Classification of employees' work alienation with machine learning

Work alienation, which can have adverse effects on various aspects of employees' performance and job satisfaction, is among the significant problems of modern work life. Work alienation emerges as a result of the complex interaction of various organizational, individual, and interpersonal factors. Therefore, predicting work alienation and implementing preventive measures against it are crucial for both individuals and organizations to maintain a healthy work environment. Considered as a significant issue for many employees, work alienation is also seen as a risk for the education sector due to its organizational structure and stressful environment. The success of the education sector in Türkiye depends on the success of employees in the education sector, especially teachers. Therefore, it is important to analyze the adverse factors that may impact individuals employed in the education sector. Based on this, the current thesis focuses on exploring ways to predict work alienation in private education institutions. The primary aim of the thesis is to use machine learning algorithms, an innovative method for organizational behaviour, to predict work alienation by job characteristics (such as skill variety, task identity, task significance, autonomy, and feedback), job stress, and locus of control variables. Models were developed using classification algorithms such as logistic regression, support vector machine, and decision tree algorithms, and prediction processes were conducted using the data set allocated for testing. The secondary aim of the research is to determine the most successful classification algorithm for classifying work alienation by examining various criteria. Data for the research were collected from employees working in the private education sector in Ankara, and the analysis was conducted with 213 participants. The data collection instruments included the Job Alienation Scale, Job Characteristics Questionnaire, Job Stress Scale, and Locus of Control Scale. The predicted variable of the research, work alienation, was dichotomized, and models were developed using classification algorithms. Models test success was evaluated using accuracy, sensitivity, specificity, and AUC criteria. The evaluation revealed that the support vector machine with a polynomial kernel function was the most successful algorithm.