| 1 |
dkralph |
1.1 |
#include "TFile.h"
|
| 2 |
|
|
#include "TNtuple.h"
|
| 3 |
|
|
#include "TMath.h"
|
| 4 |
|
|
|
| 5 |
|
|
#include "TMVA/Factory.h"
|
| 6 |
|
|
#include "TMVA/Tools.h"
|
| 7 |
|
|
#include "TMVA/Config.h"
|
| 8 |
|
|
|
| 9 |
|
|
#include <unistd.h>
|
| 10 |
|
|
#include <iostream>
|
| 11 |
|
|
#include <fstream>
|
| 12 |
|
|
#include <string.h>
|
| 13 |
|
|
#include <map>
|
| 14 |
|
|
#include <cassert>
|
| 15 |
|
|
|
| 16 |
dkralph |
1.2 |
#include "KinematicsStruct.h"
|
| 17 |
|
|
#include "HistHeaders.h"
|
| 18 |
|
|
|
| 19 |
dkralph |
1.1 |
using namespace std;
|
| 20 |
|
|
|
| 21 |
|
|
void setFractions(float mH, float mH_lo, float mH_hi, float &frac_lo, float &frac_hi);
|
| 22 |
|
|
void setNevents(int nEvtMax, int nSigTrainMax, int nSigTestMax, int nBkgTrainMax, int nBkgTestMax, TString &nSig, TString &nBkg);
|
| 23 |
|
|
|
| 24 |
|
|
int main(int argc, char** argv) {
|
| 25 |
|
|
|
| 26 |
|
|
if( argc < 7 ) {
|
| 27 |
|
|
cerr << "usage: makeBDTWeights.exe <config> <outputrootfile> <weightname> <varstring> <mH> [wL] [wH] [mH_lo] [mH_hi]" << endl;
|
| 28 |
|
|
cerr << "\t... where [wL,wH] are the requested mass window and [mH_lo,mH_hi] are used for interpolation" << endl;
|
| 29 |
|
|
return 1;
|
| 30 |
|
|
}
|
| 31 |
|
|
cout << "argc: " << argc << endl;
|
| 32 |
|
|
for(int iarg=0; iarg<argc; iarg++)
|
| 33 |
|
|
cout << "argv[" << iarg << "]: " << argv[iarg] << endl;
|
| 34 |
|
|
|
| 35 |
|
|
char * config = argv[1];
|
| 36 |
|
|
cout << "config: " << config << endl;
|
| 37 |
|
|
|
| 38 |
|
|
// TString ntupledir;
|
| 39 |
|
|
vector<TString> classes,fileNames;
|
| 40 |
|
|
vector<double> fileFracs,fileMasses;
|
| 41 |
|
|
vector<TFile*> inFiles;
|
| 42 |
|
|
vector<TTree*> inTrees;
|
| 43 |
|
|
bool multiclass=false;
|
| 44 |
|
|
bool multisigs=false;
|
| 45 |
dkralph |
1.2 |
bool noFakeTest=false;
|
| 46 |
dkralph |
1.1 |
TString NTrees(""),NNodesMax(""),Shrinkage(""),nCuts("");
|
| 47 |
dkralph |
1.2 |
int nMax=1000000;
|
| 48 |
dkralph |
1.1 |
ifstream ifs(config);
|
| 49 |
|
|
assert(ifs.is_open());
|
| 50 |
|
|
string line;
|
| 51 |
|
|
while(getline(ifs,line)) {
|
| 52 |
|
|
if(line[0]=='#') continue;
|
| 53 |
|
|
stringstream ss(line);
|
| 54 |
|
|
if(line[0]=='^') {
|
| 55 |
|
|
TString dummy;
|
| 56 |
|
|
// if(TString(line).Contains("^ntupledir")) ss >> dummy >> ntupledir;
|
| 57 |
|
|
if(TString(line).Contains("^multiclass")) { cout << "multiclass: true" << endl; multiclass = true; }
|
| 58 |
|
|
if(TString(line).Contains("^multisigs")) { cout << "multisigs: true" << endl; multisigs = true; }
|
| 59 |
dkralph |
1.2 |
if(TString(line).Contains("^noFakeTest")) { cout << "noFakeTest: true" << endl; noFakeTest = true; } // use all the events for training... we have an orthogonal sample for testing, anyway, and we need the stat
|
| 60 |
dkralph |
1.1 |
