报告题目：Response Best-subset Selector for Multivariate Regression with Large-Scale Response Variables

主讲人：Hu Jianhua

单  位：上海财经大学

时  间：12月6号10:00-12:00

腾讯ID：643 895 444

密码：123456

摘 要：This work investigates the statistical problem of response variable selection with exponentially large-scale response variables and fixed or diverging number of predictor variables in the frame of multivariate linear regression. A response best-subset selection model is proposed by introducing a 0-1 selection indictor to each response variable and a response best-subset selector is developed by introducing a separation parameter and a novel penalty least square function. The developed procedure can perform simultaneously response variable selection and regression coefficient estimation. The proposed response best-subset selector has model consistency under mild conditions both for fixed and diverging cases of predictor variables. In addition, consistency and asymptotic normality of the fixed case are presented for the corresponding regression coefficient estimators. It is discovered that the Bonferroni test procedure turns to be a special response best-subset selector. In finite-sample simulation studies our response best-subset selector has stronger competitive advantages in keeping balance between higher accurate rates of necessary and unnecessary response variables over four existing competitors in the sense of the Matthews correlation coefficient. A real data analysis demonstrates the effectiveness of the response best-subset selector in an application of identifying dosage-sensitive genes.

简介：胡建华，现为上海财经大学数据科学与统计研究院研究员，博士生导师。自2007年以来，为上海财经大学985经济创新平台海归教师， 2009年上海市浦江人才获得者。1996年曾任职中南大学副教授。1999年曾访问菲尔兹研究所。2000年获中南大学理学博士学位，2007年获加拿大University of Windsor统计学博士学位。长期从事与统计理论与方法、高维数据分析、多元数据分析、空间计量、投资组合与运输管理等相关的科学研究工作。已主持参加完成包括中国国家自然科学基金（重点或面上）和上海市自然科学基金等在内的科研项目十多项。在包括《Biometrika》、《Bernoulli》、《Statistica Sinica》和《中国科学数学》等在内的国际国内著名学术杂志发表学术论文四十多篇，专著两本。现为国际统计学权威杂志《Journal of Multivariate Analysis》副主编，中国环境统计学会大数据科学分会常务理事，多个国际统计学会会员。