CODEWITHSUNDEEP
rmachine-learningparallel-processingsvmr-caret
hachondeoro
hachondeoro

Reputation: 43

Caret with SVM in R: Error in names(resamples) <- gsub("^\\.", "", names(resamples)) : attempt to set an attribute on NULL

I am trying to train a SVM model in R, using caret and doMC. Here is a reproducible example:

library(mlbench)
library(caret)
library(doMC)
registerDoMC(cores = 8)
training <- mlbench.cassini(5000)
Fitsvm<-train(classes ~ .,data=training,
                    preProcess=c('scale', 'center'),
                    method="svmRadial",
                    tuneGrid=expand.grid(sigma=0.5,C=c(0.01,0.05,0.1,0.5,1)))

But when I run this code, I get the following message: 

Error in names(resamples) <- gsub("^\\.", "", names(resamples)) : 
  attempt to set an attribute on NULL

This error only appears when I try to run in parallel the SVM with caret, and only with the SVM model. When I run it with GBM or RF, the codes functions normally. Could you help me figure out what is wrong or how can I get it to run? Does caret support SVM parallelisation? Thank you.

I am running it on a Macbook pro, mid 2013 on 4 cores.

Upvotes: 3

Views: 839

Answers (1)

phiver
phiver

Reputation: 23598

Your training object is a list, which you are passing as a data.frame.

If you change your training object into a data.frame it should work.

training <- data.frame(mlbench.cassini(5000))

Upvotes: 1

Related Questions