Abstract

In the ever-evolving landscape of 21st-century education, this research seeks to understand the challenges of on-time graduation for students in two related computing majors. In particular, we focus on the Universidad Cat & oacute;lica del Norte computing engineering programs in Chile, specifically the "Computing and Informatics Civil Engineering" (ICCI) and "Computing and Informatics Execution Engineering" (IECI) programs. We developed a machine-learning-based model using random forests to predict delays in submissions of the final report of graduation projects, the key step in the graduation pipeline of the analyzed students. We had access to a data set comprised of 209 students in the period from 2013 to 2017, out of these students, only 111 completed all their graduation requirements. Thus, we focused on this subset of students for the analysis. Our analyses of results indicate that individual advisors minimally contribute to predicting timely or late submissions, emphasizing the need for a holistic approach. In contrast, the specific major, graduation modality, and time in the program play crucial roles, with GPA emerging as the most influential factor (24.06%). Notably, the "Professional Work" modality exhibits a moderate positive correlation with late submissions, contextualized by students' employment commitments. The study's predictive model offers actionable insights for educators and administrators, identifying at-risk students and advocating for personalized support strategies. This research contributes to the ongoing dialogue on enhancing educational outcomes by integrating data-driven approaches tailored to diverse student profiles.