5

I'm trying to use the EM cluster algorithm, provided by the software Weka, to classify my data and it only finds one cluster.

  • Could I interpret this as there are no ways to distinguish the instances in my sample?

This is a result that is coherent with others analysis that I'm doing to the data, but I don't know if I could use a cluster analysis to state that.

Rafael
  • 185
  • 4

1 Answers1

1

Two assumptions here: 1) Weka's finding the number of clusters (k) without issues, and 2) I believe EM uses mixtures of Gaussians which means the clusters need to be round/elliptical.

So, given that Weka's algorithm is finding the best k, the answer would be that using round/elliptical clusters, the most likely clustering is one group. That doesn't mean that your data doesn't cluster at all (using other shapes, essentially).

Wayne
  • 19,981
  • 4
  • 50
  • 99