財務危機預測方法與比較

Abstract

本研究針對財務危機預測，開發了一種使用機器學習技術及AI方法的財務危機預警模型。考慮到財務危機的複雜性，本研究結合了財務比率、公司治理指標、市場數據和總體經濟指標等多方面指標，並運用了羅吉斯迴歸、隨機森林、支援向量機、神經網路在內等多種機器學習演算法及AI方法來建立預測模型。 研究結果顯示，單純貝氏分類器在各演算法中表現最佳，特別是在召回率方面具有顯著優勢，能有效辨識財務危機公司，降低金融機構的風險。進一步指出，經濟成長率和公司治理指標是影響財務危機預測的關鍵因素。本研究的貢獻在於提供了一個綜合多方面思考之財務危機預測模型，為金融機構在風險管理和決策方面提供了有效之工具。

This study develops a financial distress early warning model that employs machine learning techniques and AI methods to predict financial distress. Given the complexity of financial distress, this study integrates a variety of indicators including financial ratios, corporate governance metrics, market data, and macroeconomic indicators. The prediction model was constructed using a variety of machine learning algorithms and AI methods, such as logistic regression, random forests, support vector machines, and neural networks. The results indicate that the Naive Bayes classifier performed the best among all algorithms, particularly in terms of recall rate, effectively identifying companies at risk of financial distress and reducing risk for financial institutions. Furthermore, it was found that economic growth rates and corporate governance metrics are critical factors affecting the prediction of financial distress. The contribution of this study lies in providing a comprehensive financial distress prediction model that incorporates multi-dimensional thinking, offering effective tools for risk management and decision-making in financial institutions.