Πρόβλεψη οικειοθελούς αποχώρησης εργαζομένων

View/ Open
Keywords
Μηχανική μάθηση ; Machine learning ; Οικειοθελής αποχώρηση προσωπικού ; Employee attrition ; Κατηγοριοποίηση ; Classification ; Logistic regression ; Decision tree ; Random tree forests ; Support vector machines ; Gradient boosting ; Ada BoostAbstract
This thesis addresses the issue of employee attrition and focuses on using machine learning methods to predict voluntary employee exits, supporting the HR departments of a large business group in Greece. The algorithms that were utilized and evaluated to assess the risk of attrition for approximately 7,000 active employees, based on data collected from the business groups ERP system, were Logistic Regression, Decision Tree, Random Tree Forest, Support Vector Machines, Gradient Boosting, and AdaBoost. The Random Tree Forest algorithm proved to be the most effective, achieving high accuracy in predicting employee exits. A Power BI interactive dashboard allows HR departments to identify employees with a high probability of leaving and take measures to retain them. Recommendations for improving algorithm accuracy and explainability are introduced.