This work is based upon the results of an evaluation process applied over data mining techniques, in order to find the most adequate ones to extract classification rules from first-year students' academic and demographic data in relation with their academic performance. As a result of this, the formulation of a predictive model for academic performance is presented; model whose construction was achieved by analyzing, selecting and defining the classification rules that properly predict the academic performance of Systems Engineering students, at Universidad El Bosque in Bogotá, Colombia. Classification rules that make up the model are analyzed from a contextualized academic point of view; consequently evaluating the pertinence of the relationships between attributes contained within these rules and their ability to predict poor academic performance (through academic risk). Also their applicability to datasets from other academic programs is contemplated, in order to generate useful strategies to prevent academic desertion, being poor academic performance one of the most influencing factors over this phenomenon.

No abstract

No abstract