Identificación de patrones de deserción y riesgo académico en carreras de la Facultad de Ingeniería de la Universidad Distrital a través de técnicas Machine Learning

Abstract

This paper focuses on the development of a predictive model to identify the risk of students at the Universidad Distrital of being placed on academic probation or dropping out of their studies. Dropout and academic risk represent crucial problems for the institution, affecting both the academic trajectory of students and the ability of the university to fulfill its educational mission. Dropout refers to when a student abandons his or her studies before completing them, while academic probation is assigned to students with poor performance, such as failing multiple subjects or repeating a course for the third time. This work's main objective is to identify patterns and factors associated with dropout and academic risk through an exploratory data analysis, based on significant variables extracted from the literature. These include socioeconomic level, academic performance, distance of residence to the university, ICFES exam results, type of school, and the grade history of each student. The research uses exploratory data analysis techniques in combination with machine learning methods, moving from the preparation and selection of variables to the application of transformations, in order to develop an accurate classification model for the prediction of dropouts. Through this methodology, it is expected to identify relevant patterns and relationships that reflect the impact of the selected variables on academic results. The proposed predictive model focuses on the use of machine learning algorithms that allow predicting the academic risk of each student individually. This will provide the university with a tool that enables the implementation of personalized interventions and supports, with the aim of improving retention rates and academic success. In conclusion, this degree work represents a significant advance towards the creation of effective student retention strategies, with a data-driven approach that allows a better understanding of the factors that affect dropout and academic risk, which contributes to the improvement of academic performance at the Universidad Distrital.