Predictive modeling based on machine learning strategies to forecast student dropout at a Peruvian university: A case study

• Autores: Aguilar Lopez, Kristelly Magdalena, Carbajal Ortega, Yuri, Martinez Hilario, Daril Giovanni, Rodriguez, Sol
• Formato: objeto de conferencia
• Fecha de Publicación: 2024
• Institución: Universidad Tecnológica del Perú
• Repositorio: UTP-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorio.utp.edu.pe:20.500.12867/14392
• Enlace del recurso: https://hdl.handle.net/20.500.12867/14392; https://doi.org/10.18687/LACCEI2024.1.1.1316
• Nivel de acceso: acceso abierto
• Materia: University dropout; Desertion; Machine learning; Predictive model; https://purl.org/pe-repo/ocde/ford#2.11.04

University students experiment different factors that bring as a consequence the abandonment of his professional career. In Perú, the dropout rate becomes a critical point of attention due to its increase since COVID-19. Despite the fact that the institutions join forces to improve student retention, these seem to be insufficient because of the root causes of the problem are not analyzed. Hence, this study aims to analyze the main causes associated to student dropout of a population of students from the academic period 2022-2 of a private university. For this purpose, three predictive models (random forest, logistic regression and decision tree) were designed to identify the main risks associated to abandonment of students. The predictive models were designed with the automatic learning method (Machine Learning) through Google Collab programming, obtaining a comparison of predicted dropout versus real dropouts, performing a model accuracy of 93% for the logistic regression model. Weighting the main risks identified, different retention strategies can be proposed to reduce the desertion rate.