Padrões sequenciais para Learning Analytics: um estudo sobre como redes neurais podem prever o desempenho de estudantes

Abstract

A study was conducted for the development of predictive models based on sequential patterns identified through student activity logs in a virtual learning environment to rate student performance before the completion of their course. The objective of this study was to compare the performance of predictive model based on decision tree algorithms and neural networks, as well as to identify whether the use of sequential patterns contributes to the performance of these models rather than to evaluate individual actions taken by students. It was concluded that both sequential patterns and neural networks can improve model performance, but some decision tree-based models were also chosen among the best.