An extension of PROMETHEE to interval clustering

A model based on the PROMETHEE I outranking method and the FlowSort sorting procedure is developed for interval clustering and some promising results are pointed out by comparing the performances of the proposed model with the P2CLUST model.

Abstract

Multicriteria clustering techniques aim to detect groups of alternatives evaluated on multiple criteria with similar profiles. The preferential partitioning of the dataset allows the decision maker to get a better understanding of the structure of his problem. In this paper, we focus on the particular case of interval clustering. This approach allows us to assign alternatives either in individual or interval clusters. For this purpose, we develop a model based on the PROMETHEE I outranking method and the FlowSort sorting procedure. The performance of the proposed model is evaluated on real-world datasets regarding the convergence, the stability and the quality of the clustering. This analysis has pointed out some promising results that we underline by comparing the performances of the proposed model with the P2CLUST model.