ABSTRACT
School dropout has significant consequences for individuals and society, including increased crime, reduced productivity, and limited economic innovation. Identifying students at risk of dropping out is crucial. This paper aims to develop a logistic regression-based tool for predicting dropout in the state’s public schools of Espírito Santo, Brazil. We utilized students’ information, such as grades, school attendance, socioeconomic data, and others, provided by Espírito Santo State Education Secretariat and the National Institute of Educational Studies and Research Anísio Teixeira. Various regularization methods were employed. We compared three model specifications for students in the first year of high school in Espírito Santo using data from 2019 to 2022. The results indicated that the models could identify at-risk students satisfactorily, highlighting the effective use of available data by educational departments to identify potential dropouts. This tool can aid educators in creating targeted interventions to minimize dropout rates.
Keywords:
education; machine learning methods; school dropout prediction