摘要
文章研究了可变窗宽的自适应N-W核回归估计,并提出了一种改进的自适应N-W核回归估计。研究表明,在三种N-W核回归估计中,具有可变窗宽的自适应N-W核回归方法比固定窗宽的N-W核回归方法的估计效果更好,对于一个自适应N-W核回归估计量来说,使用算术均值得到的窗宽比使用几何均值得到的窗宽,在估计效果上有更大的优势。
This paper makes a study on the adaptive N-W kernel regression estimation with variable window width, and proposes an improved adaptive N-W kernel regression estimator. The study result shows that in the three kinds of N-W kernel regression estimations, the effect of the adaptive N-W kernel regression method with variable window width is better than that of the N-W kernel regression method with fixed window width, and that for an adaptive N-W kernel regression estimation, the window width obtained by using the arithmetic mean has much greater advantages than that by using the geometric mean in terms of estimation effect.
作者
张颖
Zhang Ying(School of Mathematical Sciences, Jinan University, Jinan 250022, Chin)
出处
《统计与决策》
CSSCI
北大核心
2018年第5期16-19,共4页
Statistics & Decision
基金
济南大学校级科研资助项目(XKY1615)
