Article ID: | iaor198826 |
Country: | Germany |
Volume: | 10 |
Start Page Number: | 195 |
End Page Number: | 212 |
Publication Date: | Nov 1988 |
Journal: | OR Spektrum |
Authors: | Fildes R. |
time series (extrapolative) forecasting procedures are widely used in business. Their importance has led to the development of new methods of forecasting and research into the evaluation of well-established methods. This paper presents an overview of the major methods of extrapolative forecasting. Secondly it considers the evidence on the relative accuracy of these methods, highlighting the unexpected conclusions arrived at by this new research. Organizations need to identify the forecasting method most appropriate to their particular circumstances. The paper discusses how such a selection should take place, commenting on how time series methods can complement managerial judgement. The paper concludes with some brief remarks on forecasting software, and a personal view of the future of time series forecasting.