org.codehaus.jet.regression
Interface InformationCriterionEstimator

All Known Implementing Classes:
AbstractInformationCriterionEstimator, AkaikeInformationCriterionEstimator, HannanQuinnInformationCriterionEstimator, SchwarzInformationCriterionEstimator

public interface InformationCriterionEstimator

An autoregressive (AR) process can be represented as

  y(t)=a_1*y(t-1)+...+a_p*y(t-p)+e_t, t=1,...,T
 
where a_i, i=,1,...,p are autoregressive parameters,e_t ~ N(0, sigma^2), and p is the lag order.

The AR lag order can selected so that it minimises certain Information Criterion (IC), such as

For notation and IC formulae see V. Khim-Sen Liew, Economics Bulletin (2004), Vol. 3, 33, 1-9.

Author:
Mauro Talevi

Method Summary
 void addData(double[] y)
          Adds sample data
 double estimateIC(int lag)
          Estimates the IC value for a given lag order
 int minimiseIC(int minLag, int maxLag)
          Minimise the IC value for a given lag order interval
 

Method Detail

addData

void addData(double[] y)
Adds sample data

Parameters:
y - the array of size T representing the y sample

estimateIC

double estimateIC(int lag)
Estimates the IC value for a given lag order

Parameters:
lag - the lag order used
Returns:
The value of the IC

minimiseIC

int minimiseIC(int minLag,
               int maxLag)
Minimise the IC value for a given lag order interval

Parameters:
minLag - the minimum lag order used
maxLag - the maximum lag order used
Returns:
The lag order that minimises the IC


Copyright © 2006-2007. All Rights Reserved.