Github user mengxr commented on the pull request:
https://github.com/apache/spark/pull/7884#issuecomment-137251694
1. `weight` vs. `sampleWeight`. I don't think `sampleWeight` solves the
problem. I would rather change `weights` in linear models to `coefficients`
instead.
2. Using `RDD[(Double, Double, Vector)]` without weights definitely
introduces storage overhead, especially when the number of nonzeros per
instance is small and the number of instances is very large. This happens in
https://issues.apache.org/jira/browse/SPARK-10354. However, I would prefer
merging the current version and test more. If we have billions of instances,
the number of features are usually not that small.
---
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]
