非参数回归函数最近邻估计误差分布的随机加权逼近

A RANDOM WEIGHTING APPROXIMATION OF ERROR DISTRIBUTION FOR NEAREST NEIGHBOR ESTIMATOR OF NONPARAMETRIC REGRESSION FUNCTION

本文研究回归函数的最近邻估计的分布逼近问题．在一定条件下得到了最近邻回归估计误差的逼近分布，且逼近的精度比正态逼近精度更高．

In this paper, we study the estimate problem of error distribution for the nearest neighbor estimator of regression function. Under certain conditions, its approximating distribution is obtained, and this approach is more precise than normal approximation based on Berry-Esseen inequality.