Building Generalizable Models for Student Retention: A Multi-Source Validation Framework for the Peruvian Higher Education Context

Cardenas G.A.E.; Pachamango I.Y.R.; Ventocilla A.J.P.; Eskenazy E.J.M. · Proceedings of the 2025 IEEE 32nd International Conference on Electronics, Electrical Engineering and Computing, INTERCON 2025 (2025) · DOI: 10.1109/intercon67304.2025.11244379

Resumen

Addressing student attrition in higher education requires predictive models that are not only accurate but also generalizable across diverse institutional contexts. This study confronts the crisis of external validity in educational data mining by developing and rigorously validating a machine learning framework for predicting undergraduate dropout. Using a public dataset of 4,424 students from a Portuguese university as our primary source, we employ Random Forest and Support Vector Classifier (SVC) models, optimized with Recursive Feature Elimination with Cross-Validation (RFECV), to identify the most salient predictors of dropout. Our analysis identifies a parsimonious set of nine key features, with 'curricular units passed in the first semester' and 'tuition fee status' emerging as the most powerful predictors. The resulting Random Forest model achieved an accuracy of 0.88 and an AUC of 0.86 on the primary dataset. The core contribution of this research lies in its multi-source validation strategy. We tested the trained model's predictive power on an external dataset from Tecnologico de' Monterrey, Mexico, where it maintained a high accuracy of 0.88, demonstrating exceptional generalizability. Furthermore, we corroborated the identified risk factors against national-level statistics from Peru's Superintendencia Nacional de Educación Superior Universitaria (SUNEDU), confirming their contextual relevance. This research provides a robust, validated model for student retention and, more importantly, offers a methodological blueprint for developing and validating generalizable predictive tools to support evidence-based interventions in higher education systems globally.

Instituciones: Universidad Tecnológica del Perú