协变量随机缺失下部分线性模型的变量选择（英文）  

Variable Selection in the Partial Linear Models with Covariate Data Missing at Random

作　　者：杨宜平[1] 

机构地区：[1]重庆工商大学数学与统计学院,重庆400067

出　　处：《工程数学学报》2014年第1期139-151,共13页Chinese Journal of Engineering Mathematics

基　　金：The National Social Science Foundation of China(11CTJ004);the National Natural Science Foundation of China(11301569;11101452);the Natural Science Foundation Project of CQCSTC(cstc2011jjA00014);the Research Foundation of Chongqing Municipal Education Commission(KJ120504)

摘　　要：考虑协变量随机缺失下部分线性模型,采用惩罚加权最小二乘提出了一种变量选择方法,研究了所提出方法的有限样本性质,证明了非零系数的估计具有Oracle性质.进一步,基于局部线性逼近方法给出了一步稀疏估计.通过模拟研究了所提出方法的有限样本性质.A partial linear model with covariate data missing at random is considered. A strategy for variable selection is proposed by penalizing the weighted least squares. The sampling properties for the proposed procedure are investigated. Furthermore, the proposed estimators of the nonzero coefficients are shown to have the asymptotic oracle property. In addition, a one-step sparse estimator is considered by the local linear approximation for the penalty function. The finite sample behavior of the proposed method is evaluated with a simulation study.

关 键 词：变量选择 最小二乘 缺失数据 Oracle性质 SCAD 

分 类 号：O212.7[理学—概率论与数理统计]