Different clustering results while running CLUTO on a sparse vs. dense matrix representing the same data.
Hi to all of you,
I want to double check if the behavior I am observing is normal or there is some mistake somewhere in my own code.
I am running CLUTO once on the dense matrix representing some data, and later on,
on a sparse matrix representing the same set, generated from the dense matrix. I get different clusters (the same number but different in size and members.) Does that look suspicious?
Thank you very much in advance.
Warm regards.
Submitted by Anonymous on Mon, 2006-12-11 12:52
»
- Login to post comments
RE: Depending on the type of
Depending on the type of clustering that you are using, this behavior is fine. For dense input matrix, Cluto uses -colmodel=none, as -colmodel=idf does not apply for dense matrices.