Geo-referenced data appear in many areas of applications including agricultural field trials, environmental and atmospheric studies (temperature and precipitaion fields, air-pollution), medical sciences (medical image data, disease mapping, gene-environment interactions), remote sensing, etc. to name a few. Analysis of such data sets requires special considerations to account for local interactions in space and time. Standard statistical methods developed under independence assumption are often inadequate to capture the effects of such spatial interactions. This is an introductory course on Spatial Statistics that aims to introduce students to the spatial aspect of geo-referenced data. The course is intended for M.S. and advanced undergraduate students in Mathematics and Statistics and related fields. The course will cover some of the basic concepts (approximately 1-2 weeks on each topic) including: 1) Form of spatial dependence and its impact on estimation accuracy, 2) Mean and covariance functions for spatial processes, 3) Kriging (or spatial prediction) and its variants, 4) Estimation of parameters and uncertainty quantification, 5) Spectral methods, 6) Simple statistical models for point source data, 7) Visualization, and 8) Applications to real datasets. Additional topics may be added if time permits. The course will emphasize statistical theory and methodology for analysis of geo-referenced data, and provide ample applications involving real datasets. Computation and visualization of spatial data will be done using R. Students are expected to do a group project to gain hands on experience with analyzing spatial data. Prerequisites: Math 493/494.
Course Attributes: FA NSMAR NSMAS NSM
Section 01Topics in Statistics
INSTRUCTOR: LahiriView Course Listing