Skip to Main Content
In order to avoid excess or shortage of inventory and to maintain appropriate inventory level, consumer-goods manufacturers have to balance demand and supply. Fluctuation of the demand of regular items is relatively stable, which allows statistical time series forecasting methods work well. As a result, the inventory level of the regular items can be maintained properly. On the other hand, accurate forecast of new products is difficult to achieve. Thus, the problem is to realize accurate forecast of new products and to enable early decision making on adjustment of production schedule. This paper proposes a new demand forecasting model that is an extension of the traditional exponential diffusion models. We examined the forecasting performance of the models just after the release of the item when the small number of model calibration data is available. This paper shows that the proposed model has the best performance and enables early decision making.