Modelo predictivo para admisión en una institución de educación superior en las etapas de asignación y aceptación de cupo.

Abstract

The student selection process has been a critical success factor for higher education institutions. Historically, universities have sought to admit qualified students who are committed to their academic programs. However, since 2023 with the implementation of the regulations of the national leveling and admission system in Ecuador, the pressure to achieve enrollment goals, equal opportunities, equity, among others, has led to institutions not having autonomy in the method of assigning places and has led to the assignment of places that could be rejected. This phenomenon has generated challenges in terms of academic quality and financial sustainability. On the other hand, when students received multiple admission offers, their final decision was influenced by a variety of factors, such as institutional reputation, the quality of academic programs, and the availability of resources. In this context, educational institutions have recognized the importance of better understanding student behavior during the selection process. By analyzing these challenges and opportunities, the need to develop tools and models that allow universities to optimize their admission processes has become evident. By better understanding the factors that influence student decisions, institutions can make more informed and strategic decisions, improving both the quality of their students and their competitive position in the educational market. With the help of machine learning models, CatBoost, Random Forest and Multilayer perceptron were implemented, compared their performance and relationship with the dataset variables, and a predictive model was implemented for the stage of acceptance of places more than 90% accuracy and prioritizing the needs of the university. Keywords: university admission, predictive model, quota acceptance