Extracurricular Activities as Dropout Prediction Factors in Higher Education Using Decision Trees

Abstract

Educational Data Mining can help predict dropout prone students and the factors institutions should observe in trying to avoid an important social problem in modern societies. However, most current predicting models use academic credit worth information from the curricula, ignoring extracurricular activities, while there is evidence from other research fields that some activities like sports can be related to academic performance. This paper studies the importance of extracurricular activities to predict dropout in students from two Bachelor of Science degrees (Engineering and Business). Data from 4.840 students was collected and two models, one including all data and another removing credits worth courses were trained and validated, showing that extracurricular activities are excellent dropout predictors.