摘要
本文研究回归函数的ｋｎ－近邻估计的渐近性质，得到了回归函数的ｋｎ－近邻估计的渐近正态性和它的Ｂｏｏｔｓｔｒａｐ统计量的相合性. 在高阶矩存在的条件下，我们证明了回归函数的ｋｎ－近邻估计的Ｂｏｏｔｓｔｒａｐ逼近比正态逼近更精确．
In this paper, asymptotic properties of kn-nearest neighbor estimator of reg ression function are investigated, its asymptotic normality and Bootstrap consistency are obtained. Under the condition that higher-order moments are existed, we prove that Boot strap approximation of the kn-nearest neighbor estimator of regression function is more presise than normal approximation.
出处
《应用数学学报》
CSCD
北大核心
2001年第3期410-417,共8页
Acta Mathematicae Applicatae Sinica
