← Glossary Term
F1 score
A single accuracy number that balances how often a system is right against how much it finds.
The F1 score combines two things that pull against each other. Precision asks: of the answers the system gave, how many were correct? Recall asks: of the correct answers that existed, how many did it find? You can make either one look great by sacrificing the other, so F1 blends them into one number.
It shows up constantly in research papers because it stops a system from cheating the metric. A model that flags everything scores perfect recall and terrible precision, and F1 punishes it accordingly. Higher is better, and 100 is a perfect score.
Mentioned in
Not referenced in any article yet.