| 1 |
benhoob |
1.1 |
/**********************************************************************************
|
| 2 |
|
|
* Project : TMVA - a Root-integrated toolkit for multivariate data analysis *
|
| 3 |
|
|
* Package : TMVA *
|
| 4 |
|
|
* Exectuable: TMVAClassificationApplication *
|
| 5 |
|
|
* *
|
| 6 |
|
|
* This macro provides a simple example on how to use the trained classifiers *
|
| 7 |
|
|
* within an analysis module *
|
| 8 |
|
|
**********************************************************************************/
|
| 9 |
|
|
|
| 10 |
|
|
#include <cstdlib>
|
| 11 |
|
|
#include <vector>
|
| 12 |
|
|
#include <iostream>
|
| 13 |
|
|
#include <map>
|
| 14 |
|
|
#include <string>
|
| 15 |
|
|
|
| 16 |
|
|
#include "TFile.h"
|
| 17 |
|
|
#include "TTree.h"
|
| 18 |
|
|
#include "TString.h"
|
| 19 |
|
|
#include "TSystem.h"
|
| 20 |
|
|
#include "TROOT.h"
|
| 21 |
|
|
#include "TStopwatch.h"
|
| 22 |
|
|
#include "TChain.h"
|
| 23 |
|
|
|
| 24 |
|
|
#include "TMVAGui.C"
|
| 25 |
|
|
|
| 26 |
|
|
#if not defined(__CINT__) || defined(__MAKECINT__)
|
| 27 |
|
|
#include "TMVA/Tools.h"
|
| 28 |
|
|
#include "TMVA/Reader.h"
|
| 29 |
|
|
#include "TMVA/MethodCuts.h"
|
| 30 |
|
|
#endif
|
| 31 |
|
|
|
| 32 |
|
|
using namespace TMVA;
|
| 33 |
|
|
|
| 34 |
|
|
void Classify_HWW( TString myMethodList = "" )
|
| 35 |
|
|
{
|
| 36 |
|
|
#ifdef __CINT__
|
| 37 |
|
|
gROOT->ProcessLine( ".O0" ); // turn off optimization in CINT
|
| 38 |
|
|
#endif
|
| 39 |
|
|
|
| 40 |
|
|
//---------------------------------------------------------------
|
| 41 |
|
|
|
| 42 |
|
|
// This loads the library
|
| 43 |
|
|
TMVA::Tools::Instance();
|
| 44 |
|
|
|
| 45 |
|
|
// Default MVA methods to be trained + tested
|
| 46 |
|
|
std::map<std::string,int> Use;
|
| 47 |
|
|
|
| 48 |
|
|
// --- Cut optimisation
|
| 49 |
|
|
Use["Cuts"] = 1;
|
| 50 |
|
|
Use["CutsD"] = 1;
|
| 51 |
|
|
Use["CutsPCA"] = 0;
|
| 52 |
|
|
Use["CutsGA"] = 0;
|
| 53 |
|
|
Use["CutsSA"] = 0;
|
| 54 |
|
|
//
|
| 55 |
|
|
// --- 1-dimensional likelihood ("naive Bayes estimator")
|
| 56 |
|
|
Use["Likelihood"] = 1;
|
| 57 |
|
|
Use["LikelihoodD"] = 0; // the "D" extension indicates decorrelated input variables (see option strings)
|
| 58 |
|
|
Use["LikelihoodPCA"] = 1; // the "PCA" extension indicates PCA-transformed input variables (see option strings)
|
| 59 |
|
|
Use["LikelihoodKDE"] = 0;
|
| 60 |
|
|
Use["LikelihoodMIX"] = 0;
|
| 61 |
|
|
//
|
| 62 |
|
|
// --- Mutidimensional likelihood and Nearest-Neighbour methods
|
| 63 |
|
|
Use["PDERS"] = 1;
|
| 64 |
|
|
Use["PDERSD"] = 0;
|
| 65 |
|
|
Use["PDERSPCA"] = 0;
|
| 66 |
|
|
Use["PDEFoam"] = 1;
|
| 67 |
|
|
Use["PDEFoamBoost"] = 0; // uses generalised MVA method boosting
|
| 68 |
|
|
Use["KNN"] = 1; // k-nearest neighbour method
|
| 69 |
|
|
//
|
| 70 |
|
|
// --- Linear Discriminant Analysis
|
| 71 |
|
|
Use["LD"] = 1; // Linear Discriminant identical to Fisher
|
| 72 |
|
|
Use["Fisher"] = 0;
|
| 73 |
|
|
Use["FisherG"] = 0;
|
| 74 |
|
|
Use["BoostedFisher"] = 0; // uses generalised MVA method boosting
|
| 75 |
|
|
Use["HMatrix"] = 0;
|
| 76 |
|
|
//
|
| 77 |
|
|
// --- Function Discriminant analysis
|
| 78 |
|
|
Use["FDA_GA"] = 1; // minimisation of user-defined function using Genetics Algorithm
|
| 79 |
|
|
Use["FDA_SA"] = 0;
|
| 80 |
|
|
Use["FDA_MC"] = 0;
|
| 81 |
|
|
Use["FDA_MT"] = 0;
|
| 82 |
|
|
Use["FDA_GAMT"] = 0;
|
| 83 |
|
|
Use["FDA_MCMT"] = 0;
|
| 84 |
|
|
//
|
| 85 |
|
|
// --- Neural Networks (all are feed-forward Multilayer Perceptrons)
|
| 86 |
|
|
Use["MLP"] = 0; // Recommended ANN
|
| 87 |
|
|
Use["MLPBFGS"] = 0; // Recommended ANN with optional training method
|
| 88 |
|
|
Use["MLPBNN"] = 1; // Recommended ANN with BFGS training method and bayesian regulator
|
| 89 |
|
|
Use["CFMlpANN"] = 0; // Depreciated ANN from ALEPH
|
| 90 |
|
|
Use["TMlpANN"] = 0; // ROOT's own ANN
|
| 91 |
|
|
//
|
| 92 |
|
|
// --- Support Vector Machine
|
| 93 |
|
|
Use["SVM"] = 1;
|
| 94 |
|
|
//
|
| 95 |
|
|
// --- Boosted Decision Trees
|
| 96 |
|
|
Use["BDT"] = 1; // uses Adaptive Boost
|
| 97 |
|
|
Use["BDTG"] = 0; // uses Gradient Boost
|
| 98 |
|
|
Use["BDTB"] = 0; // uses Bagging
|
| 99 |
|
|
Use["BDTD"] = 0; // decorrelation + Adaptive Boost
|
| 100 |
|
|
//
|
| 101 |
|
|
// --- Friedman's RuleFit method, ie, an optimised series of cuts ("rules")
|
| 102 |
|
|
Use["RuleFit"] = 1;
|
| 103 |
|
|
// ---------------------------------------------------------------
