Predicting the Work-Life Balance of Employees Based on the Ensemble Learning Method

Abstract

Working life has a great impact on other areas of people's lives. Efforts made at work lead to attrition, exhaustion, and health problems. Employers need to take the necessary steps to keep employees motivated by helping them balance work and personal lives. Employers can use many different techniques to measure and analyse the work-life balance of their employees, such as questionnaires and machine learning techniques. This research was conducted to group workers based on turnover levels using effort and work-life balance parameters. Machine learning, including ensemble learning techniques, is used to achieve this. One ensemble learning algorithm, Random Forest, performed almost as well as Support Vector Machine with the highest score of 95%. Almost all algorithms, regardless of whether they are part of ensemble learning or not, achieved an f-score of 86%. However, one of the ensemble learning models, xGBoost, performed poorly with the lowest f-score of 69%. All algorithms predicted the lowest and highest work-life balance scores but were confused when predicting the middle scores (class 2 and class 3).