Further Statistical Methods, HT07

The practical exercise for week 6 will be concerned with graphical models and based on use of the software MIM which can be downloaded from www.hypergraph.dk.

In addition it will be based on mimR which is an R-package which provides interface between R and MIM, so that MIM can be run from R and the results of MIM-computations can be saved as R-objects.

First part of exercise is here, together with the associated MIM-file, should you prefer to analyse the data in MIM.

For the second part, you should perform a similar analysis of data taken from Christensen (1990). Log-linear models. Springer.

It concerns an American study of attitudes towards abortion in relation to age, sex and race. The following information was recorded for the participants in the study:

R: Race (White,non-white)
S: Sex (Male, female),
O: Opinion (For, against, undecided),
A: Age (18-25, 26-35, 36-45, 46-55, 56-65, 66+).

Here is the MIM data file and the raw data for reading into R, for example using this script and here is a solution sketch. Here is some general feedback prepared by the marker.

mimR needs MIM to function. In addition to the standard R distribution for Windows, it depends on a number of other packages that can all be downloaded from the links below. If you wish to use mimR on your own machines, you need to install MIM separately, and mimR and the other packages using the package installation facility of R.

You may wish to download the R-script and the MIM-data file from Wednesday's demonstration.

The zip file below for mimR2.0.3 does also contain some documentation, but the addendum in the zipfile is out of date and should be replaced with this mimR-addendum.

The packages are either on CRAN or the bioconductor websites (mimR might not quite be there yet as it has been uploaded today), but you may also download the necessary packages from here:

  1. mimR
  2. rcom
  3. Rgraphviz (which depends on the packages below)
  4. annotate
  5. Biobase
  6. geneplotter
  7. graph
  8. RColorBrewer

The author of mimR, Søren Højsgaard, is happy to receive any error reports, provided that you read and follow the instructions for doing so given at the R web site. This includes in particular that he is not your course instructor...

Course page


Last updated: Thursday, 03 May 2007 09:23Steffen L. Lauritzen