|
| 104 |
|
|
Use["Plugin"] = 0;
|
| 105 |
|
|
Use["Category"] = 0;
|
| 106 |
|
|
Use["SVM_Gauss"] = 0;
|
| 107 |
|
|
Use["SVM_Poly"] = 0;
|
| 108 |
|
|
Use["SVM_Lin"] = 0;
|
| 109 |
|
|
|
| 110 |
|
|
std::cout << std::endl;
|
| 111 |
|
|
std::cout << "==> Start TMVAClassificationApplication" << std::endl;
|
| 112 |
|
|
|
| 113 |
|
|
// Select methods (don't look at this code - not of interest)
|
| 114 |
|
|
if (myMethodList != "") {
|
| 115 |
|
|
for (std::map<std::string,int>::iterator it = Use.begin(); it != Use.end(); it++) it->second = 0;
|
| 116 |
|
|
|
| 117 |
|
|
std::vector<TString> mlist = gTools().SplitString( myMethodList, ',' );
|
| 118 |
|
|
for (UInt_t i=0; i<mlist.size(); i++) {
|
| 119 |
|
|
std::string regMethod(mlist[i]);
|
| 120 |
|
|
|
| 121 |
|
|
if (Use.find(regMethod) == Use.end()) {
|
| 122 |
|
|
std::cout << "Method \"" << regMethod
|
| 123 |
|
|
<< "\" not known in TMVA under this name. Choose among the following:" << std::endl;
|
| 124 |
|
|
for (std::map<std::string,int>::iterator it = Use.begin(); it != Use.end(); it++) {
|
| 125 |
|
|
std::cout << it->first << " ";
|
| 126 |
|
|
}
|
| 127 |
|
|
std::cout << std::endl;
|
| 128 |
|
|
return;
|
| 129 |
|
|
}
|
| 130 |
|
|
Use[regMethod] = 1;
|
| 131 |
|
|
}
|
| 132 |
|
|
}
|
| 133 |
|
|
|
| 134 |
|
|
// --------------------------------------------------------------------------------------------------
|
| 135 |
|
|
|
| 136 |
|
|
//-----------------------------------
|
| 137 |
|
|
//select samples to run over
|
| 138 |
|
|
//-----------------------------------
|
| 139 |
|
|
|
| 140 |
|
|
vector<char*> samples;
|
| 141 |
|
|
samples.push_back("WWTo2L2Nu");
|
| 142 |
benhoob |
1.2 |
samples.push_back("GluGluToWWTo4L");
|
| 143 |
|
|
samples.push_back("WZ");
|
| 144 |
|
|
samples.push_back("ZZ");
|
| 145 |
|
|
samples.push_back("TTJets");
|
| 146 |
|
|
samples.push_back("tW");
|
| 147 |
|
|
samples.push_back("WJetsToLNu");
|
| 148 |
|
|
samples.push_back("DY");
|
| 149 |
|
|
samples.push_back("Higgs130");
|
| 150 |
benhoob |
1.1 |
// samples.push_back("Higgs160");
|
| 151 |
|
|
// samples.push_back("Higgs200");
|
| 152 |
|
|
|
| 153 |
|
|
const unsigned int nsamples = samples.size();
|
| 154 |
|
|
|
| 155 |
|
|
for( unsigned int i = 0 ; i < nsamples ; ++i ){
|
| 156 |
|
|
|
| 157 |
|
|
// --- Create the Reader object
|
| 158 |
|
|
|
| 159 |
|
|
TMVA::Reader *reader = new TMVA::Reader( "!Color:!Silent" );
|
| 160 |
|
|
|
| 161 |
|
|
// Create a set of variables and declare them to the reader
|
| 162 |
|
|
// - the variable names MUST corresponds in name and type to those given in the weight file(s) used
|
| 163 |
|
|
// Float_t var1, var2;
|
| 164 |
|
|
// Float_t var3, var4;
|
| 165 |
|
|
// reader->AddVariable( "myvar1 := var1+var2", &var1 );
|
| 166 |
|
|
// reader->AddVariable( "myvar2 := var1-var2", &var2 );
|
| 167 |
|
|
// reader->AddVariable( "var3", &var3 );
|
| 168 |
|
|
// reader->AddVariable( "var4", &var4 );
|
| 169 |
|
|
|
| 170 |
|
|
Float_t lephard_pt;
|
| 171 |
|
|
Float_t lepsoft_pt;
|
| 172 |
|
|
Float_t dil_dphi;
|
| 173 |
|
|
Float_t dil_mass;
|
| 174 |
|
|
|
| 175 |
|
|
reader->AddVariable( "lephard_pt" , &lephard_pt );
|
| 176 |
|
|
reader->AddVariable( "lepsoft_pt" , &lepsoft_pt );
|
| 177 |
|
|
reader->AddVariable( "dil_dphi" , &dil_dphi );
|
| 178 |
|
|
reader->AddVariable( "dil_mass" , &dil_mass );
|
| 179 |
|
|
|
| 180 |
|
|
// Spectator variables declared in the training have to be added to the reader, too
|
| 181 |
|
|
// Float_t spec1,spec2;
|
| 182 |
|
|
// reader->AddSpectator( "spec1 := var1*2", &spec1 );
|
| 183 |
|
|
// reader->AddSpectator( "spec2 := var1*3", &spec2 );
|
| 184 |
|
|
|
| 185 |
|
|
Float_t Category_cat1, Category_cat2, Category_cat3;
|
| 186 |
|
|
if (Use["Category"]){
|
| 187 |
|
|
// Add artificial spectators for distinguishing categories
|
| 188 |
|
|
// reader->AddSpectator( "Category_cat1 := var3<=0", &Category_cat1 );
|
| 189 |
|
|
// reader->AddSpectator( "Category_cat2 := (var3>0)&&(var4<0)", &Category_cat2 );
|
| 190 |
|
|
// reader->AddSpectator( "Category_cat3 := (var3>0)&&(var4>=0)", &Category_cat3 );
|
| 191 |
|
|
}
|
| 192 |
|
|
|
| 193 |
|
|
// --- Book the MVA methods
|
| 194 |
|
|
|
| 195 |
benhoob |
1.3 |
//--------------------------------------------------------------------------------------
|
| 196 |
|
|
// tell Classify_HWW where to find the weights dir, which contains the trained MVA's.
