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144期讲座-杨羿(7月2日)

报告题目：New Techniques for ModelingNon-life Insurance Claims

报告时间：2018年7月2日(周一)上午9:30—11:00

报告人：

杨羿 现就职于加拿大麦吉尔大学数学与统计学系，担任助理教授，明尼苏达大学统计学博士，师从邹辉教授。主要研究领域为机器学习、计算统计、高维统计推断、非参数分类和回归问题、精算统计、商业统计等。其编写的精算软件被美国和加拿大多个保险公司使用。

摘要：

Tweedie’s Compound Poisson model is a popularmethod to model data with probability mass at zero and non-negative, highlyright-skewed distribution. Motivated by wide applications of the Tweedie modelin various fields such as actuarial science, we investigate a grouped elasticnet method and a boosted nonparametric method for the Tweedie model in thecontext of the generalized linear model.

For the grouped elastic net method, in order toefficiently compute the estimation coefficients, we devise a two-layeralgorithm that embeds the blockwise majorization descent method into aniteratively re-weighted least square strategy. In together with the strongrule, the proposed algorithm is implemented in an easy-to-use R packageHDtweedie, and is shown to compute the whole solution path very efficiently.

On the other hand, the linear form of thelogarithmic mean in the Tweedie GLM sometimes can be too rigid for manyapplications. As a better alternative, we propose a boosted nonparametricTweedie model for pure premiums and use a profile likelihood approach toestimate the index and dispersion parameters. To our knowledge, there is noexisting nonparametric Tweedie method available before this work. Our method iscapable of fitting a flexible nonlinear Tweedie model and capturing complexinteractions among predictors. We have also implemented this method in auser-friendly R package that includes a nice visualization tool forinterpreting the fitted model.

报告地点：中央财经大学学术会堂南楼506（精算院会议室）

欢迎各位老师和同学积极参加！