Abstract
Cluster analysis has achieved growing recognition as a useful tool in the analysis of large sets of multivariable chemical data. A new family of clustering algorithms is discussed that appear to offer several advantages over more traditional approaches. These algorithms are based upon the concept of permitting data samples to possess partial membership in different clusters, thus defining a so-called “fuzzy” partition of the data. By exploiting the fuzzy-set interpretation of the algorithms, researchers can gain valuable insight into the structure of the data.