Combining expert systems and the Bayesian approach to support forecasting

Abstract

An approach to integrating expert system models (for information regarding the present and future problem states) with a Bayesian forecasting model (for information based on the historical data) is presented. Using this approach, past, present, and future information can be processed coherently in an automated system to support forecasting in general. It is shown that this approach is a practical way to provide support to both qualitative and quantitative information processing.<>