if(TString(line).Contains("^NTrees")) ss >> dummy >> NTrees;
|
| 61 |
|
|
if(TString(line).Contains("^NNodesMax")) ss >> dummy >> NNodesMax;
|
| 62 |
|
|
if(TString(line).Contains("^Shrinkage")) ss >> dummy >> Shrinkage;
|
| 63 |
|
|
if(TString(line).Contains("^nCuts")) ss >> dummy >> nCuts;
|
| 64 |
|
|
if(TString(line).Contains("^nMax")) ss >> dummy >> nMax;
|
| 65 |
|
|
continue;
|
| 66 |
|
|
}
|
| 67 |
|
|
|
| 68 |
|
|
TString cls,fname;
|
| 69 |
|
|
double frac,fileMass;
|
| 70 |
|
|
ss >> cls >> frac >> fname >> fileMass;
|
| 71 |
|
|
assert(cls=="sig" || cls=="sig2" || cls=="sig3" || cls=="bkg");
|
| 72 |
|
|
classes.push_back(cls);
|
| 73 |
|
|
fileNames.push_back(/*ntupledir+"/"+*/fname);
|
| 74 |
|
|
fileFracs.push_back(frac);
|
| 75 |
|
|
fileMasses.push_back(fileMass);
|
| 76 |
|
|
cout << " pushing back: " << cls << " " << setw(12) << frac << " " << fname << " " << fileMass << endl;
|
| 77 |
|
|
inFiles.push_back(TFile::Open(fileNames.back()));
|
| 78 |
|
|
if(!inFiles.back()->IsOpen()) {
|
| 79 |
|
|
cout << fileNames.back() << " not open!" << endl;
|
| 80 |
|
|
assert(0);
|
| 81 |
|
|
}
|
| 82 |
|
|
inTrees.push_back((TTree*)inFiles.back()->Get("zznt"));
|
| 83 |
|
|
assert(inTrees.back());
|
| 84 |
|
|
}
|
| 85 |
|
|
ifs.close();
|
| 86 |
|
|
// if(multisigs) assert(multiclass);
|
| 87 |
|
|
|
| 88 |
|
|
char * outputrootfile = argv[2];
|
| 89 |
|
|
char * weightname = argv[3];
|
| 90 |
|
|
char * varstring = argv[4];
|
| 91 |
|
|
float mH = strtof(argv[5],NULL);
|
| 92 |
|
|
float wH=900, wL=100;
|
| 93 |
|
|
if( argc > 6 ) {
|
| 94 |
|
|
wL = strtof(argv[6], NULL);
|
| 95 |
|
|
wH = strtof(argv[7], NULL);
|
| 96 |
|
|
}
|
| 97 |
|
|
float mH_lo=0,mH_hi=0;
|
| 98 |
|
|
float frac_lo=1,frac_hi=1;
|
| 99 |
|
|
map<double,double> mH_fracs;
|
| 100 |
|
|
mH_fracs[-1] = 1; // this means zz
|
| 101 |
|
|
assert( argc > 8);
|
| 102 |
|
|
mH_lo = strtof(argv[8],NULL);
|
| 103 |
|
|
mH_hi = strtof(argv[9],NULL);
|
| 104 |
|
|
if(mH_lo!=0 && mH_hi!=0) {
|
| 105 |
|
|
// set the fraction to use for each input file, with more events from the closer mass point
|
| 106 |
|
|
// NOTE: we only use these fractions if the fraction from the config file is -1
|
| 107 |
|
|
setFractions(mH,mH_lo,mH_hi,frac_lo,frac_hi);
|
| 108 |
|
|
mH_fracs[mH_lo] = frac_lo;
|
| 109 |
|
|
mH_fracs[mH_hi] = frac_hi;
|
| 110 |
|
|
} else {
|
| 111 |
|
|
mH_fracs[mH] = 1;
|
| 112 |
|
|
}
|
| 113 |
|
|
|
| 114 |
|
|
cerr
|
| 115 |
|
|
<< "name: " << weightname << endl
|
| 116 |
|
|
<< "output: " << outputrootfile << endl
|
| 117 |
|
|
<< "variables: " << varstring << endl
|
| 118 |
|
|
<< "mH: " << mH << endl
|
| 119 |
|
|
<< "mass window:" << endl
|
| 120 |
|
|
<< " wL: " << wL << endl
|
| 121 |
|
|
<< " wH: " << wH << endl
|
| 122 |
|
|
<< "masses from which to interpolate:" << endl
|
| 123 |
|
|
<< " mH_lo: " << mH_lo << endl
|
| 124 |
|
|
