用Cp统计量选择多元线性模型及估计形式

MULTIVARIATE LINEAR MODELS AND ESTIMATIONS BY Cp STATISTICS

本文根据风险函数,提出了一个多变量场合的Cp统计量作为风险的估计,按Cp最小的原则给出了一种选择多元线性模型及估计形式的方法,重点讨论多变量正交试验的情况。

In this paper, a multivariate Cp statistics as the estimation for a risk function Rp hasbeen proposed. By using which, we can chose an optimal multivariate linear model, particlarly theorthognal design model, and solve some estimations by matrix differential technique.