|
| 197 |
|
|
// In this example, the weights dir is located at [path]/[dir]
|
| 198 |
|
|
// and the output root file is written to [path]/[output]
|
| 199 |
|
|
//--------------------------------------------------------------------------------------
|
| 200 |
|
|
|
| 201 |
benhoob |
1.1 |
//TString dir = "weights/";
|
| 202 |
benhoob |
1.2 |
//TString path = "Trainings/H130_WWTo2L2Nu/";
|
| 203 |
|
|
//TString path = "Trainings/H130_WWTo2L2Nu_WJetsToLNu/";
|
| 204 |
|
|
TString path = "Trainings/H130_allbkg/";
|
| 205 |
benhoob |
1.1 |
TString dir = path + "weights/";
|
| 206 |
|
|
TString outdir = path + "output/";
|
| 207 |
|
|
|
| 208 |
|
|
TString prefix = "TMVAClassification";
|
| 209 |
|
|
|
| 210 |
|
|
// Book method(s)
|
| 211 |
|
|
for (std::map<std::string,int>::iterator it = Use.begin(); it != Use.end(); it++) {
|
| 212 |
|
|
if (it->second) {
|
| 213 |
|
|
TString methodName = TString(it->first) + TString(" method");
|
| 214 |
|
|
TString weightfile = dir + prefix + TString("_") + TString(it->first) + TString(".weights.xml");
|
| 215 |
|
|
reader->BookMVA( methodName, weightfile );
|
| 216 |
|
|
}
|
| 217 |
|
|
}
|
| 218 |
|
|
|
| 219 |
|
|
// Book output histograms
|
| 220 |
benhoob |
1.2 |
UInt_t nbin = 1000;
|
| 221 |
benhoob |
1.1 |
TH1F *histLk(0), *histLkD(0), *histLkPCA(0), *histLkKDE(0), *histLkMIX(0), *histPD(0), *histPDD(0);
|
| 222 |
|
|
TH1F *histPDPCA(0), *histPDEFoam(0), *histPDEFoamErr(0), *histPDEFoamSig(0), *histKNN(0), *histHm(0);
|
| 223 |
|
|
TH1F *histFi(0), *histFiG(0), *histFiB(0), *histLD(0), *histNn(0),*histNnbfgs(0),*histNnbnn(0);
|
| 224 |
|
|
TH1F *histNnC(0), *histNnT(0), *histBdt(0), *histBdtG(0), *histBdtD(0), *histRf(0), *histSVMG(0);
|
| 225 |
|
|
TH1F *histSVMP(0), *histSVML(0), *histFDAMT(0), *histFDAGA(0), *histCat(0), *histPBdt(0);
|
| 226 |
|
|
|
| 227 |
|
|
if (Use["Likelihood"]) histLk = new TH1F( "MVA_Likelihood", "MVA_Likelihood", nbin, -1, 1 );
|
| 228 |
|
|
if (Use["LikelihoodD"]) histLkD = new TH1F( "MVA_LikelihoodD", "MVA_LikelihoodD", nbin, -1, 0.9999 );
|
| 229 |
|
|
if (Use["LikelihoodPCA"]) histLkPCA = new TH1F( "MVA_LikelihoodPCA", "MVA_LikelihoodPCA", nbin, -1, 1 );
|
| 230 |
|
|
if (Use["LikelihoodKDE"]) histLkKDE = new TH1F( "MVA_LikelihoodKDE", "MVA_LikelihoodKDE", nbin, -0.00001, 0.99999 );
|
| 231 |
|
|
if (Use["LikelihoodMIX"]) histLkMIX = new TH1F( "MVA_LikelihoodMIX", "MVA_LikelihoodMIX", nbin, 0, 1 );
|
| 232 |
|
|
if (Use["PDERS"]) histPD = new TH1F( "MVA_PDERS", "MVA_PDERS", nbin, 0, 1 );
|
| 233 |
|
|
if (Use["PDERSD"]) histPDD = new TH1F( "MVA_PDERSD", "MVA_PDERSD", nbin, 0, 1 );
|
| 234 |
|
|
if (Use["PDERSPCA"]) histPDPCA = new TH1F( "MVA_PDERSPCA", "MVA_PDERSPCA", nbin, 0, 1 );
|
| 235 |
|
|
if (Use["KNN"]) histKNN = new TH1F( "MVA_KNN", "MVA_KNN", nbin, 0, 1 );
|
| 236 |
|
|
if (Use["HMatrix"]) histHm = new TH1F( "MVA_HMatrix", "MVA_HMatrix", nbin, -0.95, 1.55 );
|
| 237 |
|
|
if (Use["Fisher"]) histFi = new TH1F( "MVA_Fisher", "MVA_Fisher", nbin, -4, 4 );
|
| 238 |
|
|
