I have 3 input variables and and 1 output variable. I should define factor in 3 levels for the output variable and use the neural network to predict the output variable in 3 levels. I mean that I should have predictions for all 3 levels for the output. Can anyone help me how can I do it in R?
Thanks
please look at the following:
http://stackoverflow.com/questions/20017966/using-nnet-for-classification
Thank you Men, but i did not mean classification. all my variables are numerical and the output has 3 levels. I am interested in using nnet or neuralnet
function in R, but i have a problem about it. I have 3 input variables
and 1 output variable ( all are numerical variables). The output
variable has factor number and it has 3 levels. I test the "neural
network" menu in SPSS software , it has dependent variable,independent
variable and the factor, then i can enter the factor variable (of the output) in
the related box. Finally the software produce the predicted values for
each record in each level of the output. My question is that if there
is a similar function in R related to neural network?
So, what do you what to do? Are you doing classification or regression using neural networks? Both nnet and neuralnet can do this. I do not use SPSS but using R and MATLAB.
Dear Gazhale
You can use the "nnet" packake. It is very simple one and good for starting.
If I understood well, you should transform the three level output variable in a new binary matrix, the simples one is a matrix with three columns, where each row will have a 1 in one column and ceros in the other ones. Please check the nnet help for that.
A thre level facttor variable can also be modeled with only two columns.
According to your output matrix, the number of output neurons will be defined.
Hope this helps
https://www.researchgate.net/post/How_can_factors_be_considered_for_Hopfield_neural_networks_in_logiciel_R
- Badges
- Science topic