学术预告:Bayesian inference and variable selection for quantile autoregressive models with explanatory variables

报告题目:Bayesian inference and variable selection for quantile autoregressive models with explanatory variables

报告时间:2023年7月10日(星期一)14:00-16:00

报告地点:理学院B311

主办单位:理学院

报告人:杨凯

报告人简介:杨凯,副教授,博士生导师,现任长春工业大学数学与统计学院统计系主任,吉林省高层次人才,曾赴日本岛根大学学术访问,兼任全国工业统计教学研究会理事,中国现场统计研究会大数据统计分会理事,中国现场统计研究会多元统计分析分会理事,中国现场统计研究会经济与金融统计分会理事,吉林省工业与应用数学学会理事。主要研究领域包括时间序列分析、高维数据分析、贝叶斯分析等。主持国家自然科学青年基金项目1项,吉林省自然科学基金面上项目1项,吉林省博士后基金择优资助项目1项,吉林省产业关键核心技术攻关项目1项(子课题负责人),吉林省教育厅科学研究项目1项,横向科研项目1项,以主要参加人身份参与国家级、省部级科研项目4项。以第一作者、通讯作者身份在Applied Mathematical Modelling,Computational Statistics&Data Analysis等杂志发表SCI论文16篇,其中二区以上论文5篇,ESI高被引论文1篇。荣获第四届全国高校数学微课程教学设计竞赛全国一等奖1项,指导学生参加学科竞赛获得国家级奖项10余项。

报告内容简介:In recent years, Bayesian variable selection methods have achieved more and more attention. This study focuses on the Bayesian inference and variable selection problems for a class of quantile autoregressive models with explanatory variables (QAR-X). We introduce three Bayesian variable selection methods for the QAR-X model. The Gibbs sampling algorithms are developed for each method by setting different priors. The numerical simulations suggest that the Gibbs sampling algorithms converge fast and Bayesian variable selection methods are reliable. A real example is given to analyze the relationship between the count of total rental bikes and five explanatory variables. Both simulations and data example indicate that the proposed methods are feasible, reliable, and appropriate for analyzing the gold price and Bike Sharing data sets.