Research on the Prediction of Employee Turnover Behavior and Its Interpretability
Zhe Tao, Cisheng Wu, Shuping Zhao · Proceedings of the 2021 5th International Conference on Electronic Information Technology and Computer Engineering · 2021
Employee turnover is a key issue that companies must consider, and predicting employee turnover behavior can help companies reduce losses. However, the current Mechanism-Driven prediction method has some problems such as simplified assumptions and low accuracy. Scholars have begun to use it the Data-Driven method establishes an employee turnover behavior prediction model, but there is a lack of interpretability research on the model, which cannot effectively explain the prediction results, it will hinder the actual implementation of the method. This research establishes the RSGSBoost-SHAP prediction and interpretation framework on the basis of combing relevant literature. The framework can directly predict employee turnover behavior based on input characteristics, and explore the mechanism of the characteristics of the turnover behavior, providing management inspiration for enterprises. In the prediction results, AUC is 0.9224, and Recall is 0.8487, reaching the expected accuracy. The interpretation results elaborated on the mechanism of employee characteristics affecting their turnover behavior.