[R] number of weights in multinom ?
franck.vermet at univ-brest.fr
Tue Nov 18 13:08:33 CET 2014
I have the book by W. Venables and B. Ripley, but I didn't find the answer.
Here is an explicit example :
ir.glm <- multinom(type ~., data=fgl)
# weights: 66 (50 variable)
initial value 383.436526
iter 10 value 259.867465
iter 20 value 184.185706
iter 30 value 146.240801
iter 40 value 139.340485
iter 50 value 136.405273
iter 60 value 132.523057
iter 70 value 131.095461
iter 80 value 130.303792
iter 90 value 128.402012
iter 100 value 127.396942
final value 127.396942
stopped after 100 iterations
I understand that there are 50 variable weights, but I still not understand the sense of the value '66' in this model.
I'm sorry to spam the system;
I'm using R-help for the first time.
Le 18 nov. 2014 à 12:48, Prof Brian Ripley a écrit :
> On 18/11/2014 11:35, Franck Vermet wrote:
>> In the function multinom (package nnet), I get the following message after training for a model with 9 inputs and 6 classes (output) :
>> # weights: 66 (50 variable)
>> I understand that there are 50 variables in the model,
>> but I don't understand the number 66.
>> How can we interpret this number ?
> That is not what the message says: 'variable' is not a plural noun but an adjective, the antonym of 'fixed'.
> Package nnet is support software for a book: you need to do your own homework by reading it.
>> Franck Vermet.
>> R-help at r-project.org mailing list
>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
> That does mean you: in the absence for a reproducible example no one can help you.
> Also, repeated posting is very strongly discouraged: you sent this on Nov 13 (and also spammed CRAN with it).
> Brian D. Ripley, ripley at stats.ox.ac.uk
> Emeritus Professor of Applied Statistics, University of Oxford
> 1 South Parks Road, Oxford OX1 3TG, UK
[[alternative HTML version deleted]]
More information about the R-help