Construcción de modelos para determinar el rendimiento académico en estudiantes de educación superior mediante técnicas no clásicas de machine learning

Abstract

Academic performance is one of the topics that have been analyzed with great interest in educational institutions, since it has been discovered that the behavior of academic performance can indicate the probability of success or failure of a student in his or her academic life. However, academic performance depends on multiple variables, so it is questionable whether it is possible to predict it. When this problem is approached in a quantitative way, we look for ways to adopt new tools that allow us to process information efficiently. To this end, neural network models have been implemented for specific contexts. This research work focuses on studying new models adapted to the characteristics of the students of the Universidad Distritral FCJ, in order to find out if this type of models are effective enough to predict academic performance and thus become a useful tool for managers when making decisions that address issues such as student dropout.