讲座题目：Regression Tree Credibility Model
主 讲 人：加拿大滑铁卢大学翁成国副教授
讲座摘要：Credibility theory is one of the cornerstones in actuarial science and has been widely applied for insurance premium prediction. In this talk I will introduce our research for an SOA-funded project jointly with Dr. Liqun Diao (University of Waterloo). We bring regression trees techniques into the credibility theory and propose a novel credibility premium formula, which we call regression tree credibility (RTC) premium. The proposed RTC method ﬁrst recursively binary partitions a collective of individual risks into exclusive sub-collectives using a regression tree algorithm based on credibility loss, and then applies the classical Buhlmann-Straub credibility formula to predict individual net premiums within each sub-collective. The proposed method eﬀectively predicts individual net premiums by incorporating covariate information in a very ﬂexible way, and it is particularly appealing to capture various non-linear covariates eﬀects and/or interaction eﬀects because no speciﬁc regression form needs to be prespeciﬁed in the method. Our proposed RTC method automatically selects inﬂuential covariate variables for premium prediction with no additional ex ante variable selection procedure required. The superiority in prediction accuracy of the proposed RTC model is demonstrated by extensive simulation studies.