Model calibration Machine Learning; Explainable AI (XAI)
noun phrase
Definition: The property or process by which a model’s predicted probabilities are aligned with observed outcome frequencies, so that stated confidence levels correspond to empirical likelihoods [Pavlovic 2025].
Example in context: “In this work, we perform a comprehensive investigation into machine learning model calibration across 7 open access engineering mechanics datasets.” [Mohammadzadeh et al. 2023]
Synonyms: probability calibration; confidence calibration
Related terms: reliability diagram; temperature scaling; calibration error; uncertainty estimation