[R] Error in solvet: apparently singular matrix

[Gad Abraham]

Hi,

Again I'm doing logistic regression using lrm from Harrell's Design 
package, but on wide matrices (28 observations, 1891 variables) so L2 
penalisation is used. I get the following error, although I've tried 
different penalties and different tolerances. The X matrix is scaled and 
centred.

 > library(Design)
 > load(url("http://www.csse.unimelb.edu.au/~gabraham/data2.Rdata"))
 > l <- lrm(y ~ X, penalty=100, tol=1e-13)
Error in solvet(info.matrix, tol = tol) : apparently singular matrix

 > traceback()
4: stop("apparently singular matrix")
3: solvet(info.matrix, tol = tol)
2: fitter(X, Y, initial = f$coef, maxit = 1, tol = tol)
1: lrm(y ~ X, penalty = 100, tol = 1e-13)

Any idea what the reason for this error is?

Thanks,
Gad




 > sessionInfo()
R version 2.7.2 (2008-08-25)
i386-apple-darwin8.11.1

locale:
en_AU.UTF-8/en_AU.UTF-8/C/C/en_AU.UTF-8/en_AU.UTF-8

attached base packages:
[1] splines   stats     graphics  grDevices utils     datasets  methods 
   base

other attached packages:
[1] Design_2.1-1    survival_2.34-1 Hmisc_3.4-3

loaded via a namespace (and not attached):
[1] cluster_1.11.11 grid_2.7.2      lattice_0.17-15


-- 
Gad Abraham
Dept. CSSE and NICTA
The University of Melbourne
Parkville 3010, Victoria, Australia
email: gabraham at csse.unimelb.edu.au
web: http://www.csse.unimelb.edu.au/~gabraham