[BioC] need help in nnet (madhurima bhattacharjee)

Tarca, Adi atarca at med.wayne.edu
Wed Dec 21 14:07:04 CET 2005


Hi Madhurima,
The confusion matrix you get in the end is a "bad" result. You should
have the values on the main diagonal higher than those on the off
diagonal.
Looking at the iteration results you may see that the nnet did not learn
too much from your data.
There are lots of possible problems:
A) there is no relation between the class variable and the predictor you
use
B) the model is not well designed (too few or too many hidden nodes for
eg.)
Send the data you want more detail.
Laurentiu 
>>Hello Everybody,

>>I would like to know how to interpret the result of nnet function of
R.
>>My result looks like this:

# weights:  24
initial  value 6.533893
iter  10 value 4.616299
iter  20 value 4.616120
iter  30 value 4.616109
iter  30 value 4.616109
final  value 4.616109
converged
    cres
true  1
   1 10
   2  3

Can anyone please help me asap?



More information about the Bioconductor mailing list