PDFTeacher Class Reference

Teacher for the PDF MVA method. More...

#include <PDF.h>

Inheritance diagram for PDFTeacher:

Public Member Functions
	PDFTeacher (const GeneralOptions &general_options, const PDFOptions &specific_options)
	Constructs a new teacher using the GeneralOptions and PDFoptions for this training.
virtual Weightfile	train (Dataset &training_data) const override
	Train PDF method using the given dataset returning a Weightfile.

Protected Attributes
GeneralOptions	m_general_options
	GeneralOptions containing all shared options.

Private Attributes
PDFOptions	m_specific_options
	Specific options of the PDF method.

Detailed Description

Teacher for the PDF MVA method.

Definition at line 63 of file PDF.h.

Constructor & Destructor Documentation

PDFTeacher()

PDFTeacher	(	const GeneralOptions &	general_options,
		const PDFOptions &	specific_options
	)

Constructs a new teacher using the GeneralOptions and PDFoptions for this training.

Parameters

general_options	defining all shared options
specific_options	defining all options unique to the PDFTeacher

Definition at line 51 of file PDF.cc.

                                                                                                    : Teacher(general_options),
      m_specific_options(specific_options) { }

Member Function Documentation

train()

Weightfile train ( Dataset &  training_data ) const

overridevirtual

Train PDF method using the given dataset returning a Weightfile.

Parameters

training_data

used to train the method

Implements Teacher.

Definition at line 54 of file PDF.cc.

    {
 
      unsigned int numberOfFeatures = training_data.getNumberOfFeatures();
      if (numberOfFeatures != 1) {
        B2ERROR("PDF method only supports exactly one feature!");
        throw std::runtime_error("PDF method only supports exactly one feature!");
      }
      unsigned int numberOfEvents = training_data.getNumberOfEvents();
 
      std::vector<float> data(numberOfEvents);
      std::vector<float> weights(numberOfEvents);
      std::vector<bool> isSignal(numberOfEvents);
      for (unsigned int iEvent = 0; iEvent < numberOfEvents; ++iEvent) {
        training_data.loadEvent(iEvent);
        data[iEvent] = training_data.m_input[0];
        weights[iEvent] = training_data.m_weight;
        isSignal[iEvent] = training_data.m_isSignal;
      }
 
      unsigned int nBins = m_specific_options.m_nBins;
 
      Binning binning;
      if (m_specific_options.m_binning == "frequency")
        binning = Binning::CreateEqualFrequency(data, weights, isSignal, nBins);
      else if (m_specific_options.m_binning == "equidistant")
        binning = Binning::CreateEquidistant(data, weights, isSignal, nBins);
      else {
        B2ERROR("Unknown binning in PDF Teacher named " << m_specific_options.m_binning);
        throw std::runtime_error("Unknown binning in PDF Teacher named " + m_specific_options.m_binning);
      }
 
      Weightfile weightfile;
      std::string custom_weightfile = weightfile.generateFileName();
      std::fstream file(custom_weightfile, std::ios_base::out | std::ios_base::trunc);
 
      std::vector<double> value(nBins, 0);
      if (m_specific_options.m_mode == "probability") {
        for (unsigned int iBin = 0; iBin < nBins; ++iBin) {
          if (binning.m_signal_pdf[iBin] > 0)
            value[iBin] = binning.m_signal_pdf[iBin] / (binning.m_signal_pdf[iBin] + binning.m_bckgrd_pdf[iBin]);
        }
      } else {
        B2ERROR("Unknown mode in PDF Teacher named " << m_specific_options.m_mode);
        throw std::runtime_error("Unknown mode in PDF Teacher named " + m_specific_options.m_mode);
      }
 
      file << nBins << std::endl;
      for (unsigned int iBin = 0; iBin < nBins; ++iBin) {
        file << value[iBin] << " " << binning.m_boundaries[iBin] << std::endl;
      }
      file << binning.m_boundaries[nBins];
 
      file.close();
 
      weightfile.addOptions(m_general_options);
      weightfile.addOptions(m_specific_options);
      weightfile.addFile("PDF_Weightfile", custom_weightfile);
      weightfile.addSignalFraction(training_data.getSignalFraction());
 
      return weightfile;
 
    }

Member Data Documentation

m_general_options

GeneralOptions m_general_options

protectedinherited

GeneralOptions containing all shared options.

Definition at line 49 of file Teacher.h.

m_specific_options

PDFOptions m_specific_options

private

Specific options of the PDF method.

Definition at line 80 of file PDF.h.

---

The documentation for this class was generated from the following files:

• mva/methods/include/PDF.h
• mva/methods/src/PDF.cc