Github user yanboliang commented on the issue:

    https://github.com/apache/spark/pull/12819
  
    @sethah It's a good questions, and we have some concerns:
    * The Dataset based implementation does not bring much performance 
improvement from the test result(even worse in some case).
    * I'm not sure the UDAF API is stable enough and have confidential 
performance.
    * This PR mainly focus on moving NB implementation from mllib to ml package 
and refactor it to support weighted instance. Let's focus on what it should be. 
We can change it to use dataset based UDAF in a separate work later if it's 
necessary. Thanks!


---
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

Reply via email to