DEVELOPMENT OF A METHODOLOGY FOR CONTINUOUS AI-MONITORING OF THE FINANCIAL STABILITY OF RETAIL TO REDUCE LOSSES FROM BANKRUPTCY

Abstract

Amid high macroeconomic volatility and the ongoing transformation of consumer behavior, traditional methods for assessing the financial stability of retail chains suffer from significant inertia and exhibit low predictive capabilities. This article proposes a methodology for the continuous monitoring of retail entities’ financial stability, leveraging artificial intelligence (AI) and machine learning (ML) technologies. A multilevel architecture for collecting and processing heterogeneous data–encompassing financial, operational, behavioral, and alternative sources–has been developed. Furthermore, the selection of ensemble algorithms and deep learning models to predict the probability of default (PD) is rigorously substantiated. Particular emphasis is placed on Explainable AI (XAI) to ensure model interpretability, as well as on the system's integration into existing business processes. This approach minimizes false alarms and substantially mitigates both direct and indirect losses arising from the bankruptcy of counterparties or internal business units.