Cross-learning with short seasonal time series

Huijing Chen, John E. Boylan, Ivan Svetunkov

Research output: Contribution to specialist publicationArticle


Since its introduction by R. G. Brown over 60 years ago, exponential smoothing, in its various flavors, has been a go-to model for many forecasting professionals. Thanks to its solid performance across 40 years of M competitions, exponential smoothing has earned a secure place in the forecaster's toolbox. The familiar Error-Trend-Seasonality (ETS) taxonomy by Hyndman and colleagues helps define how components of a time series interact with each other, and this new research by Chen, Boylan, and Svetunkov provides an enhanced taxonomy that can aid in cross-learning from similar time series with very short histories
Original languageEnglish
Number of pages7
Specialist publicationForesight: International Journal of Applied Forecasting
Publication statusPublished - 1 Sept 2023

Cite this