What is the best way to fit the ARMA model?

I have having problem with autocorrelation errors in the ARMA model and I am wondering what is the best/fastest way to find the ARMA model that best fits my data.

[José-Luis]

Your question is too vague to be answered efficiently. "I have a problem" is not the most descriptive ---or helpful--- statement.

The ARMA model that makes most sense might not be the one that best fits your data. Given enough degrees of freedom, you might get a perfect fit for a model that makes no sense.

What is it you are trying to simulate?

Answer 1

[Roger Wohlwend]

To remove the autocorrelation you have to add more AR or MA terms. Make sure that you work with stationary data.

The fastest way to find the best fit? People often use an information criterion to find the best model. You will find extensive information on that in the Matlab documention.

However, it is important to check if an ARMA model is appropriate. Check if the time series exhibits autocorrelation. Only if it does an ARMA model can be applied. ARMA models are popular, they are easy to use, that is why everybody uses them, but it seems to me that hardly anyone checks if it is appropriate. So if you get bad results or you cannot make the autocorrelation disappear you have to consider other kinds of models.