How to choose the right optimization options for seasonal ARIMA process?

Hello everyone!

I want to create a seasonal arima (3,1,2)x(3,1,2) process with seasonality 24. I use the default optimization algorithm sequential quadratic programming. When I run the code I get the following warning:

 Nonlinear inequality contraints are active; standard errors may be inaccurate.

After searching the internet I found out that I can choose a wide range of other optimization algorithms. How can I find out which one is the right one for my problem?

Are there any other possibilities to get rid of the warning?

Thanks a lot!

[Matt J]

When I run the code

You should mention what code it is you're running so we're all on the same page.

[MC3105]

My code is:

 % model architecture and optimization options
 options = optimset('fmincon');
 options = optimset(options,'Algorithm','sqp','TolCon',1e-7,'MaxFunEvals',3000,'Display','final','Diagnostics','on');
 model =  arima('Constant',0,'D',1,'Seasonality',24,'ARLags',1:3,'SARLags',1:3,'MALags',1:2,'SMALags',1:2); 

 for t=v1:24:v2

 % database for model estimation
 Y=pv(1+t-v1:t-1);

 % model calibration
 [fit,EstParamCov,LogL,info] = estimate(model,Y,'options',options);   

 % forecast for the next 24 hours
 [Yf,YMSE] = forecast(fit,24,'Y0',Y);
 pred_sarima(t:t+23)=Yf;

 end

Now matlab tells me that it cannot find a feasible solution... any idea what the problem might be?

Answer 1

[Matt J]

The warning is only consequential if you are using the error estimates returned by the routine. The warning isn't saying that the solution is inaccurate, just that the standard error estimates might be.