|
Hello:
I am using S-Plus 4.5 on a
Windows 2000 machine. I have been trying to use the multinom() function
from Prof. Ripley's nnet library. Using the "housing" dataset from the
MASS library, I type:
house.mult <- multinom(Sat ~ Infl+Type+Cont,
weights=Freq, data="">
I received the following message:
# weights: 24 (14 variable)
Error in
nnet.default(X, Y, w, mask = mask, size = 0, skip = T, softm..: Compiled code
has not been dynamically loaded
Dumped
Is it because this library does not work with
S-Plus 4.5 or have I loaded the library incorrectly? I appreciate any
suggestions.
Secondly, from a statistical point of view, what
are the different assumptions (and strengths, limitations) involved in using a
multinomial logit vs. probit model?
Thanks,
Michael
************************************
Michael
Ian Westphal, PhD
Candidate
Department of Environmental Science, Policy, &
Mgt
University of California at
Berkeley
The Ecology Centre
Department of Zoology and
Entomology
University of Queensland
St. Lucia, QLD 4072
http://www.ecology.uq.edu.au
Tel:
61 7 3365 1671
Fax: 61 7 3365 1655
Home Tel: 61 7 3719
5638
|