International Conference on Economics, Finance & Business, Rome

LEADERSHIP AND EMPLOYEE SATISFACTION: AN EMPIRICAL STUDY BASED ON FEATURE ANALYSIS WITH MACHINE LEARNING

IVANA FOSIĆ, DUBRAVKA PEKANOV, IVONA VIDOVIĆ

Abstract:

Employee satisfaction is an important indicator of organizational success and leadership effectiveness. The aim of this study is to investigate employees’ perceptions of leadership in organizations and to determine its influence on job satisfaction. The focus is on key leadership dimensions such as trust in the employer, fairness, communication, understanding of employees and autonomy at work. The research comprised a primary quantitative study conducted through an online survey of employees from different sectors using a standardized and structured questionnaire. A total of 120 responses were collected, of which 103 were valid for analysis. The data was analyzed using the Random Forest algorithm for machine learning, with the interpretation of features through SHAP values. Leadership was defined as a predictor variable, while job satisfaction was analyzed at two levels – as an overall dependent variable reflecting general employee satisfaction and as a dependent sub-variable providing insight into specific aspects of job satisfaction. The results showed that variables relating to fairness, employee age and communication had the greatest impact on job satisfaction, while variables such as gender and education level had a significantly lower predictive value. The model achieved moderate recall (0.75) but lower overall accuracy (0.55) and a moderate F1 score (0.68), suggesting the need for further model optimization. The study highlights specific leadership aspects that significantly influence employee satisfaction. Based on the results, guidelines can be formulated for developing more effective organizational practices, with a focus on promoting fairness, clear and open communication, and building mutual trust in the workplace. In addition, further research and refinement of predictive models of employee satisfaction are recommended to improve the quality of organizational decision making and more accurately identify factors that contribute to positive work outcomes.

Keywords: leadership, employee satisfaction, predictive model, random forest, SHAP analysis