Github user avulanov commented on the pull request:
https://github.com/apache/spark/pull/1155#issuecomment-48035031
@mengxr Thanks for your review! I've addressed all suggestions, except: 1)I
used Experimental annotation for the whole class only. Do I need to annotate
each method? 2)I removed the prefix "micro" from 3 methods, but I didn't remove
recall and fMeasure. I think it might be confusing that there is precision and
no recall and fMeasure methods despite their equality.
I think handling the models with raw score should be done in a separate
class, that knows the model output. Some models output the number between 0 and
1 (Bayesian), some output either 0 either 1 (Logistic Regression), the other -1
or 1 (SVM). Ultimately, I would like that all models return class or no_class
by default and the raw score optionally. What do you think?
Could you review my pull request for multilabel measures?
https://github.com/apache/spark/pull/1270
---
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.
---
