Github user srowen commented on the issue:
https://github.com/apache/spark/pull/14551
Anyway, back to the actual content -- the actual test assertions in the
failing test are correct, in that it's obvious that the first two points and
second two points should cluster together. I expect that just picking another
random seed or two would make it pass. Although that's a little clunky as a
solution, because the test should probably be more robust, it's at least no
less robust for your change. You can also try a tighter convergence tolerance
or more iterations. Anything like that is all that I think we should ask of
this PR.
Does the GaussianMixture doctest also fail? I think similar logic would
apply. In that test, clearly the 3 successive pairs of points are meant to
cluster together. Actually, the first assertion looks wrong because it says the
first two don't cluster together. The others look right. Perhaps your fix
actually makes the corrected assertion pass, if that was the failure.
---
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]
