When the aim is to evaluate the solution of a fuzzy clustering algorithm, the computation of the adjusted version of the Rand index requires converting the soft partitions to hard partitions. Furthermore, in comparing two fuzzy partitions from two different clustering methods, an external validation index should satisfy two desirable properties: (i) reflexivity, and (ii) a proper interpretation of correction for agreement due to chance. In this paper, we show an extension of the commonly used adjusted Rand index to fuzzy partitions based on normalized degree of concordance.
A PROPOSAL TO EVALUATE THE SOLUTION OF FUZZY CLUSTERING ALGORITHMS / Iorio, Carmela; Pandolfo, Giuseppe; D'Ambrosio, Antonio. - (2023), pp. 520-523. (Intervento presentato al convegno 14th Scientific Meeting of the Classiffcation and Data Analysis Group (CLADAG 2023) tenutosi a Salerno nel 11-13 Settembre 2023).
A PROPOSAL TO EVALUATE THE SOLUTION OF FUZZY CLUSTERING ALGORITHMS
Carmela Iorio
Conceptualization
;Giuseppe PandolfoConceptualization
;Antonio D'AmbrosioConceptualization
2023
Abstract
When the aim is to evaluate the solution of a fuzzy clustering algorithm, the computation of the adjusted version of the Rand index requires converting the soft partitions to hard partitions. Furthermore, in comparing two fuzzy partitions from two different clustering methods, an external validation index should satisfy two desirable properties: (i) reflexivity, and (ii) a proper interpretation of correction for agreement due to chance. In this paper, we show an extension of the commonly used adjusted Rand index to fuzzy partitions based on normalized degree of concordance.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
