probable change point of the given timer series. Confidence is calculated with the help of bootstrapping.

Parameters:

begin - Subset begin

end - Subset end

thresh - Threshold

Returns:

confidence for most probable change point