<< " mH_hi: " << mH_hi << endl
|
| 125 |
|
|
<< "weights for interpolation:" << endl
|
| 126 |
|
|
<< " frac_lo: " << frac_lo << endl
|
| 127 |
|
|
<< " frac_hi: " << frac_hi << endl
|
| 128 |
|
|
<< endl;
|
| 129 |
|
|
|
| 130 |
|
|
TFile * outputFile = new TFile(outputrootfile, "RECREATE");
|
| 131 |
|
|
|
| 132 |
|
|
map<string,bool> varmap;
|
| 133 |
|
|
char * dovar = strtok(varstring, ":");
|
| 134 |
|
|
if( dovar != NULL ) varmap[string(dovar)]=true;
|
| 135 |
|
|
while( (dovar = strtok(NULL, ":")) != NULL ) {
|
| 136 |
|
|
varmap[string(dovar)]=true;
|
| 137 |
|
|
}
|
| 138 |
|
|
|
| 139 |
|
|
char name[256];
|
| 140 |
|
|
sprintf(name, weightname );
|
| 141 |
|
|
TMVA::Tools::Instance();
|
| 142 |
|
|
TString facStr("!V:!Silent:!DrawProgressBar");
|
| 143 |
|
|
if(multiclass) facStr = facStr+":AnalysisType=multiclass";
|
| 144 |
|
|
TMVA::Factory* factory = new TMVA::Factory( name, outputFile, facStr );
|
| 145 |
dkralph |
1.2 |
// if( varmap[string("costheta1" )] ) factory->AddVariable( "costheta1", 'F',-1, 1);
|
| 146 |
|
|
// if( varmap[string("costheta2" )] ) factory->AddVariable( "costheta2", 'F',-1, 1);
|
| 147 |
|
|
// if( varmap[string("costhetastar" )] ) factory->AddVariable( "costhetastar", 'F',-1, 1 );
|
| 148 |
|
|
// if( varmap[string("Phi" )] ) factory->AddVariable( "Phi", 'F',-TMath::Pi(), TMath::Pi() );
|
| 149 |
|
|
// if( varmap[string("Phi1" )] ) factory->AddVariable( "Phi1", 'F',-TMath::Pi(), TMath::Pi() );
|
| 150 |
|
|
if( varmap[string("costheta1" )] ) factory->AddVariable("costheta1", 'F');
|
| 151 |
|
|
if( varmap[string("costheta2" )] ) factory->AddVariable("costheta2", 'F');
|
| 152 |
|
|
if( varmap[string("costhetastar" )] ) factory->AddVariable("costhetastar", 'F');
|
| 153 |
|
|
if( varmap[string("Phi" )] ) factory->AddVariable("Phi", 'F');
|
| 154 |
|
|
if( varmap[string("Phi1" )] ) factory->AddVariable("Phi1", 'F');
|
| 155 |
|
|
if( varmap[string("mZ1" )] ) factory->AddVariable("mZ1", 'F');
|
| 156 |
|
|
if( varmap[string("mZ2" )] ) factory->AddVariable("mZ2", 'F');
|
| 157 |
|
|
if( varmap[string("pt4l" )] ) factory->AddVariable("ZZpt/m4l", 'F');
|
| 158 |
|
|
if( varmap[string("zzdotz1" )] ) factory->AddVariable("ZZdotZ1/(m4l*mZ1)", 'F');
|
| 159 |
|
|
if( varmap[string("zzdotz2" )] ) factory->AddVariable("ZZdotZ2/(m4l*mZ2)", 'F');
|
| 160 |
|
|
// if( varmap[string("dphi1" )] ) factory->AddVariable("ZZptCosDphiZ1pt", 'F', -TMath::Pi(), TMath::Pi() );
|
| 161 |
|
|
// if( varmap[string("dphi2" )] ) factory->AddVariable("ZZptCosDphiZ2pt", 'F', -TMath::Pi(), TMath::Pi() );
|
| 162 |
|
|
if( varmap[string("dphi1" )] ) factory->AddVariable("ZZptCosDphiZ1pt", 'F');
|
| 163 |
|
|
if( varmap[string("dphi2" )] ) factory->AddVariable("ZZptCosDphiZ2pt", 'F');
|
| 164 |
|
|
if( varmap[string("Z1pt" )] ) factory->AddVariable("Z1pt/m4l", 'F');
|
| 165 |
|
|
