A data mining approach to employee turnover prediction (case study: Arak automotive parts manufacturing)

Training and adaption of employees are time and money consuming. Employees’ turnover can be predicted by their organizational and personal historical data in order to reduce probable loss of organizations. Prediction methods are highly related to human resource management to obtain patterns by historical data. This article implements knowledge discovery steps on real data of a manufacturing plant. We consider many characteristics of employees such as age, technical skills and work experience. Different data mining methods are compared based on their accuracy, calculation time and user friendliness. Furthermore the importance of data features is measured by Pearson Chi-Square test. In order to reach the desired user friendliness, a graphical user interface is designed specifically for the case study to handle knowledge discovery life cycle.