* assign objects into groups (called clusters) so that objects (cases) from the same cluster are more similar to each other than objects from different clusters.