TMVA::Reader prediction

Hi to everybody. I would like to use TMVA to “reshape” the background into the signal.

At the moment, I am using the classical approach: I am evaluating the TMVA Method result with
reader->EvaluateMVA("method") and selecting events above a given cut.
What I would like to know is if it is possible to have weights that can change the shape of the background to the signal one. Let me rephrase it: I would like to know if there is something similar to the Keras “predict” method. I couldn’t find anything similar in TMVA documentation.

Thank you!


The return value of reader->EvaluateMVA("method") would be the same as when using predict in Keras.
For BDT you can use the new RReader::Compute interface, see the tutorial


Thank you for the help, Lorenzo!

I’ll try to use the reader output as weights, then. Just a question: is this new RReader::Compute interface working with other MVA methods (e.g. MLP and DL)?

Ciao Pasquale,

This new interface, RReader should work also for the pother TMVA methods, since it uses internally the old TMVA::Reader class. If you have any issue using it with other methods, please let me know



1 Like