Machine Learning Model for Predicting Academic Success Through Multi-Criteria Admissions

Arana M.; Arana M.; Camborda M.; Gamarra A. · Proceedings - 9th International Symposium on Accreditation of Engineering and Computing Education, ICACIT 2023 (2023) · DOI: 10.1109/icacit59946.2023.10403678

Resumen

Academic success has been under study for several years. University students are vulnerable to problems affecting their academic background, in some cases even dropping out. The present research analyzed the academic success of those entering the UNCP taking aspects of the school trajectory, sociodemographic level, academic competencies, and extracurricular achievements. The methodology selected was the deductive-inductive method, applied type, explanatory-predictive level, nonexperimental longitudinal cohort study design; the population chosen was 11466 admissions in the academic periods 2020-I, 2020-II, 2021-I, and 2021-II. The prediction was made using three machine learning algorithms: linear regression, decision tree, and classification; where the result was 95.44% prediction with the classification model. The indicators taken into account for the model included school trajectory (entrance grade, modalities of admission, and academic preparation), socio-demographic level (age of the entrant), academic competencies and extracurricular achievements (sports and artistic participation upon entering university), the approved credits and the average, which were obtained after the end of the first academic semester. It was possible to predict the academic success of UNCP students using machine learning tools, allowing them to detect students who might face problems in their university studies.

Instituciones: Universidad Continental; Universidad Peruana Los Andes; Universidad Nacional del Centro del Perú