if( varmap[string("Z2pt" )] ) factory->AddVariable("Z2pt/m4l", 'F');
|
| 166 |
|
|
if( varmap[string("y4l" )] ) factory->AddVariable("ZZy", 'F');
|
| 167 |
|
|
if( varmap[string("nJets" )] ) factory->AddVariable("nJets", 'F');
|
| 168 |
|
|
if( varmap[string("mjj" )] ) factory->AddVariable("mjj", 'F');
|
| 169 |
|
|
if( varmap[string("dEta" )] ) factory->AddVariable("dEta", 'F');
|
| 170 |
|
|
if( varmap[string("etaProd" )] ) factory->AddVariable("etaProd", 'F');
|
| 171 |
|
|
if( varmap[string("dphiJ1HiPtZ" )] ) factory->AddVariable("dphiJ1HiPtZ", 'F');
|
| 172 |
|
|
if( varmap[string("dphiJ1LoPtZ" )] ) factory->AddVariable("dphiJ1LoPtZ", 'F');
|
| 173 |
|
|
if( varmap[string("dEtaJ1HiPtZ" )] ) factory->AddVariable("dEtaJ1HiPtZ", 'F');
|
| 174 |
|
|
if( varmap[string("dEtaJ1LoPtZ" )] ) factory->AddVariable("dEtaJ1LoPtZ", 'F');
|
| 175 |
|
|
if( varmap[string("J1dotHiPtZ" )] ) factory->AddVariable("J1dotHiPtZ", 'F');
|
| 176 |
|
|
if( varmap[string("J1dotLoPtZ" )] ) factory->AddVariable("J1dotLoPtZ", 'F');
|
| 177 |
|
|
if( varmap[string("dphiJ2HiPtZ" )] ) factory->AddVariable("dphiJ2HiPtZ", 'F');
|
| 178 |
|
|
if( varmap[string("dphiJ2LoPtZ" )] ) factory->AddVariable("dphiJ2LoPtZ", 'F');
|
| 179 |
|
|
if( varmap[string("dEtaJ2HiPtZ" )] ) factory->AddVariable("dEtaJ2HiPtZ", 'F');
|
| 180 |
|
|
if( varmap[string("dEtaJ2LoPtZ" )] ) factory->AddVariable("dEtaJ2LoPtZ", 'F');
|
| 181 |
|
|
if( varmap[string("J2dotHiPtZ" )] ) factory->AddVariable("J2dotHiPtZ", 'F');
|
| 182 |
|
|
if( varmap[string("J2dotLoPtZ" )] ) factory->AddVariable("J2dotLoPtZ", 'F');
|
| 183 |
dkralph |
1.1 |
|
| 184 |
|
|
if( varmap[string("m4l")] ) factory->AddVariable( "m4l", 'F' );
|
| 185 |
|
|
else factory->AddSpectator("m4l", 'F' );
|
| 186 |
|
|
|
| 187 |
|
|
char buf[256];
|
| 188 |
|
|
sprintf( buf, "(m4l>%f&&m4l<%f)", (float)wL, (float)wH );
|
| 189 |
|
|
TCut mycuts(buf);
|
| 190 |
|
|
TString cutStr(buf);
|
| 191 |
|
|
cout << "cutStr: " << cutStr << endl;
|
| 192 |
|
|
|
| 193 |
|
|
vector<TString> sigFiles,bkgFiles;
|
| 194 |
|
|
vector<double> passFracs,sigFileFracs,bkgFileFracs,nSigPassCutv,nBkgPassCutv;
|
| 195 |
|
|
for(unsigned ifile=0; ifile<fileNames.size(); ifile++) {
|
| 196 |
|
|
double nPassCut = double(inTrees[ifile]->GetEntries(cutStr));
|
| 197 |
|
|
if(fileFracs[ifile] == -1)
|
| 198 |
|
|
fileFracs[ifile] = mH_fracs[fileMasses[ifile]];
|
| 199 |
|
|
passFracs.push_back( nPassCut / inTrees[ifile]->GetEntries() );
|
| 200 |
|
|
cout << classes[ifile] << ": "
|
| 201 |
|
|
<< setw(8) << nPassCut << " / " << setw(8) << inTrees[ifile]->GetEntries() << " events pass cut in file: " << fileNames[ifile] << ", filefrac: " << fileFracs[ifile] << endl;
|
| 202 |
|
|
if(classes[ifile].Contains("sig")) {
|
| 203 |
|
|
sigFiles.push_back(fileNames[ifile]);
|
| 204 |
|
|
sigFileFracs.push_back(fileFracs[ifile]);
|
| 205 |
|
|
nSigPassCutv.push_back( nPassCut );
