基于删失数据的变系数模型的小波估计

摘要

Varying-coefficient models with censored data are studied. For censored data, the usual statistical techniques are not directly applicable to this model. First, a general trans formation method is used to transform the response observation into conditionally unbiased modification. Then, wavelet estimator of the coefficient functions is proposed as complete data by wavelet procedure. Asymptotic normality of the proposed estimator is established under the weaker assumptions about the smoothness of the coefficient functions.%研究了删失数据下的变系数模型．由于数据删失，常用的统计方法不能直接应用于此模型．首先利用一种变换方法对响应变量观测值进行条件无偏修正；然后，正如完全数据一样，利用小波方法，给出了系数函数的小波估计．在关于系数函数光滑性的较弱的假设条件下建立了该估计的渐近正态性．