Github user mengxr commented on the pull request:
https://github.com/apache/spark/pull/11018#issuecomment-178934878
@davies Thanks for plotting the histogram! The mean estimator is apparently
not good here due to outliers. Median seems more stable and I don't think we
have enough numbers to confidently tell whether best time or median is better.
Given the current data scale, it seems that `a` (or `b`) are significantly
large for us to ignore `W` in estimating the ratio `a/b`. So I would go with
median and increase the data size if necessary in order to reduce the effect of
`W`. Does it sound good?
---
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]
