In Information retrieval, precision and recall are performance metrics measuring the quality of a retrieval set. Precision is the fraction of retrieved items that are relevant; recall is the fraction of relevant items that are retrieved.

Note that the term “precision” is ambiguous because it has an unrelated meaning; see below.

For binary classification

Both precision and recall have equivalent metrics for binary classification.

Precision is equivalent to the probability that a positive prediction corresponds to a positive outcome; i.e., it is the positive predictive value:

Recall is equivalent to the probability that a positive outcome is predicted as such; i.e., it is equivalent to sensitivity:

For multi-class classification

See main article.

The term “precision” is ambiguous

Precision used in this context should not be confused with precision in relation to accuracy. There is no deep connection; it’s just an unfortunate naming convention.