[R] Problem with multinom ?

From: Marc Girondot <marc.girondot_at_ese.u-psud.fr>
Date: Sat 11 Jun 2005 - 18:06:20 EST

Thanks for your response.
OK, multinom() is a more logical in this context.

But similar problem occurs:

Let these data to be analyzed using classical glm with binomial error:

m   f   factor   m theo              f theo 
-Ln L model        -Ln L full          interecept 
f
10  12  1.2      0.452494473  0.547505527 
1.778835688 1.778648963 2.632455675 -2.034882223
14  14  1.3      0.503222759  0.496777241  1.901401922  1.900820284
15  12  1.4      0.553884782  0.446115218  1.877062369  1.876909821

                                 Sum -Ln L 
5.557299979  5.556379068  Residual deviance
                                 Deviance 
11.11459996 11.11275814 0.001841823

If I try to use multinom() function to analyze these data, the fitted parameters are correct but the residual deviance not.

> dt<-read.table('/try.txt'. header=T)
> dt

   output factor n

1      m    1.2 10
2      f    1.2 12
3      m    1.3 14
4      f    1.3 14
5      m    1.4 15
6      f    1.4 12

> dt.plr <- multinom(output ~ factor. data=dt. weights=n. maxit=1000)
# weights: 3 (2 variable)
initial  value 53.372333
iter  10 value 53.115208
iter  10 value 53.115208
iter  10 value 53.115208

final value 53.115208
converged
> dt.plr

Call:
multinom(formula = output ~ factor. data = dt. weights = n. maxit = 1000)

Coefficients:
(Intercept) factor

   -2.632443 2.034873

Residual Deviance: 106.2304
AIC: 110.2304
> dt.pr1<-predict(dt.plr. . type="probs")
> dt.pr1

         1 2 3 4 5 6 0.4524948 0.4524948 0.5032229 0.5032229 0.5538846 0.5538846

Probability for 2. 4 and 6 are not correct and its explain the non-correct residual deviance obtained in R.
Probably the problem I have is due to an incorrect data format... could someone help me... Thanks

(I know there is a simple way to analyze binomial data. but in fine I want to use multinom() for 5 categories of outputs.

Thanks a lot

Marc

-- 

__________________________________________________________
Marc Girondot, Pr
Laboratoire Ecologie, Systématique et Evolution
Equipe de Conservation des Populations et des Communautés
CNRS, ENGREF et Université Paris-Sud 11 , UMR 8079
Bâtiment 362
91405 Orsay Cedex, France

Tel:  33 1 (0)1.69.15.72.30   Fax: 33 1 (0)1 69 
15 56 96   e-mail: marc.girondot@ese.u-psud.fr
Web: http://www.ese.u-psud.fr/epc/conservation/Marc.html
Skype: girondot
Fax in US: 1-425-732-6934

______________________________________________
R-help@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
Received on Sat Jun 11 18:10:04 2005

This archive was generated by hypermail 2.1.8 : Fri 03 Mar 2006 - 03:32:31 EST