Abstract:
|
There is a growing awareness amongst decision makers that single number and scenario-based demographic forecasts are inadequate for their purposes. In parallel, developers of forecasting methodology have increasingly adopted Bayesian approaches which can coherently summarise uncertainty. However, the utilisation of the resulting probabilistic forecasts in the decision-making processes remains underdeveloped. How do we then help users to fully utilise the probabilistic forecasts in their decision making process? One possible answer consists of eliciting the loss functions from the decision makers and applying Bayesian decision theory to deliver bespoke decision advice and support. In this paper, we describe a framework to develop a two-way communication with decision makers, whose priorities inform the loss function, and to whom the resulting forecasts are communicated back in a user-friendly form. We illustrate our approach with a real-world decision example and contrast it with a scenario-based alternative.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.