New Classification algorithm plus bug fix

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

New Classification algorithm plus bug fix

luca.gerevini
Hi, I've implemented the pyod.models.knn (python) algorithm in weka3.8.3 (it's a knn version for the anomaly detection). I would like to know how can i made a public release in orther to share the new classificator with the community, even an unofficial release would be great.
Furthermore working on the implementation of the "anomaly Knn" i found out a bug that didn't allow to select any others distance function but Euclidean for the BallTree nearest neighbour search algorithm.
I fixed the bug and now i would like to share it.

Thanks :)
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: New Classification algorithm plus bug fix

Peter Reutemann
> Hi, I've implemented the pyod.models.knn (python) algorithm in weka3.8.3 (it's a knn version for the anomaly detection). I would like to know how can i made a public release in orther to share the new classificator with the community, even an unofficial release would be great.

You can create a package of your algorithm:
https://waikato.github.io/weka-wiki/packages/

Then you can do a pull request on the "unofficial.md" file from the
Weka wiki repo:
https://github.com/Waikato/weka-wiki
https://github.com/Waikato/weka-wiki/blob/master/docs/packages/unofficial.md

> Furthermore working on the implementation of the "anomaly Knn" i found out a bug that didn't allow to select any others distance function but Euclidean for the BallTree nearest neighbour search algorithm.
> I fixed the bug and now i would like to share it.

I'm not familiar with the BallTree neighborhood search, but its
assumptions may be based on using the Euclidean distance. At least the
set-method in the BallTreeConstructor class is called
"setEuclideanDistanceFunction". Maybe Eibe knows more about that?

Cheers, Peter
--
Peter Reutemann
Dept. of Computer Science
University of Waikato, NZ
+64 (7) 577-5304
http://www.cms.waikato.ac.nz/~fracpete/
http://www.data-mining.co.nz/
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html