AIC for ARIMA

7 views (last 30 days)
Marina
Marina on 2 May 2012
Hi all. I am modeling a time series with ARIMA model, and I have to test th AIC creterion to know wich is the best model. aic(model), but what to put in place of "model"?
  3 Comments
Walter Roberson
Walter Roberson on 2 May 2012
http://www.mathworks.com/matlabcentral/answers/29922-why-your-question-is-not-urgent-or-an-emergency
Marina
Marina on 2 May 2012
I use ssytem identification toolbox

Sign in to comment.

Answers (1)

Rajiv Singh
Rajiv Singh on 3 May 2012
In System Identification Toolbox, use the armax command to generate a model. Suppose x represents your time series data: z = iddata(x,[]); model = armax(z,[na nc], 'IntegrateNoise', true)
The "IntegrateNoise" option is available in R2012a, but not before that. In older releases, you can do something like this: zd = iddata(diff(x),[]); model = armax(zd,[na nc]) model.a = conv([1 -1],model.a)
rajiv

Categories

Find more on Linear Model Identification in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!