Recall
ELI5 — The Vibe Check
Recall asks: 'Of all the actual YES cases in the world, how many did the AI catch?' High recall means the model finds almost everything it should. A cancer screening model should have very high recall — missing a real case is catastrophic. The tradeoff: to catch everything, you might also flag a lot of false positives.
Real Talk
Recall (sensitivity) is TP / (TP + FN) — the fraction of actual positives correctly identified. High recall indicates low false negative rate. Recall is prioritized when missing a positive is costly (e.g., disease detection, security threats). The precision-recall tradeoff is tuned via classification threshold.
When You'll Hear This
"Recall is more important here — missing a real threat is dangerous." / "We sacrificed precision to maximize recall."
Related Terms
Accuracy
Accuracy is the simplest way to score a model — what percentage of predictions were correct.
Classification
Classification is teaching an AI to sort things into categories. Is this email spam or not? Is this image a cat, dog, or bird?
F1 Score
The F1 Score is the balanced average of precision and recall — a single number that captures both.
Precision
Precision asks: 'Of all the times the AI said YES, how often was it actually right?