Title: Different fuzzy cluster validity indexes for the evaluation of the quality of the resulting partitioning

Authors: Paola Perchinunno; Silvestro Montrone

Addresses: Department of Business and Law Studies, University of Bari, Via C. Rosalba 53, 70100 Bari, Italy ' Department of Business and Law Studies, University of Bari, Via C. Rosalba 53, 70100 Bari, Italy

Abstract: Fuzzy clustering is useful to mine complex and multi-dimensional datasets, where the members have partial or fuzzy relations. The procedure of evaluating the results of a fuzzy clustering algorithm is known under the term cluster validity. There are three principal approaches to investigate cluster validity: external, internal and relative criteria (Theodoridis and Koutroubas, 1999). These methods give an indication of the quality of the resulting partitioning and, so, they can be considered as an instrument available to experts in order to assess the results of fuzzy clustering. In this work we apply some indexes to evaluate the quality of the results obtained from a case study.

Keywords: fuzzy clustering; cluster validity indexes; socio-economic hardship; partitioning quality; data mining.

DOI: 10.1504/IJICA.2016.077594

International Journal of Innovative Computing and Applications, 2016 Vol.7 No.2, pp.84 - 90

Received: 05 Feb 2016
Accepted: 03 Mar 2016
Published online: 06 Jul 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article