Application of fuzzy logic as an alternative to classical approaches in automated decision-making systems of credit organizations

Abstract

In Ukraine, banks and financial companies are actively switching to digital technologies, which gives a chance to implement new decision-making methods. This can make lending more profitable without losing control over risks. Automated systems have become an essential part of lending activities, helping financial institutions process a significant volume of loan applications almost without human involvement in a matter of seconds. But conventional mathematical models are not always suitable, since they struggle to capture all the nuances of lending and the unpredictability of the financial market. This study is focused on the application of fuzzy logic to assess creditworthiness using the example of a microfinance organization. Fuzzy logic is an effective way to enhance accuracy and quality of automated decisions when there is a lack of information or uncertainty. Research shows that systems based on fuzzy logic have certain advantages compared to conventional methods, especially when it is necessary to consider qualitative factors and work with language variables. The main advantage of fuzzy logic in automated credit scoring is the ability to transform uncertainty and expert opinions into understandable rules, which makes the model more robust to inaccuracies, missing data, and hidden information, compared to classical statistical approaches that are more demanding on data and less flexible in doubtful situations. The relevance of this study is further enhanced by the regulatory challenges that Ukrainian financial institutions have been facing in recent years. The implementation of IFRS 9 standards and the gradual approach to Basel III requirements create a need for more complex models for estimating expected credit losses (ECL) and increase the requirements for the explainability of algorithmic decisions in lending. The National Bank of Ukraine is strengthening control over the quality of risk models used in the financial sector for calculating provisions and making credit decisions. These factors emphasize the importance of developing methods that combine statistical accuracy with regulatory transparency