Group Abstract Group Abstract

Message Boards Message Boards

How to understand the optional method in TimeSeriesForecast?

GROUPS:
Members of the group I hope the is the right place fo my question.

I have been using several of the functions in the  Time Series toolbox of Mathematica. I am particularly interested in identifying models for electricity price forecasting.  
TimeSeriesForcast is a Mathematica function that I am using to perform forecast for several steps ahead.  An option for this function is the method, for instance, Method->"AR".
How can I know the order of the model being used in the approximation? I cannot find these kind of details in the help section.

Thanks for any help you could give

Jesus Rico-Melgoza 
Answer
1 year ago
As you noted, the documentation does not seem to show a way to find the order of the AR process used by TimeSeriesForecast to estimate the result.

I would suggest using EstimatedProcess to estimate the values of the parameters for the AR process of the order you are interested in. Once you have that, you can use it as the first argument to TimeSeriesForecast.
POSTED BY: Sean Clarke
Answer
1 year ago
Thanks for the suggestion

I still finding confusing. Say, you have an estimated ARIMA process and you use 
TimeSeriesForecast[ARIMA,data,Method->"AR"]
Is it not contradictory using a simple AR model for the forecast? I just can not understand why the need or perhaps the possibility for changing the model at this point.

Best regards

Jesus
Answer
1 year ago