Definition
Plain language
A single score that balances how often you find what's there against how often you raise false alarms.
As stated in the literature
The harmonic mean of precision and recall, used as a single-number summary of classification quality.
Also called: F1 score
Why it matters: It captures the tension between missing things and crying wolf in one number, which makes it the default summary metric for classification tasks.
For example, a spam filter that catches 90% of spam but flags lots of real mail as spam will have a lower F1 than one that catches 80% but rarely misfires.
Heard on the show
“And the secondary measure, slot F1 — how many individual facts it pinned correctly — went from seventy-nine to ninety-one.”Episode 181 — How to Backpropagate Blame Through a Team of Chatbots — And When It Backfires