Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

There is a debate in selecting the smoothign constant in Single Exponentioan Smoothing method by practitioner or considering it as a process parameter?

Could you please provide your opinion regarding this issue?

share|improve this question
Perhaps you could link to somewhere on the web where the debate is so we can see what the various parties are proposing, and the context? See my comment to @StephanKolassa. – Peter Ellis Dec 22 '12 at 19:12

I am unaware of any expert recommending that users choose their own smoothing constant. The consensus, as far as I understand it, is to optimize it in some way. See, e.g., here.

Of course, older software may require the user to set the constant himself, but this is not supported by evidence and probably due to the software development process and the constraints it faces.

share|improve this answer
Cowpertwaite and Metcalfe in Chapter 3 of Introductory Time Series Using R (Springer) seem to recommend that a forecaster may want to adjust upwards a smoothing constant that has been optimised for a long and stable time series (which will give a very low estimate of the smoothing constant under most optimization routines) if there is reason to believe that recent events should be taken more into account - citing video tape sales as an example. But perhaps this just shows the limitiations of any automatic forecasting during or just after a game changing alteration in the environment. – Peter Ellis Dec 22 '12 at 19:10

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.