Admissible Variable-Selection Procedures When Fitting Regression Models by Least Squares for Prediction
针对正态线性回归模型，证明了所有变量选择程序在选择最小二乘拟合子集时，在预测平方误差损失下都是可容许的。
Given data with a single dependent variable arising from a normal linear regression model, a variable-selection procedure completely specifies a least squares fit by choosing the subset of the independent variables to include in the model. For loss equal to squared error of prediction, we prove that all variable-selection procedures are admissible for choosing among least-squares fits of the regression model.