Resumen de Comparative of Clustering Techniques for Academic Advice and Performance Measurement

Maite García-Ordás, José Antonio López Vázquez, Héctor Alaiz Moretón, José-Luis Casteleiro-Roca, David Yeregui Marcos del Blanco, Roberto Casado Vara, José Luis Calvo-Rolle

• This article presents an innovative proposal for improving personalized student performance counselling. The methodology implemented applies clustering techniques in order to obtain group profiles of students with similar features. The research has been performed utilizing anonymized real academic grades from student data sets of the Polytechnic School of the University of A Coruñaa. The ultimate purpose for the proposed tool is to be dynamic and adaptive to different data sets. Therefore, only the most representative, universal variables are considered. Overall, three techniques have been evaluated for clustering, with two additional ones for dimensional reduction with very promising results.