[jira] [Commented] (MADLIB-1017) OPTICS and/or DBSCAN clustering algo

["Frank McQuillan (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MADLIB-1017?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15485667#comment-15485667 ] 

Frank McQuillan commented on MADLIB-1017:
-----------------------------------------

I emailed Maxence to ask him to describe where the project left off, and he replied: 

"The k-medoids project was completed, save for one small bug in HAWQ, which does not display the correct number of iterations.
Unfortunately, right after the end of the GSoC I moved to Siberia for my studies, and did not have time to investigate the reason of this bug.
The copy of Madlib I worked on is still available here [0].
If I remember correctly, save for this bug in HAWQ, the code is ready to be merged: docs and basic tests have been written. I didn't check if the code still merges correctly with the main repo, though."

[0] http://git.viod.eu/viod/madlib

[~riyer] may be able to add more color as well.

We are interested in other clustering algorithms.  If someone in the community (yourself?) would like to pick this JIRA up then I would certainly support them to do so.  The Pivotal MADlib team may pick this up too but I do not have a timeline for that.

Frank





> OPTICS and/or DBSCAN clustering algo
> ------------------------------------
>
>                 Key: MADLIB-1017
>                 URL: https://issues.apache.org/jira/browse/MADLIB-1017
>             Project: Apache MADlib
>          Issue Type: New Feature
>            Reporter: eric peterson
>
> As a data scientist, I'd like to be able to use DBSCAN or OPTICS clustering algorithms instead of k-means so that I don't have to compute K. There seems to have been some initial work on this, but dont see any final result/progress. 
> https://blog.pivotal.io/pivotal/features/updating-postgresql-based-open-source-machine-learning-library-from-google-summer-of-code
> Any plans to include clustering algorithms beyond k-means?



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)