It's unlikely, but possible, that k-means will generate a poor clustering. One approach for dealing with this is to modify method Cluster so that it returns a value that represents how well the data's ...