Unified Matrix Framework for Classifier Evaluation

This framework unifies discrimination, calibration, and ranking metrics into a single algebraic language using indicator matrices and cell operators. It extends Boolean operations to soft labels via t-norms, bridging the gap between metrics like F1, Brier score, and AUC.