Prediction of Course Failure in University Students Using Machine Learning Techniques: Comparison of Supervised Models
OnBoard Knowledge Journal · 30.07.2026 · 1-13
DOI: 10.70554/objk2026.v02n02.01 ↗Kurz gefasst
Fachartikel in der Zeitschrift OnBoard Knowledge Journal, online erschienen am 30.07.2026. Verfasst von Andrés Solano-Barliza und 2 weiteren Autor:innen. Die Arbeit nennt 24 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
The use of digital platforms among Colombian college students continues to grow, and with that, the question of whether it affects retention and academic performance has resurfaced. In Riohacha, this relationship had not been studied using artificial intelligence tools at the institutional level. For this study, a dataset was constructed based on a survey administered to 300 undergraduate students at the University of La Guajira. Twenty-one sociodemographic, behavioral, and digital perception variables were captured and organized using the CRISP-DM methodology. The target variable is binary: whether the student has failed at least one course. Three supervised classification algorithms (Decision Tree, KNN, and Random Forest) were compared, integrated into pipelines with SMOTE and five-fold stratified cross-validation. Random Forest yielded the best results, with 62% accuracy, an F1-Score of 0.606, and an ROC AUC of 0.682. Feature importance analysis identified the level of concentration while studying, daily hours of cell phone use, and preferred social network as the most significant predictors.
Zitieren
APA
Solano-Barliza, A., Figueroa-Sarmiento, I., & Romero-Gómez, F. (2026). Prediction of Course Failure in University Students Using Machine Learning Techniques: Comparison of Supervised Models. OnBoard Knowledge Journal, 1-13. https://doi.org/10.70554/objk2026.v02n02.01
Vancouver
Solano-Barliza A, Figueroa-Sarmiento I, Romero-Gómez F. Prediction of Course Failure in University Students Using Machine Learning Techniques: Comparison of Supervised Models. OnBoard Knowledge Journal. 2026:1-13. doi:10.70554/objk2026.v02n02.01
BibTeX
@article{SolanoBarliza2026,
author = {Solano-Barliza, Andrés and Figueroa-Sarmiento, Inés and Romero-Gómez, Fabio},
title = {Prediction of Course Failure in University Students Using Machine Learning Techniques: Comparison of Supervised Models},
journal = {OnBoard Knowledge Journal},
year = {2026},
pages = {1--13},
doi = {10.70554/objk2026.v02n02.01}
}
EUVIMED