Computer-Based Bayesian Forecasting Methodologies

Over the past several decades a concerted effort has been made to advance the art of forecasting to a science, ¹ While debate rages on regarding the success of this effort, members of the university, industrial, and governmental communities continue to develop and test “artistic” and “scientific” forecasting methodologies. ² To a great extent, this distinction between artistic and scientific forecasting may be understood on epistemological grounds. ³ Artistic or subjective forecasting methodologies assume one knowledge acquisition/production bias while quantitative-empirical, or objective, methodologies assume another. Neither bias is “right"; nor is it the purpose of this chapter to argue the costs and benefits of each--the forecasting literature manages this debate quite nicely. Rather, it is the intention here to examine a number of subjective Bayesian forecasting methodologies which, in recent years, have been refined and computerized, i.e., implemented on small minicomputers and microcomputers.