报告题目:Estimation and inference for ultra-high dimensional quasi-likelihood models based on data splitting
报 告 人:蒋学军 副教授
所在单位:南方科技大学
报告时间:2023年5月12日 星期五 15:00-16:00
报告地点:腾讯会议147655560
校内联系人:朱复康 zhufk@126.com
报告摘要:In this article, we develop a valid framework for inference of ultra-high dimensional quasi-likelihood models, based on a novel weighted estimation approach and a data splitting technique. The optimal weight is obtained by maximizing the efficiency of the estimator. Using the weighted estimator, we construct confidence intervals for some group components of the regression coefficient vector and perform the Wald test for a linear structure of the group components. Theoretically, we establish asymptotic normality of the weighted estimator, and asymptotic distributions of the Wald statistic under the null and alternative hypotheses, without assuming model selection consistency. We highlight the advantages of the proposed test over some competitive tests through theoretical and empirical comparisons, which demonstrates the local optimality of the proposed test. Furthermore, we prove that when variable selection consistency is achieved, the proposed Wald test is asymptotically equivalent to the oracle test which knows the true model. The superior finite sample performance of our proposed test is demonstrated via extensive simulations. Finally, we use a breast cancer dataset to illustrate the use of our methodology.
报告人简介:蒋学军,南方科技大学统计与数据科学系副教授(长聘),博士生导师,于2009年博士毕业于香港中文大学统计系,2009-2010在港中文从事博士后研究,2010-2013任中南财经政法大学副教授,于2013年07月加入南方科技大学,入选深圳市海外高层次人才孔雀计划(2016),曾获南方科技大学杰出教学奖(2018),深圳市优秀教师(2018),主持和完成国家(广东省)自然科学基金、深圳市基础研究面上项目及技术委托开发项目等10余项。其主要研究方向包括分位数回归、变量选择、假设检验、高维统计推断,金融统计与计量等,在数理统计领域和金融与计量的交叉领域做出了比较深入的研究,已在统计学主流期刊和相关金融、经济等交叉学科期刊上发表SCI&SSCI论文50余篇,授权专利1项,并出版英文教材一部。国内学会任职主要有中国现场统计研究会-教育统计与管理分会副理事长,多元分析应用专业委员会秘书长等。