if (Use["FisherG"]) histFiG = new TH1F( "MVA_FisherG", "MVA_FisherG", nbin, -1, 1 );
|
| 239 |
|
|
if (Use["BoostedFisher"]) histFiB = new TH1F( "MVA_BoostedFisher", "MVA_BoostedFisher", nbin, -2, 2 );
|
| 240 |
|
|
if (Use["LD"]) histLD = new TH1F( "MVA_LD", "MVA_LD", nbin, -2, 2 );
|
| 241 |
|
|
if (Use["MLP"]) histNn = new TH1F( "MVA_MLP", "MVA_MLP", nbin, -1.25, 1.5 );
|
| 242 |
|
|
if (Use["MLPBFGS"]) histNnbfgs = new TH1F( "MVA_MLPBFGS", "MVA_MLPBFGS", nbin, -1.25, 1.5 );
|
| 243 |
|
|
if (Use["MLPBNN"]) histNnbnn = new TH1F( "MVA_MLPBNN", "MVA_MLPBNN", nbin, -1.25, 1.5 );
|
| 244 |
|
|
if (Use["CFMlpANN"]) histNnC = new TH1F( "MVA_CFMlpANN", "MVA_CFMlpANN", nbin, 0, 1 );
|
| 245 |
|
|
if (Use["TMlpANN"]) histNnT = new TH1F( "MVA_TMlpANN", "MVA_TMlpANN", nbin, -1.3, 1.3 );
|
| 246 |
|
|
if (Use["BDT"]) histBdt = new TH1F( "MVA_BDT", "MVA_BDT", nbin, -0.8, 0.8 );
|
| 247 |
|
|
if (Use["BDTD"]) histBdtD = new TH1F( "MVA_BDTD", "MVA_BDTD", nbin, -0.8, 0.8 );
|
| 248 |
|
|
if (Use["BDTG"]) histBdtG = new TH1F( "MVA_BDTG", "MVA_BDTG", nbin, -1.0, 1.0 );
|
| 249 |
|
|
if (Use["RuleFit"]) histRf = new TH1F( "MVA_RuleFit", "MVA_RuleFit", nbin, -2.0, 2.0 );
|
| 250 |
|
|
if (Use["SVM_Gauss"]) histSVMG = new TH1F( "MVA_SVM_Gauss", "MVA_SVM_Gauss", nbin, 0.0, 1.0 );
|
| 251 |
|
|
if (Use["SVM_Poly"]) histSVMP = new TH1F( "MVA_SVM_Poly", "MVA_SVM_Poly", nbin, 0.0, 1.0 );
|
| 252 |
|
|
if (Use["SVM_Lin"]) histSVML = new TH1F( "MVA_SVM_Lin", "MVA_SVM_Lin", nbin, 0.0, 1.0 );
|
| 253 |
|
|
if (Use["FDA_MT"]) histFDAMT = new TH1F( "MVA_FDA_MT", "MVA_FDA_MT", nbin, -2.0, 3.0 );
|
| 254 |
|
|
if (Use["FDA_GA"]) histFDAGA = new TH1F( "MVA_FDA_GA", "MVA_FDA_GA", nbin, -2.0, 3.0 );
|
| 255 |
|
|
if (Use["Category"]) histCat = new TH1F( "MVA_Category", "MVA_Category", nbin, -2., 2. );
|
| 256 |
|
|
if (Use["Plugin"]) histPBdt = new TH1F( "MVA_PBDT", "MVA_BDT", nbin, -0.8, 0.8 );
|
| 257 |
|
|
|
| 258 |
|
|
// PDEFoam also returns per-event error, fill in histogram, and also fill significance
|
| 259 |
|
|
if (Use["PDEFoam"]) {
|
| 260 |
|
|
histPDEFoam = new TH1F( "MVA_PDEFoam", "MVA_PDEFoam", nbin, 0, 1 );
|
| 261 |
|
|
histPDEFoamErr = new TH1F( "MVA_PDEFoamErr", "MVA_PDEFoam error", nbin, 0, 1 );
|
| 262 |
|
|
histPDEFoamSig = new TH1F( "MVA_PDEFoamSig", "MVA_PDEFoam significance", nbin, 0, 10 );
|
| 263 |
|
|
}
|
| 264 |
|
|
|
| 265 |
|
|
// Book example histogram for probability (the other methods are done similarly)
|
| 266 |
|
|
TH1F *probHistFi(0), *rarityHistFi(0);
|
| 267 |
|
|
if (Use["Fisher"]) {
|
| 268 |
|
|
probHistFi = new TH1F( "MVA_Fisher_Proba", "MVA_Fisher_Proba", nbin, 0, 1 );
|
| 269 |
|
|
rarityHistFi = new TH1F( "MVA_Fisher_Rarity", "MVA_Fisher_Rarity", nbin, 0, 1 );
|
| 270 |
|
|
}
|
| 271 |
|
|
|
| 272 |
|
|
// Prepare input tree (this must be replaced by your data source)
|
| 273 |
|
|
// in this example, there is a toy tree with signal and one with background events
|
| 274 |
|
|
// we'll later on use only the "signal" events for the test in this example.
