Application of predictive algorithms to improve retention and academic success in higher education

Abstract

This study examines the application of predictive algorithms to improve retention and academic success in higher education. The main objective is to evaluate the effectiveness of various algorithms in early identification of students at risk of dropout or poor academic performance. The methodology employed consists of a systematic literature review and documentary analysis, exploring recent studies on the use of academic, socioeconomic, and interaction data in educational platforms. The results reveal a trend towards more sophisticated machine learning models, with emphasis on neural networks and decision trees. It was found that multidimensional approaches, integrating academic, socioeconomic, and online behavior data, achieve up to 85% accuracy in predicting academic risk. Additionally, a positive correlation was evidenced between participation in practical and interdisciplinary activities and academic success. The conclusions underscore the significant potential of predictive algorithms to improve retention and academic success in higher education, while pointing out the need for a multifaceted approach that considers academic, socioeconomic, technological, and ethical factors in their implementation.