Towards the development of a classification service for predicting students' performance

Choosing a suitable classifier for a given data set is an important part of a data mining process. Since a large variety of classification algorithms are proposed in literature, nonexperts, as teachers, do not know which method should be used in order to achieve a good pattern. Hence, a recommender service which guide on the process or automatize it is welcome. In this paper, we rely on meta-learning in order to predict the best algorithm for a data set given. More specifically, our work analyses what meta-features are more suitable for the problem of predicting student performance and also evaluates the viability of the recommender.