报告人：余晗助理教授

        Department of Applied Statistics and Research Methods at University of Northern Colorado

题目：A Nonparametric Assessment of Model Adequacy with Generalized Relative Entropy

时间：2018年07月11日下午15:3 0

地点：海韵实验楼105

摘要：Maximum likelihood ratio test statistics may not exist in general in nonparametric function estimation and inference. In this talk a new class of generalized likelihood ratio (GLR) tests is proposed for nonparametric goodness-of-fit testing via the asymptotic variant of the minimax approach. The proposed nonparametric tests are developed to be asymptotically distribution-free based on latent variable representations. The nonparametric tests are ameliorated to be appropriately complex so that they are analytically tractable and numerically feasible. They are well applicable for the “adaptive” study of hypothesis testing problems of growing dimensions. To assess the proposed GLR tests, the asymptotic properties are derived. The procedure can be viewed as a novel nonparametric extension of the classical parametric likelihood ratio test as a guard against possible gross misspecification of the data-generating mechanism. Simulations of the proposed minimax-type GLR tests are investigated for the small sample size performance and show that the GLR tests have appealing small sample size properties.

报告人简介：Han Yu, PhD in Statistics from Florida State University, is an assistant professor in the Department of Applied Statistics and Research Methods at University of Northern Colorado. His research focuses on the modern semi-parametric causal inference under realistic assumptions defining non-parametric structural equation models (NPSEM) with spatio-temporal network data based on modern advanced empirical processes theory by fully incorporating the state of the art in machine learning tools. His independent methodological research integrated with his cross-interdisciplinary collaborative applied research activities including social science, criminology, sport sciences, information sciences, management and marketing brings together statistical methodology, theory and computation for the hidden stochastic processes of function-valued outcomes, endogenous and heterogeneous treatment effects, and very many control variables in infinite-dimensional structural models with applications in large, complex and heterogeneous high-dimensional data (“BigData”).

联系人：黄荣坦副教授

欢迎广大师生参加！