Abstract

TORRES LOPEZ, Casio Aurelio; PACHAS-HUAYTAN, Jorge Vladimir  and  ORTIZ FERNANDEZ, Jaime Humberto. Predictive analytics using logistic regression for early intervention in engineering. e-Rev. Multidiscip. Saber [online]. 2025, vol.3, e-RMS10112025.  Epub Dec 10, 2025. ISSN 2960-2467.  https://doi.org/10.61286/e-rms.v3i.299.

Low academic performance and student dropout rates constitute a critical and costly problem in engineering programs, affecting educational quality and institutional efficiency. Traditional descriptive metrics offer an insufficient retrospective view for early and proactive intervention. Therefore, the integration of Predictive Analytics is proposed as a managerial imperative for diagnosing academic risk in advance. It is argued that Logistic Regression (LR) is the ideal statistical technique for this purpose. This is justified by its mathematical rigor in modeling binary outcomes (success/failure) and its capacity to transform a combination of multidimensional factors (academic, psychosocial, and socioeconomic) into a quantified probability bounded between 0 and 1. Crucially, LR allows for the interpretation of the Odds Ratio, a statistic that translates complexity into a managerial metric, quantifying how much each predictor increases or decreases a student's probability of success. This interpretability is vital for decision-making, allowing university administrators to prioritize intervention resources in a cost-effective manner. The creation of an Early Warning System with differentiated risk thresholds guides personalized support strategies (intensive tutoring, socio-emotional support). It is concluded that logistic regression represents the optimal balance between statistical robustness and managerial efficiency, serving as the standard of interpretability against which future, more complex machine learning models should be compared, ensuring that analytics remains an ethical and supportive tool for improving student outcomes.

Keywords : logistic regression; academic performance; student dropout; predictive analytics; odds ratio; higher education.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )