变系数模型中的逐元B-Spline估计法  

Componentwise B-Spline Estimation for Varying Coefficient Models

作　　者：唐庆国[1] 王金德[2] 

机构地区：[1]解放军理工大学理学院,南京210007 [2]南京大学数学系,南京210093

出　　处：《数学年刊（A辑）》2007年第5期611-620,共10页Chinese Annals of Mathematics

基　　金：国家自然科学基金(No.10671089)资助的项目

摘　　要：给出了一种用于估计变系数模型中未知函数的逐元B-Spline方法,建立了估计量的局部渐近偏差,方差和渐近正态分布,开发了一种快速选择估计量窗宽的方法,通过Monte Carlo模拟研究了估计量的有限样本性质.A componentwise B-Spline method is proposed for estimating the unknown functions in the varying-coefficient models. Explicit expressions for the asymptotic pointwise bias and variance of the estimators are obtained, asymptotic normality of the estimators is also established. A procedure is developed for selecting fast the smoothing parameters of the estimators. Finite sample properties of the procedures are studied through Monte Carlo simulations.

关 键 词：变系效模型 逐元B-Spline估计量 渐近正态性 

分 类 号：O212.1[理学—概率论与数理统计]