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

Is R-square an important measure in Holt-Winters method?

share|improve this question
Measure of what? – chl Nov 22 '12 at 8:45

From my experience in forecasting: no.

The problem is that (as with any data analysis method) by increasing the complexity of the model, one can reduce R squared, even if the more complex model does not improve the predictive power of the model. In Exponential Smoothing, this would involve going from Single Exponential Smoothing to smoothing with a trend (Holt's method) to smoothing with trend and seasonality (Holt-Winters). In-sample fit and thus R squared will improve with each step in complexity - but forecasts may very easily get worse.

Conversely, it is hard to get a decent concept of "degrees of freedom" in smoothing, so calculating an "adjusted R squared" to account for this spurious improvement in in-sample fit is hard.

Bottom line: I have been forecasting for years now, but R squared is not something I see applied to smoothing often. Better to keep a holdout sample and compare forecasts for that holdout sample for different models.

share|improve this answer

Not usually. The main concern in forecasting is likely to be a measure of likely forecast error such as root mean squared error (or MAPE, MPE, or others).

Of course, with the original variance and the mean squared error of the forecast model it is easy to compute R squared, but that's usually not the emphasis.

share|improve this answer

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.