报告题目:General matching quantiles M-estimation
报 告 人:Yuehua Wu教授,加拿大约克大学
报告时间:2023年11月21日下午14:30
报告地点:曲江校区beat365官方最新版报告厅(教九楼617)
报告摘要:In this talk, we first describe the background/motivation of matching quantiles estimation (MQE). After it, we briefly introduce M-estimation which is a robust alternative to the ordinary least-squares. Then we present a matching quantiles M-estimation (MQME) method. Since the number of variables may be large in many real problems, a `sparse' representation is highly desirable. We thus integrate the proposed MQME with adaptive Lasso for selecting informative variables. In order to compute MQME, we develop an iterative algorithm. Moreover, we show that the (sparse) matching quantiles M-estimator possesses the property of asymptotic consistency. Our simulations demonstrate the efficient finite-sample performance of the proposed method. We end this talk by presenting an illustrative real case study. (Joint work with Dr. S. Qin)
报告人简介: Yuehua Wu,加拿大约克大学统计系教授,1989年获美国匹兹堡大学的统计学博士学位,师从世界著名统计学家C. R. Rao。研究领域广泛,包括空间统计、M-估计、模型选择、变点检测、非参数估计、金融统计等,以及在环境科学、信息科学、计量经济学、生物医学等领域中的应用,目前是国际统计学会的会员。在PNAS(美国国家科学院院刊),Computational Statistics & Data Analysis(计算统计与数据分析)、The Canadian Journal of Statistics(加拿大统计期刊)、Statistica Sinica(泛华统计)、Journal of Multivariate Analysis(多元统计分析)等期刊发表学术论文100多篇。承担加拿大环境署(Environment Canada)、加拿大自然科学基金(NSERC Discovery Grant)等多项重要科研项目。