Metadata
Recommender Systems Evaluation
  • Offline evaluation
  • ROC-AUC
  • PR-AUC

Recall @K

It gives a measure of how many of the relevant items are present in top K out of all the relevant items, where K is the number of recommendations generated for a user.

Precision @K

It gives a measure of “out of K” items recommended to a user and how many are relevant, where K is the number of recommendations generated for a user.

F1 @K

F1 Score is a combination of Precision and Recall using harmonic mean. This is the same as the regular F1 score and does not differ in the context of the recommendation systems.


Considerations:

  • Popularity Bias
  • Position bias
  • Degenerate Feedback loop