Github user jkbradley commented on the issue:
https://github.com/apache/spark/pull/15211
You have a good point about setting user expectations about speed and
scalability. I don't think that the average user needs to understand the
underlying implementation, but performance expectations are important. Let's
go with what you have (LinearSVC).
Side note: If we wanted to match libsvm/liblinear, then we would not add a
new class but would just add hinge loss support to GeneralizedLinearRegression.
I hesitate to do that since GLMs technically require natural exponential
families and adding hinge loss would make it harder to explain algorithm
behavior (such as which evaluation stats are available).
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