|
| 275 |
|
|
//
|
| 276 |
|
|
|
| 277 |
benhoob |
1.3 |
char* prefix = "babies/v3";
|
| 278 |
benhoob |
1.1 |
TChain *ch = new TChain("Events");
|
| 279 |
|
|
|
| 280 |
|
|
if( strcmp( samples.at(i) , "DY" ) == 0 ){
|
| 281 |
benhoob |
1.3 |
ch -> Add( Form("%s/DYToMuMuM20_PU_testFinal_baby.root",prefix) );
|
| 282 |
|
|
ch -> Add( Form("%s/DYToMuMuM10To20_PU_testFinal_baby.root",prefix) );
|
| 283 |
|
|
ch -> Add( Form("%s/DYToEEM20_PU_testFinal_baby.root",prefix) );
|
| 284 |
|
|
ch -> Add( Form("%s/DYToEEM10To20_PU_testFinal_baby.root",prefix) );
|
| 285 |
|
|
ch -> Add( Form("%s/DYToTauTauM20_PU_testFinal_baby.root",prefix) );
|
| 286 |
|
|
ch -> Add( Form("%s/DYToTauTauM10To20_PU_testFinal_baby.root",prefix) );
|
| 287 |
benhoob |
1.1 |
}
|
| 288 |
|
|
else if( strcmp( samples.at(i) , "Higgs130" ) == 0 ){
|
| 289 |
benhoob |
1.3 |
ch -> Add( Form("%s/HToWWTo2L2NuM130_PU_testFinal_baby.root",prefix) );
|
| 290 |
|
|
ch -> Add( Form("%s/HToWWToLNuTauNuM130_PU_testFinal_baby.root",prefix) );
|
| 291 |
|
|
ch -> Add( Form("%s/HToWWTo2Tau2NuM130_PU_testFinal_baby.root",prefix) );
|
| 292 |
benhoob |
1.1 |
}
|
| 293 |
|
|
else if( strcmp( samples.at(i) , "Higgs160" ) == 0 ){
|
| 294 |
benhoob |
1.3 |
ch -> Add( Form("%s/HToWWTo2L2NuM160_PU_testFinal_baby.root",prefix) );
|
| 295 |
|
|
ch -> Add( Form("%s/HToWWToLNuTauNuM160_PU_testFinal_baby.root",prefix) );
|
| 296 |
|
|
ch -> Add( Form("%s/HToWWTo2Tau2NuM160_PU_testFinal_baby.root",prefix) );
|
| 297 |
benhoob |
1.1 |
}
|
| 298 |
|
|
else if( strcmp( samples.at(i) , "Higgs200" ) == 0 ){
|
| 299 |
benhoob |
1.3 |
ch -> Add( Form("%s/HToWWTo2L2NuM200_PU_testFinal_baby.root",prefix) );
|
| 300 |
|
|
ch -> Add( Form("%s/HToWWToLNuTauNuM200_PU_testFinal_baby.root",prefix) );
|
| 301 |
|
|
ch -> Add( Form("%s/HToWWTo2Tau2NuM200_PU_testFinal_baby.root",prefix) );
|
| 302 |
benhoob |
1.1 |
}
|
| 303 |
|
|
else{
|
| 304 |
benhoob |
1.3 |
ch -> Add( Form("%s/%s_PU_testFinal_baby.root",prefix,samples.at(i)) );
|
| 305 |
benhoob |
1.1 |
}
|
| 306 |
|
|
|
| 307 |
|
|
|
| 308 |
|
|
|
| 309 |
|
|
// --- Event loop
|
| 310 |
|
|
|
| 311 |
|
|
// Prepare the event tree
|
| 312 |
|
|
// - here the variable names have to corresponds to your tree
|
| 313 |
|
|
// - you can use the same variables as above which is slightly faster,
|
| 314 |
|
|
// but of course you can use different ones and copy the values inside the event loop
|
| 315 |
|
|
//
|
| 316 |
|
|
|
| 317 |
|
|
TTree *theTree = (TTree*) ch;
|
| 318 |
|
|
|
| 319 |
|
|
std::cout << "--- Using input files: -------------------" << std::endl;
|
| 320 |
|
|
|
| 321 |
|
|
TObjArray *listOfFiles = ch->GetListOfFiles();
|
| 322 |
|
|
TIter fileIter(listOfFiles);
|
| 323 |
|
|
TChainElement* currentFile = 0;
|
| 324 |
|
|
|
| 325 |
|
|
while((currentFile = (TChainElement*)fileIter.Next())) {
|
| 326 |
|
|
std::cout << currentFile->GetTitle() << std::endl;
|
| 327 |
|
|
}
|
| 328 |
|
|
|
| 329 |
|
|
Float_t lephard_pt_;
|
| 330 |
|
|
Float_t lepsoft_pt_;
|
| 331 |
|
|
Float_t dil_dphi_;
|
| 332 |
|
|
Float_t dil_mass_;
|
| 333 |
|
|
Int_t event_type_;
|
| 334 |
|
|
Float_t met_projpt_;
|
| 335 |
|
|
Int_t jets_num_;
|
| 336 |
|
|
Int_t extralep_num_;
|
| 337 |
|
|
Int_t lowptbtags_num_;
|
| 338 |
|
|
Int_t softmu_num_;
|
| 339 |
|
|
Float_t event_scale1fb_;
|
| 340 |
|
|
|
| 341 |
|
|
theTree->SetBranchAddress( "lephard_pt_" , &lephard_pt_ );
|
| 342 |
|
|
theTree->SetBranchAddress( "lepsoft_pt_" , &lepsoft_pt_ );
|
| 343 |
|
|
theTree->SetBranchAddress( "dil_dphi_" , &dil_dphi_ );
|
| 344 |
|
|
theTree->SetBranchAddress( "dil_mass_" , &dil_mass_ );
|
| 345 |
|
|
