报告题目：Frequentist model averaging for Poisson and zero-inflated Poisson regression models

报告人：喻达磊教授

报告摘要：This paper considers frequentist model averaging for estimating the unknown parameters in that have received widespread attention in many applications, the Poisson and zero-inflated Poisson regression models. In Poisson regression model, we propose a weight choice method based on the minimisation of an unbiased estimator of an expected Kullback-Leibler-type loss function. For its zero-inflated counterpart, our proposed weight choice procedure is based on the minimisation of an unbiased estimator of a conditional quadratic loss function. We prove that the resulting model average estimators enjoy optimal asymptotic properties and improved finite sample properties over a range of model selection estimators. Our proposed methods are illustrated by two real data examples.

报告时间：12月8日（周四）下午16:10—17:10

报告地点：腾讯会议562-346-020

报告人简介：喻达磊，博士（香港城市大学），云南财经大学教授，博士生导师。研究领域为随机效应模型、混合模型以及空间计量模型的模型选择、模型平均和估计理论等。已在包括JRSS-B，JASA，JBES和中国科学：数学在内的国内外统计学期刊上发表论文十余篇。主持国家自然科学基金项目三项，入选了云南省中青年学术和技术带头人后备人才和云南省“万人计划”的“青年拔尖人才”专项。