R caret nnet package

Question

I have two R objects as below.

matrix "datamatrix" - 200 rows and 494 columns: these are my x variables

dataframe Y. Y$V1 is my Y variable. I have converted column V1 to a factor I am building a classification model.

I want to build a neural network and I ran below command.

model <- train(Y$V1 ~ datamatrix, method='nnet', linout=TRUE, trace = FALSE,
               #Grid of tuning parameters to try:
               tuneGrid=expand.grid(.size=c(1,5,10),.decay=c(0,0.001,0.1)))

I got an error - " argument "data" is missing, with no default"

Is there a way for caret package to understand that I have my X variables in one R object and Y variable in other? I dont want to combined two data objects and then write a formula as the formula will be too long

Y~x1+x2+x3.................x199+x200....x493+x494

thanks...seems to be working...pls post your comment as an answer — user2543622, Apr 16 '15 at 16:56

score 3 · Accepted Answer · edited May 23 '17 at 10:28

The argument "data" is missing error is addressed by adding a data = datamatrix argument to the train call. The way I would do it would be something like:

datafr <- as.data.frame(datamatrix)

# V1 is the first column name if dimnames aren't specified
datafr$V1 <- as.factor(datafr$V1)

model <- train(V1 ~ ., data = datafr, method='nnet', 
               linout=TRUE, trace = FALSE,
               tuneGrid=expand.grid(.size=c(1,5,10),.decay=c(0,0.001,0.1)))

Now you don't have to pull your response variable out separately.

The . identifier allows inclusion of all variables from datafr (see here for details).

R caret nnet package

1 Answers1