To overcome the limitation of time series models which can not incorporate unexpected future events and other qualitative factors, a knowledge-assisted adjustment of statistical forecasts is proposed and a prototype UNIK-FCST is being developed. This ...
To overcome the limitation of time series models which can not incorporate unexpected future events and other qualitative factors, a knowledge-assisted adjustment of statistical forecasts is proposed and a prototype UNIK-FCST is being developed. This paper presents five types of judgmental factors and proposes schemes to identify and evaluate the impacts of historical events by regression and analogy, case-based reasoning based on limited similar cases, and non-monotonic reasoning for the management of multiple judgmental factors. The UNIK-FCST is applied to the demand forecasting of oil products.