銀行信用風險評分應用資料探勘技術之比較研究

A Comparative Study of Data Mining Techniques for Credit Scoring in Banking

黃世禎(Shih-Chen Huang)

指導教授 ： 戴敏育

淡江大學/商管學院/資訊管理學系碩士班/碩士(2013年)

https://doi.org/10.6846/TKU.2013.00868

摘要

信貸對於銀行機構是重要收入來源，過去研究指出信用風險評分模型以邏輯斯迴歸和類神經網路分類方法較佳。本研究主要目的為提出較合適的信用風險評分模型以降低信貸風險並分析比較各分類模型正確率。
本研究提出利用企業資料探勘軟體建構四種信用風險評分模型，分別為決策樹法(Decision Tree)、邏輯斯迴歸(Logistic Regression)、類神經網路(Neural Network)、支持向量機(Support Vector Machine; SVM)，並進一步詳細比較17種分類模型之正確率，實驗結果顯示，支持向量機分類模型有較高正確率。本研究主要貢獻為利用資料探技術建立各種銀行信用風險評分之分類模型並比較其正確率，並證實支持向量機分類方法皆優於傳統分類方法。

關鍵字

分類方法 ； 資料探勘 ； 信用風險評分 ； 支持向量機 ； 賽仕企業採礦工具

並列摘要

Credit is becoming one of the most important sources of income for the banking institutions. Prior studies indicated that logistic regression and neural network had been performed better on credit risk scoring. The major purpose of the present study is to propose appropriate credit risk scoring models to reduce credit risk and compare the accuracy of various classification models. The study proposed using enterprise data mining software to construct four classifications predictive models, such as decision tree, logistic regression, neural network and support vector machine, and further compared their accuracy of 17 classification models. The experimental results show that support vector machine classification models perform better in terms of high accuracy. The main contribution of this paper is that we use data mining techniques to construct various classification models for credit scoring in banking and compare their accuracy, and evidence shows that support vector machine outperforms traditional classification methods.

並列關鍵字

Classification Method ； Data mining ； Credit Risk Score ； Support Vector Machine (SVM) ； SAS Enterprise Miner (SAS EM)

參考文獻

李逢嘉, 2010, "特徵選取為基礎之複合分類預測模式-以信用資料為例," 清華大學工業工程與工程管理學系學位論文2010年。

Berkson, J. 1944, "Application of the Logistic Function to Bio-Assay," Journal of the American Statistical Association (39:227), pp. 357-365.

Chen, Y., and C. Lin. 2006, "Combining SVMs with various Feature Selection Strategies,” Feature Extraction pp. 315-324.

Cho, B. H., H. Yu, K. Kim, T. H. Kim, I. Y. Kim, and S. I. Kim. 2008, "Application of Irregular and Unbalanced Data to Predict Diabetic Nephropathy using Visualization and Feature Selection Methods," Artificial Intelligence in Medicine (42:1), pp. 37-54.

Dash, M., and H. Liu. 1997, "Feature Selection for Classification," Intelligent Data Analysis (1:1-4), pp. 131-156.