Github user sethah commented on the issue:
https://github.com/apache/spark/pull/15450
The cases you enumerated are the ones I was thinking of. The changes
introduced here would alleviate those problems, I agree. What I'm wondering is
if this problem still exists in other cases. If Derrick had 1.3M data points
and asked for 10k clusters, and got only 1k unique cluster centers, why did it
happen? Is it common/possible for clusters that start in different locations to
converge to the same point? I'd be interested to replicate this issue, not sure
if I will have the time.
Also, I'm slightly inclined to match scikit or R unless we're certain of a
clear benefit not to.
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
