主题：Optimal response-free cluster subsampling for longitudinal data under measurement constraints
主讲人：王磊 南开大学
主持人：王国长 暨南大学
时间：2026年4月22日（周三）下午16:40-17:40
地点：暨南大学石牌校区经济学院大楼（中惠楼）503室
摘要
Under measurement constraints, where covariates are always accessible but obtaining responses is costly or restricted, we propose a unified response-free cluster subsampling framework for massive longitudinal data, focusing on two aspects. First, when the dimension of covariates is fixed and small, to account for within-subject correlation, we consider cluster subsampling and formulate a response-free weighted quasi-score to obtain the subsample estimator with consistency and asymptotic normality. An optimal cluster subsampling scheme is obtained by optimizing a general criterion that encompasses both A-optimality and L-optimality criteria. To enhance the estimation efficiency, a response-free unweighted estimator is subsequently constructed based on the optimal subsample and a two-step algorithm is devised to facilitate practical implementation. Second, when the dimension of covariates is comparable to or exceeds the subsample size, we further construct a response-free weighted quasi decorrelated score for the preconceived low-dimensional parameter of main interest and derive the optimal subsampling schemes. The resulting unweighted estimator and a two-step algorithm are also proposed. Extensive simulation studies, along with a real-data application,are conducted to empirically demonstrate the effectiveness of the proposed methods.
主讲人简介
王磊，南开大学统计与数据科学学院教授、博导、百名青年学科带头人。研究方向是统计学习和复杂数据分析，已在统计学期刊Biometrika,JMLR,IEEE TIT,AOAS,Bernoulli,JCGS,Statistica Sinica等发表学术论文多篇，主持3项国家自然科学基金和1项天津市自然科学基金项目。
校对|王国长
责编| 彭毅
初审| 姜云卢
终审发布| 何凌云
（来源：暨南大学经济学院微信公众号）
