随机删失半参数回归模型参数估计的渐近性质

The Asymptotic Properties of Parametric Estimation in Semiparametric Regression Model under Random Censorship

考虑半参数回归模型Yi=Xiβ+g(Ti)+ei,i=1,2…,n,β∈Rd为未知回归参数,g(·)为[0,1]上的未知Borel函数.利用偏残差法并综合最小二乘法,定义了右删失数据情形下参数β、g(T)的核估计^β 、^g (T),在一定条件下,证明了^β 的渐近正态性,同时得到了^g (T)的最优收敛速度.

For semiparametric regression model,Y_i=X_iβ+g(T_i)+e_i,i=1,2,…,n,we use the least squares and partial residual methods to define the kernel estimations and (T) for β and g(T) under random censored data, the convergence rates of kernel estimations are obtained under suited conditions.