k-Medoids on large datasets - CLARANS ?

Hi,

As k-Medoids is very computationally expensive, especially on large datasets. Please can the CLARANS implementation of k-Medoids be added to the KNIME toolset to give an extra option of being able to do k-Medoid type clustering with big data.

CLARANS uses O(n2) complexity as opposed to O(p(n-p)2), making it much more viable.

Thanks

Simon.

Hi Simon, 

Clarans is on our radar already.  I'm not sure when we will be able to squeeze it in, but I've flagged this post to add a bit of bouyancy. 

Cheers,

Aaron

 

Thanks for the update Aaron, most appreciated.

simon.