|
| 206 |
|
|
} else if(classes[ifile].Contains("bkg")) {
|
| 207 |
|
|
bkgFiles.push_back(fileNames[ifile]);
|
| 208 |
|
|
bkgFileFracs.push_back(fileFracs[ifile]);
|
| 209 |
|
|
nBkgPassCutv.push_back( nPassCut );
|
| 210 |
|
|
} else assert(0);
|
| 211 |
|
|
}
|
| 212 |
|
|
|
| 213 |
|
|
double nMin=999999;
|
| 214 |
|
|
double nSig=0;
|
| 215 |
|
|
if(sigFiles.size()==2) {
|
| 216 |
|
|
cout << " signal comparison: " << nSigPassCutv[0]*sigFileFracs[1] << " " << nSigPassCutv[1]*sigFileFracs[0] << endl;
|
| 217 |
|
|
if(nSigPassCutv[0]*sigFileFracs[1] > nSigPassCutv[1]*sigFileFracs[0]) { // statistics are limited by second signal file
|
| 218 |
|
|
cout << " use second signal as min signal" << endl;
|
| 219 |
|
|
nMin = nSig = nSigPassCutv[1] * ( 1 + sigFileFracs[0] / sigFileFracs[1] );
|
| 220 |
|
|
} else { // limited by first one
|
| 221 |
|
|
cout << " use first signal as min signal" << endl;
|
| 222 |
|
|
nMin = nSig = nSigPassCutv[0] * ( 1 + sigFileFracs[1] / sigFileFracs[0] );
|
| 223 |
|
|
}
|
| 224 |
|
|
} else if(sigFiles.size()==1 && mH_lo==0 && mH_hi==0) { // not interpolating
|
| 225 |
|
|
nMin = nSig = nSigPassCutv[0];
|
| 226 |
|
|
} else assert(multisigs);
|
| 227 |
|
|
cout << "nMin after signal check: " << nMin << endl;
|
| 228 |
|
|
|
| 229 |
|
|
long nBkg=0,nBkg2=0;
|
| 230 |
|
|
for(unsigned ibkg=0; ibkg<bkgFiles.size(); ibkg++) {
|
| 231 |
|
|
if(multiclass && bkgFiles[ibkg].Contains("fakes"))
|
| 232 |
|
|
nBkg2 += nBkgPassCutv[ibkg];
|
| 233 |
|
|
else
|
| 234 |
|
|
nBkg += nBkgPassCutv[ibkg];
|
| 235 |
|
|
}
|
| 236 |
|
|
cout << "nBkg: " << nBkg << "\tnBkg2: " << nBkg2 << endl;
|
| 237 |
dkralph |
1.2 |
if(!multisigs && nBkg < nMin) nMin = nBkg; // don't include fakes in this comparison. Also, for multisigs we train two signal samples against each other, so we don't care if there is less bkg
|
| 238 |
dkralph |
1.1 |
cout << "nMin after bkg check: " << nMin << endl;
|
| 239 |
|
|
|
| 240 |
|
|
double nTarget = min(nMin,double(nMax));
|
| 241 |
|
|
cout << "nTarget: " << nTarget << endl;
|
| 242 |
|
|
|
| 243 |
|
|
map<TString,double> fracUsed;
|
| 244 |
|
|
if(nTarget < nSig) fracUsed["sig"] = double(nTarget) / nSig; else fracUsed["sig"] = 1;
|
| 245 |
|
|
if(nTarget < nBkg) fracUsed["bkg"] = double(nTarget) / nBkg; else fracUsed["bkg"] = 1;
|
| 246 |
|
|
cout << "frac Used sig: " << fracUsed["sig"] << ", bkg: " << fracUsed["bkg"] << endl;
|
| 247 |
|
|
if(multisigs) {
|
| 248 |
|
|
fracUsed["sig2"] = fracUsed["sig3"] = fracUsed["sig"];
|
| 249 |
|
|
}
|
| 250 |
|
|
|
| 251 |
|
|
vector<TTree*> TreeCopies; // copies of the input trees, but with only the number of events we want
|
| 252 |
|
|
TList *sigList = new TList;
|
| 253 |
|
|
TList *bkgList = new TList;
|
| 254 |
|
|
TList *bkg2List = new TList;
|
| 255 |
|
|
TList *sig2List = new TList;
|
| 256 |
|
|
