Comparison of Boosting Machine Learning Models for Classifying Students in Blended Learning Programs in Higher Education During The Period 2020–2024

Abstract

This paper presents a comparative analysis of the performance of four supervised machine learning algorithms based on Boosting models: XGBoost, LightGBM, CatBoost, and Gradient Boosting Machine. The objective is to classify students according to their enrollment modality: face-to-face or blended learning at a National University of Education in Peru during the period 2020–2024. A dataset from the Peruvian government's national open data platform was used, which includes academic and demographic information about students. The four machine learning models were then trained using the stratified cross-validation technique to ensure that each class was adequately represented. To analyze the performance of the evaluated machine learning models, the following evaluation metrics were used: Accuracy, Precision, Recall, F1-Score, and Area Under the Curve (AUC). Of all the models evaluated, the CATBOOST algorithm stood out for offering a balance between precision and efficiency. These final findings highlight the importance and usefulness of implementing automated systems that support enrollment planning and management in both face-to-face and blended learning modalities in Peruvian higher education.