Feature Importance ranks the relative predictive power or impact of input features in a model. Techniques include measuring how much a feature contributes to reducing impurity in trees, or how model performance degrades when a feature is permuted. It helps interpret and refine models.