多元线性模型系数岭估计的改进研究

Research on improvement of ridge estimate of coefficients in the multiple linear model

文章对病态的多元线性模型Yn×q^(Xn×pBp×q,Vq×q In×n)提出了c-k型估计,证明了利用Stein式压缩技术可以改进岭估计;同时给出了参数最优值满足的条件,证明了c-k型估计的可容许性。

This paper considers the sick multiple linear model Yn×q-（Xn×pBp×q,Vq×q ×In×n） and gives the c-k class of estimation. The ridge estimation is improved by applying the idea of the James-Stein regression to this class of biased estimation. And the satisfactory condition of the optimal value of the parameter is given. This class of biased estimation is proved to be admissible.