Methods for diagnostics and forecasting SMEs creditworthiness using artificial intelligence

Abstract

Introduction. The impact of multidirectional external macroeconomic and regional factors of the economic environment in conditions of uncertainty and increased risks causes significant difficulties in diagnosing, assessing and forecasting the creditworthiness of financial and credit support recipients and borrowers (micro, small and medium-sized enterprises) in the Russian Federation. Theoretical analysis. The author systematized the basic mathematical methods and models for assessing and forecasting the level of creditworthiness of micro, small and medium-sized businesses in foreign and Russian practice, using modern systems and technologies of artificial intelligence and machine learning methods. Empirical analysis. The author proposed the results of approbation of methodological approach for express diagnostics of the financial and economic condition and forecasting the creditworthiness of SMEs in the Krasnodar krai for the period of 2014–2017, based on expert assessment methods, economic analysis and fuzzy logic systems, which form the credit rating of SMEs considering their industry affiliation. Results. In this study, the author has determined the advantages and disadvantages of methods and models for diagnosing creditworthiness and credit scoring from the perspective of their application for various categories of SMEs. As it is shown that the most promising and mathematically reliable models for credit scoring and risk assessment of financial support and lending to various enterprises in the SME sector at different stages of their life cycle both in Russia and abroad are computerized models and expert systems, based on such methods and technologies of Artificial Intelligence, as fuzzy logic systems, artificial neural networks, support vector machines, ensemble methods (random forest method), as well as intelligent information systems, hybrid models and hybrid systems. The study reveals that their combination with each other will allow to achieve synergistic and system effects in the interaction between lenders and borrowers (SMEs) and timely avoid their bankruptcy.

Keywords

• creditworthiness
• smes
• diagnostic and forecasting methods and models
• artifi cial intelligence
• machine learning
• management decisions