This study aims to identify key challenges in applying random forest to turnover intentions prediction research and to provide practical guidance for the use of random forest in human resource management (HRM).
To address these challenges, the study conducted two sets of analyses using the Human Capital Corporate Panel (HCCP) dataset collected in Korea. First, we evaluated two random forest models, one trained on scale-level data and the other on item-level data. Second, we compared the results of random forest and multiple regression models to assess their relative utility for predicting and interpreting turnover intentions.
The findings highlight the importance of carefully selecting data, handling constructs in line with research purpose and cautiously interpreting feature importance scores when applying random forest to turnover intentions research. The results also provide practical guidelines and directions for future research.
This study extends prior turnover intentions research by emphasizing three challenges that have received limited attention in HRM research. By offering both methodological guidance and practical implications, it contributes to a more informed and effective use of random forest in turnover intentions research.
