Precision at K

P@K

$P@K=\frac{|\text{relevant documents in top }K|}{K}$

Simple, intuitive: “of the top K results, how many are relevant?” Ignores documents below rank K and doesn’t consider the order within the top K.

Appears In

• IR-L04 - Evaluation
• RS-L01 - Course Overview & Introduction
• RS-L02 - Evaluation Beyond Accuracy
• RS-L03a - Sequential Recommendation Models
• RS-L03b - From LLMs to LRMs
• RS-L04 - Generative Recommendation