Estimating Seasonal Relatives Using Simple Averaging: Not Just for Stationary Data
William J. Stevenson and A. Erhan Mergen
Saunders College of Business, Rochester Institute of Technology, USA
Volume 8: 2014, pp. 125-138; ABSTRACT
A popular technique that is used to teach students about forecasting for seasonality is to illustrate the computation of seasonal relatives using simple averaging, with the admonition that averaging cannot be used if the data include trend. However, the authors have found an interesting relationship between the accuracy of estimates of seasonal relatives when a linear trend is present and the ratio of intercept to slope: when the ratio is relatively high, fairly good estimates of seasonal relatives can be made, even though trend is present. Moreover, that same relationship exists for a conventional method for estimating seasonal relatives, such as a centered moving average approach. Beyond the obvious benefit of expanding the use of simple averaging, the nature of the relationship can lead to an insightful classroom discussion on why the ratio of intercept to slope affects accuracy.
ORDER ARTICLE PERMISSIONS/REPRINTS/OFFPRINTS
To order permissions to include this case study in textbooks, edited volumes, course booklets, online/digital course packs, etc., and/or to order multiple individual hard copies for classroom use, please use the appropriate form available on the Order Forms page or alternatively, contact the Publishing Editor pneilson@neilsonjournals.com directly.
