摘要
本文采用压缩最小二乘估计Ｂ∧（ｍ）来估计设计阵呈病态时的增长曲线模型回归系数阵Ｂ．通过ｍ值的选取，可使β＾（ｍ）＝Ｖｅｃ（Ｂ∧（ｍ））的均方误差小于β＝Ｖｅｃ（Ｂ）的ＬＳＥβ∧的均方误差．证明了β∧（ｍ）具有可容许性、抗干扰性和有效性，并给出了实际应用中选取ｍ值的方法．
In this paper, compression least squares estimate B∧(m) of the regression coefficient B is considered when the design matrix presents ill condition in growth curve model. The mean square error of the estimate β∧(m)=Vec(B(m)) is less than the mean square error of least squares estimate β∧ of the regression coefficient β =Vec(B) by choosing the parameter m. Admissibility, numerical stability and relative efficiency of β∧(m)are proved. The method of determining in value for practical use is also given.
出处
《广西民族大学学报（自然科学版）》
CAS
1997年第2期13-18,52,共7页
Journal of Guangxi Minzu University ：Natural Science Edition
关键词
增长曲线模型
最小二乘估计
压缩最小二乘估计
均方误差
growth curve model
least squares estimate
compression least squares estimate
mean square error
