Interview Guides
You're training a supervised learning model on tabular data and have many candidate predictors, some redundant, noisy, or weakly useful. You want a principled way to reduce the feature set without hurting generalization.
What techniques do you use for feature selection?
