A Versatile Family of Forecasting Systems

Computer packages for short-term forecasting and statistical stock control usually offer largely self-contained systems with no facility for including special management or market knowledge directly in the control system. This paper describes an empirical solution developed under the limitations on time and resources commonly experienced in the commercial environment. It was successfully implemented in applications in the volume engineering and food industries. Possible applications to the shoe industry are also considered. The basic model uses a Box-Jenkins predictor with transformation factors for seasonality, forward pattern of promotions and known step increases in demand.