Factors Influencing Voluntary Turnover Among Young College Graduates Using the XGBoost with Bagging Aggregation Algorithm: Findings from Nationwide Survey in South Korea

This study aims to analyze the factors influencing turnover among young professionals aged 20 to 30 and compare the predictive performance of various machine learning models using data from the 2019 Graduates Occupational Mobility Survey (n=11,605). The XGBoost with Bagging model was selected for its ability to handle complex interactions and large datasets. The dataset includes demographic information, job characteristics, job satisfaction scores, and other relevant variables. Data preprocessing involved handling missing values, one-hot encoding for categorical variables, and normalization. The model’s performance was compared to KNN, Logistic Regression, SVM, and Bagging using metrics such as Area Under the Curve (AUC) and F1-score. The XGBoost with Bagging model demonstrated superior performance, with an AUC of 0.85 and an F1-score of 0.86, outperforming the other models. Key features influencing turnover intentions included permanent employment status, salary, job satisfaction, job security, and career advancement. These findings provide actionable insights for human resource management strategies aimed at reducing employee turnover. The study concludes that the XGBoost with Bagging model is a robust tool for predicting turnover intentions and recommends future research to integrate additional features and apply the model to different age groups and industries for further validation.