摘要
对于GMANOVA-MANOVA模型Y=X1B1X'2+ZB2+ε(1)Cov(Y)=∑ In.(2)本文通过两阶段抽样构造出了其均值参数矩阵B_1,B_2的一种估计B(_1N1),B_(2N2),使得这种估计在二次损失下,其风险小于任意预先给定的ε> 0,并且证明了这种估计的停时数(stop number)是渐近有效的(在Chow and Robbins (1965)意义下).
For regression coefficient matrices of the GMANOVA-MANOVA model, this paper gives gives two- stage estimators, such that their risk functions related to arbitrary quadratic loss are bounded above a preassigned constant.Their asymptotic effciency are also discussed.
出处
《应用概率统计》
CSCD
北大核心
1999年第1期83-91,共9页
Chinese Journal of Applied Probability and Statistics
基金
国家自然科学基金