A general feature engineering wrapper for sklearn estimators

View the Project on GitHub


Feature Engineering Wrapper

Few is a feature engineering wrapper that pairs with any ML estimator to generate a readable representation that facilitates learning. It is implemented to pair easily with any scikit-learn-base estimator.

Few uses genetic programming to generate, search and update engineered features. It incorporates feedback from the ML process to select important features, while also scoring them internally.


If you use Few, please reference our publication:

La Cava, W., and Moore, J. A general feature engineering wrapper for machine learning using epsilon-lexicase survival. Proceedings of the 20th European Conference on Genetic Programming (EuroGP 2017), Amsterdam, Netherlands.

(A preprint is available here.)


Check out few_example.py to see how to apply FEW to a regression dataset.