TList *sig3List = new TList;
|
| 257 |
|
|
cout << "making tree copies: " << endl;
|
| 258 |
|
|
for(unsigned ifile=0; ifile<fileNames.size(); ifile++) {
|
| 259 |
|
|
// if(mH_fracs.find(fileMasses[ifile]) == mH_fracs.end()) { cout << "filemasses[ifile]: " << fileMasses[ifile] << endl; assert(0); }
|
| 260 |
|
|
|
| 261 |
|
|
long nToRequest;
|
| 262 |
|
|
nToRequest = /*fileFracs[ifile] * */inTrees[ifile]->GetEntries(); // / passFracs[ifile];
|
| 263 |
|
|
if(!(multiclass && classes[ifile].Contains("bkg") && fileNames[ifile].Contains("fakes")))
|
| 264 |
|
|
nToRequest *= fracUsed[classes[ifile]];
|
| 265 |
|
|
cout << " adding " << setw(12) << passFracs[ifile]*nToRequest << " from file " << fileNames[ifile]
|
| 266 |
|
|
<< ", which means asking for: " << nToRequest << endl;
|
| 267 |
|
|
TreeCopies.push_back(inTrees[ifile]->CopyTree(cutStr,"",nToRequest));
|
| 268 |
|
|
if(classes[ifile].Contains("sig")) {
|
| 269 |
|
|
if(!multisigs) {
|
| 270 |
|
|
sigList->Add(TreeCopies.back());
|
| 271 |
|
|
} else {
|
| 272 |
|
|
if(classes[ifile]=="sig") sigList->Add(TreeCopies.back());
|
| 273 |
|
|
else if(classes[ifile]=="sig2") { cout << "added to sig2list" << endl; sig2List->Add(TreeCopies.back()); }
|
| 274 |
|
|
else if(classes[ifile]=="sig3") { cout << "added to sig3list" << endl; sig3List->Add(TreeCopies.back()); }
|
| 275 |
|
|
else assert(0);
|
| 276 |
|
|
}
|
| 277 |
|
|
}
|
| 278 |
|
|
if(classes[ifile].Contains("bkg")) {
|
| 279 |
|
|
if(multiclass && fileNames[ifile].Contains("fakes"))
|
| 280 |
|
|
bkg2List->Add(TreeCopies.back());
|
| 281 |
|
|
else
|
| 282 |
|
|
bkgList->Add(TreeCopies.back());
|
| 283 |
|
|
}
|
| 284 |
|
|
}
|
| 285 |
|
|
|
| 286 |
|
|
TTree *sigTreeAll = TTree::MergeTrees(sigList);
|
| 287 |
|
|
TTree *sig2TreeAll = TTree::MergeTrees(sig2List);
|
| 288 |
|
|
TTree *sig3TreeAll = sig3List->GetSize() ? TTree::MergeTrees(sig3List) : 0;
|
| 289 |
|
|
TTree *bkgTreeAll = TTree::MergeTrees(bkgList);
|
| 290 |
|
|
TTree *bkg2TreeAll = multiclass ? TTree::MergeTrees(bkg2List) : 0;
|
| 291 |
|
|
cout << "sigTreeAll entries: " << sigTreeAll->GetEntries() << endl;
|
| 292 |
|
|
cout << "bkgTreeAll entries: " << bkgTreeAll->GetEntries() << endl;
|
| 293 |
|
|
if(multiclass) {
|
| 294 |
|
|
cout << "bkg2TreeAll entries: " << bkg2TreeAll->GetEntries() << endl;
|
| 295 |
|
|
if(multisigs) {
|
| 296 |
|
|
cout << "sig2TreeAll entries: " << sig2TreeAll->GetEntries() << endl;
|
| 297 |
|
|
if(sig3TreeAll)
|
| 298 |
|
|
cout << "sig3TreeAll entries: " << sig3TreeAll->GetEntries() << endl;
|
| 299 |
|
|
}
|
| 300 |
|
|
}
|
| 301 |
|
|
|
| 302 |
dkralph |
1.2 |
// // add variables to the trees that aren't already in them
|
| 303 |
|
|
// float J1dotHiPtZ;
|
| 304 |
|
|
// KinematicsStruct kine;
|
| 305 |
|
|
// sigTreeAll->SetBranchAddress("kinematics", &kine);
|
| 306 |
|
|
