摘要
本文提出了增长曲线模型回归系数的一种新的有偏估计——组合主成分估计,在一定条件下证明了此估计优于最小二乘估计且是可容许估计,进一步,给出了这种估计与文献[4]中提出的主成分估计的关系。
In this paper, a new biased estimate, which is called the combinatorial principal components estimate (CPCE), is proposed for regression coefficient in growth curve model. It is shown that CPCE is superior to least squares estimate, and is admissible eatimate. Furthermore, a relation between CPCE and principal components estimate given in [4] is derived.
出处
《广西民族大学学报(自然科学版)》
CAS
1997年第1期4-7,共4页
Journal of Guangxi Minzu University :Natural Science Edition
关键词
增长曲线模型
均方误差
容许性
组合主成分估计
Growth curve model
Mean square error
Admissibility
Combinatorial principal components estimate