Real-Time Bayesian Inference for Latent Ability Models

• Activity Number: 589
• Type: Topic Contributed
• Date/Time: Wednesday, August 12, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Ruby Chiu-Hsing Weng*
• Companies: National Chengchi University
• Keywords: Bayesian inference ; item response theory ; Kalman filter ; latent ability model ; online inference ; Stein's identity

Abstract

Latent ability models relate a set of observed variables to a set of latent ability variables. It has a variety of applications, including the item response theory (IRT) models and the paired comparison models. The IRT models have been widely used in modeling educational test data, and the paired comparison models such as Bradley-Terry-Luce models are popular in modeling competition data. In this talk, I first review a Bayesian approximate method for online gaming analysis with paired comparison models, and then present an application to Internet ratings data using item response theory models. I will also briefly discuss how the Bayesian technique can be related to the Kalman filter. (This is joint work with Steve Coad.)