多元广义岭估计确定偏参数的两种准则

本文给出广义岭估计β（Ｋ）的偏参数Ｋ选择的Ｄ（Ｑ）准则和均方误差无偏估计的极小化准则，以及偏参数Ｋ的最优解公式．它使广义岭估计β（Ｋ）的均方误差（ＭＳＥ）明显小于 ＬＳ估计β的 ＭＳＥ．从而提高了广义岭估计的拟合精确度．最后给出一实例．

Two criterions for the determination of partial parameter of multivariate generalized ridge estimation

In this paper, the criterion D(Q) which is used for choosing partial parameter K of generalized ridge estimation β(K), the minimization criterion for the deviation-free unbiassed estimation of mean square error and the optimal solution of partial parameter K are given. By our new method, the mean square error of generalized ridge estimationβ(K) is considerably less than that of LS estimation. Thus, we get the generalized ridge estimation more accurate. Finally, we give a concrete example.