Github user srowen commented on the issue:
https://github.com/apache/spark/pull/12896
OK. I can see that this is a bit of a band-aid, but might still be
implemented as part of the beginnings of a more general 'cold start'
configuration. As long as the API itself makes sense in the long term, and it
seems so.
"drop" really drops any prediction that's NaN, not necessarily only for
unknown users and items. I suppose the assumption is that those are the same
cases, which is probably true in practice now.
What would it mean if a method was later added to make point predictions?
you'd get nothing back, or maybe this would be considered synonymous with NaN
in this case and get NaN back.
It's a little stretch to implement the required behavior here as a
cold-start strategy but not terrible. If you're OK with it go for it. I think
there are a few more comments to resolve about the `@since` tags
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
