GIRO 2022, Session E4, 16.35
IFoA Machine Learning in Reserving WP aims to advance advanced analytics methods within non-life claims reserving. One major stumbling block is the availability of granular data. While the ultimate goal is to apply machine learning on real claims data, simulated data may be beneficial in the intermediate steps. First, researchers may create specific trends that the algorithms should capture. Second, training the algorithm on a simulated data set may allow one to separate subjective effects.