Identifying the numbers of AR or MA terms in matlab

How do I identify the order of ARMA model for given time series in matlab. In R, there is a function ar, auto.arima that searches for the optimal number of lags for AR and MA models. But I haven't been able to find a similar function in matlab. The AR function in matlab estimates parameters for a 'specified' number of lags. Any ideas?

Secondly, is there a function similar to arima.sim in matlab to simulate arima processes?

[Wayne King]

Do you have the System Identification Toolbox?

Answer 1

[Rajiv Singh]

Look into arxstruc, selstruc and n4sid's order selection feature.

[Yujiang Wu]

the arxstruc computes the loss function. Is this loss function based on the prediction focus or based on the simualtion focus? I assume the default is the prediction focus. Then how do I change to simulation focus?

I can estimate a ARX model for simulation focus using

opt = arxOptions;
opt = arxOptions('Focus','simulation');
NN = struc(2:6,2:5,2:5,[0 1],[0 1]);
V = arxstruc(de,dv,NN);
nns = selstruc(V)% interactively select system order
M_arx_sim = arx(de,nns,opt);% estimate arx model using simulation focus

But if during the model order seletion stage, I am computing the loss function using the prediction focus, then the model order selected might not be the optimum choice? Any instructions?