// for(unsigned ientry=0; ientry<sigTreeAll->GetEntries(); ientry++) {
|
| 307 |
|
|
// sigTreeAll->GetEntry(ientry);
|
| 308 |
|
|
// sigTreeAll->SetScanField(10);
|
| 309 |
|
|
// sigTreeAll->Scan("ptJet1");
|
| 310 |
|
|
// cout << kine.l1pt << endl;
|
| 311 |
|
|
// if(ientry>10) assert(0);
|
| 312 |
|
|
// }
|
| 313 |
|
|
|
| 314 |
|
|
|
| 315 |
dkralph |
1.1 |
factory->AddTree( sigTreeAll, "Signal", 1, "((evt%2)==0)", TMVA::Types::kTraining);
|
| 316 |
|
|
factory->AddTree( sigTreeAll, "Signal", 1, "((evt%2)!=0)", TMVA::Types::kTesting);
|
| 317 |
|
|
if(multisigs) {
|
| 318 |
|
|
if(multiclass) {
|
| 319 |
|
|
factory->AddTree( sig2TreeAll, "Signal2", 1, "((evt%2)==0)", TMVA::Types::kTraining);
|
| 320 |
|
|
factory->AddTree( sig2TreeAll, "Signal2", 1, "((evt%2)!=0)", TMVA::Types::kTesting);
|
| 321 |
|
|
} else {
|
| 322 |
|
|
factory->AddTree( sig2TreeAll, "Background", 1, "((evt%2)==0)", TMVA::Types::kTraining);
|
| 323 |
|
|
factory->AddTree( sig2TreeAll, "Background", 1, "((evt%2)!=0)", TMVA::Types::kTesting);
|
| 324 |
|
|
}
|
| 325 |
|
|
if(sig3TreeAll) {
|
| 326 |
|
|
factory->AddTree( sig3TreeAll, "Signal3", 1, "((evt%2)==0)", TMVA::Types::kTraining);
|
| 327 |
|
|
factory->AddTree( sig3TreeAll, "Signal3", 1, "((evt%2)!=0)", TMVA::Types::kTesting);
|
| 328 |
|
|
}
|
| 329 |
|
|
} else {
|
| 330 |
|
|
factory->AddTree( bkgTreeAll, "Background", 1, "((evt%2)==0)", TMVA::Types::kTraining);
|
| 331 |
|
|
factory->AddTree( bkgTreeAll, "Background", 1, "((evt%2)!=0)", TMVA::Types::kTesting);
|
| 332 |
|
|
if(multiclass) {
|
| 333 |
dkralph |
1.2 |
if(noFakeTest) {
|
| 334 |
|
|
factory->AddTree( bkg2TreeAll, "Background2", 1, "", TMVA::Types::kTraining);
|
| 335 |
|
|
factory->AddTree( bkg2TreeAll, "Background2", 1, "", TMVA::Types::kTesting); // just make 'em the same, should work...
|
| 336 |
|
|
} else {
|
| 337 |
|
|
factory->AddTree( bkg2TreeAll, "Background2", 1, "((evt%2)==0)", TMVA::Types::kTraining);
|
| 338 |
|
|
factory->AddTree( bkg2TreeAll, "Background2", 1, "((evt%2)!=0)", TMVA::Types::kTesting);
|
| 339 |
|
|
}
|
| 340 |
dkralph |
1.1 |
}
|
| 341 |
|
|
}
|
| 342 |
|
|
|
| 343 |
|
|
if(NTrees=="") NTrees = "10000";
|
| 344 |
|
|
if(NNodesMax=="") NNodesMax = "10";
|
| 345 |
|
|
if(Shrinkage=="") Shrinkage = "0.05";
|
| 346 |
|
|
if(nCuts=="") nCuts = "20";
|
| 347 |
|
|
factory->PrepareTrainingAndTestTree( "", "SplitMode=Random:V:NormMode=NumEvents");
|
| 348 |
|
|
factory->BookMethod( TMVA::Types::kBDT, "BDTG", "!H:V:NTrees="+NTrees
|
| 349 |
|
|
+":NNodesMax="+NNodesMax
|
| 350 |
|
|
+":BoostType=Grad:Shrinkage="+Shrinkage
|
| 351 |
|
|
+":UseBaggedGrad:GradBaggingFraction=0.50:nCuts="+nCuts
|
| 352 |
|
|
+":IgnoreNegWeights");
|
| 353 |
|
|
|
| 354 |
|
|
// KH params:
|
| 355 |
|
|
// factory->BookMethod( TMVA::Types::kBDT, "BDTG","!H:V:NTrees=2000:BoostType=Grad:Shrinkage=0.10:UseBaggedGrad:GradBaggingFraction=0.50:nCuts=20:UseYesNoLeaf=False");//NNodesMax=8
