Hi everybody, I found an R package calle NbClust which calculates the "optimal" number of clusters based on several indices for (just) the Kmeans and Hierarchical clustering algorithms. I am integrating the result with knime and I beg if someone can validate if the options used in the workflow vs R code are correct.
Thanks for pointing out NbClust. I haven't heard of that before, but it looks like a great tool!
Regarding your question: Yes, that is the correct way to use the results of NbClust in KNIME. Just one thing: You want to use a k-means, but in KNIME, you use k-medoids instead. Is there a reason for that? If you're doing it because you want to specify a distance measure, that's not necessary, because the k-means uses euclidean distance anyway :)