Actuarial intelligence in auto insurance: claim frequency modeling with driving behavior features and improved boosted trees (Q2172034)

The paper focuses on the frequency of usage-based insurance (UBI) and related claims, proposing an approach based on machine learning algorithms together with classic actuarial methodologies. After presenting the UBI dataset, the authors describe the boosted models, instead of traditional linear models, and the forecasting algorithm. The study deepens the effectiveness and robustness of the model on two auto insurance datasets and the role of the specific driving behaviour is explored, with quantitative implications. Technical details of the algorithms used are provided in the Appendix.