R help - Jun 2007 - Multinomial models

Hello,

I am VERY new to R (one week) and I am trying to run a multinomial logit model.
The model I am using is
> model1 <- multinom(Y ~ X1 + X2 + , ..., Xn)
if I put in
> summary(model1)
I get

#Error in function (classes, fdef, mtable)  :
        unable to find an inherited method for function "fitted", for
signature "multinom"

and if I put in
> coef(model1)
I get the coefficients. I would like, however, to get the
coefficients, estimates, Std Erros and t-Ratios and/or z or P>[z]

Also can you change the speficiation of the base category without recoding?

Thanks.

Nicole

Hi,

since you are very new to R, just a small advice: try to give a minimal 
reproducible, self-contained example. This helps quite often to spot 
some mistakes yourself. This is my own experience and has been 
experienced by others, see, for example:
library(fortunes)
fortune("Ripleyed")

  :-)

For example, mention that you are (probably?) using the package 'nnet'.

nicole baerg wrote:
> I get the coefficients. I would like, however, to get the
> coefficients, estimates, Std Erros and t-Ratios and/or z or P>[z]
Check
names(model1) #if model1 is your fitted model as you write

then you will see a list of the components the fitted object should 
have. I guess the list should look similar to this:
  [1] "n"             "nunits"        "nconn"     
"conn"
  [5] "nsunits"       "decay"         "entropy"   
"softmax"
  [9] "censored"      "value"         "wts"       
"fitted.values"
[13] "residuals"     "lev"           "call"       
"terms"
[17] "weights"       "deviance"      "rank"       
"coefnames"
[21] "vcoefnames"    "contrasts"     "xlevels"    
"edf"
[25] "AIC"


> 
> Also can you change the speficiation of the base category without recoding?
> 
Check
?relevel

I hope this helps,
Roland

On Thu, 21 Jun 2007, nicole baerg wrote:
> Hello,
>
> I am VERY new to R (one week) and I am trying to run a multinomial logit
model.
> The model I am using is
>
>> model1 <- multinom(Y ~ X1 + X2 + , ..., Xn)
>
> if I put in
>
>> summary(model1)
>
> I get
>
> #Error in function (classes, fdef, mtable)  :
>        unable to find an inherited method for function "fitted",
for
> signature "multinom"
You have some other package attached (you are using 'nnet' without
mention
or credit) that has broken summary() (the correct summary method does not 
use fitted()).  The R posting guide asks for the output of sessionInfo(), 
and that would have helped us here. All I can tell is that the culprit is 
an S4-using package.

Try again with only the smallest possible set of packages attached.
> and if I put in
>
>> coef(model1)
>
> I get the coefficients. I would like, however, to get the
> coefficients, estimates, Std Erros and t-Ratios and/or z or P>[z]
>
> Also can you change the speficiation of the base category without recoding?
?contr.treatment
?relevel
> Thanks.
>
> Nicole
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
PLEASE do

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595