Brier score
Brier score: The mean squared difference between predicted probabilities and observed outcomes used to assess probabilistic accuracy. It includes calibration and resolution but should be compared with baselines and prevalence. Lower values indicate better forecasts. This perspective connects the definition with practical behaviour in artificial intelligence systems.
In italiano: Brier ScoreLegge V · Overfitting, bias e collasso dei modelliIt includes calibration and resolution but should be compared with baselines and prevalence. Lower values indicate better forecasts.