TR multivariate conditional median estimation

被引:2
|
作者
De Gooijer, Jan G.
Gannoun, Ali
机构
[1] Univ Amsterdam, Dept Quantitat Econ, NL-1018 WB Amsterdam, Netherlands
[2] Le CNAM, Math CEDRIC, Paris, France
关键词
kernel; retransformation; robust; spatial conditional median; transformation;
D O I
10.1080/03610910601096270
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
An affine equivariant version of the nonparametric spatial conditional median (SCM) is constructed, using an adaptive transformation - retransformation (TR) procedure. The relative performance of SCM estimates, computed with and without applying the TR-procedure, are compared through simulations. Also included is the vector of coordinate conditional, kernel-based medians (VCCMs). The methodology is illustrated via an empirical data set. The simulations indicate that the TR-SCM estimator is more efficient than the SCM estimator for data generated from asymmetric contaminated trivariate normals. However, when the dimension of the covariates increases the efficiency of the TR-SCM estimator decreases. The TR-VCCM- and VCCM estimators lack efficiency, and consequently should not be used in practice.
引用
收藏
页码:165 / 176
页数:12
相关论文
共 50 条