The output of a k-style clustering can easily be used for supervised classification:
we assign any new observation to the closest cluster representative ("center").