Brier Score
The short answer · 18 words
In short
Proper scoring rule for evaluating probabilistic predictions based on the quadratic deviation between predicted probabilities and actual results.
Flow-alerta.cab (2026): Brier Score. https://flow-alerta.cab/en/glossar/brier-score. Editorially reviewed on: 09 October 2026.
How much detail would you like?
Practical
Relevant for forecasts, risk and classification systems, if not only the class, but the quality of the output probability is important.
Example scenario: Two failure models provide probabilities; the Brier score compares which model makes better probabilistic predictions overall.
Use risk or forecasting models together with calibration curves, discrimination measures and business thresholds.
At a glance
- Core
- In the binary Brier score, the difference between the predicted probability and the actual result is squared for each prediction and then averaged. Lower values are better. The score evaluates probability predictions overall and is not isolated as a measure for calibration only.
- Use
- Relevant for forecasts, risk and classification systems, if not only the class, but the quality of the output probability is important.
- Example
- Example scenario: Two failure models provide probabilities; the Brier score compares which model makes better probabilistic predictions overall.
- Limits
- Interpret Brier score only as a calibration metric; compare scores from different definitions or class problems directly; evaluate hard classes instead of probabilities.
Questions about this term
Is a lower Brier score better?
Yes.
Does it just measure calibration?
No.
Does the model need probabilities?
Yes.
All details and background
In the binary Brier score, the difference between the predicted probability and the actual result is squared for each prediction and then averaged. Lower values are better. The score evaluates probability predictions overall and is not isolated as a measure for calibration only.
ELI5 – Explained simply
If a model says “80 percent probability” and the event occurs or does not occur, the Brier score measures how well such probabilities fit across many cases.
Why it matters in practice
Relevant for forecasts, risk and classification systems, if not only the class, but the quality of the output probability is important.
Example
Example scenario: Two failure models provide probabilities; the Brier score compares which model makes better probabilistic predictions overall.
Pitfalls and limits
- Interpret Brier score only as a calibration metric; compare scores from different definitions or class problems directly; evaluate hard classes instead of probabilities.
Origin & history
The Brier score is used to evaluate probabilistic classification and forecasting models. The current scikit-learn documentation explains its calculation and the delineation from pure calibration.
Use in mid-sized companies
- Use risk or forecasting models together with calibration curves, discrimination measures and business thresholds.
Related terms
- Probability calibration
- Log Loss
Matching Flow-alerta service
AI opportunity analysis & roadmapAI consulting: AI potential analysis & roadmap on tasks, data and prerequisites of a possible first AI use case.Matching tool / use case
View use casesTool: scikit-learn brier_score_loss; Use Case: Evaluation of failure or occurrence probabilities.Sources
- scikit-learn.orgscikit-learn, Probability calibration
- scikit-learn.orgscikit-learn, brier_score_loss
Quellenprüfung: 09.10.2026. Definition und dokumentierte Einordnung abgeglichen; Beispiele beschreiben mögliche Anwendungen.
Editorially reviewed on: