benfred / implicit

Fast Python Collaborative Filtering for Implicit Feedback Datasets
https://benfred.github.io/implicit/
MIT License
3.57k stars 612 forks source link

What is a good enough Precision@K and MAP@K value? #709

Open akash-kwiff opened 9 months ago

akash-kwiff commented 9 months ago

Not really an issue, I am just trying to understand what Precision@K and MAP@K values are considered good enough? I have tried tuning the hyper-parameters and experimented with cleaning my data (removing user-item interactions that are really low and very high) but the precision@10 never seem to go beyond 0.23 and MAP@10 - 0.15

I understand "what is good" depends on my scenario, but is it possible to achieve higher results for Precision@K / MAP@K?