Because they explicitly recognize and measure uncertainty about the future. probability forecasts ("the probability is 0.6 that the product will be successful") are more informative and more valuable to decision makers that categorical forecasts ("the product will be successful"). Previous work on forecast evaluation has focused extensively on categorical forecasts. The goal of this research is to develop new procedures that can improve forecasting and decision making in practice by improving how we evaluate probability forecasts. This will help in the determination of how useful different forecasts are and the comparison of the sources of the forecasts (which could be data, models, or experts). It will also provide important diagnostic guidance as to how future forecasts might be improved and enable users to tailor evaluation schemes to particular decision-making problems. Finally, it will provide a better understanding of how combining probabilities from different sources can yield better forecasts.
