删失数据下回归函数的加权局部复合分位数回归估计

Weighted local composite quantile regression estimation in non-parametric regression model under right-censored data

在右删失数据下,研究了误差具有异方差结构的非参数回归模型,利用局部多项式方法构造了回归函数的加权局部复合分位数回归估计,并得到了该估计的渐近正态性结果,最后通过模拟,当误差为重尾分布时,该估计比局部多项式估计以及核估计表现得更好.

In this paper, the nonparametric regression model with heteroscedastic error is considered under right-cesored data. Based on the local polynomial method, a weighted local composite quantile regression estimator of regression function is constructed. Under appropriate assumptions, the asymptotic normality of the estimator is also established. The simulation studies show that the paper's estimators perform better than the local polynomial estimator and the kernel estimation when the error is the heavy tail distribution.