文字
缩放
增大字体
减小字体
英国约克大学 张文扬教授:Identification and Estimation in GVC Models with unspecified link functions

([西财新闻] 发布于 :2018-06-19 )

光华讲坛——社会名流与企业家论坛第5013期

 

主题Identification and Estimation in GVC Models with unspecified link functions

主讲人:英国约克大学 张文扬教授

主持人:统计学院 林华珍教授

时间:2018621日(星期四)下午3:00-4:00

地点:www.bte365.com柳林校区弘远楼408会议室

主办单位:统计研究中心  统计学院  科研处

 

主讲人简介:

张文扬教授是英国约克大学统计学首席教授,统计学顶级期刊JASAAssociate Editor,IMS(国际数理统计学会)教材-专著系列的编委会委员,英国皇家统计学会科研分会委员(历史上仅有三位华人担任该委员会委员)。他在非参数统计、时间序列、统计计算、生存分析、结构方程模型等领域发表了很多高质量的研究论文,其中有12篇文章发表在统计学顶级期刊AOS,JASA,JRSSBBiometrika上。他和合作者1999年合作发表在AOS的文章,已经成为半参数变系数模型(VCM)研究中的经典工作,几乎被每一篇有关VCM文章所引用,文章的引用次数超过600次(Google citations)。他2002年发表在Genetics上另一篇论文研究Bayesian近似计算算法,已经被引用1850次。

具体详情请见其个人主页:https://www.york.ac.uk/maths/staff/wenyang-zhang/

主要内容:

The generalised varying coefficient (GVC) models have attracted much attention in modern data analysis. Most existing works are based on the assumption that the link function is given, however, in reality, this assumption does not hold.  To replace the specified link function in GVC models by an unknown function would fundamenatlly change the concept of GVC, it creates serious identification problem, and makes the estimation procedure very hard.  In this talk, I will present a very weak, possibly the weakest, identification condition for GVC models, and build an estimation procedure based on this identification condition.  An algorithm will also be introduced to implement the proposed estimation procedure. I will also show the asymptotic properties of the estimators obtained by the proposed estimation procedure, and some simulation study results.  Finally, I will use the GVC models with unspecified link function together with the proposed estimation procedure to analyse a real dataset, and show some interesting findings.

 


☆该新闻已被浏览: 次★

打印本文】 【关闭窗口