Clever Geek Handbook

Using all input variables in a neural network in R

I am trying to create a simple neural network in R using the nerualnet package. Instead of entering all 784 input variables, I just use a. as suggested in this thread: neural network using all input variables?

But I am getting this error

> digitnet <- neuralnet(label ~ ., trainingset, hidden = 4)

Error in terms.formula (formula): '.' in formula and without 'data' Argument

+3


source to share


1 answer


I don't know why this doesn't work, but you can always use the following:

myform <- as.formula(paste0('label ~ ', 
                             paste(names(trainingset[!names(trainingset) %in% 'label']),
                     collapse = ' + ')))

and then:



digitnet <- neuralnet(myform, trainingset, hidden = 4)

And it will use all 784 input variables in the neural network model.

+2


source






More articles:


All Articles