Analyzing spatial data locally
报告人： Tailen Hsing, Michael B. Woodroofe Collegiate Professor of Statistics, University of Michigan, Co-editor of Annals of Statistics
时间：2017-05-18 14:00 ~ 15:00
地点：Room 217, Guanghua Building 2
Stationarity is a common assumption in spatial models. The justification is often that stationarity is a reasonable approximation if data are collected in a sufficiently small region. In this talk we first review various known approaches for modeling nonstationary spatial data. We then examine the notion of local stationarity in more detail. In particular, we will consider a general nonstationary spatial model whose covariance behaves like the Matern covariance locally and a corresponding inference approach for the model based on dense gridded data.
About the Speaker:
Prof. Tailen Hsing obtained his bachelor’s degree in Mathematics from National Taiwan University in 1978, and his Ph.D. degree in statistics from University of North Carolina in 1984. Prof. Hsing’s research interests include extreme value theory, limit theory under dependence, functional data and spatial data. Currently, his focus is on the last two areas. For functional data, he is considering the estimation of high-dimensional parameters, such as the cross covariance, that characterize dependence between functional variables. For spatial data, he is interested in the inference of spatial processes that are locally stationary, especially in the context of densely observed data.