讲座主题:Data-enriched Prediction of Insurance Risk
主讲人:尹晔博士(北京大学经济学院风险管理与保险学系)
时间:2022年10月19日 13:00
地点:腾讯会议(129304840)
主讲人简介:
尹晔,北京大学经济学院风险管理与保险学专业博士生。他的研究领域是保险经济学与保险科技,关注科技如何提高保险业的风险管理水平以及覆盖范围,并在《保险研究》《上海保险》《保险理论与实践》等期刊发表论文多篇。他曾在《中国银行保险报》等媒体刊物上发表文章,并直接或间接提交中央和上海市有关部门智库报告多篇。
We study the impact of big data and big data techniques on insurance risk assessment. We show that both big data obtained from digital technology and the Lasso-style approach, a modern big data selection technique, effectively improve the accuracy of insurance risk assessment. We find the OLS after Lasso not only select the most effective predictors but also improve the out-sample prediction. We further modify the adaptive group Lasso approach to identify the fields with the highest predictability of insurance risk, which can guide the collection of underwriting information and improve the accuracy of risk prediction.