How does the gbm function from the caret package in R handle categorical input attributes when dealing with a classification dataset? Can it handle them natively? Does it require any encoding (such as one-hot) or it encondes the attributes automatically? Is there any source where this can be consulted?
I have searched the internet and haven't found anything definitive on the subject. Here, I have been able to read that GBM handles them natively, but in this other thread, it appears to suggest that the GBM algorithm in the caret package uses dummy encoding automatically. However, I'm not sure where that information comes from or if it's actually true.
Thank you very much.