Caret error using GBM, but not without caret

Question

I\'ve been using gbm through caret without problems, but when removing some variables from my dataframe it started to fail. I\'ve tried wit

Answer 1

same problem with the glm method. Solved when I remove the VERBOSE option...

Answer 2

For now, casting a dataframe from plyr/dplyr to a normal dataframe with as.data.frame() fixes the problem.

train(as.data.frame(issueDataframe), issueResponse, method="gbm")

See this issue.

Answer 3

With some of the caret methods this problem will arise when the user tries to predict with multinomial classification and only binary {0,1} outcomes are allowed in the algorithm, or with the current set of parameters.