Aplicación de técnicas de minería de datos en el contexto del rendimiento académico en la Universidad de Cuenca

Abstract

The University of Cuenca has a large amount of data related to academic performance, student desertion and the socio-economic status of its students that have been collected since 2008 to the present, based on this is it is possible to extract significant information that is not at first sight, that helps the fulfillment of the strategic objectives of institutional development of the University of Cuenca and of the metrics of the quality model used by the Council for quality assurance of higher education for the evaluation of higher education. Problem: Within this context it is desirable to have an analysis of academic performance and dropout, in order to have a preventive approach that identifies subjects in situations of risk at an early stage of the career or study cycle and that the actors involved in these processes can make better informed decisions based on historical records. Solution: This work applies data mining to develop classification and prediction models, using the CRISP-DM methodology, three specific problems are proposed to solve: desertion prediction, prediction of failure by study cycle and prediction of failure by subject. It was possible to conclude that the obtained models had a good percentage of correct answers (about 73 %) and constitute a very valuable tool for the university.