Some diagnostic results in nonparametric density estimation

被引：10
|
作者：
Kim, C
Kim, W
机构：
[1] Pusan Natl Univ, Dept Stat, Pusan 609735, South Korea
[2] Seoul Natl Univ, Dept Comp Sci & Stat, Seoul 151742, South Korea
来源：
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS | 1998年 / 27卷 / 02期
关键词：
bandwidth; Cook's distance; cross-validation; influential observations;
D O I：
10.1080/03610929808832096
中图分类号：
O21 [概率论与数理统计]; C8 [统计学];
学科分类号：
020208 ; 070103 ; 0714 ;
摘要：
Kernel method is most widely used in nonparametric density estimation and data-driven bandwidth selection such as least squares cross-validation is frequently used to estimate the corresponding bandwidth parameter. In this paper, we define a version of Cook's distance (Cook 1977) to investigate the influence of few observations on the overall shape of the curve, and suggest useful formula to detect influential observations on the estimator of bandwidth. An example based on a real data set is given.
引用
收藏
页码：291 / 303
页数：13
相关论文
共 50 条
12345