|
| 356 |
|
|
// decorrelate them first:
|
| 357 |
|
|
// factory->BookMethod( TMVA::Types::kBDT, "BDTG","!H:V:NTrees=2000:BoostType=Grad:Shrinkage=0.10:UseBaggedGrad:GradBaggingFraction=0.50:nCuts=20:UseYesNoLeaf=False:VarTransform=D");//NNodesMax=8
|
| 358 |
|
|
|
| 359 |
|
|
// josh:
|
| 360 |
|
|
// factory->BookMethod( TMVA::Types::kBDT, "BDTG_josh" , "!H:V:NTrees=2000:BoostType=Grad:Shrinkage=0.1:UseBaggedGrad=F:nCuts=2000:MaxDepth=3:NNodesMax=100000:UseYesNoLeaf=F:nEventsMin=10000:");
|
| 361 |
|
|
// factory->BookMethod( TMVA::Types::kBDT, "BDTG_josh" , "!H:V:NTrees=2000:BoostType=Grad:Shrinkage=0.1:UseBaggedGrad=F:nCuts=2000:MaxDepth=3:NNodesMax=100000:UseYesNoLeaf=F:nEventsMin=1000:");
|
| 362 |
|
|
// 'phil' params:
|
| 363 |
|
|
// factory->BookMethod( TMVA::Types::kBDT, "BDTG_newparams","!H:V:NTrees=200:BoostType=Grad:Shrinkage=0.01:!UseBaggedGrad:nCuts=2000:nEventsMin=100:NNodesMax=5:MaxDepth=100");
|
| 364 |
|
|
// small number of very dep trees
|
| 365 |
|
|
// ntrees should be smaller (200-500)
|
| 366 |
|
|
// max depth 100
|
| 367 |
|
|
// shrink 0.01
|
| 368 |
|
|
|
| 369 |
|
|
factory->TrainAllMethods();
|
| 370 |
|
|
factory->TestAllMethods();
|
| 371 |
|
|
factory->EvaluateAllMethods();
|
| 372 |
|
|
|
| 373 |
|
|
std::cerr << "closing ... " << std::endl;
|
| 374 |
|
|
outputFile->Close();
|
| 375 |
|
|
}
|
| 376 |
|
|
//----------------------------------------------------------------------------------------
|
| 377 |
|
|
void setFractions(float mH, float mH_lo, float mH_hi, float &frac_lo, float &frac_hi)
|
| 378 |
|
|
{
|
| 379 |
|
|
double delta = mH_hi - mH_lo;
|
| 380 |
|
|
frac_lo = (mH_hi - mH)/delta;
|
| 381 |
|
|
frac_hi = (mH - mH_lo)/delta;
|
| 382 |
|
|
}
|
| 383 |
|
|
//----------------------------------------------------------------------------------------
|
| 384 |
|
|
void setNevents(int nEvtMax, int nSigTrainMax, int nSigTestMax, int nBkgTrainMax, int nBkgTestMax, TString &nSig, TString &nBkg)
|
| 385 |
|
|
{
|
| 386 |
|
|
// we want nEvtMax, but if there's fewer than this in any of the trees, we reduce our expectations
|
| 387 |
|
|
if(fabs(nSigTrainMax-nSigTestMax)/double(nSigTrainMax) > 0.1) {
|
| 388 |
|
|
assert(0);
|
| 389 |
|
|
}
|
| 390 |
|
|
if(fabs(nBkgTrainMax-nBkgTestMax)/double(nBkgTrainMax) > 0.1) {
|
| 391 |
|
|
assert(0);
|
| 392 |
|
|
}
|
| 393 |
|
|
|
| 394 |
|
|
nEvtMax = min(nEvtMax,nSigTrainMax);
|
| 395 |
|
|
nEvtMax = min(nEvtMax,nSigTestMax);
|
| 396 |
|
|
nEvtMax = min(nEvtMax,nBkgTrainMax);
|
| 397 |
|
|
nEvtMax = min(nEvtMax,nBkgTestMax);
|
| 398 |
|
|
stringstream ss;
|
| 399 |
|
|
ss << nEvtMax;
|
| 400 |
|
|
ss >> nSig;
|
| 401 |
|
|
nBkg = nSig;
|
| 402 |
|
|
// ss << min(nSigTrainMax,nSigTestMax) << " " << min(nBkgTrainMax,nBkgTestMax);
|
| 403 |
|
|
// ss >> nSig >> nBkg;
|
| 404 |
|
|
}
|
