回归函数核估计的随机加权法

RANDOM WEIGHTING METHOD FOR THE KERNEL ESTIMATOR OF REGRESSION FUNCTION

本文利用随机加权法的思想，构造了回归函数 ｇ（ｘ）的核估计ｈｎ（ｘ）的随机加权统计量，证明了用随机加权统计量的分布逼近原估计量的误差分布的精度可达到或，其中０＜Ｍｎ→作为应用，我们也构造了（ｘ）和ｇ（ｘ）的置信区间．

In this paper, the random weighting statistic of the kernel estimator of regression function g(x) is constructed by the random weighting method. We prove that the random weighting approximation can attain the accuracy of , where We also construct the confidence intervals of and g(x).