Clustering Users Based on the Capacity to Solve Questions in an Educational Platform

Mariana G. M. MacedoCarmelo J. A. Bastos-Filho

The intense daily use of educational platforms results in high volumesof data. Because of this gigantic pile of data, the users' needs frequently remainunnoticed and thus are not enhanced. This study characterised students' profiles based on an educational database. To achieve these goals, we assessed theapplication of the K-means and C-means algorithms for clustering. The numberof profiles was chosen according to Davies-Bouldin metrics and Gap Statistic.After the experiments, K-means revealed to be inefficient for the database. Onthe other hand, C-means reached a satisfactory result, sustained by Spearman'srank correlation coefficient.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: