org.wso2.siddhi.extension.timeseries.linreg

Class LengthTimeRegressionCalculator

java.lang.Object

org.wso2.siddhi.extension.timeseries.linreg.LengthTimeRegressionCalculator

Direct Known Subclasses:

LengthTimeMultipleLinearRegression, LengthTimeSimpleLinearRegressionCalculator

public abstract class LengthTimeRegressionCalculator
extends Object

Field Summary

Fields

Modifier and Type	Field and Description
protected int	batchSize
protected int	calcInterval
protected double	confidenceInterval
protected long	currentEventTime
protected long	duration
protected int	eventCount
protected Queue<Long>	expiryTimeList
protected int	incCounter
protected int	xParameterCount

Constructor Summary

Constructors

Constructor and Description
LengthTimeRegressionCalculator(int paramCount, long timeWindow, int lengthWindow, int calcInt, double ci) Constructor

Method Summary

Modifier and Type	Method and Description
protected abstract void	addEvent(Object[] data, long expiryTime) Abstract method for accumulating events
Object[]	calculateLinearRegression(Object[] data, long expiryTime) Method which returns linear regression results
protected abstract Object[]	processData() Abstract method for performing linear regression
protected abstract void	removeExpiredEvents() Abstract method for removing events which have expired based on time and/or length

Methods inherited from class java.lang.Object

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

Field Detail

duration

protected long duration

batchSize

protected int batchSize

calcInterval

protected int calcInterval

confidenceInterval

protected double confidenceInterval

eventCount

protected int eventCount

xParameterCount

protected int xParameterCount

incCounter

protected int incCounter

currentEventTime

protected long currentEventTime

expiryTimeList

protected Queue<Long> expiryTimeList

Constructor Detail

LengthTimeRegressionCalculator

public LengthTimeRegressionCalculator(int paramCount,
                                      long timeWindow,
                                      int lengthWindow,
                                      int calcInt,
                                      double ci)

Constructor

Parameters:

paramCount - Number of X variables + 1 (Y variable)

timeWindow - Time window which is used to constrain number of events

lengthWindow - Length window which is used to constrain number of events

calcInt - Frequency of regression calculation

ci - Confidence interval

Method Detail

addEvent

protected abstract void addEvent(Object[] data,
                                 long expiryTime)

Abstract method for accumulating events

Parameters:

data - Array which holds Y and X values

expiryTime - Time at which each event should expire (based on time window)

removeExpiredEvents

protected abstract void removeExpiredEvents()

Abstract method for removing events which have expired based on time and/or length

processData

protected abstract Object[] processData()

Abstract method for performing linear regression

Returns:

Array containing standard error and beta values

calculateLinearRegression

public Object[] calculateLinearRegression(Object[] data,
                                          long expiryTime)

Method which returns linear regression results

Parameters:

data - Array which holds Y and X values

expiryTime - Time at which each event should expire (based on time window)

Returns:

Array containing standard error and beta values