FORECAST.ETS: how does Excel calculate the Alpha, Beta, Gamma parameters in the AAA-ETS algorithm?

Apologies for the late reply. I was searching the Microsoft Documentation website as you suggested. Unfortunately, I have not found any details on my question, i.e., which approach Excel's FORECAST.ETS takes to calculate the smoothing parameters.

I bet the implementation follows an optimization approach (maximum likelihood or minimize mean squared error). It would be very helpful if there is a reference to which approach is taken.

Apologies for the late reply. I was searching the Microsoft Documentation website as you suggested. Unfortunately, I have not found any details on my question, i.e., which approach Excel's FORECAST.ETS takes to calculate the smoothing parameters.

I bet the implementation follows an optimization approach (maximum likelihood or minimize mean squared error). It would be very helpful if there is a reference to which approach is taken.

Hi holg3000,

Besides the FORECAST.ETS function article you've already mentioned any other documentation with regards to this function can only be found in our Microsoft Documentation website.

Another avenue to get information for this is our Office VBA support and feedback page which states that VBA programming questions can be better answered by the people at Stack Overflow -- a question and answer website for professional and enthusiast programmers. You can also post this issue of yours in our MSDN Forums which is geared for software developers and testers.

Both options should give you more specific help on the issue you've described in your first post above.

Thanks again for choosing Microsoft! We’re here when you need us.

Sincerely,

Glenn Rue

Consumer M365 Forum Moderator

Microsoft Community

Hi holg3000,

We’re glad that you reached out to the community for support on this issue of yours.  We’ve already responded to your query, however, we have not heard a reply from you in the last 48 hours confirming if any of our suggestions have been helpful or not.

I personally understand that finding time to work on this matter can be a little difficult to manage.  So please accept my sincerest gratitude for even entertaining this follow-up of mine.

At any rate, please post on the same thread if you need further help.

Hope to hear from you soon.

Respectfully,

Glenn Rue

Consumer M365 Forum Moderator

Microsoft Community

It is clear that FORECAST.ETS uses an AAA version of the Exponential Smoothing (ETS) algorithm. However, I did not found any reference which specific approach is applied to determine the base, trend and seasonality parameters Alpha, Beta, Gamma?

These parameters are shown in the FORECAST.ETS.STAT function, but it's unclear how Excel calculates them. There exist multiple approaches on how to set these parameters in the literature and in other software packages (see e.g. https://robjhyndman.com/hyndsight/estimation2/ for approaches available in R).

Is there a reference explaining which approach is used in FORECAST.ETS?

Hi holg3000,

First and foremost, thanks for reaching out to our Microsoft Community Forum and please accept our sincerest hope that all is well.

We understand that you're looking for additional information with regards to FORECAST.ETS besides that support article you referred on your post above.

If that is the case, you'd probably want to checkout our Microsoft Documentation website which host extensive and detailed documentations on such tops.

Hope that helps you find what you need.

Sincerely,

Glenn Rue

Consumer M365 Forum Moderator

Microsoft Community