Classifying Relevant Causes of Employee Absenteeism Using Machine Learning

Employee absenteeism has a significant impact on the organization as it imposes high costs that can be reflected in the organization’s economic results. The impact can be seen in productivity, sustainability, competitiveness, profitability, and inter-organizational relations. Considering the wide range of causes of absenteeism, this paper provides a systematic overview of the causes considering different theoretical approaches. A primary survey was conducted (N=420), and the causes of absenteeism were divided into two groups of causes (personal causes and organizational causes). Unsupervised and supervised machine learning methods were applied to the data set. In the unsupervised machine learning phase, the Elbow data grouping method was applied, and the K-Means algorithm was used to assign each record in the dataset to a specific cluster. In supervised machine learning, the data were divided into learning and testing parts in a 70/30 ratio, and the Random Forest (RF), Decision Tree (DT), Support Vector Machines (SVM), and k-Nearest Neighbors (K-NN) classification algorithms were used. The best performance and area under the ROC curve (AUC) values are obtained using the RF classification algorithm. Using the machine learning algorithm RF, the main causes of absenteeism in both groups of causes of absenteeism were identified. From the group of personal causes (personal illness as well as illness of family members), from the group of organizational causes (mobbing, conflict with work colleagues) stand out the most important causes of absenteeism.