ARIMA models and Box-jenkins function

Question

antonio tirri 2012-1-4

Hi, I have to forecast some value of a time series using an ARIMA(5,1,3) model.

I saw in Matlab there isn't a function for ARIMA models because ARIMA models are a type of Box-Jenkins models. But how to set parameters?

In the Box-Jenkins models

m = bj(data,[nb nc nd nf nk])

How to set nb, nc, nd, nf and nk in order to have an ARIMA(p,d,q) model?

Thanks

Answer 1

Rajiv Singh 2012-3-17

In R2012a, use the ARMAX function with "IntegrateNoise" attribute specified using name-value pair. See ARMAX documentation for more info.

For example, model = armax(data, [na nc], 'IntegrateNoise', true) estimates an ARIMA model of time series "data".