[R] Variable selection.

[Sokratis Alikhanidi]

  Dear colleagues,

Please point me out to a variable selection approach or package in the R environment.
I am particularly interesting in evolutionary/genetic algorithms for optimization of PLS and SVM models.

---
My knowledge is limited by the follows packages:

Package "Subselect" does not seem to be valid, because it handles with the correlation/covariance matrix only.
Therefore, is it applicable to linear models (am I right?).
Can it be used for the non-linear learning approaches? What is its area of application?

Package "gafit" can optimize only continuous parameters in a function, but not the feature selection.
---

I will be glad for any recommendations.

Thank you very much.

Sokratis Alikhanidi.