Fuzzy clustering and consensus methods: an application to the analysis of Italian banks (Q2711707)

In fuzzy clustering each unit presents a set of membership grades indicating the extent to which the unit belongs to each cluster. A wide class of methods to obtain fuzzy partitions is presented by the fuzzy \(G\)-means algorithm. A different approach to the problem was proposed for an extension to spatial data.NEWLINENEWLINENEWLINEThis paper deals with the problem of aggregating different partitions of a set of units, in order to obtain an overall classification summarizing the aspects considered in each partition. A new method of fuzzy clustering and the consensus of classifications are considered, and the two approaches in a theoretical and applicative framework are compared. Previous methods are applied in a research about the features of the most important Italian banks using a set of eight indicators.