Home › Glossary › Evaluation › Average precision at k

Intermediate · Evaluation

Average precision at k

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. A metric evaluating ranked results by averaging precision scores at different recall levels up to k, measuring relevance in ordered lists.

Technical Definition

A metric evaluating ranked results by averaging precision scores at different recall levels up to k, measuring relevance in ordered lists.

How it works

Average Precision at k (AP@k) is a performance metric used for tasks that produce ranked lists, such as recommendation systems. It calculates the average of precision values obtained at each point a relevant item is encountered in the ranked list, up to the k-th position. This metric offers a nuanced view of how well a model ranks relevant items.

Further Reading

  • Google ML Glossary