Propuesta de un Modelo Predictivo utilizando Aprendizaje Profundo para el análisis de deserción estudiantil en Universidades Colombianas Virtuales

Abstract

Student dropout is a reality and a complex phenomenon in the country. In this paper, a predictive model is proposed that serves as a support to the colombian universities for dropout analisys on
students, main, on virtual undergraduate programs. A predictive model, can help to organizations
generate profits and avoid future losses, taking historical data so show results expected to be analized and support on decisions. This model is developed, taking historical events with different variables of type: social, academic, personal, work, login to e-learning platforms, etc, next, this variables deep learning algorithms are applied. The dropout probabilities prediction for each student
is expected, then, can be alerted to apply early preventive measures with the student population
Keywords: deep learning; educational institutions; electronic learning; student dropout; predictive models