Beyond plain F1
F1 is the harmonic mean of precision and recall, weighting them equally. But many tasks care more about one. The F beta score adds a dial.
The formula in words
F beta is a weighted harmonic mean where beta sets how much recall matters relative to precision.
- beta equals 1 gives the familiar F1, equal weight
- beta greater than 1 weights recall more, for example F2
- beta less than 1 weights precision more, for example F0.5
A useful intuition: beta is the factor by which recall is considered more important than precision.
Picking beta
- A disease alert system uses F2 because missing a case is worse than a false alarm
- A product recommendation that must not annoy users uses F0.5 to favor precision
Why harmonic mean
The harmonic mean punishes imbalance. A model with high precision but near zero recall gets a low F score, unlike a plain average.
Key idea
F beta is one knob that blends precision and recall. Pick beta from how much worse a miss is than a false alarm.