The ratio of correctly predicted positive observations to the total predicted positives. It answers: 'Of all the instances the model claimed were positive, how many actually were?'
Critical for spam detection (avoiding false positives).
Used when the cost of a false alarm is high.