Lopez, Manuel Ignacio, et al. "Classification via clustering for predicting final marks based on student participation in forums." International Educational Data Mining Society (2012).

This paper explores a classification via clustering approach to predict final marks in a university course based on forum participation data. The study aims to determine if forum participation can predict final marks and if the clustering approach achieves accuracy comparable to traditional classification algorithms. Using data from first-year university students, several clustering algorithms were compared with traditional classification methods. Results indicate that the Expectation-Maximisation (EM) clustering algorithm provides results similar to the best classification algorithms, especially with selected attributes. The paper also describes the centroids of the EM clusters to illustrate the relationship between the clusters and student performance.