# Are there any search algorithms for feature optimization similar to RFE, but which consider all possible combinations?

Does anyone know any good search algorithms for feature optimization that search through every possible combination to find the optimal combination of features for maximum predictive power? (Permutations are not important).

So far I have been using Recursive Feature Elimination (RFE), which trains a model many times over and each time removes a feature with the least ranking. It is good but not perfect. Say for example we have a,b,c, it then goes to a,b and then a, but does not consider a,c.

There are hundreds of algorithms, but if you know one such, I would really appreciate it! Computational power is not that important, as I only need to run it once!

This is implemented in mlxtend as ExhaustiveFeatureSelector: docs.