Github user MLnick commented on the pull request:
https://github.com/apache/spark/pull/12577#issuecomment-216645240
@sethah @holdenk @jkbradley I thought about this some more. I can't
realistically think of a scenario apart from the ALS one where handling NaNs in
the evaluator is desirable.
So actually I think this should rather go into ALS itself - I'll call the
param something like `unknownUserItemStrategy`. The default can be to return
NaN as it does currently. We can make an option (perhaps called "skip" or
"filter") that filters out NaNs in the prediction DF. This would allow this
option to be used in cross-validation. This will also make it extensible for
future potential additions such as using the "average user" factor, or
whatever other strategy.
---
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]
