Hi all,
I would like to preform an autologistic regression analysis on my thesis data.
This is a form of logistic regression that accounts for spatial autocorrelation
within and between predictor variables. To this end I have received some script
from Dr. Fred Huffer which will allow for the analysis in R. It consists of an
R script (autolog.R) that calls a compiled Fortran script (combo.o). I am
supposed to use "dyn.load' to load the fortran script. To do this in R
I realize that I need to create a dll for combo.o and that this is supposed to
be done using "Rcmd SHLIB combo.f". This is where I run into
problems.
I have read the "readmepackages.txt", and done all that I believe
applies to me: installed PERL, MingGW (fortran compiler), Prof. Ripley's
Tools package and set PATH environment variable for all. Then when I ran
"Rcmd SHLIB combo.f" I got as far as:
Rcmd SHLIB combo.f
ar cr combo.a *.o
ranlib combo.a
gcc --shared -s -o combo.dll combo.def omo.a
-LC:/PROGRA~1/R/rw1070/src/gnuwin32 -lg2c -lR
Warning: .drectve '%.*s' unrecognized
So I tried to reinstall R to C:\rw1070, instead of C:\program files, because I
read that spaces in file names may cause errors. Now I don't even get that
far. Now it reads:
g77 -02 -Wall -c combo.f combo.o
application has requested runtime to terminate in unusual way
make: *** [combo.o] Error 3
Has anyone on a Windows box had any success with the "Rcmd SHLIB"
command and a Fortrn script?? I have struggled with this problem for about
three weeks and could really use some help.
Thanx
Mark Piorecky
[[alternative HTML version deleted]]
