Data-Driven and Predictive Modeling of Institutional Well-Being Based on the SPIRE Model: Evidence from a Latin American Public University

Abstract

Well-being and happiness are increasingly recognized as key outcomes in higher education, yet few Latin American public universities systematically assess them at the institutional level. This study reports the results of an institutional measurement of well-being and happiness, based on the SPIRE model (Spiritual, Physical, Intellectual, Relational, and Emotional) and NPS, applied to the entire academic community. The database was collected through a digital survey grounded in the SPIRE dimensions, administered to students, faculty, staff, and contractors at a Colombian public university. A total of 877 complete responses were obtained from a population of 8362 individuals. Data were analyzed using descriptive and inferential statistics, psychometric analysis, clustering, and supervised classification techniques. The findings reveal differences in well-being across institutional roles, with faculty and staff reporting higher levels of overall well-being than students, particularly in the physical and emotional dimensions. The analysis also highlights patterns in perceived life purpose, interpersonal relationships, and intellectual well-being. The Random Forest classifier achieved an accuracy of 95.8% in reproducing membership in data-driven well-being clusters derived from the SPIRE dimensions. This classification performance reflects the reproducibility of the cluster structure within the same institutional dataset and should not be interpreted as evidence of future well-being prediction or independent predictive validity.