Admissible clustering procedures

Abstract

SUMMARY The present paper extends the results of Fisher & Van Ness (1971). Table 1 more than doubles the size of the corresponding earlier table. In addition a theorem is proved showing an incompatibility between sum of squares methods and the very appealing well-structured admissibility. Finally, a section is included discussing equal admissibility which arises from the idea that it is impossible to distinguish two clusters which have the same population distributions.