数学科学学院学术报告[2025] 103号

(高水平大学建设系列报告1205号）

报告题目: Robust group identification and membership prediction

报告人：陈雪蓉 教授（西南财经大学）

报告时间：2025年 10月 30日 上午10:00-11:00

报告地点：腾讯会议#280-405-138

报告摘要：In this paper, we propose a novel quantile regression modeling framework with a latent group structure, allowing samples drawn from a population consisting of groups with different conditional quantiles along with certain covariates. Different from most conventional modeling approaches for group identification, such as finite mixture models and threshold models, our new model is distribution free, allows the group numbers and group structure of regression coefficients to be the same or different for different covariates. We identify the potential group structure for the quantile regression coefficients using a regularization method and achieve group boundaries recovery through support vector machine (SVM) method, after artificially assigning appropriate labels to different groups. The computational burden of our approach is significantly lower than the pairwise fused regularization method. Moreover, unlike existing regularization methods, our method can analyze and explain the reasons for grouping and predict the group membership of new individuals based on the estimated group boundary. We establish the theoretical properties of the proposed estimators for group parameters and boundary parameters. Simulation studies and real data analysis illustrate that the proposed methods perform well.

报告人简介：  陈雪蓉，西南财经大学“光华杰出学者计划”青年杰出教授、博士生导师，国家级青年人才计划入选者，省级高层次人才入选者。论文发表于JASA, Biometrics, Journal of Business and Economic Statistic等统计学、生物统计学、计量经济学权威期刊。主持国家自然科学基金面上项目2项、青年项目、国家自然科学基金重点项目子课题2项、国家重点研发计划课题子课题1项。曾荣获教育部“第八届高等学校科学研究优秀成果奖青年成果奖”。

欢迎师生参加！

邀请人：王江洲

数学科学学院

2025年10月29日