Variable selection in robust regression models for longitudinal data

JOURNAL OF MULTIVARIATE ANALYSIS

卷:109

Article

Fan, Yali^1,2  Qin, Guoyou^3,4  Zhu, Zhongyi¹ 

[1] Fudan Univ, Dept Stat, Shanghai 200433, Peoples R China

[2] Shanghai Univ Sci & Technol, Coll Sci, Shanghai 200093, Peoples R China

[3] Fudan Univ, Sch Publ Hlth, Dept Biostat, Shanghai 200032, Peoples R China

[4] Fudan Univ, Key Lab Publ Hlth Safety, Minist Educ China, Shanghai 200433, Peoples R China

关键词: Longitudinal data;    Penalized estimating equation;    Robust method;    Variable selection;   

DOI  :  10.1016/j.jmva.2012.03.007

来源: Elsevier

【 摘 要 】

In this article, we consider variable selection in robust regression models for longitudinal data. We propose a penalized robust estimating equation to estimate the regression parameters and to select the important covariate variables simultaneously. Under some regularity conditions, we show the oracle properties of the proposed robust variable selection methods. A simulation study shows the robustness of the proposed methods against outliers. Moreover, it is found by the simulation study that incorporating the correlation structure into the procedure of variable selection will lead to better performance than ignoring the correlation structure for longitudinal data. In the end, the proposed methods are illustrated in the analysis of a real data set. (C) 2012 Elsevier Inc. All rights reserved.

【 授权许可】

Free