theTree->SetBranchAddress( "event_type_" , &event_type_ );
|
| 346 |
|
|
theTree->SetBranchAddress( "met_projpt_" , &met_projpt_ );
|
| 347 |
|
|
theTree->SetBranchAddress( "jets_num_" , &jets_num_ );
|
| 348 |
|
|
theTree->SetBranchAddress( "extralep_num_" , &extralep_num_ );
|
| 349 |
|
|
theTree->SetBranchAddress( "lowptbtags_num_" , &lowptbtags_num_ );
|
| 350 |
|
|
theTree->SetBranchAddress( "softmu_num_" , &softmu_num_ );
|
| 351 |
|
|
theTree->SetBranchAddress( "event_scale1fb_" , &event_scale1fb_ );
|
| 352 |
|
|
|
| 353 |
|
|
// Efficiency calculator for cut method
|
| 354 |
|
|
Int_t nSelCutsGA = 0;
|
| 355 |
|
|
Double_t effS = 0.7;
|
| 356 |
|
|
|
| 357 |
|
|
std::vector<Float_t> vecVar(4); // vector for EvaluateMVA tests
|
| 358 |
|
|
|
| 359 |
|
|
std::cout << "--- Processing: " << theTree->GetEntries() << " events" << std::endl;
|
| 360 |
|
|
TStopwatch sw;
|
| 361 |
|
|
sw.Start();
|
| 362 |
|
|
|
| 363 |
|
|
int npass = 0;
|
| 364 |
|
|
|
| 365 |
|
|
for (Long64_t ievt=0; ievt<theTree->GetEntries();ievt++) {
|
| 366 |
|
|
|
| 367 |
|
|
if (ievt%1000 == 0) std::cout << "--- ... Processing event: " << ievt << std::endl;
|
| 368 |
|
|
|
| 369 |
|
|
theTree->GetEntry(ievt);
|
| 370 |
|
|
|
| 371 |
benhoob |
1.3 |
//-------------------------------------------------------
|
| 372 |
|
|
// event selection
|
| 373 |
|
|
//-------------------------------------------------------
|
| 374 |
|
|
|
| 375 |
|
|
//em
|
| 376 |
|
|
if( event_type_ == 1 || event_type == 2 ){
|
| 377 |
|
|
if( met_projpt_ < 20. ) continue;
|
| 378 |
|
|
}
|
| 379 |
|
|
//ee/mm
|
| 380 |
|
|
if( event_type_ == 0 || event_type == 3 ){
|
| 381 |
|
|
if( met_projpt_ < 35. ) continue;
|
| 382 |
|
|
}
|
| 383 |
|
|
|
| 384 |
benhoob |
1.1 |
if( lephard_pt_ < 20. ) continue;
|
| 385 |
benhoob |
1.2 |
//if( lepsoft_pt_ < 10. ) continue;
|
| 386 |
|
|
if( lepsoft_pt_ < 20. ) continue;
|
| 387 |
benhoob |
1.1 |
if( jets_num_ > 0 ) continue;
|
| 388 |
|
|
if( extralep_num_ > 0 ) continue;
|
| 389 |
|
|
if( lowptbtags_num_ > 0 ) continue;
|
| 390 |
|
|
if( softmu_num_ > 0 ) continue;
|
| 391 |
|
|
if( dil_mass_ < 12. ) continue;
|
| 392 |
|
|
|
| 393 |
|
|
float weight = event_scale1fb_ * 0.0355;
|
| 394 |
|
|
|
| 395 |
|
|
lephard_pt = lephard_pt_;
|
| 396 |
|
|
lepsoft_pt = lepsoft_pt_;
|
| 397 |
|
|
dil_mass = dil_mass_;
|
| 398 |
|
|
dil_dphi = dil_dphi_;
|
| 399 |
|
|
|
| 400 |
|
|
npass++;
|
| 401 |
|
|
// var1 = userVar1 + userVar2;
|
| 402 |
|
|
// var2 = userVar1 - userVar2;
|
| 403 |
|
|
|
| 404 |
|
|
// --- Return the MVA outputs and fill into histograms
|
| 405 |
|
|
|
| 406 |
|
|
if (Use["CutsGA"]) {
|
| 407 |
|
|
// Cuts is a special case: give the desired signal efficienciy
|
| 408 |
|
|
Bool_t passed = reader->EvaluateMVA( "CutsGA method", effS );
|
| 409 |
|
|
if (passed) nSelCutsGA++;
|
| 410 |
|
|
}
|
| 411 |
|
|
|
| 412 |
|
|
if (Use["Likelihood" ]) histLk ->Fill( reader->EvaluateMVA( "Likelihood method" ) , weight);
|
| 413 |
|
|
if (Use["LikelihoodD" ]) histLkD ->Fill( reader->EvaluateMVA( "LikelihoodD method" ) , weight);
|
| 414 |
|
|
if (Use["LikelihoodPCA"]) histLkPCA ->Fill( reader->EvaluateMVA( "LikelihoodPCA method" ) , weight);
|
| 415 |
|
|
if (Use["LikelihoodKDE"]) histLkKDE ->Fill( reader->EvaluateMVA( "LikelihoodKDE method" ) , weight);
|
| 416 |
|
|
if (Use["LikelihoodMIX"]) histLkMIX ->Fill( reader->EvaluateMVA( "LikelihoodMIX method" ) , weight);
|
| 417 |
|
|
if (Use["PDERS" ]) histPD ->Fill( reader->EvaluateMVA( "PDERS method" ) , weight);
|
| 418 |
|
|
if (Use["PDERSD" ]) histPDD ->Fill( reader->EvaluateMVA( "PDERSD method" ) , weight);
|
| 419 |
|
|
