Desarrollo de un modelo para predecir el rendimiento académico de estudiantes de la EPN en base a su nivel de acceso a TICS y factores socioeconómicos.

Abstract

The aim of this research was to stablish the academic performance of university students by identifying the relationship between socioeconomic factors and the level of access to the Information and Communication Technologies (ICTs). The study took place in a public university, one of the largest and recognized higher education institutions in Ecuador, South America. Academic records, socioeconomic background and level access to ICT was considered for the research.

The study considers Action Research and CRISP DM as a framework; undertaking the steps of data analysis, data preparation, cleaning phases, and finally consider several machine learning algorithms.

The study revealed a solid dataset with 15 variables that reflect its correlation and affect academic performance. year of beginning of studies, age of beginning of studies, career that the student is studying, faculty, credits for registered subject, percentage of credits accumulated, number of members in the family environment, father´s salary, mother´s salary, father's social security affiliation, mother's social security affiliation, properties owned by the student, if the student has a vehicle, the student's IPF code, if the student's computer is for personal use, have a direct influence on the final grade and academic performance. The result of this research represents an important information for academic control authorities in the higher education; in the task of improving the conditions of students, providing them with all the needs for success. KEY WORDS: Academic Performance, Artificial Intelligence, Machine Learning, CRISP DM.