半变系数模型的M估计  

M-estimation on Semi-varying Coefficient Models

作　　者：富伯亭[1] 郑刘锁[2] 

机构地区：[1]山西工程职业技术学院基础部,太原030006 [2]太原理工大学数学系,太原030024

出　　处：《工程数学学报》2009年第4期609-617,共9页Chinese Journal of Engineering Mathematics

摘　　要：本文讨论半变系数模型的稳健M估计。先用局部线性方法给出未知可测函数的一步估计,并讨论其弱一致性和渐近正态性;通过Back-fitting技巧,给出未知参数向量的一般M估计,然后再通过局部M方法给出未知函数的两步估计,并讨论它们的渐近正态性。The robust M-estimates of the Semi-varying coefficient models are discussed. The one-step estimators of the unknown measurable functions are proposed by local linear method, then their weak consistence and the asymptotic normality are given. After the generic M-estimators of the unknown parameters coefficient are given by the back-fitting technique, the two step estimators of the unknown functions are given by local M method. Finally, the asymptotic normality of their M-estimate is investigated.

关 键 词：半变系数模型 M估计 Back-fitting技巧 局部线性方法 渐近正态性 

分 类 号：O212.7[理学—概率论与数理统计]