Search code examples
rcross-validation

Error in newdata[, object$model.list$variables] : subscript out of bounds"

When I am running this code, I am getting this error "Error in newdata[, object$model.list$variables] : subscript out of bounds" I am not getting how to solve it.

install.packages("boot")
install.packages("plyr")

library(boot)
library(plyr)

set.seed(50)
k=10
RMSE.NN=NULL

List=list( )

for(j in 10:65){
  for(i in 1:k){
    index=sample(1:nrow(data2),j)
    
    trainNN=d2[index,]
    testNN=d1[-index,]
    dataset=data2[-index,]
    
    NN=neuralnet(quality~.,d2,hidden=10,linear.output = T)
    predict_testNN=compute(NN,d1[,c(1:5)])
    predict_testNN=(predict_testNN$net.result*(max(data2$quality)-min(data2$quality)))+min(data2$quality)
    
    RMSE.NN[i]<-(sum((data2$quality-predict_testNN)^2)/nrow(dataset))^0.5
  }
  List[[j]]=RMSE.NN
}

(matrixRMSE=do.call(cbind, List))
matrixRMSE
Solution

  • The wine data you shared has variable names with spaces in them.

    R can get confused when variables have spaces in them; such names require quoting with backticks etc to refer to

    # i.e. my_num would be simple but `my num` would be needed to maintain the space.

    Easiest thing for you to do is use library(janitor) and clean_names on the dataset you want to work with.