COMPEL Glossary / recall
Recall
Recall is a model performance metric measuring the proportion of actual positive cases that the model correctly identifies -- in other words, of all the real positives, how many did the model catch? High recall means few false negatives (missed cases).
What this means in practice
Recall is critical in applications where missing a positive case is dangerous or costly: a disease screening tool with low recall misses patients who need treatment, a fraud detection system with low recall allows fraudulent transactions through, or a quality inspection system with low recall passes defective products. Recall and precision often have an inverse relationship, and the appropriate balance depends on the business context. In high-stakes applications like healthcare or safety, organizations typically prioritize recall even at the cost of more false positives.
Why it matters
Recall is critical in high-stakes applications where missing a positive case is dangerous or costly. A disease screening tool with low recall misses patients who need treatment; a fraud detection system with low recall allows fraudulent transactions through. Understanding the recall-precision tradeoff helps organizations set appropriate performance thresholds that reflect the real-world consequences of missed detections.
How COMPEL uses it
During the Model stage, recall thresholds are defined alongside precision as part of use case success criteria, with the tradeoff explicitly approved by business stakeholders. The Evaluate stage measures recall against acceptance thresholds and monitors for degradation over time. COMPEL requires that high-stakes applications prioritize recall even at the cost of more false positives, documented in the governance artifacts.
Related Terms
Other glossary terms mentioned in this entry's definition and context.