PubH 8472: Spatial Biostatistics (Spring 2008)

• Course information and syllabus in pdf format

• Copies of the course notes and the most recent midterm exam used in class

• Homework assignments made to date, with due dates

• A list of typos in the first and second printings of the Banerjee, Carlin, and Gelfand text (corrected in the third and subsequent printings)

• Most of the data sets and WinBUGS code used in the book

• The BUGS software package for carrying out the Bayesian approaches described in the book

• A page written by Brad and Joe Koopmeiners ( josephk@biostat.umn.edu) on a new and easy-to-use (but sadly still largely undocumented) collection of R routines for calling WinBUGS from R (written by WinBUGS founding father Andrew Thomas) called BRugs.

• If this doesn't work, you may want to instead try Andrew Gelman's software for calling WinBUGS from R

• A page written by Ms. Yue Cui on getting boundaries from ArcView (version 8.x or lower) into WinBUGS , so that the latter can build the adjacency matrix and enable MCMC sampling and posterior mapping over any set of regional boundaries available in ArcView!

• Another page by Cui on drawing 3-d drop-line scatterplots with geographic boundaries on the same axes in R , like on the cover of the Banerjee et al. text

• The exciting new spatialBayes (spBayes) package for doing hierarchical modeling of point-level spatial data that is far more general than that in geoR and far, far faster than WinBUGS! This package is now downloadable directly from within R; see these instructions for help. You can also check out several examples stolen from Sudipto Banerjee's Duke Summer Institute short course:
  • Code illustrating the ggt.sp function for univariate spatial regression - a simulated data example.
  • Code illustrating a univariate spatial regression example.
  • Code illustrating the ggt function

Here are some files you'll need for the homework and lab demos:

• booknew.R, the commands from the Chapter 2 R tutorial (pp.52-65)
• MinnesotaChoropleth.R, and TA_Lab_hw3_EIREdata.R, R tutorials that are similar (but not identical) to the Chaper 3 S+SpatialStats tutorial (pp. 88-95)
• SpatialPredictionsNew.odc and NuggetPredictionsNew.odc, the two Bayesian kriging examples using WinBUGS 1.4
• SpatialPredictions.odc and NuggetPredictions.odc, the two Bayesian kriging examples using WinBUGS 1.3.1
• SpatialPredictionsOld.odc and NuggetPredictionsOld.odc, the two "old" Bayesian kriging examples, also using WinBUGS 1.3.1
• BuildYourBUGS.txt and BuildYourBUGSNew.txt, the two versions where you actually build the kriging code yourself (as in the book), using WinBUGS 1.3.1 and 1.4, respectively
• BatonRougeGeostat.odc, the WinBUGS code for the Baton Rouge geostat model (Ch 5, problem 7)
• hull.odc, the full WinBUGS code for the Mike Hull Interstim (multivariate) example
• Lipsbrad.odc, the full WinBUGS code for the Scottish lip cancer example
• coloreccols.dat, the colorectal late detection data in standard column format
• colorecbugs.dat, the colorectal late detection data and screening covariate data in WinBUGS format
• poly.S and poly.R, the S (or R) routine that extracts the county polygons for any state in the US
• MACfrailty.odc, the WinBUGS code for the MAC frailty example
• MACspatialfrailty.odc, the spatial version of this code!