线性回归模型的一种新估计

A New Estimation on the Linear Regressive Model

为了解决最小二乘估计的设计矩阵之列向量间存在近似线性关系时而出现的性质不稳定,结合广义岭估计,用均方误差矩阵的方法,给出了一种新的有偏估计K-d型估计,并讨论了它的性质,得出两个主要结论.

In order to overcome the instability occurred when the approximately linear relation exists among all the column vectors of the design matrix for least squares estimation, a new biased estimate, i.e. K-d-type estimate, was presented by means of the mean square error matrix and combining the generalized ridge estimate, the property of this biased estimates was discussed, and the two main conclusions drawn.