Highly fluctuating demand during the peak sale season may be modeled by a continuous time stochastic process. In this paper, a Bayesian approach to demand estimation is outlined for the cases of large fluctuating seasonal demand. The proposed Bayesian model is based on physical observable quantities and demand factors between the slow and peak demand period. The model predicts the probability of the future demand expressed explicitly conditional on the observed demand prior to the peak season. Forecast is compared with real data and a widely implemented adaptive Holt-Winters (H-W) seasonal forecasting model. Results show that the proposed method improves the forecast, and reduces the inventory quantity during the busy sale season than do adaptive H-W model.
展开▼
