MACHINE LEARNING APPLICATIONS IN ACTUARIAL RISK ASSESSMENT AND PRICING

Abstract

This study explores the potential of applying machine learning models in actuarial premium calculations by classifying policyholders based on their risk levels.
Among the various insurance products available in the market, health insurance known for its high loss ratio was selected as the focus of this research. Based on population morbidity data, the most significant variables were identified and used to develop and evaluate machine learning models that classify risk levels according to each category of the International Classification of Diseases (ICD). Furthermore, for each ICD category, the study examines the distribution patterns and descriptive statistics of the corresponding treatment cost data. Based on this analysis and the machine learning outputs, risk assessment recommendations are proposed using the z-score methodology.