Application Of Data Mining Techniques For The Analysis Of Work Absenteeism At In The Human Resources Records Of An Academic Institution

Abstract

Objective: Identify trends and patterns in employee absenteeism in the Human Resources Department at the Academic Institution, based on data analysis using the CRISP-DM methodology, in order to provide strategic input that strengthens institutional human talent management.

Method: Quantitative, descriptive, and exploratory design, based on data mining, from CRISP-DM. The population corresponded to a total of 1,406 disability records in the period 2012-2024. Cleaning, preparation, attribute selection, and modeling processes were used, employing statistical and computational techniques.

Results: General illness is the main cause of absences; administrative workers account for the highest proportion of absences; prolonged absences are associated with middle and advanced ages. Furthermore, the incorporation of ensemble methods revealed more stable and accurate predictive patterns across these groups, reinforcing the utility of advanced analytical approaches for understanding the dynamics of absenteeism.

Discussions: While the prevalence of general illness is consistent with what has been documented in the literature, the concentration of absenteeism among administrative staff raises a particular issue in the university context. Organizational and contractual factors influence this phenomenon. The usefulness of data mining as a predictive and interpretive tool for understanding workplace absenteeism is highlighted. Conclusions: The application of the CRISP-DM methodology made it possible to identify relevant patterns and transform them into strategic inputs to strengthen institutional human talent management. Additionally, the results demonstrate the value of ensemble methods, which outperformed individual classifiers and provided more robust predictive capacity, highlighting their potential as reliable analytical tools for supporting evidence-based decision-making