Objective: Machine learning is emerging as a new alternative in various scientific fields and is potentially a new method of interpretation. Using the Light Gradient Boosting Machine (LightGBM), we analyzed the factors that influence the re-choice of emergency medicine responders. The survey is a cross-sectional study which provides an accurate understanding of a responder's current status. However, the results may vary depending on the composition, format, and question, and the relationship between the answers may be unclear. Methods: This study evaluated the modified 2020 Korean Emergency Physician Survey raw data. We applied the preferred model for random relationship check, random forest, support vector machine, and LightGBM models. The stacking ensemble model was used for the final decision process. Results: ‘It is fun working in an emergency room’was the most selected response factor for re-choice, followed by ‘interesting major’. The physical burden of age and lack of identity had a negative impact, whereas burnout and emotional stress factors had a lesser effect. Anxiety caused by the coronavirus disease 2019 (COVID-19) is thought to have a significant impact on this decision making. Conclusion: Establishing the identity of emergency medicine and being faithful to its fundamental mission is a way to increase the rate of re-choice. Decreasing the burden of workload modified according to age is recommended to establish career longevity. The method of machine learning presents us with a new possibility of checking the relevance of survey results quickly and easily.