Machine Learning Approaches for Insurance Pricing: A Case Study of Public Liability Coverage in Morocco

Abstract

As data-driven technologies continue to advance, the importance of machine learning (ML) techniques in enhancing the accuracy of insurance premium calculations cannot be overstated. That is especially crucial because traditional actuarial methods often fail to incorporate individual risk factors fully. This limitation significantly impacts the insurance industry's capacity to accurately determine premium prices, which in turn affects financial stability and customer satisfaction. This research seeks to assess the efficiency insurance premium calculations through different regression models in ML, including polynomial, decision tree, random forest, and gradient boosting. The study employs rigorous analysis techniques using a comprehensive dataset from a Moroccan vehicle insurance company, including claim history and insurance categories. The dataset is split into training and testing sets to assess the accuracy of the ML models using metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and the coefficient of determination (R-squared). Initial findings suggest that ML models greatly surpass traditional actuarial methods, indicating the potential for machine learning to transform premium pricing strategies. That could result in more customized and financially sustainable outcomes within the insurance industry. The findings of this paper are likely to contribute to national insurance premium policy and expand the existing literature on this subject in Morocco.