Precision and Recall
/prɪˈsɪʒən ənd rɪˈkɔːl/
Metrics for classification: Precision is correct positives / predicted positives; Recall is correct positives / actual positives.
In italiano: Precision e RecallPrecision measures accuracy of positive predictions (avoiding false positives). Recall measures completeness (avoiding false negatives). The F1-score combines both into a single metric.
Examples
- Medical diagnosis (high recall)
- Spam detection (high precision)
- Information retrieval