org.hawkular.datamining.forecast

Class AutomaticForecaster.ConceptDriftStrategy

java.lang.Object

org.hawkular.datamining.forecast.AutomaticForecaster.ConceptDriftStrategy

Direct Known Subclasses:

AutomaticForecaster.ErrorChangeStrategy, AutomaticForecaster.PeriodicIntervalStrategy

Enclosing class:

AutomaticForecaster

public abstract static class AutomaticForecaster.ConceptDriftStrategy
extends Object

Strategy used for dealing with concept drift (statistical properties of modelled time series change over time) For instance originally monotonic time series has changed to trend stationary or seasonal pattern has showed up.

Field Summary

Fields

Modifier and Type	Field and Description
protected AutomaticForecaster	forecaster

Constructor Summary

Constructors

Constructor and Description
ConceptDriftStrategy()

Method Summary

Modifier and Type	Method and Description
abstract boolean	shouldSelectNewModel(int learnSize)

Methods inherited from class java.lang.Object

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

Field Detail

forecaster

protected AutomaticForecaster forecaster

Constructor Detail

ConceptDriftStrategy

public ConceptDriftStrategy()

Method Detail

shouldSelectNewModel

public abstract boolean shouldSelectNewModel(int learnSize)

Parameters:

learnSize - size of dataSet from learn method

Returns:

true if model should be selected - concept drift is present