报告题目： Bayesian Tree Models for Insurance Pricing

报告人：Georgios Aivaliotis

邀请人：王璐

讲座时间：2026年3月26日（星期四）10:30-11:30

讲座地点：犀浦校区三号教学楼X30312

报告摘要：Insurance pricing requires the mapping of factors to claims size and frequency to be accurate as well as explainable. In this talk, I will describe some recent work on using Bayesian Trees for claims frequency and size as well as compound (total) claims modelling. I will introduce the main principles of BCARTs, their adaptation for suitable distributions for claims modelling and an appropriate model assessment measure, an adaptation of DIC. Performance of the models will be examined on both simulated and real claims data. The talk will be based on the two paper published with Y. Zhang, L, Ji and C. Taylor.

Zhang, Y., Ji, L., Aivaliotis, G., Taylor, C., 2024. Bayesian CART models for insurance claims frequency. IME, 114, 108-131.

Zhang, Y., Ji, L., Aivaliotis, G., Taylor, C., 2025. Bayesian CART models for aggregate claim modeling. IME, 103-136.

主讲人简介：Georgios Aivaliotis is Associate Professor in Statistics and Financial Mathematics at the University of Leeds. He started his research career as a probabilist working in stochastic control. He has since turned into statistical methodology, where he exploits probabilistic ideas in the intersection of statistics and ML. His current research interests are in Bayesian methods for trees, explainability of machine learning, functional models for complex data (like networks) and non-linear survival models.