              .            .        .
        ___  /|   . _  _  /|  __   /| __
       /  \   |  /|/ \/ \  |  __\   |/  \
       \__/   |   |  |  |  | /  |   |    )
       /__    |   |  |  |  | \__|_  |___/
    == /  \ ===================================
       \__/ 

glmlab is a set of m-files for using  MATLAB  for analysing generalised linear
models.  It originally was developed to replace GLIM in a small way at the 
University where I work (the University of Southern Queensland, Australia).
It has since grown, and now contains quite a lot of the features found in GLIM.
It enables models such as multiple regression, probit models, logistic
regression and log-linear models (among others) to be fitted.

glmlab has these features:
* a graphical user interface with drop down menu to select distribution, link, etc...
* a choice of error distribution (normal, gamma, inverse Gaussian, Possion, binomial)
* a choice of links (identity, log, reciprocal, square root, power, logistic,
  probit, complementary log-log)
* a choice of errors to calculate (Pearson, deviance, quantile)
* some residual plots
* the ability to define your own error distributions and link functions
* prior weights, offsets and interactions
* the ability to save models between sessions

glmlab can be obtained from the MathWorks user-contributed files page
(http://www/mathworks.com/ftp/statv5.shtml), or indirectly from the 
glmlab home page (http://www.sci.usq.edu.au/staff/dunn/glmlab.html)

Peter Dunn
02 March 1998
