A Comparison of Three Predictor Selection Techniques in Multiple Regression
- 1 June 1970
- journal article
- Published by Cambridge University Press (CUP) in Psychometrika
- Vol. 35 (2) , 257-271
- https://doi.org/10.1007/bf02291267
Abstract
Three methods for selecting a few predictors from the many available are described and compared with respect to shrinkage in cross-validation. From 2 to 6 predictors were selected from the 15 available in 100 samples ranging in size from 25 to 200. An iterative method was found to select predictors with slightly, but consistently, higher cross-validities than the popularly used stepwise method. A gradient method was found to equal the performance of the stepwise method only in the larger samples and for the largest predictor subsets.Keywords
This publication has 26 references indexed in Scilit:
- Selection of the Best Subset in Regression AnalysisTechnometrics, 1967
- Selection of Variables for Fitting Equations to DataTechnometrics, 1966
- On multiple regression analysisStatistica Neerlandica, 1962
- Statistical principles in experimental design.Published by American Psychological Association (APA) ,1962
- Geometrical Representation of Two Methods of Linear Least Squares Multiple CorrelationPsychometrika, 1961
- Some Multiple Correlation and Predictor Selection MethodsPsychometrika, 1960
- Item Selection Procedures for Item Variables with a Known Factor StructurePsychometrika, 1959
- II. Approximate Linear Restraints and Best Predictor WeightsEducational and Psychological Measurement, 1951
- The Discrimination of Two Racial SamplesPsychometrika, 1950
- The Square Root Method and its Use in Correlation and RegressionJournal of the American Statistical Association, 1945