Academic Report
Title: An Algorithm and A Strategy for Nonparametric Inference under Imperfect Rankings in Ranked-Set Sampling
Reporter: Prof.Yimin Zhang (Villanova University)
Time: May 31,2017 PM 14:00-15:00
Location: 408#room, Research and education building
Contact: WANG Xiaoguang(tel: 84708351-8013)
Abstract: Ranked set sampling has been successfully applied in fields such as forestry, entomology, medicine and environmental monitoring. In this talk, I will introduce a simple general strategy for approximate nonparametric inference in ranked-set sampling. This strategy involves of fitting a one-parameter model for imperfect rankings and then using an inference procedure that offers guaranteed good performance under the fitted model. The imperfect rankings model is fitted through a nonparametric maximum likelihood scheme that is made possible by a new algorithm for computing the probability that specified independent mixtures of order statistics have a particular ordering. I will illustrate the general strategy by using it to obtain confidence bands for the distribution function and confidence intervals for population quantiles. It will be shown through simulation that in both cases, the new strategy performs as well or better than existing methods and it performs well even when the data-generating model differs from the one-parameter model that is fitted.
The brief introduction to the reporter: Dr. Yimin Zhang is an assistant professor in Department of Mathematics and Statistics at Villanova University, US. She received her B.S. in Computational Mathematics from Dalian University of Technology, China, and Ph.D. in Statistics from Oklahoma State University. Dr. Zhang’s statistical methodological research interests include multiple comparisons, inference with ranked-set sampling, and statistical learning. Her collaborative work spans agriculture, transportation, ecology, and education.
