Discrimination
The ability of a probabilistic prediction system to distinguish between easy and difficult samples, that is, whether it can effectively separate questions of differing difficulty. It is an evaluation dimension independent of calibration.
Timeline (last 90 days)
Calibration and discrimination are two independent metrics: discrimination measures the model's ability to rank positives ahead of negatives (such as ROC-AUC), while calibration measures whether the model's output probabilities match true frequencies; commonly used calibration measures include Expected Calibration Error (ECE) and reliability diagrams, and correction methods include Platt scaling, isotonic regression and temperature scaling.
A system can have high overall accuracy but be completely uncalibrated, and it can be well calibrated but have low discrimination power; calibration and discrimination power are two independent dimensions.
All Facts (2)
Calibration and discrimination are two independent metrics: discrimination measures the model's ability to rank positives ahead of negatives (such as ROC-AUC), while calibration measures whether the model's output probabilities match true frequencies; commonly used calibration measures include Expected Calibration Error (ECE) and reliability diagrams, and correction methods include Platt scaling, isotonic regression and temperature scaling.
50%UnverifiedA system can have high overall accuracy but be completely uncalibrated, and it can be well calibrated but have low discrimination power; calibration and discrimination power are two independent dimensions.
50%