Github user srowen commented on the pull request:
https://github.com/apache/spark/pull/6761#issuecomment-112537749
Yes, would be good to check that the predicted class is where the max
probability occurs, that they sum to 1 too. I don't know if there's a standard
epsilon but use the surrounding code as a guideline to what seems like the
general practice for this test. In some cases, maybe the toy example in a test
means you know exactly what the probabilities should be and you can verify them
directly, because you do want to check that they're not just some set of
numbers that happens to sum to 1 and is in the right order. If that's not easy
maybe try to make a really trivial example model where the output probabilities
are easy to know.
---
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]
