Predictive Analysis of Student Dropout in Higher Education

This study presents a predictive model of student dropout in higher education, developed using preprocessing techniques and a Support Vector Machine (SVM) model. A dataset from Tecnológico de Monterrey, which includes demographic, academic and financial information of students, was used. The data preparation process included the cleaning and normalization of key variables, such as gender, academic level and types of scholarships, as well as the elimination of irrelevant columns. Subsequently, the data set was divided into training, validation, and test subsets, following standard predictive modeling practices to ensure accuracy and generalizability of the model. Preliminary results suggest that the SVM model is effective in predicting student dropout risk, providing a basis for the development of more personalized educational interventions.