Can SAS Enterprise Miner - Cluster node - take coordinate matrix as input?

Question

I am using SAS proc distance to create a distance matrix. I wanted to know if SAS EM cluster node can use this matrix to create perform K mean clustering?

Answer 1

K-means must be used with squared euclidean. Because technically, it minimizes variances, not distances.

Furthermore, it can obviously not be used with distance matrixes, because it does not need object-to-object distances, but object-to-mean distances (and the means change).