EMPIRICAL COMPARISON OF MODELS FOR SHORT RANGE FORECASTING.

Abstract

Results of a simulation study of the short-range forecasting effectiveness of exponentially smoothed and selected Box-Jenkins models for sixty-three monthly sales series are presented. Tabular data are presented to show the average percent error, and the ranking of the models for both the one-month forecast and the cumulative six-month forecast for the sixty-three actual sales series. The average value of the smoothing parameter is shown for the models smoothed with the recurrence equations. The average percentage of errors for each model for five synthetic data series, plus forecasting results on the series, are also described.