mlp tmva

void AddData(TMVA::Factory* factory, vector<vector<double> > &data, double learnTestRatio, const char *className)
{
    int firstTestEventIndex = learnTestRatio * data.size();
    for(int i = 0; i < data.size(); ++i)
    {
        if (i < firstTestEventIndex)
        {
            factory->AddTrainingEvent(className, data[i], 1);
        }
        else
        {
            factory->AddTestEvent(className, data[i], 1);
        }
    }
}

// .......


gROOT->SetBatch(true);

vector<vector<double> > alphaData;
vector<vector<double> > electronData;
vector<vector<double> > gammaData;
vector<vector<double> > muonData;

LoadData(m_AlphaStorage, alphaData);
LoadData(m_ElectronStorage, electronData);
LoadData(m_GammaStorage, gammaData);
LoadData(m_MuonStorage, muonData);

TFile outputFile("tmva_multi.root", "RECREATE");
TMVA::Factory* factory = new TMVA::Factory("multi", &outputFile, "V");

//add variables to the factory
int numDim = alphaData[0].size();
for(int i=0; i<numDim; i++)
{
    char name[10];
    sprintf(name,"var%d",i+1);
    factory->AddVariable(name, 'F');
}
OutputDebugStringF("Number of dimensions for LDA learning: %d\n", numDim);

AddData(factory, alphaData, 0.3, "Alpha");
AddData(factory, electronData, 0.5, "Electron");
AddData(factory, gammaData, 0.5, "Gamma");
AddData(factory, muonData, 0.5, "Muon");
factory->PrepareTrainingAndTestTree("", "");

factory->BookMethod(TMVA::Types::kMLP, "MLP", "HiddenLayers=N+3,N");
factory->TrainAllMethods();
factory->TestAllMethods();
factory->EvaluateAllMethods();

outputFile.Close();