Typically defect prediction models rely on supervised learners, which use a labeled training dataset to learn the association between measured entity properties (e.g., metrics calculated on files or methods), with the defect proneness of these entities.