L'algorithme OKM pour la classification recouvrante: la détermination du nombre de classes

Abstract

The goal of this report is to adapt two methods for estimating the number optimal of clusters to the overlapping clustering and to attach them to the OKM algorithm. These two methods are the method gamma and a modified version of the Gap Statistic. In this report, we study the performance of this algorithm on a real data set and we attach to it the two methods for estimating the number optimal of clusters. We apply then these methods on different kinds of artificial or real data and we analyze the results.

Date of Award	5 Sept 2011
Original language	French
Supervisor	Andre Hardy (Supervisor), Renaud Lambiotte (Jury), Marcel Remon (Jury) & Anne Lemaitre (Jury)