Why is it important to verify the integrity of results, i.e. length(labels)==length(results)?
Isn't it possible to use nmi for clustering with k that is different from number of labels?
I changed the script for my use so that it will accept this. should I upload?