Factors Associated with Dropout Intention in Engineering Education: A Learning Interpretable Modeling Approach.

Dropout in higher education remains a persistent challenge with academic, institutional, and social consequences. Identifying students' dropout intention before formal withdrawal occurs may provide useful evidence for preventive support strategies. This study examines dropout intention among undergraduate engineering students at a Chilean university. Using survey data from 189 students, five variables were operationalized: personal self-regulation, mental health, institutional perception, socio-economic conditions, and academic motivation. Two interpretable models were estimated: multivariable logistic regression and a shallow decision tree. The logistic regression model showed satisfactory explanatory capacity and identified socio-economic conditions and personal self-regulation as significant protective predictors of dropout intention, while mental health, institutional perception, and academic motivation were not significant after adjustment. The decision tree provided a complementary rule-based segmentation and confirmed the prominence of socio-economic conditions and personal self-regulation, while also indicating a secondary contribution of mental health. These findings suggest that dropout intention in engineering education is shaped primarily by structural conditions and self-regulatory resources rather than by institutional perception or academic motivation alone. The study provides empirically grounded evidence to support early identification and student support strategies in higher education.