A Model to Predict Low Academic Performance at a Specific Enrollment Using Data Mining

• Autores: Camilo Ernesto López Guarín, Elizabeth Leon Guzmán, Fabio A. González Osorio
• Localización: Revista Iberoamericana de Tecnologías del Aprendizaje: IEEE-RITA, ISSN 1932-8540, Vol. 10, Nº. 3, 2015, págs. 119-125
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • This paper presents the results of applying an educational data mining approach to model academic attrition (loss of academic status) at the Universidad Nacional de Colombia. Two data mining models were defined to analyze the academic and nonacademic data; the models use two classification techniques, naïve Bayes and a decision tree classifier, in order to acquire a better understanding of the attrition during the first enrollments and to assess the quality of the data for the classification task, which can be understood as the prediction of the loss of academic status due to low academic performance. The models aim to predict the attrition in the student's first four enrollments. First, considering any of these periods, and then, at a specific enrollment. Historical academic records and data from the admission process were used to train the models, which were evaluated using cross-validation and previously unseen records from a full academic period. Experimental results show that the prediction of the loss of academic status is improved when the academic data are added.