stats352 - spatial statistics

Instructor:	Prof. J. Taylor Sequoia Hall #137 Email 723-9230
Schedule:	TTh 12:50-2:05
Location:	Hewlett 102
Office Hours:	M 10:00-12:00 or by appointment
Reference:	Statistics for Spatial Data. N. Cressie
TA & Office Hours:	Patrick Perry Office hours: Th: 2:05-4:05, Sequoia Hall #238
General Outline:	This course is about spatial statistics, focusing on geostatistics, Markov random fields and point processes. Tentative topics include: Geostatistics: stationary random fields; variogram / covariance function; kriging; misspecification of covariance function; fixed-domain asymptotics; kriging and splines; spatio-temporal modelling; correcting non-isotropy. Markov random fields: conditionally specified Gaussian and binary fields; simulation; parameter estimation; image analysis and segmentation; disease mapping. Point processes: Poisson processes; stationary point processes; tests for complete spatial randomness; other models of point processes; scan statistics.
Prerequisites:	A course in stochastic processes, such as STATS 217, as well as a course in mathematical statistics, such as STATS 200.
Evaluation:	Assignments: 50% Presentation or project: 50% Note: I originally planned to have students do presentations but this will depend on the number of students we stabilize at. If not presentations, then students will have to do projects.
Assignments:	Assignment 1 Assignment 2 Assignment 3
Notes:	Notes on random fields Notes on Ising models
R examples:	Kriging and variograms Disease mapping Point processes Point processes (cluster models) Point processes (Gibbs models) Point processes (marked models)