Modelo de clasificación para la deserción estudiantil en las universidades públicas del Perú

Villarreal-Torres H.; Ángeles-Morales J.; Marín-Rodriguez W.; Cano-Mejía J. · Revista de Ciencias Sociales (2024) · DOI: 10.31876/rcs.v30i1.41667

Resumen

Information and communication technologies play a relevant role in different fields of knowledge. Currently, there is a greater capacity to identify patterns and anomalies in an organization’s data using artificial intelligence; The study aimed to develop a classification model for student dropout by applying machine learning with the autoML method of the H2O.ai framework, taking into account the dimensionality of socioeconomic and academic characteristics. The methodology used was predictive and non-experimental, observational and prospective in design; To this end, a 20-item questionnaire was applied to 237 students from the Graduate School enrolled in master’s degree programs in education. The research resulted in a supervised machine learning model, gradient boosting machine, to classify student dropout, thus identifying the main associated factors that influence dropout, obtaining a Gini coefficient of 92.20%, AUC of 96.10% and a LogLoss of 24.24% representing a model with efficient performance. It is concluded that the model is appropriate for its performance metrics, offering advantages such as working with unbalanced data, cross validation and making predictions in real time. © (2024), (Universidad del Zulia). All Rights Reserved.

Instituciones: Universidad Nacional José Faustino Sánchez Carrión