1. Finding a set of natural clusters, and the corresponding description of each cluster. This is relevant if there is the belief
that there are natural groupings in the data. In some cases such as fraud detection, there is interest in finding segmentations
that include outlier clusters (e.g. [1]). For some other cases there may be a preference in finding a segmentation
that includes a pair of clusters that provide the lowest mean and highest mean for each variable. And still for other cases
there may be a preference for finding segmentations that include certain specified variables as important discriminating
variables between the clusters. For each of these cases, it is possible that multiple segmentations could apply. Thus it
might not be appropriate to use a single clustering algorithm and/or parameter setting to find the appropriate set of ‘natural’
clusters