A novel fuzzy credit risk assessment decision support system based on the python web framework

ABSTRACT

When a financial institute constructs a credit risk assessment model, qualitative, and quantitative data must be considered at the same time due to differences in the attributes of each form of data. While qualitative data are usually relied on professional appraisers’ judgment, 2-tuple fuzzy linguistic representation could assist to organize appraisers’ scores to avoid any loss of information. In practice, a financial institute needs to regularly update its credit risk assessment model to maintain correct assessment results. However, every update involves a lot of numerical experiments using multiple systems or even software packages for evaluating the effect of different sampling methods and classifiers to construct a suitable model for the updated dataset. Such an assessment process is time-consuming with many repetitive processes. This study applied the latest web-based technology to develop a fuzzy decision support system (DSS) that used logistic regression as the classifier combined with different sampling methods and model threshold settings to make data preprocessing and model fitting process more structured and efficient. This DSS was written by Django, a Python web framework, using the RESTful architecture that has good database imaging mechanism and good interaction between the front-end (user side) and the back-end (service side). After verification of multiple actual data set, the average time spent on constructing one model combination per dataset is less than 1 min, which indeed is significantly shorter than the original time required for credit risk assessment and provided a practical tool for the related work of credit risk assessment.

KEYWORDS:

• Risk assessment
• decision support system
• python web framework2-tuple fuzzy linguistic representation model
• logistic regression

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Funding

This work was supported by the Ministry of Science and Technology, Taiwan [MOST 107-2410-H-145-001, MOST 108-2410-H-009-050, MOST 108-2410-H-145-001].

Notes on contributors

Yung-Chia Chang

Yung-Chia Chang is a professor in the Department of Industrial Engineering and Management at National Chiao Tung University, Taiwan. She received her Ph.D. in Industry Engineering from Texas A&M University, College Station, Texas, USA. Her research interests include Supply Chain Management, Total Quality Management, Data Mining and Smart Manufacturing.

Kuei-Hu Chang

Kuei-Hu Chang is a professor in the Department of Management Sciences at ROC Military Academy, Taiwan. He received her Ph.D. in Industrial Engineering and Management from National Chiao-Tung University in 2008. His research is mainly in the fields of Fuzzy Logic, Soft Computing, Data Analytics, Data Mining, and Reliability.

Yi-Hsuan Huang

Yi-Hsuan Huang is a graduate student in the Department of Industrial Engineering and Management at Ntaional Chiao Tung University, Taiwan. She received her Bachelor’s degree in Industrial Engineering from Chung Yuan Christian University, Taiwan, and is working on her Master degree. Her research interests include Data Analytics, Data Mining, and Applied Statistics.

(Received April 2019; revised February 2020; accepted May 2020)