企業破產分析是財務統計與應用經濟領域的重要研究課題。隨著企業財務結構與營運環境日益複雜，傳統以線性財務比率為基礎的模型往往無法充分描述變數間的非線性關係與高維交互效應。本研究以探索性資料分析的觀點，發展一套能揭示破產資料中多層次依存結構的統計方法。
研究資料取自 UCI Machine Learning Repository 之台灣企業破產資料集，包含 6,819 家公司與 95 個財務指標。首先，以熵為基礎的關聯量測建構變數間的相似矩陣，並以階層式群集分析歸納出具經濟意涵的特徵群組。其次，透過列聯表分析探討群組變數與破產狀態之間的條件分佈與勝算比，以評估主要效果與交互作用。為檢驗結果的穩健性，本研究比較在對立（Alternative）與虛無（Null）列聯表族群下的隨機性分佈差異，以有限樣本精確度（finite-sample precision）為依據判定顯著關聯。結果顯示，多組變數間的高階交互作用可有效區分破產風險，凸顯在多變量資料中探討依存結構的重要性。

Corporate bankruptcy analysis is an important topic in applied statistics and financial economics. As corporate financial structures become more complex, traditional ratio-based linear models often fail to capture nonlinear relationships and higher-order interactions among variables. This study adopts an exploratory statistical framework to reveal multi-level dependency structures in bankruptcy data.
The dataset, obtained from the Taiwanese Bankruptcy Prediction database in the UCI Machine Learning Repository, includes 6,819 firms and 95 financial indicators. An entropy-based association measure constructs a variable similarity matrix, and hierarchical clustering identifies interpretable feature groups. Contingency-table analysis examines conditional distributions and odds ratios between grouped variables and bankruptcy outcomes to detect main and interaction effects. For robustness, the randomness of Alternative and Null contingency-table ensembles is compared, using finite-sample precision to confirm significant associations. The findings reveal higher-order interactions that effectively differentiate bankruptcy risk, underscoring the value of exploring dependency structures in multivariate financial data.

Edward I. Altman. Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. The Journal of Finance, 23(4):589–609, 1968. doi: 10.1111/j.1540-6261. 1968.tb00843.x.
Flávio Barboza, Herbert Kimura, and Edward I. Altman. Machine learning models and bankruptcy prediction. Expert Systems with Applications, 83:405–417, 2017.
William H. Beaver, Maureen F. McNichols, and Jung-Wu Rhie. Have financial statements become less informative? evidence from the ability of financial ratios to predict bankruptcy. Review of Accounting Studies, 10(1):93–122, 2005. doi: 10.1007/ s11142-004-6341-9.
Sreedhar T. Bharath and Tyler Shumway. Forecasting default with the merton distance to default model. Review of Financial Studies, 21(3):1339–1369, 2008. doi: 10.1093/rfs/ hhn044.
John Y. Campbell, Jens Hilscher, and Jan Szilagyi. In search of distress risk. The Journal of Finance, 63(6):2899–2939, 2008. doi: 10.1111/j.1540-6261.2008.01416.x.
Darrell Duffie, Andreas Eckner, Guillaume Horel, and Leandro Saita. Frailty correlated default. The Journal of Finance, 64(5):2089–2123, 2009. doi: 10.1111/j.1540-6261. 2009.01495.x.
H. Fushing and T. Roy. Complexity of possibly-gapped histogram and analysis of histogram (anoht). Royal Society Open Science, 2018.
H. Fushing, E. P.-T. Chou, and T.-L. Chen. Multiscale major factor selections for complex system data with structural dependency and heterogeneity. Physica A: Statistical Mechanics and its Applications, 630:129227, 2023.
H. Fushing, H.-W. Kao, and E. P.-T. Chou. Topological risk-landscape in metric-free categorical database. IEEE Access, 2024.
Stephen A. Hillegeist, Elizabeth K. Keating, Donald P. Cram, and Kyle G. Lundstedt. Assessing the probability of bankruptcy. Review of Accounting Studies, 9(1):5–34, 2004. doi: 10.1023/B:RAST.0000013627.90884.b7.
Andrei N. Kolmogorov. On logical foundations of probability theory. In Lecture Notes in Mathematics. Springer, 1983.
Stefan Lessmann, Bart Baesens, Hsin-Vonn Seow, and Lyn C. Thomas. Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research. European Journal of Operational Research, 247(1):124–136, 2015.
Deron Liang, Chih-Chou Lu, Chia-Fei Tsai, and Grace A. Shih. Financial ratios and corporate governance indicators in bankruptcy prediction: A comprehensive study. European Journal of Operational Research, 252(2):561–572, 2016. doi: 10.1016/j.ejor.2015.12. 018.
James A. Ohlson. Financial ratios and the probabilistic prediction of bankruptcy. Journal of Accounting Research, 18(1):109–131, 1980. doi: 10.2307/2490395.
A.-H. Omotayo, H. Fushing, Y. Wei, P. Guzmán-Delgado, G. Marino, and B. BlancoUlate. Evolving topological color landscape unravels the final stages of pistachio nut development and the incidence of blank nuts. Journal of the Royal Society Interface, 2025.
Duc-Tai Pham, Minh-Tri Nguyen, and Van-Hieu Tran. Hybrid smote-enn and feature selection for taiwanese corporate bankruptcy prediction. International Journal of Technology, 15(1):1–14, 2024.
Tyler Shumway. Forecasting bankruptcy more accurately: A simple hazard model. The Journal of Business, 74(1):101–124, 2001. doi: 10.1086/209665.
Taiwanese Bankruptcy Prediction. Taiwanese bankruptcy prediction [dataset], 2020. URL https://doi.org/10.24432/C5004D.
Junjie Wang, Xiaohui Li, and Yufeng Chen. Handling class imbalance in bankruptcy prediction: A f2-measure-based undersampling framework. PLOS ONE, 16(3):e0248700, 2021.