APPLICATION OF MACHINE LEARNING TECHNIQUES TO CLASSIFY ACADEMIC PERFORMANCE IN ENEM: A STUDY WITH DATA FROM MARANHÃO

Abstract

This study aims to apply and compare machine learning techniques to classify the performance of students from Maranhão in the 2023 National High School Exam (ENEM), with a focus on the Mathematics test. Using a dataset with over 112,000 participants, three predictive models were employed: Logistic Regression, Random Forest, and XGBoost. After data preprocessing and conversion of categorical variables, the models were trained to classify students into two categories: good and poor, based on a cutoff score of 500 points. The analysis revealed regional patterns of educational inequality and highlighted socioeconomic variables such as income and age as the main predictors of performance. Logistic Regression achieved the highest accuracy, while XGBoost demonstrated a better balance between precision and recall. The results reinforce the usefulness of machine learning for educational analysis and provide important support for public policies aimed at improving equity and the quality of education in the state of Maranhão.

DOI: https://doi.org/10.56238/sevened2025.030-001