2022-03-22 信息来源：数学科学学院sx226
举办单位：数学科学学院
负责人：牛奉高
电话：13994212988
活动主题：testing and estimation of change-point in mean of dependent data---a trimmed-mean-based cusum approach
形式：学术报告
内容摘要：it is well-known how to test and estimate the change-point in the heavy-tail time series is greatly open. this paper proposes a trimmed-mean-based cusum approach to test and estimate the change-point in mean of dependent data. it is shown that the self-normalized cusum test converges to the maxima of a function of standard brownian motion and has a power approaching to 1, asymptotically. furthermore, we show that the trimmed-mean-based least squares estimator (lse) of change point is stochastically bounded and its limiting distribution is the minimizer of a random walk. simulation study is carried to assess the performance of our new methodology and two real examples are given. our approach can be applied for both heavy-tailed and the finite variance time series and provide a robust inference of the change-point in the data.
主讲人基本情况：凌仕卿教授, 香港科技大学数学系讲座教授(chair professor)及金融科技课程联合主任, journal of econometrics 杂志fellow, ims (institute of mathematical statistics) 学会fellow, mssanz (modelling and simulation society of australia and new zealand) 学会fellow, 荣获mssanz学会2013双年度勋章, 目前正担任其研究领域主要国际期刊《journal of time series analysis》联合主编,以及《statistica sinica》, 《计量经济学报》 与其他三个杂志的副主编。连续三年(2019, 2020, 2021)入选美国斯坦福大学(john p. a. ioannidis教授团队)发布全球前2%终身科学影响力排行榜，凌仕卿教授主要研究领域是经济与金融时间序列分析.
听众范围：数学科学学院统计学专业学生
举办时间：2022年3月23日
举办地点：线下数学科学学院sx226，线上腾讯会议
报告类型：理科类