IMPROVING INSURANCE TARIFFS BASED ON ACTUARIAL AND MACHINE LEARNING MODELS FOR INSURANCE RISK ASSESSMENT

This article examines the theoretical, methodological, and practical foundations of improving insurance tariffs through the integration of traditional actuarial methods and modern machine learning algorithms. The increasing complexity of insurance risks, growing availability of large-scale datasets, and rapid development of artificial intelligence technologies require insurance companies to adopt more accurate, transparent, and adaptive approaches to premium calculation. Traditional actuarial models, including generalized linear models, credibility theory, and frequency–severity modeling, provide a statistically grounded framework for estimating expected insurance losses. However, these methods may face limitations when dealing with nonlinear relationships, complex interactions, and high-dimensional risk factors. Machine learning techniques, such as random forests, gradient boosting, artificial neural networks, and extreme gradient boosting, offer additional opportunities to improve risk classification and predictive accuracy. The study proposes an integrated actuarial–machine learning framework for insurance tariff optimization, incorporating risk assessment, expected claim frequency, claim severity, operating expenses, and profitability requirements. Particular attention is given to predictive performance, model interpretability, fairness, regulatory compliance, and the economic implications of risk-based pricing. The article also presents an illustrative comparative analysis of actuarial and machine learning approaches and develops recommendations for implementing hybrid insurance pricing systems. The findings suggest that combining actuarial principles with machine learning techniques can strengthen risk differentiation and pricing efficiency, provided that the resulting models are appropriately validated, calibrated, and governed.