删失数据下部分线性模型变量选择方法及应用  

Variable selection method and applicationof partial linear model under censored data

作　　者：王纯杰 王畅 蒋京京 WANG Chun-jie;WANG Chang;JIANG Jing-jing(School of Mathematics and Statistics,Changchun University of Technology,Changchun 130012,China)

机构地区：[1]长春工业大学数学与统计学院,吉林长春130012

出　　处：《吉林师范大学学报(自然科学版)》2025年第2期67-73,共7页Journal of Jilin Normal University:Natural Science Edition

基　　金：国家自然科学基金项目(12301332);吉林省重大科技专项项目(20220301031GX)。

摘　　要：在右删失数据下研究了部分线性模型的变量选择问题.基于Kaplan-Meier估计转化删失响应变量,利用限制性立方样条估计非线性函数,并使用Lasso和弹性网惩罚方法,对线性参数进行了变量选择;通过数值模拟研究评估了参数估计量和非参数函数估计量的性能,并应用于骨髓瘤实例数据中.The variable selection problem for partially linear models under right-censored data was studied.The Kaplan-Meier estimator was used to transform the censored response variable,and restricted cubic splines were applied to estimate the nonlinear function.Meanwhile,Lasso and Elastic Net penalization methods were used for variable selection of the linear parameters.The performance of parameter estimators and nonparametric function estimators was evaluated through numerical simulations.Finally,the method was applied to a multiple myeloma case dataset.

关 键 词：AFT部分线性模型 限制性立方样条 Lasso 弹性网 变量选择 

分 类 号：O212[理学—概率论与数理统计]