A note on Krafft's maximin linear estimator for linear regression parameters
作者:
J. Pilz,
期刊:
Statistics
(Taylor Available online 1986)
卷期:
Volume 17,
issue 1
页码: 9-14
ISSN:0233-1888
年代: 1986
DOI:10.1080/02331888608801903
出版商: Akademie-Verlag
关键词: Primary 62 J 05;secondary 62 C 20;BATES and minimax linear estimators;least favourable priors;restricted para¬meter space;convex design theory
数据来源: Taylor
摘要:
This note reconsiders, from a BAYESian viewpoint, the maximin approach to linear regression estimation with restricted parameter space adopted by KRAFFT [4]. It is shown that krafft's maximin turns out to be minimax whenever there exists a degenerate (one-point) measure which is least favourable among all prior distributions on the parameter space.
点击下载:
PDF (1346KB)
返 回