企業信用評等模型－以營造業為例

本研究目的，是將資料採礦進行的分析流程，導入企業信用評等模型的建置程序，針對內部評等法中的企業型暴險，根據新版巴塞爾資本協定與金管會的準則，建立信用評等模型，投入模型的變數，分為財務變數以及總體經濟變數，最後模型評估結果決定使用羅吉斯迴歸模型。本研究所建構出的信用評等系統分為8個評等等級，違約的機率隨評等遞增，以第8等作為違約戶的評等結果，本研究建構之模型具有一定的穩定性與預測效力，並且皆通過新巴塞資本協定與金管會的各項規範，顯示本研究之信用評等模型能夠在銀行授信流程實務中加以應用。

關鍵字

資料採礦；信用評等；違約機率

並列摘要

The purpose of this research is to introduce the analysis procedure of data mining into the corporate credit rating model and to use the financial variable and the economic variable to create a credit evaluation model that aimed at the risk exposed in the corporations. The credit evaluation is based on the guidance set forth by the New Basel Capital Accord and the Standard of Financial Supervisory Commission. After the final evaluation of the model, the study decided to use the Logistic Regression model. The credit rating system is categorized into eight levels where the eighth level has the highest probability of being default. The resulting model formulated by this study exhibits stability and predictive capability that is within the accordance of the New Basel Accord and the Standard of Financial Supervisory Commission and has demonstrated its ability for its application in a real case in the corporations.