Aplicación de árboles de clasificación a la detección precoz de abandono en los estudios universitarios de administración y dirección de empresas

Abstract

Dropouts in university occur mainly in the first academic year, with an average for Spain of 25%. High dropout rates lead to prejudice against educational institutions, it harms their reputation in terms of low quality. In order to help the processes of tutoring students in the university, our work analyzes if it is feasible to get a profile of the student who is at risk of having a low academic performance in his first year in three different moments: when the admission takes place, at the beginning of the academic year, and after the first examinations. This study has used the classification tree technique based on the CART and QUEST algorithms and has used data from 844 first year students enrolled in the Business Administration Licentiate Degree at the Universidad Pontificia Comillas. We have obtained a 56% percentage of correct classified observations for those students who end up presenting low academic performance, with the information available at the end of the first semester.Keywords: Student withdrawing, Low academic performance, Classification trees, Spanish university system.