The subselect R package

The subselect package addresses the issue of variable selection in different statistical contexts, among which exploratory data analyses; univariate or multivariate linear models; generalized linear models; principal components analysis; linear discriminant analysis, canonical correlation analysis. Selecting variable subsets requires the definition of a numerical criterion which measures the quality of any given variable subset as a surrogate for the full set of variables. The current version of the subselect package provides eight different criteria. For each available criterion, the package provides a function that computes the criterion value of any given subset. More significantly, the package provides efficient search functions that seek the best subsets of any given size, for a specified criterion.