Do feature selection in 4 different ways: 1) rank models by AICc, 2) leave one out or k-fold cross validation, 3) Bayesian variable selection, and 4) use caret's recursive feature elimination. For better or worse, they all give different answers (at least predicting mpg with mtcars dataset).
RobertSWalker/feature-selection-4ways
Do variable selection in 4 different ways: 1) rank models by AICc, 2) leave one out or k-fold cross validation, 3) Bayesian variable selection, and 4) use caret's recursive feature elimination.
R