[jira] [Commented] (MAHOUT-2028) Silhoutte indexing for clustering evaluation

["Trevor Grant (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MAHOUT-2028?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16271526#comment-16271526 ] 

Trevor Grant commented on MAHOUT-2028:
--------------------------------------

This sounds good to me, probably would go in with the "algorithms framework"

"beginner issue" might be a little ambitious, but if you want to give it a shot go ahead.

Anything I can do to help?

> Silhoutte indexing for clustering evaluation
> --------------------------------------------
>
>                 Key: MAHOUT-2028
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-2028
>             Project: Mahout
>          Issue Type: Proposal
>          Components: Algorithms, Clustering
>    Affects Versions: 0.13.0, 0.14.0, 0.13.1, 0.13.2
>            Reporter: Stefano Dalla Palma
>            Priority: Minor
>              Labels: beginner
>
> What about implementing clustering evaluation measure like Silhouette indexing? We could use one or more of the currently implemented distance measures and some clustering algorithm, like k-means, and start measuring the goodness of that clustering, i.e. with Silhouette in this case.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)