Using Data Mining Techniques to Perform School Dropout Prediction: A Case Study

Abstract

School Dropout is a severe problem for educational institutions. Institutions need to be able to measure and reduce dropout rates. Currently, annual expenses with dropout reach R$ 415 million in Brazilian currency. The purpose of this article is to identify the factors that affect students who drop out of the University of Brasilia (UnB) and Machine Learning to provide a model for predicting which students will drop out of undergraduate courses. With this, actions can be taken to reduce the dropout rate. The result of this work demonstrates that the courses with the most credits (workload), longer time to complete (5–6 year courses) and student’s poorer academic performance (poor grades) influences student dropout rate. Also, social factors, such as quota holders or non-quota holders, also influence the dropout rate of undergraduate students at the University of Brasília (UnB).