Overlapping Clustering of Statistical Software Packages for PC
Cluster analysis is specially concerned with algorithms for computing non-overlapping classifications, for example partitions or hierarchies, on given object sets. For several economic problems the determination of non-overlapping classifications representing the structure of data is too specific and narrow. In opposition to that given natural overlappings should not be suppressed because the construction of overlapping clusters gives a better insight into the structure of data. In this paper several principles of constructing overlapping clusters e.g. maximal cliques, fuzzy clustering, quasi-hierarchies and pyramidal classification are presented. The advantages and disadvantages of these clumping techniques are discussed in an overlapping clustering of selected software packages.
KeywordsFuzzy Cluster Maximal Clique Partial Graph Fuzzy Intersection Overlap Cluster
Unable to display preview. Download preview PDF.
- BEZDEK, J. (1973): Fuzzy Mathematics in Pattern Classification. PhD Thesis, Cornell University, Ithaca, N.Y.Google Scholar
- JARDINE, N., SIBSON, R. (1971): Mathematical Taxonomy. Wiley, New York.Google Scholar
- LASCH, R. (1993): Pyramidaie Darstellung multivariater Daten. Josef Eul, Ber- gisch Gladbach, Köln.Google Scholar