search for




 

Study on credit rating model using explainable AI
Journal of the Korean Data & Information Science Society 2021;32:283-95
Published online March 31, 2021;  https://doi.org/10.7465/jkdi.2021.32.2.283
© 2021 Korean Data and Information Science Society.

Ye Eun Chun1 · Se Bin Kim2 · Ja Yun Lee3 · Ji Hwan Woo4

1234AI Competency Center, Shinhan Bank
4Korea University School of Management of Technology
Correspondence to: 1AI Competency Center, Shinhan Bank, 55, Sejong-daero, Jung-gu, Seoul 04513, Korea.
2AI Competency Center, Shinhan Bank, 55, Sejong-daero, Jung-gu, Seoul 04513, Korea.
3AI Competency Center, Shinhan Bank, 55, Sejong-daero, Jung-gu, Seoul 04513, Korea.
4Corresponding author: AI Competency Center, Shinhan Bank, 55, Sejong-daero, Jung-gu, Seoul 04513, Korea. E-mail: jihwan woo@korea.ac.kr
Received December 21, 2020; Revised January 21, 2021; Accepted January 27, 2021.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract
As artificial intelligence technology develops, cases of applying it to the financial industry are increasing. However, the biggest drawback is that understanding the process how the results are derived is not intuitive because most of its relationships are non-linear. Therefore, the process of deriving results using artificial intelligence is sometimes expressed as a black box. Recently, the EU created a new privacy regulation, guaranteeing the right of customers to request service providers for explanations about the results obtained by artificial intelligence algorithms. In other words, to apply artificial intelligence technology in the financial industry, not only high precision but also the ability to explain the results must be considered. In this paper, using various externally disclosed credit information data, an artificial intelligence-based credit rating algorithm was proposed. Also, for the results derived by artificial intelligence, we introduced an algorithm to calculate and distinguish which of the various characteristics of the data has the most significant effect. Finally, we further expanded this by applying the method of explaining the modified result to the financial data to explain when there is a change in the result derived by artificial intelligence. This research has great significance as it confirms that the proposed method can provide explanatory power when introducing artificial intelligence technology in financial services.
Keywords : Credit rating model, digital finance, explainable AI, XAI.