摘要
设{(Xi,Yi),i≥1}是从取值于Rd×R的总体i≥1中抽取的严平稳、α-混合样本。回归函数m(x)=E(Y|X=x)改良的递归核估计定义为:m2n(x)=∑ni=1YiI(|Yi|<bi)hi-dKx-hi Xi∑nj=1hj-dKx-hj Xj,在适当的条件下,讨论了m2n(x)的渐近正态性。
Let {(Xi, Yi),i≥l}be a strictly stationary and a-mixing sample sequence from i≥l in R^d×R.The improved kernel estimator for regression function m(x)=(Y|X=x) is defined by m2n^-(x)=∑j=1^n hj^-d K(hj^-x-Xj)^--∑i=1^n Yi I(|Yi|〈bi)hi^-d K(hi^-x-Xi)Under suitable conditions, we prove the asymptotic normality of m2n^—(x).
出处
《金陵科技学院学报》
2008年第2期1-6,共6页
Journal of Jinling Institute of Technology