Evaluation
Recall
Recall is the fraction of actual positives that the system finds.
How it works
It answers how much of the positive class was recovered. Recall matters when missed positives are costly.
A concrete example
Illustrative: 72 true positives and 8 false negatives give recall 72/80 = 0.90.
Do not confuse it with Precision
recall measures coverage of positives; precision measures purity of positive predictions.