Github user manishamde commented on the pull request:
https://github.com/apache/spark/pull/79#issuecomment-36755664
Thanks Sean.
Multi-class classification and feature importances are important features
that will be added soon. We implemented a minimal feature set since we wanted
to focus on functional accuracy and (weak and strong) scaling. Now that we are
satisfied on that front, I am sure these features will be added soon. It's a
fairly big PR in terms of code size so I prefer to avoid adding any more
features to the basic implementation.
Also, we have plans to add ensemble trees (random decision forests,
boosting, etc.) soon to mllib.
Finally, even though mllib lacks this functionality just yet, one could
always implement a bank of one-versus-all classifiers as a workaround to handle
the multi-class classification problem. At the same time, I agree its important
to add this functionality to the classification algorithm itself and will be
added soon.
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
