Can I employ item "masking" in model evaluation to eliminate cold start scenarios?
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 5.1k
- Forks
- 724
- PR merge metrics
- No merged PRs in 30d
Description
What I want to do is generate a train/test split that splits off random user/item interactions into the test in such a way that if that'd leave any given user without any interactions in the training data they'll simply be dropped from the model entirely or disregarded for the purposes of calculating the evaluation metrics. Is this supported somehow?
If not, how hard would that be to do?
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
The issue names no file, test, or entry point to begin with. First determine whether LightFM's train/test split or evaluation entry points support excluding users or items absent from training, then define the expected split and metric behavior as the completion criteria.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100