INTELLIGENT METHODS OF ANALYSIS AND FORECASTING OF ENTERPRISE FINANCIAL STABILITY BASED ON MACHINE LEARNING

This article examines intelligent methods of analyzing and forecasting the financial stability of enterprises based on machine learning. The relevance of the study lies in the need to improve the accuracy of financial assessments of companies in the context of economic instability in 2023–2025. Using data from 187 Russian manufacturing companies for the period 2020–2025, we compared five machine learning models: logistic regression, random forest, gradient boosting (XGBoost), support vector machines, and multilayer perceptron. Ensemble methods (random forest and XGBoost) demonstrated the highest predictive accuracy, with F1 scores reaching 0.89 and 0.91, respectively, and areas under the receiver operating characteristic (ROC) curves (AUC-ROC) of 0.94 and 0.96, respectively. The current liquidity ratio, return on assets, debt burden, and industry index were identified as key predictors of financial instability. The developed predictive tools enable the identification of pre-bankruptcy states of companies with a lead time of up to two quarters, as confirmed by the results of out-ofsample testing on 2025 data. The practical significance of this work lies in the development of a methodological approach that can be integrated into the financial monitoring and risk management systems of commercial banks and enterprise analytical departments.