基于奇异设计矩阵的岭估计及其性质

Ridge estimator and its properties on the singular design matrix

基于设计矩阵是奇异矩阵的线性模型,讨论线性模型系数参数岭估计的优良性。对于可估函数在均方误差意义下,得到岭估计优于最小二乘估计的性质。而且在二次损失函数下岭估计具有可容许性。

Optimality of ridge estimator of linear model coefficient parameters on the singular design matrix is explored. For the estimable function, it is shown that the properties of ridge estimator superior to least squares estimator under mean square error. Ridge estimator is admissibility under quadratic loss function.