Github user mengxr commented on the pull request:
https://github.com/apache/spark/pull/8546#issuecomment-141347838
@HuJiayin This basically reverts the behavior back to 1.2. The changes we
made in 1.3 is to avoid recomputing distances between old centers and input
points during initialization. That is why we need `newCenters`. If you test the
current version with a large `k`, you will see the performance difference. Base
on our discussion offline, I think there are not much work to do here. The case
when the new implementation introduces overhead is when the dataset is really
tall and skinny, but we haven't heard negative feedback from practical use
cases yet. Do you mind closing this PR for now? Thanks!
---
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]
