A comparison of extrapolation models on yearly sales forecasts |
| |
| Institution: | 1. Faculty of Veterinary and Agricultural Sciences, The University of Melbourne, Victoria 3121, Australia;2. CSIRO Land and Water, GPO Box 1666, Canberra, ACT 2601, Australia;1. Department of Sanitary and Environmental Engineering, Federal University of Santa Catarina, Brazil;2. Embrapa, Swine and Poultry, Brazil;1. Departamento de Producción Agraria, Universidad Politécnica de Madrid, E.T.S. Ingenieros Agrónomos de Madrid, Ciudad Universitaria s/n, 28040 Madrid, Spain;2. Instituto de Ciencia y Tecnología Animal, Universitat Politècnica de València, Camino de Vera s/n, 46022 Valencia, Spain;3. Centro de Investigación y Tecnología Animal, Instituto Valenciano de Investigaciones Agrarias, Pol. La Esperanza 100, 12400 Segorbe, Castellón, Spain |
| |
| Abstract: | This study compares the performance of seven extrapolation models and three combinations of models on nearly 1,500 forecasts of annual, unit sales series. The focus is on two important issues — the identification of those instances when forecasting with extrapolation models is likely to be successful (irrespective of the model selected), and which extrapolation approaches offer the greatest potential. This study replicates many of the finding of the M-competition. It also adds some new evidence. For example, Gardner's dampened smoothing was found to perform especially well on the series tested. Also, extrapolation forecasts of durable good series and ‘mature’ product categories were rarely superior to forecasts made using a simple random-walk model. |
| |
| Keywords: | |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