if (Use["PDERSPCA" ]) histPDPCA ->Fill( reader->EvaluateMVA( "PDERSPCA method" ) , weight);
|
| 420 |
|
|
if (Use["KNN" ]) histKNN ->Fill( reader->EvaluateMVA( "KNN method" ) , weight);
|
| 421 |
|
|
if (Use["HMatrix" ]) histHm ->Fill( reader->EvaluateMVA( "HMatrix method" ) , weight);
|
| 422 |
|
|
if (Use["Fisher" ]) histFi ->Fill( reader->EvaluateMVA( "Fisher method" ) , weight);
|
| 423 |
|
|
if (Use["FisherG" ]) histFiG ->Fill( reader->EvaluateMVA( "FisherG method" ) , weight);
|
| 424 |
|
|
if (Use["BoostedFisher"]) histFiB ->Fill( reader->EvaluateMVA( "BoostedFisher method" ) , weight);
|
| 425 |
|
|
if (Use["LD" ]) histLD ->Fill( reader->EvaluateMVA( "LD method" ) , weight);
|
| 426 |
|
|
if (Use["MLP" ]) histNn ->Fill( reader->EvaluateMVA( "MLP method" ) , weight);
|
| 427 |
|
|
if (Use["MLPBFGS" ]) histNnbfgs ->Fill( reader->EvaluateMVA( "MLPBFGS method" ) , weight);
|
| 428 |
|
|
if (Use["MLPBNN" ]) histNnbnn ->Fill( reader->EvaluateMVA( "MLPBNN method" ) , weight);
|
| 429 |
|
|
if (Use["CFMlpANN" ]) histNnC ->Fill( reader->EvaluateMVA( "CFMlpANN method" ) , weight);
|
| 430 |
|
|
if (Use["TMlpANN" ]) histNnT ->Fill( reader->EvaluateMVA( "TMlpANN method" ) , weight);
|
| 431 |
|
|
if (Use["BDT" ]) histBdt ->Fill( reader->EvaluateMVA( "BDT method" ) , weight);
|
| 432 |
|
|
if (Use["BDTD" ]) histBdtD ->Fill( reader->EvaluateMVA( "BDTD method" ) , weight);
|
| 433 |
|
|
if (Use["BDTG" ]) histBdtG ->Fill( reader->EvaluateMVA( "BDTG method" ) , weight);
|
| 434 |
|
|
if (Use["RuleFit" ]) histRf ->Fill( reader->EvaluateMVA( "RuleFit method" ) , weight);
|
| 435 |
|
|
if (Use["SVM_Gauss" ]) histSVMG ->Fill( reader->EvaluateMVA( "SVM_Gauss method" ) , weight);
|
| 436 |
|
|
if (Use["SVM_Poly" ]) histSVMP ->Fill( reader->EvaluateMVA( "SVM_Poly method" ) , weight);
|
| 437 |
|
|
if (Use["SVM_Lin" ]) histSVML ->Fill( reader->EvaluateMVA( "SVM_Lin method" ) , weight);
|
| 438 |
|
|
if (Use["FDA_MT" ]) histFDAMT ->Fill( reader->EvaluateMVA( "FDA_MT method" ) , weight);
|
| 439 |
|
|
if (Use["FDA_GA" ]) histFDAGA ->Fill( reader->EvaluateMVA( "FDA_GA method" ) , weight);
|
| 440 |
|
|
if (Use["Category" ]) histCat ->Fill( reader->EvaluateMVA( "Category method" ) , weight);
|
| 441 |
|
|
if (Use["Plugin" ]) histPBdt ->Fill( reader->EvaluateMVA( "P_BDT method" ) , weight);
|
| 442 |
|
|
|
| 443 |
|
|
// Retrieve also per-event error
|
| 444 |
|
|
if (Use["PDEFoam"]) {
|
| 445 |
|
|
Double_t val = reader->EvaluateMVA( "PDEFoam method" );
|
| 446 |
|
|
Double_t err = reader->GetMVAError();
|
| 447 |
|
|
histPDEFoam ->Fill( val );
|
| 448 |
|
|
histPDEFoamErr->Fill( err );
|
| 449 |
|
|
if (err>1.e-50) histPDEFoamSig->Fill( val/err , weight);
|
| 450 |
|
|
}
|
| 451 |
|
|
|
| 452 |
|
|
// Retrieve probability instead of MVA output
|
| 453 |
|
|
if (Use["Fisher"]) {
|
| 454 |
|
|
probHistFi ->Fill( reader->GetProba ( "Fisher method" ) , weight);
|
| 455 |
|
|
rarityHistFi->Fill( reader->GetRarity( "Fisher method" ) , weight);
|
| 456 |
|
|
}
|
| 457 |
|
|
}
|
| 458 |
|
|
|
| 459 |
|
|
std::cout << npass << " events passing selection" << std::endl;
|
| 460 |
|
|
|
| 461 |
|
|
// Get elapsed time
|
| 462 |
|
|
sw.Stop();
|
| 463 |
|
|
std::cout << "--- End of event loop: "; sw.Print();
|
| 464 |
|
|
|
| 465 |
|
|
// Get efficiency for cuts classifier
|
| 466 |
|
|
if (Use["CutsGA"]) std::cout << "--- Efficiency for CutsGA method: " << double(nSelCutsGA)/theTree->GetEntries()
|
| 467 |
|
|
<< " (for a required signal efficiency of " << effS << ")" << std::endl;
|
| 468 |
|
|
|
| 469 |
|
|
if (Use["CutsGA"]) {
|
| 470 |
|
|
|
| 471 |
|
|
// test: retrieve cuts for particular signal efficiency
