[
https://issues.apache.org/jira/browse/MATH-897?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Thomas Neidhart updated MATH-897:
---------------------------------
Attachment: MATH-897-review.patch
Hi Reid,
please find attached a review of your patch with the following changes:
* minor javadoc updates
* use more specific exceptions
* the data points are now an input to the cluster method instead of the ctor
(similar to the kmeans++ clusterer)
* fix the expandCluster to match the algorithm on wikipedia (last if + the way
how to determine if a point is already part of a cluster)
* change the visited set to a map to mark also if a point is part of a
cluster, see above
* improve the merge method
* make the call to cluster thread-safe, similar to the kmeans++ clusterer.
This may not be necessary, but I prefer it like this usually.
What do you think about the changes?
btw. for the future, if there are multiple clustering algorithms, we should
think about a unifying interface.
> Add DBScan clustering algorithm
> -------------------------------
>
> Key: MATH-897
> URL: https://issues.apache.org/jira/browse/MATH-897
> Project: Commons Math
> Issue Type: Sub-task
> Reporter: Thomas Neidhart
> Priority: Minor
> Fix For: 3.2
>
> Attachments: MATH-748.txt, MATH-897-review.patch
>
>
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
