统计与数据科学藕舫讲坛:特邀南京审计大学孔新兵教授作学术报告

发布单位:数学与统计学院(公共数学教学部)创建者:贾继红发布时间:2024-12-04浏览量:10

报告题目:Semi-parametric Quantile Matrix Factor Model

报告人: 孔新兵 教授

报告时间:2024125日(周四)16:00-17:00

报告地点:藕舫楼724

主持人: 曹春正 教授

报告摘要:In this paper, we introduce a semi-parametric quantile factor model to characterize the quantiles of the entries of the matrix observations. The row and column factor loading matrices are allowed to depend on some closely related covariates in nonparametric way, with the unexplained parts modeled by constant loading matrices. The nonparametric loading functions, constant loadings, factor matrices are estimated via minimizing the unsmoothed or smoothed check loss function under constraints. The asymptotic results are derived. For model checking purposes, we present two novel specification tests, one for the existence of a level positive loadings and one for the necessity of a nonparametric component that is related to extra covariates. Our real data analysis demonstrates that the factor structure varies at different quantiles, and the nonparametric component does exist showing different patterns across quantiles.

报告人简介:

孔新兵,南京审计大学教授、博士生导师,入选国家高层次青年人才计划,国际统计协会(ISI)当选会员、国际数理统计学会会员,主要研究兴趣为高频与髙维数据统计推断与机器学习,在统计学与计量经济学顶级期刊Annals of StatisticsJournal of the American Statistical AssociationBiometrikaJournal of EconometricsJournal of Business and Economic Statistics等发表论文二十余篇,主持国家自然科学基金项目4项。获第一届统计科学技术进步奖一等奖,江苏省教育系统先进个人(优秀教师)、江苏省“双创博士”、江苏省“青蓝工程”中青年学术带头人。

 

数学与统计学院

江苏省应用数学(南京信息工程大学)中心

江苏省系统建模与数据分析国际合作联合实验室

江苏省统计科学研究基地

2024124