Improving the prediction of school dropout with the support of the semi-supervised learning approach

Abstract

School dropout is a phenomenon characterized by being influenced by several variables. This research used Machine Learning techniques, especially in the context of the semi-supervised learning strategy, to predict the risk of dropout in undergraduate courses at a Brazilian higher education institution. Two phases of experiments were conducted, the first using Feature Selection techniques and the second applying a semi-supervised learning strategy to improve performance metrics collected from the increase in the number of instances of students labeled as Graduated. As a main result, we obtained a model capable of classifying dropout with 90% accuracy and 86% Macro-F1.