报告人: 张文杨 (University of York，UK)

时  间: 2018年4月16日  15:00---16:00

地  点: 理科楼LA106

摘  要:In multilevel modelling for clustered survival data, to account for the difference among different clusters, a commonly used approach is to introduce cluster effects, either random or fixed, in the modelling. The modelling with random effects may lead to difficulties in the implementation of the estimation procedure for the unknown parameters of interest, because numerical computation for multiple integral may become unavoidable when the cluster effects are not scalars.  On the other hand, if fixed effects are used, there would be a danger of having estimators with big variances, because there are too many nuisance parameters involved in the model used. In this talk, I will show a new approach of multilevel modelling for clustered survival data. The proposed modelling does not have the potential computational problem which the modelling with random effects does, it also involves far less unknown parameters than the modelling with fixed effects. Some asymptotic properties of the proposed modelling and intensive simulation study results will be presented to demonstrate the advantage of the proposed method. Finally, I will apply the proposed method to analyse a data set about the second-birthinterval in Bangladesh. The findings are quite interesting.

报告人简介: 张文扬教授是英国顶尖大学约克大学的统计学首席教授，主要从事大数据分析，金 融数据分析，高维数据分析，非参数建模、时间序列分析、空间数据分析，多层次建模，生存 分析，结构方程模型等方向的研究。曾先后在英国伦敦政治经济学院、英国 Kent 大学、英国 Bath 大学、英国 York 大学任教，现为英国 York 大学统计学首席教授。他曾是英国皇家统计学 会科研委员会委员(历史上仅有三位华人担任该委员会委员)，曾经连续担任三届统计学三大国 际顶尖期刊之一 Journal of the American Statistical Association 的副主编。

学院联系人: 穆春来

欢迎广大师生积极参与！