2

I am new to R, I would like to ask if there is a way to set the minimum number of observation per cluster in R. I am currently using k-means.

Sometimes my cluster, looks like this:

Clusers: 1    2   3    4
         762 24  553  4013

But I want the distribution per cluster to be a little more balance or at least have a minimum of 100 observation per cluster.

Thanks in advance.

jbest
  • 640
  • 1
  • 10
  • 28
  • Does this make sense?? In your example suppose the centroids of clusters 1, 3, and 4 were closer to each other than to the centroid of cluster 2. Then imposing your condition would result in one cluster (e.g., no clustering at all). – jlhoward Nov 20 '14 at 18:16
  • It may not "make sense" from a clustering standpoint (a group could have items in it that are closer to what would be another cluster), but if you don't care about that, it could nonetheless be useful. – Eden Dec 07 '22 at 17:34

0 Answers0