教育部人文社科重点研究基地中央财经大学中国精算研究院学术活动

精算论坛第188期讲座

(2021年7月2日)

讲座主题：Data or Models – Which is More Important? Evidences from Claims Frequency Modelling with Telematics Driving Data

主讲人：高光远

   中国人民大学统计学院副教授、应用统计科学研究中心研究员。主要研究领域包括非寿险准备金评估方法，贝叶斯统计和MCMC，车险定价模型，车联网大数据分析，copulas，死亡率预测模型等。在精算的顶尖期刊发表多篇论文，如《ASTIN Bulletin：theJournalofInternationalActuarialAssociation》，《Insurance：Mathematics and Economics》、《MachineLearning》等；由Springer出版独著《Bayesian claims reserving methods in non-life insurance with Stan》；参与编著多本教材；建设慕课《金融数学》、《非寿险精算学》；主持国家自科青年项目，Society of Actuaries科研项目等；参与国家社科重大项目等。

Abstract

There is a long debate on the relative importance of data and models. In this study, we provide evidences from claims frequency modelling with telematics driving data. Under the setting of claims frequency prediction, we study the benefits brought by advanced machine learning algorithms and telematics car driving data. For a small portfolio of exposure without telematics driving data, the generalized linear model performs as well as the gradient boosting and XGBoost. With telematics driving data, given that we have done sufficient feature engineering of telematics data, machine learning algorithms are only marginally better than the generalized linear model. In this particular case study, compared with traditional regression modelling, the benefit of machine learning algorithms is automatically feature engineering but at the cost of a heavier computing burden and more difficult interpretation. In this particular case study, the claims frequency prediction is largely improved due to the telematics driving data using either the generalized linear model or machine learning algorithms. 

讲座时间：7月2日 上午9：00-10：30 

报告地点：腾讯会议（会议ID：853 470 461）

邀 请 人：韦晓

欢迎各位老师和同学积极参加！