benfred / implicit

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

fix train_only_size #652

Open bos1988 opened 1 year ago

bos1988 commented 1 year ago

With direct multiplication of train_only_mask & candidate_mask, the proportion of train_only_size is overestimated To save the share of train_only_size, you need to filter by users from candidate_mask with an adjusted selection share