11.5 Exponential smoothing and ETS

Exponential smoothing methods update forecasts by giving greater weight to recent observations. Simple exponential smoothing is appropriate for a series with a relatively stable level and no clear trend or seasonality.

Holt’s method adds a changing trend. Holt-Winters methods add seasonality. ETS provides a general framework in which error, trend, and seasonal components can be additive or multiplicative. Additive seasonality has a roughly constant size, while multiplicative seasonality changes in proportion to the series level.

The method should be selected through time-aware validation and diagnostic review, not solely by accepting software defaults.