Large sample results for varying kernel regression estimates
Large sample results for varying kernel regression estimates
复制标题
不同核回归估计的大样本结果
DOI：
10.1080/10485252.2013.810742
复制
发表时间：
2013
影响因子：
1.2
通讯作者：
Weixing Song
中科院分区：
文献类型：
--
作者：
H. Koul;Weixing Song
变核密度估计是针对正随机变量设计的。与常用的对称核密度估计不同，变密度核估计不受边界问题的影响。当随机变量为正时，本文建立了变密度核估计的渐近正态性和一致几乎处处收敛结果。当协变量为正时，对于回归函数的变核非参数估计也得到了类似的结果。通过模拟研究，讨论了变核回归估计的优缺点。
The varying kernel density estimates are particularly designed for positive random variables. Unlike the commonly used symmetric kernel density estimates, the varying kernel density estimates do not suffer from the boundary problem. This paper establishes asymptotic normality and uniform almost sure convergence results for a varying kernel density estimate when the underlying random variable is positive. Similar results are also obtained for a varying kernel nonparametric estimate of the regression function when the covariate is positive. Pros and cons of the varying kernel regression estimate are also discussed via a simulation study.