FEED-FORWARD NEURAL NETWORKS: AN APPLICATION TO THE PREDICTION OF STUDENTS' PERFORMANCE

During the last few years, the application of artificial intelligence in education has grown very fast and has enabled the development of more sophisticated and more efficient models for predicting students' performances. These models allow to detect a broader range of students' behavior, permitting to distinguish in an early stage lower performers from higher ones. In this context, neural nets, especially static ones with feed-forward architecture, have been extensively studied. In this paper we briefly describe neural nets from a theoretical point of view and we show an application of neural nets to the results of the exercises done by the students of one of the courses of Financial Mathematics with the platform M.In.E.R.Va. (already presented at EDULEARN 14) and we try to predict students' performances in the intermediate exam. The analysis is performed using the statistical software R, and in particular the package "neuralnet".