随机删失场合半参数回归模型的误差方差估计  

Variance Estimation of Error in a Semiparametric Regression Model with Censored Data

作　　者：万树平[1] 刘亚芳[2] 涂国平[1] 

机构地区：[1]南昌大学数理与管理科学学院,江西南昌330047 [2]南昌职业技术师范学院,江西南昌330013

出　　处：《南昌水专学报》2002年第1期6-11,共6页Journal of Nanchang College of Water Conservancy and Hydroelectric Power

基　　金：国家自然科学基金资助项目(7996003).

摘　　要：考虑半参数回归模型Y=X'β+g(T)+e,其中(X,T)为取值RP×犤0,1犦上的随机向量(p≥1),β为p×1的未知参数向量,g为定义于犤0,1犦上的未知函数,e为随机误差,Ee=0,Ee2=σ2>0,且与(X,T)独立,Y被一个与之独立的随机变量V所截,此时仅能观察到Z=min(Y,V),δ=I(YV).综合最近邻法和最小二乘法,定义了β、g和σ2的估计量,在适当的条件下证明了σ2的估计量的渐进正态性.This paper studies the parameter esimation in a semipatametric regression model with censored data.The semiparametric regression model:Y=X′β+g(T)+e,was taken into account,where(X,T)is a random vector on RP×,βis p-dimensional unknown parameter,(g) is a unknown function on,e is radom error with meaning 0 and variance σ2,(X,T) is independent of e.and Y may be censored and therefore not completely observable,where the censoring variable V and the variable Yare independent mutally,We can estimate β,gand σ2 by the nearest neighbour and least square methods.As a result,the asymptotic normality of σ2 is obtained under some suitable conditions.

关 键 词：半参数回归模型 删失数据 渐进正态性 误差 方差估计 

分 类 号：O212.1[理学—概率论与数理统计]