|
| 472 |
|
|
// CINT ignores dynamic_casts so we have to use a cuts-secific Reader function to acces the pointer
|
| 473 |
|
|
TMVA::MethodCuts* mcuts = reader->FindCutsMVA( "CutsGA method" ) ;
|
| 474 |
|
|
|
| 475 |
|
|
if (mcuts) {
|
| 476 |
|
|
std::vector<Double_t> cutsMin;
|
| 477 |
|
|
std::vector<Double_t> cutsMax;
|
| 478 |
|
|
mcuts->GetCuts( 0.7, cutsMin, cutsMax );
|
| 479 |
|
|
std::cout << "--- -------------------------------------------------------------" << std::endl;
|
| 480 |
|
|
std::cout << "--- Retrieve cut values for signal efficiency of 0.7 from Reader" << std::endl;
|
| 481 |
|
|
for (UInt_t ivar=0; ivar<cutsMin.size(); ivar++) {
|
| 482 |
|
|
std::cout << "... Cut: "
|
| 483 |
|
|
<< cutsMin[ivar]
|
| 484 |
|
|
<< " < \""
|
| 485 |
|
|
<< mcuts->GetInputVar(ivar)
|
| 486 |
|
|
<< "\" <= "
|
| 487 |
|
|
<< cutsMax[ivar] << std::endl;
|
| 488 |
|
|
}
|
| 489 |
|
|
std::cout << "--- -------------------------------------------------------------" << std::endl;
|
| 490 |
|
|
}
|
| 491 |
|
|
}
|
| 492 |
|
|
|
| 493 |
|
|
// --- Write histograms
|
| 494 |
|
|
cout << "dir " << dir << endl;
|
| 495 |
|
|
char* mydir = outdir;
|
| 496 |
|
|
TFile *target = new TFile( Form("%s/%s.root",mydir,samples.at(i) ) ,"RECREATE" );
|
| 497 |
|
|
cout << "Writing to file " << Form("%s/%s.root",mydir,samples.at(i) ) << endl;
|
| 498 |
|
|
|
| 499 |
|
|
if (Use["Likelihood" ]) histLk ->Write();
|
| 500 |
|
|
if (Use["LikelihoodD" ]) histLkD ->Write();
|
| 501 |
|
|
if (Use["LikelihoodPCA"]) histLkPCA ->Write();
|
| 502 |
|
|
if (Use["LikelihoodKDE"]) histLkKDE ->Write();
|
| 503 |
|
|
if (Use["LikelihoodMIX"]) histLkMIX ->Write();
|
| 504 |
|
|
if (Use["PDERS" ]) histPD ->Write();
|
| 505 |
|
|
if (Use["PDERSD" ]) histPDD ->Write();
|
| 506 |
|
|
if (Use["PDERSPCA" ]) histPDPCA ->Write();
|
| 507 |
|
|
if (Use["KNN" ]) histKNN ->Write();
|
| 508 |
|
|
if (Use["HMatrix" ]) histHm ->Write();
|
| 509 |
|
|
if (Use["Fisher" ]) histFi ->Write();
|
| 510 |
|
|
if (Use["FisherG" ]) histFiG ->Write();
|
| 511 |
|
|
if (Use["BoostedFisher"]) histFiB ->Write();
|
| 512 |
|
|
if (Use["LD" ]) histLD ->Write();
|
| 513 |
|
|
if (Use["MLP" ]) histNn ->Write();
|
| 514 |
|
|
if (Use["MLPBFGS" ]) histNnbfgs ->Write();
|
| 515 |
|
|
if (Use["MLPBNN" ]) histNnbnn ->Write();
|
| 516 |
|
|
if (Use["CFMlpANN" ]) histNnC ->Write();
|
| 517 |
|
|
if (Use["TMlpANN" ]) histNnT ->Write();
|
| 518 |
|
|
if (Use["BDT" ]) histBdt ->Write();
|
| 519 |
|
|
if (Use["BDTD" ]) histBdtD ->Write();
|
| 520 |
|
|
if (Use["BDTG" ]) histBdtG ->Write();
|
| 521 |
|
|
if (Use["RuleFit" ]) histRf ->Write();
|
| 522 |
|
|
if (Use["SVM_Gauss" ]) histSVMG ->Write();
|
| 523 |
|
|
if (Use["SVM_Poly" ]) histSVMP ->Write();
|
| 524 |
|
|
if (Use["SVM_Lin" ]) histSVML ->Write();
|
| 525 |
|
|
if (Use["FDA_MT" ]) histFDAMT ->Write();
|
| 526 |
|
|
if (Use["FDA_GA" ]) histFDAGA ->Write();
|
| 527 |
|
|
if (Use["Category" ]) histCat ->Write();
|
| 528 |
|
|
if (Use["Plugin" ]) histPBdt ->Write();
|
| 529 |
|
|
|
| 530 |
|
|
// Write also error and significance histos
|
| 531 |
|
|
if (Use["PDEFoam"]) { histPDEFoam->Write(); histPDEFoamErr->Write(); histPDEFoamSig->Write(); }
|
| 532 |
|
|
|
| 533 |
|
|
// Write also probability hists
|
| 534 |
|
|
if (Use["Fisher"]) { if (probHistFi != 0) probHistFi->Write(); if (rarityHistFi != 0) rarityHistFi->Write(); }
|
| 535 |
|
|
target->Close();
|
| 536 |
|
|
|
| 537 |
|
|
delete reader;
|
| 538 |
|
|
|
| 539 |
|
|
std::cout << "==> TMVAClassificationApplication is done with sample " << samples.at(i) << endl << std::endl;
|
| 540 |
|
|
}
|
| 541 |
|
|
}
|
