Title（问题）：Finding Needles in a Haystack: Detecting Tight Clusters and Outliers Via Cross-Validating Predictive Distributions
Speaker（报告人）：Professor George C. Tiao
The University of Chicago
北京大学信用教授
Time（时间）：2011年6月13日（周一）下昼2:00 — 3:00
Place（所在）：伟易博新楼217课堂
Abstract（摘要）：This work presents a procedure for detecting heterogeneities in a sample with respect to a given model. It can be applied to find if a univariate or a multivariate sample has been generated by different distributions, or if a regression model is really a mixture of different regression lines. Based on some special features of cross-validating predictive distributions, the idea of the procedure is first to split the sample into more homogeneous groups and second to recombine the observations in order to form homogeneous clusters. These two phases, splitting and recombining, form the core of the procedure. The proposed procedure is particularly efficient in detecting tight clusters in large samples and can be applied to study heterogeneities in any statistical model. The performance of the procedure is illustrated using univariate and multivariate data sets.
