统计与管理学院2015年学术报告第21期
【主 题】 A Critical View of Asymptotic Studies in Statistics
【报告人】 Hao Zhang 博士
Department of Statistics，Purdue University
【时 间】 2015年5月28日（星期四）15:00－16:00
【地 点】 上海财经大学统计与管理学院大楼1208室
【语 言】 英文
【摘 要】 Although asymptotic theories can provide deep insights into a statistical model or method, I will provide a critical view of the current status of asymptotic studies in the context of big data. Perhaps no area serves as a better example to show the roles of asymptotics than spatial statistics. First, there are different asymptotic frameworks (e.g., infill, increasing domain and mixed asymptotic framework). Second, asymptotic results are different under the different asymptotic frameworks. Third, for any given problem where only a finite sample is available, nobody knows which asymptotic framework is involved. Hence comes a legitimate question: How are the asymptotics help with the data analysis? This talk will be conducted around this question and will demonstrate through some known asymptotic results that asymptotics could be very helpful to provide insight into an inference problem and to enhance our understanding of a statistical method. On the other hand, we may also question the relevance of some asymptotic results, particularly in the context of big data.
