客戶屬性分組與個人信用風險評分卡模型之建構

在實務的應用上，往往需要將連續型變數轉換成類別型變數，但如何在損失最少原始資料資訊的條件下，將連續型變數加以分組便成為很重要的課題，為了探討該議題，本研究採用等距與分量分組方法，將連續型變數各分成四組、五組及六組，本研究除了先以該銀行所提供的所有變數為考量的情況下進行模式建構，另外再分別以證據權數(weight of evidence, WOE)/訊息值(information value,IV)、逐步選取法、刪除異常變數、相關係數等四種方法來選取變數，並將這五種篩選出的變數組合使用邏輯斯迴歸來建立信用風險評分模型，最後則採用吉尼係數、AUROC與正確率這三種測量指標來對建構的三十種模型進行比較與評估。本研究實證結果顯示，等距分組方法比起分量分組方法有較好的區別能力。

關鍵字

信用風險評分模型；邏輯斯迴歸；正確率；吉尼係數

並列摘要

Logistic regression model has been more commonly adopted by the credit card industry due to its interpretable feature in credit scoring. The main purpose of the research is to build a credit scoring model for personal loans with the logistic regression model using 2 different classification rules on attributes. Intuitively, equal-length and equal-proportion rules are adopted in this research to the group assigning. The features used are the original variables provided by the credit card department in Taiwan financial holding company. Taking the precision as well as parsimony into consideration, additional feature selections are performed using different criteria. It includes the stepwise procedure through the logistic regression model, weight of evidence/ information value, abnormal deletion and correlation coefficients criteria. The performance of the models are evaluated by population stability index and AUROC, area under the receiver operating characteristic, or equivalently, and gini coefficient. The empirical evidence supports that equal length classification rule outperforms the equal-proportion classification rule.

並列關鍵字

credit scoring model ； logistic regression ； AUROC ； gini coefficient

參考文獻

analysis and the prediction of corporate bankruptcy.

[6]Beaver, W. (1966). Financial ratios as prediction of

failure. Empirical research in accounting: selected

studies. Journal of Accounting Research, 4, 71–111.

for scorecard construction in credit scoring. Journal of

國際替代計量

客戶屬性分組與個人信用風險評分卡模型之建構

全文下載

主題瀏覽