Data transformation in tmva

I have come across two types of variable transformation:

  1. at TMVA::Factory *factory = new TMVA::Factory( "TMVAClassification", outputFile"Transformations=I;D;P;G,D" );
  2. at factory->BookMethod( dataloader, TMVA::Types::kBDT, "BDT","...VarTransform=N" );

I have consulted the user manual, and it says the first one just for test and display. So is it that the first one does not transform the data and pass the transformed data for training , whereas the second one do transform the data for the training ?

So if I want only the data being transformed before the training, is it OK to turn off all the transformation at the first one?

_ROOT Version: 6.24 (PyROOT via conda)
_Compiler: gcc9


This is for @moneta, thank you!