Predicting Employee Turnover in High-Tech Enterprises Using Machine Learning: Based on the Psychological Contract Perspective

High-tech enterprises are boosting technological innovation and economic growth in countries worldwide. Compared with general enterprises, high-tech enterprises are characterized by technology-intensive and high employee turnover rates, relying more on human capital, especially researchers with core technical expertise. However, high turnover rates and unexpected departures of key employees place a huge financial burden on enterprises, along with the risk of technology leakage. Therefore, this study establishes a theoretical model of voluntary employee turnover based on psychological contract theory and previous theoretical studies. We also categorize employee turnover characteristics into four dimensions: Individual conditions, Material incentives, Development opportunities, and Environmental support. Given that previous related studies lacked the combination of theory and data-driven methods, this study applies the IBM HR dataset and selects features for each dimension through the PCA method, for which machine learning models are constructed, including logistic regression, random forests, SVMs, decision trees, and XGBoost, and their performances are evaluated. In addition, the importance of different dimensions is analyzed, and it is found that material incentives have the greatest impact on employee turnover.