Importància de les dades en la predicció de l'abandonament universitari

Abstract

[en] Currently, university dropout is a problem in most studies at Catalan universities. This paper is an investigation of which information about a university student is most important in predicting whether the student will drop out. For this purpose, a comparison has been made between two academic datasets, one from the University of Barcelona (UB) and the other from the Polytechnic Institute of Portalegre (PIP). The UB dataset has only students' grades. The PIP dataset has a large number of characteristics about each student, which are analysed to define the importance of these in predicting dropout and to make a proposal at UB about what data should be considered in addition to grades. This study has been carried out using data science and machine learning models.