ABSTRACT
Cluster validity is the evaluation of (fuzzy) partitions in order to establish which partition best explains the unknown cluster structure in a data set X or in a fuzzy sub-class C of X. The partitions which have to be ranked may arise as a result of:
( i) clustering X (or C) with one algorithm at various values of its parameters;
( ii) clustering X (or C) with different algorithms and fixed parameters;
(iii) clustering X (or C) with different algorithms at various algorithmic parameters.
