【主讲】Roger Koenker Professor (Department of Economics, University of Illinois)
【主题】Unobserved Heterogeneity in Longitudinal Data: An Empirical Bayes Perspective
【时间】2014年5月28日 (周三) 10：00-11：30
【地点】上海财经大学经济学院楼710室
【语言】英文
【摘要】Empirical Bayes methods for Gaussian and Binomial compound decision problems involving longitudinal data are considered. A new convex optimization formulation of the nonparametric (Kiefer-Wolfowitz) maximum likelihood estimator for mixture models is used to construct nonparametric Bayes rules for compound decisions. The methods are illustrated with some simulation examples as well as an application to predicting baseball batting averages. Comparisons with nonparametric Bayesian methods based on Dirichlet process prior are also provided.
