Handling multicriteria preferences in cluster analysis |
| |
Authors: | Eduardo Fernandez Jorge Navarro Sergio Bernal |
| |
Institution: | aFaculty of Engineering, Autonomous University of Sinaloa, CP 80040, Culiacan, Mexico;bFaculty of Computer Science, Autonomous University of Sinaloa CP 80040, Culiacan, Mexico;cEmphasys Software, 8550 NW 33rd Street, Suite 200 Doral, FL 33122, USA |
| |
Abstract: | In the framework of multicriteria decision aid, a lot of interest has been devoted to sorting problems, in which the set of categories is pre-defined. Besides, preference oriented multicriteria clustering has received little attention. Usual geometric and related metrics are not well suited for this problem. Here, we propose a clustering method based on a valued indifference relation inspired by outranking methods. We suggest a method (based on comparing cluster centers and an average net flow score of clusters) to build a complete ranking of the set of clusters, that is, a way of defining a set of ordered categories for sorting purposes. The new approach performs very well in some examples. |
| |
Keywords: | Data mining Clustering Multicriteria analysis Outranking methods |
本文献已被 ScienceDirect 等数据库收录! |
|