XGTimeNet

Abstract

Employee behavior analysis and performance prediction have become critical tasks in data-driven human resource management for internet enterprises, but existing methods often insufficiently capture the joint influence of static employee attributes and dynamic behavioral evolution. To address this limitation, this study proposes XGTimeNet, a hybrid machine learning framework designed to improve employee behavior analysis and performance prediction by integrating structured feature learning and temporal sequence modeling. The results indicate that the proposed static-temporal fusion strategy effectively improves classification discrimination and prediction reliability compared with models that rely on single-source features or simple fusion. Overall, XGTimeNet provides a practical and effective framework for intelligent employee performance prediction and offers useful support for data-driven human resource decision-making in internet enterprises.