Department Seminar by Chi-Kuang Yeh

Please Note: This seminar will be given online.

Evaluating the performance of continuously updated probabilistic forecasts with application to National Basketball Association outcome prediction.

Making accurate predictions for an uncertain outcome is often a common humankind desire. As a result, probabilistic predictions and forecasts are ubiquitous in modern society, and many individuals consider and make decisions based on them. Over time the number and scope of probabilistic forecasts readily accessible to the public has increased at a steady pace, and now covers prediction of phenomena ranging from various fields. In the past, the available information for making the predictions is fixed and unchanging throughout the decision making process. Nowadays, many such forecasts are made initially well before the event in question occurs, and are then continuously updated as new information becomes available. Along with domain expertise, decision-makers rely heavily on these forecasts, so the forecast quality is crucial. Also, amid the sea of the available predictive models, one ought to ask: which model is significantly better than others?  To answer these questions, we develop new tools for measuring the quality of continuously updated probabilistic forecasts, including a significance test and simple graphical summaries. We conclude this talk with an application to National Basketball Association.