COMPARATIVE ANALYSIS OF MACHINE LEARNING MODELS IN THE TASK OF ASSESSING THE FINANCIAL STABILITY OF ENTERPRISES: ACCURACY, INTERPRETABILITY, AND PRACTICAL APPLICABILITY

Modern machine learning models are increasingly used in financial analysis, particularly in assessing the stability of enterprises. The purpose of this article is to compare logistic regression, random forest, XGBoost, and neural network models in terms of predictive accuracy, interpretability, and practical applicability in evaluating corporate financial stability. The methodological framework includes a theoretical overview of models, comparison across key criteria (data requirements, noise robustness, explainability), and identification of conditions under which each algorithm is most effective. The study emphasizes that model accuracy must be balanced with transparency and usability for decisionmaking. Based on the findings, the article offers recommendations on model selection depending on the objectives of analysis, data quality, and the analytical maturity of the organization. A phased and hybrid implementation of intelligent risk assessment systems is proposed as a strategic approach.