TMVAClassifier

Based on $ROOTSYS/tutorials/tmva/TMVAClassification.C

Run with: chanser_mva RunTrainSignalID.C

auto input = GetClassificationFile();

auto *signalTree     = dynamic_cast<TTree*>(input->Get("TreeS"));
auto *background     = dynamic_cast<TTree*>(input->Get("TreeB"));

TrainSignalID train("TMVAClassificationTut");

train.SetOutDir("/work/dump/tmva/");
train.IgnoreBranches("");//Any branches in tree not used must be flagged!
train.AddSignalTree(signalTree);
train.AddBackgroundTree(background);
train.SetNTrainTest(1500,1500);
train.PrepareTrees();

//Can Book methods either via standard TMVA::Factory interface...
train.BookMethod(TMVA::Types::kBDT,"BDT",
  "!H:!V:NTrees=850:MinNodeSize=2.5%:MaxDepth=3:BoostType=AdaBoost:
  AdaBoostBeta=0.5:UseBaggedBoost:BaggedSampleFraction=0.5:
  SeparationType=GiniIndex:nCuts=20");

//..or predefined methods (See HSMVA::MethodConfigure.h)
train.BookMethod(chanser::mva::Meths.MLP);

train.DoTraining();
train.DrawResponses();
train.DrawROCCurve();//Got to come after responses