How to use
- Choose whether to enter counts or direct precision and recall.
- Set beta to decide how much extra weight recall should receive relative to precision.
- Compare F-beta with F1 before you report one score as the headline metric.
Wave 7 classification metrics
Weighted harmonic mean of precision and recall
This page helps when teams already know precision and recall but need a single score that favors one side more than the other.
Inputs
Run a calculation to review F-beta and F1.
Beta controls which side matters more
F-beta is a weighted harmonic mean of precision and recall. Beta greater than 1 rewards recall more heavily. Beta below 1 rewards precision more heavily. Beta equal to 1 becomes F1.
When to use direct-rate mode
Use direct-rate mode when another report already gives precision and recall. Use count mode when you want the page to derive those rates from TP, FP, and FN.
When not to replace the confusion matrix
F-beta is useful as a headline number, but it should not replace the underlying confusion-matrix counts when operational trade-offs must stay visible.
Frequently asked questions
What happens when beta equals 1?
Beta equal to 1 gives the standard F1 score. Precision and recall are weighted equally in that case.
What does beta greater than 1 mean?
Beta greater than 1 gives recall more weight than precision. This is useful when missing positives is more costly than raising false alarms.
Can I enter precision and recall directly?
Yes. Use direct-rate mode when precision and recall are already known. Use count mode when you want the page to compute them from TP, FP, and FN.
Does the share URL include my counts?
No. The share URL stores only lightweight settings such as beta, decimal places, and the selected input mode. Counts stay in your browser.
Related
Comments (optional)
To reduce load, comments are fetched only when needed.