# F1 Score

> One score that blends precision and recall. It stays low if either one is low.

**Field:** Artificial Intelligence  
**Category:** Evaluation & Benchmarks  
**Also known as:** F1, F-Score, F-Measure

## Think of it like

It is like a grade for a team of two. If one does badly, the team grade drops a lot.

## Definition

The F1 score is one number that blends precision and recall. It is their harmonic mean, a kind of average that stays low if either one is low. It runs from 0 to 1, where 1 means perfect precision and perfect recall. A model with 0.9 precision and 0.1 recall gets an F1 of only 0.18. It is often used when the cases are lopsided and accuracy would fool you.

## Example

Lucy compares two fraud models. Model A has an F1 of 0.62 and model B 0.71, so she picks B, then checks its precision and recall one by one.

## Why it matters

It gives one number to compare models with. And it cannot be gamed by pushing just one side up.

## Common mistake

Using F1 when misses and false alarms cost very different amounts. Then look at precision and recall on their own.

## Three principles

- **Blends two scores.** Precision and recall.
- **Low if one is low.** No hiding.
- **0 to 1.** 1 is perfect.

## Related terms

- [Precision (Machine Learning)](https://learn.benibix.com/term/ai/precision)
- [Recall (Machine Learning)](https://learn.benibix.com/term/ai/recall)
- [Accuracy (Machine Learning)](https://learn.benibix.com/term/ai/accuracy)
- [Confusion Matrix](https://learn.benibix.com/term/ai/confusion-matrix)

![Three models with bars for precision, recall and F1. The model with 0.9 precision and 0.1 recall gets an F1 of only 0.18.](https://learn.benibix.com/diagrams/ev-f1-score.svg)

---

Source: https://learn.benibix.com/term/ai/f1-score
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-19
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
