Precision and Recall
Precision and recall are two sides of accuracy. Precision asks: of the things the system flagged, how many were right? Recall asks: of the things it should have flagged, how many did it catch? They trade off against each other, so which one matters depends on whether false positives or misses cost you more.
Also known as: precision, recall
Precision and recall pull apart what “accuracy” hides. Precision is how many of the system’s positive calls were actually correct — high precision means few false alarms. Recall is how many of the real positives the system actually found — high recall means few misses. A spam filter with high precision rarely flags good mail as spam; one with high recall rarely lets spam through.
The two trade off. Tune a system to catch more (higher recall) and it flags more things wrongly (lower precision), and vice versa. Which you favor is a product decision, not a math one: for a fraud detector a missed fraud may cost more than a false alarm, so you push recall; for a content filter that annoys users with false blocks, you push precision. The F1 score combines the two into one number, but the honest move is to know which side your use case can’t afford to lose.