org.openimaj.ml.timeseries.processor

Class WindowedLinearRegressionProcessor

java.lang.Object

org.openimaj.ml.timeseries.processor.WindowedLinearRegressionProcessor

All Implemented Interfaces:

TimeSeriesProcessor<double[],Double,DoubleTimeSeries>

public class WindowedLinearRegressionProcessor
extends Object
implements TimeSeriesProcessor<double[],Double,DoubleTimeSeries>

An implementation of an autoregressive model such that Xt = b*X{t-offset-window,t-offset} + c

Author:

Sina Samangooei (ss@ecs.soton.ac.uk)

Constructor Summary

Constructors

Constructor and Description
WindowedLinearRegressionProcessor() Calculate the regression from the same time series inputed
WindowedLinearRegressionProcessor(DoubleTimeSeries yearFirstHalf, int i)
WindowedLinearRegressionProcessor(DoubleTimeSeries yearFirstHalf, int windowsize, int offset)
WindowedLinearRegressionProcessor(int windowsize) Perform regression s.t.
WindowedLinearRegressionProcessor(int windowsize, int offset) Perform regression s.t.
WindowedLinearRegressionProcessor(LinearRegression reg) Use reg as the linear regression to predict.

Method Summary

Modifier and Type	Method and Description
LinearRegression	getRegression()
void	holdreg(boolean regdefined)
void	process(DoubleTimeSeries series)

Methods inherited from class java.lang.Object

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

Constructor Detail

WindowedLinearRegressionProcessor

public WindowedLinearRegressionProcessor()

Calculate the regression from the same time series inputed

WindowedLinearRegressionProcessor

public WindowedLinearRegressionProcessor(int windowsize)

Perform regression s.t. y = Sum(w_{0-i} * x_{0-i}) + c for i from 1 to windowsize

Parameters:

windowsize -

WindowedLinearRegressionProcessor

public WindowedLinearRegressionProcessor(int windowsize,
                                         int offset)

Perform regression s.t. y = Sum(w_{0-i} * x_{0-i}) + c for i from 1 to windowsize

Parameters:

windowsize -

offset -

WindowedLinearRegressionProcessor

public WindowedLinearRegressionProcessor(LinearRegression reg)

Use reg as the linear regression to predict. The process(DoubleTimeSeries) function simply calls LinearRegression.predict(Matrix) with the times in the series as input

Parameters:

reg -

WindowedLinearRegressionProcessor

public WindowedLinearRegressionProcessor(DoubleTimeSeries yearFirstHalf,
                                         int i)

WindowedLinearRegressionProcessor

public WindowedLinearRegressionProcessor(DoubleTimeSeries yearFirstHalf,
                                         int windowsize,
                                         int offset)

Method Detail

process

public void process(DoubleTimeSeries series)

Specified by:

process in interface TimeSeriesProcessor<double[],Double,DoubleTimeSeries>

Parameters:

series - alter this time series in place

holdreg

public void holdreg(boolean regdefined)

Parameters:

regdefined - if true, process holds its last LinearRegression

getRegression

public LinearRegression getRegression()

Returns:

the WindowedLinearRegressionProcessor's underlying LinearRegression model