F-beta Score Calculator
Calculate the F-beta score with a configurable weight between precision and recall.
Inputs
Beta > 1 weighs recall more heavily; beta < 1 weighs precision more heavily. Beta = 1 is the standard F1 score.
F-Beta Score
0.6316
Beta²
4.0000
Step by step
Beta squared: beta²
2²
= 4.0000
F-beta: (1 + beta²) × (P × R) / (beta² × P + R)
(1 + 4.00) × (0.8 × 0.6) ÷ (4.00 × 0.8 + 0.6)
= 0.6316
How it works
The F-beta score generalizes F1 into a weighted harmonic mean of precision and recall: F_β = (1 + β²) × (P × R) / (β² × P + R). Setting β = 1 recovers the standard F1 score (equal weight). Setting β > 1 (commonly β = 2) weighs recall more heavily than precision, useful when missing positives is costlier than false alarms. Setting β < 1 (commonly β = 0.5) weighs precision more heavily, useful when false positives are costlier.
Formula
F_beta = (1 + beta^2) * (precision * recall) / (beta^2 * precision + recall)
- P
- Precision
- R
- Recall
- \beta
- Beta weight parameter
Frequently Asked Questions
What's the difference between F0.5, F1, and F2?
F0.5 weighs precision twice as much as recall, F1 weighs them equally, and F2 weighs recall twice as much as precision — choose based on which error type (false positive vs. false negative) is more costly for your use case.
Why would I use F2 instead of F1?
F2 is common in domains like medical screening or fraud detection where missing a true positive (a false negative) is much worse than a false alarm, so recall is weighted more heavily in the combined score.
Can I compute F-beta directly from TP/FP/FN instead of precision/recall?
Yes — first compute precision = TP/(TP+FP) and recall = TP/(TP+FN) using the Precision and Recall Calculators, then enter those values here.