现在位置:首页 > 学术报告
 

 

Academy of Mathematics and Systems Science, CAS
Colloquia & Seminars

Speaker:

Prof. Jun Liu, Dept of Statistics, Harvard University and Center for Statistical Science, Tsinghua University

Inviter: 王启华
Title:
Bayesian Aggregation of Rank Data with Covariates and Heterogeneous Rankers
Time & Venue:
2018.1.11 16:00-17:00 N613
Abstract:
Rank aggregation is the combining of ranking results from different sources to generate a ``better'' ranking list. In our applications, the rank data contain covariate information of ranked entities and incomplete ranking lists for non-overlapping subgroups. Since most existing rank aggregation methods do not handle covariate information of the ranked entities as well as the rankers' heterogeneity, we propose the Bayesian Aggregation of Rank-data with Covariates (BARC), its weighted version (BARCW), and its extension to mixture models (BARCM). All three methods employ latent variable models to account for the covariate information and heterogeneity of rankers. Specifically, BARC assumes identical opinion of all rankers with the same quality; BARCW extends it by allowing varying qualities of rankers, while BARCM clusters heterogeneous ranking opinions with a Dirichlet process mixture model. Moreover, we use a parameter-expanded Gibbs sampler to draw posterior samples, and generate aggregated ranking lists with credible intervals quantifying their uncertainty. Simulation studies show the superior performance of our methods compared with other existing methods in a variety of scenarios.Finally, we exploit our proposed method to solve two real-data problems.
 

 

附件下载:
 
 
【打印本页】【关闭本页】
电子政务平台   |   科技网邮箱   |   ARP系统   |   会议服务平台   |   联系我们   |   友情链接