Condition Numbers and Minimax Ridge Regression Estimators
Condition Numbers and Minimax Ridge Regression Estimators
复制标题
条件数和极小极大岭回归估计器
DOI：
10.1080/01621459.1985.10478180
复制
发表时间：
1985
影响因子：
3.7
通讯作者：
G. Casella
中科院分区：
文献类型：
--
作者：
G. Casella
摘要岭回归最初是为了两个目标而提出的：改善均方误差和系数估计的数值稳定性。给出了极小极大岭回归估计也能改善数值稳定性的条件，这个量可以用要求逆的矩阵的条件数来度量。文中还讨论了用数值稳定性换取极小值的后果。
Abstract Ridge regression was originally formulated with two goals in mind: improvement in mean squared error and numerical stability of the coefficient estimates. Conditions are given under which a minimax ridge regression estimator can also improve numerical stability, a quantity that can be measured with the condition number of the matrix to be inverted. The consequences of trading numerical stability for minimaxity are also discussed.