Key points are not available for this paper at this time.
Abstract The problem of selecting the best subset of predictor variables in a linear regression model has led to the development of a number of criteria for choosing between contending subsets. Unfortunately, the properties of these criteria are generally not understood and the relative merits of contending criteria are not clear. In this note, a number of criteria are reviewed and a cooperative effort to resolve the problem is recommended. Key Words: Regression AnalysisElimination of VariablesOptimal Subsets
R. R. Hocking (1972) studied this question.