报告人：王明秋(曲阜师范大学)

时间：2023年11月17日 15:00-

地点：理科楼LA108

摘要：Individualized modeling has gained increasing importance in various applications. However, most of existing studies fail to accommodate the skewed or asymmetrically distributed data with heteroscedasticity. In the context of longitudinal data, this paper investigates individualized modeling by combining the asymmetric least squares loss with the MDSP method. This method enables the identification of subgroups where individuals share similar effects of covariates and allows for the selection of different important variables for different individuals. Meanwhile, the proposed method effectively addresses heteroscedasticity issues, and captures a more comprehensive distribution characteristic compared to ordinary least squares regression. The paper establishes the theoretical properties, including the consistency of parameter estimator and its oracle property. To optimize our approach, we develop an algorithm that combines the cyclic coordinate descent and alternating direction method of multipliers algorithm. Simulation studies and a practical example are conducted to demonstrate the superiority of our proposed approach.

简介：王明秋，博士，曲阜师范大学统计与数据科学学院教授, 研究生导师。主要研究方向高维数据分析、大数据子抽样。中国现场统计研究会统计调查分会常务理事、试验设计分会理事、数据科学与人工智能分会理事。先后主持国家自然科学基金面上项目、青年基金和多项省部级基金。

邀请人：夏小超

欢迎广大师生积极参与！