머신러닝 기반 해운기업 재무 건전성 예측 연구: Z-Score와 SHAP 분석의 결합

A Machine Learning Approach to Predicting the Financial Stability of Korean Shipping Companies: Integrating Z-Score and SHAP Analysis

Purpose - This study evaluates the financial stability of Korean shipping companies by integrating traditional Z-score models with advanced machine learning techniques to improve bankruptcy prediction accuracy and identify key risk factors. Design/Methodology/Approach - Companies were categorized into risk, caution, and safe groups based on Z-scores. Machine learning models, including CatBoost and LightGBM, were applied to each group, and SHAP analysis was used to interpret the impact of each variable on financial risk. Findings - Financial expense ratios and accounts receivable turnover are critical indicators of financial health across all groups. High-risk companies should focus on reducing financial expenses and managing debt, while safe companies can enhance asset efficiency and profitability. Research Implications - This study offers practical guidance to develop tailored risk management strategies specific to each financial stability group within the Korean shipping industry. By identifying and understanding key risk factors, shipping companies can implement more effective financial management practices. Furthermore, the integration of Z-score models with machine learning techniques provides a robust framework to predict financial stability, which can be extended to other capital-intensive and debt-dependent industries. The findings support the resilience of the Korean shipping industry in navigating a volatile economic landscape, contributing to its long-term sustainability and financial health.