Github user mengxr commented on the pull request:
https://github.com/apache/spark/pull/916#issuecomment-44885870
@concretevitamin This is certainly faster for large sample size. The size
of the candidate set is about `s + O(sqrt(s))` with this implementation, while
the previous implementation uses `c s`.
---
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.
---
