Knowledge Maxxing← Back to topic library
ML MaxxingBeginner60 sec
Recall
Recall is the proportion of all actual positive cases that a model correctly identifies as positive.
Prerequisite: Precision
What matters
Recall matters most when missing a real positive is costly, since it measures how completely the model catches every true case, regardless of false alarms.
Example
If there are 100 real fraud cases and a model correctly flags 80 of them, its recall is 80%, even if it also raised some false alarms.
Common misconception
High recall does not mean the model's positive predictions are trustworthy; a model can achieve perfect recall by flagging everything as positive.
Recall question
How could a model achieve 100% recall without being useful?
Keywords and related concepts
recallsensitivityfalse negativesprecision