Identification of fuzzy models to predict students performance in an e-learning environment

This brief study uses the Fuzzy Inductive Reasoning methodology to predict the final mark of the users of a real virtual campus and to determine relevant features involved in the evaluation process, reducing considerably the complexity of the evaluated process and minimizing the teachers' workload.

Abstract

Nowadays, e-learning systems such as virtual campus environments have established themselves as a strong alternative to traditional distance education. In turn, the Internet allows the gathering of information on many aspects of students' online behavior in nearly real time. The knowledge extracted from this information can be used to discover students' learning behavior and usability patterns of the courses, as well as to anticipate future behavior related with the teaching-learning process. In this brief study we use the Fuzzy Inductive Reasoning methodology to predict the final mark of the users of a real virtual campus and to determine relevant features involved in the evaluation process. Experiments carried out with the available data indicate that the final mark can be predicted with a low error and that the number of relevant features identified is small, reducing considerably the complexity of the evaluation process and minimizing the teachers' workload.