Hi all, is there a way to insert the spectator variables inside the correlation matrix for signal and background?Thanks in advance!
Hi,
Do you mean adding the spectator variables for the computation of the overall correlation matrix that is performed at the beginning in TMVA when building the internal data set ?
There it is not possible to use the spectator, the correlation is computed using only the standard variables.
You can however compute yourself the correlation/.covariance matrix with some code, either from the TNVA DataSet, see for example:
https://root.cern/doc/master/DataSetFactory_8cxx_source.html#l00579
Or if you have the input data in an array you can also use the TPrincipal class and TPrincipal::GetCovarianceMatrix.
Cheers
Lorenzo
Thanks a lot for the response! The fastest way I found so far is plotting the correlation plots from the TrainTree using the className flag to select Signal and Background.