Preferred Label : Model Calibration;
Définition CISMeF : The process of adjusting predicted probabilities generated by an ML model to ensure
that they accurately reflect the observed frequencies of events or outcomes in the
real world. For example, if a model is well calibrated and predicts 20% probability
of breast cancer for a patient, then the observed frequency of breast cancer should
be approximately 20 out of 100 patients that were given such a prediction by the model.
Source: Adapted from Chen, W., Sahiner, B., Samuelson, F., Pezeshk, A., & Petrick,
N. (2018) Calibration of medical diagnostic classifier scores to the probability of
disease. Statistical Methods in Medical Research, 27(5). https://doi.org/10.1177/0962280216661371;
Origin ID : M000766690;
Related record
The process of adjusting predicted probabilities generated by an ML model to ensure
that they accurately reflect the observed frequencies of events or outcomes in the
real world. For example, if a model is well calibrated and predicts 20% probability
of breast cancer for a patient, then the observed frequency of breast cancer should
be approximately 20 out of 100 patients that were given such a prediction by the model.
Source: Adapted from Chen, W., Sahiner, B., Samuelson, F., Pezeshk, A., & Petrick,
N. (2018) Calibration of medical diagnostic classifier scores to the probability of
disease. Statistical Methods in Medical Research, 27(5). https://doi.org/10.1177/0962280216661371
