Many companies study factors that affect retirement in order to prevent outflow of excellent personnel and induce long-term employment. Most of the related studies collected and analyzed survey-based data influenced by the subjective opinions of retirees, or used limited data used abroad. In this study, factors affecting retirement were analyzed based on objective data such as personality evaluation data used in recruitment process of domestic financial IT companies and personnel evaluation and HR data recorded during employment. After comparing the performance of various machine learning models through variable selection and hyperparameter optimization, an optimal retiree prediction model with excellent predictive performance was developed. In addi- tion, through the feature importance analysis, the main factors affecting early retirement were identified. Using this, it is expected that if a systematic personnel policy is established and man- agement efforts to maintain excellent personnel are effectively carried out, it will be possible to prevent retirement and induce long-term service.