You are working on a supervised learning model and have more candidate features than you are confident keeping. Some are correlated, some may be noisy, and some may add little signal.
How would you approach feature selection for a machine learning model?
