def __initializeSample(self,sample):
self.sample = sample
tchain = ROOT.TChain(self.treeName)
sampleDirectory = '{0}/{1}'.format(self.ntupleDirectory,sample)
summedWeights = 0.
for f in glob.glob('{0}/*.root'.format(sampleDirectory)):
tfile = ROOT.TFile.Open(f)
summedWeights += tfile.Get("summedWeights").GetBinContent(1)
tfile.Close()
tchain.Add(f)
self.sampleLumi = float(summedWeights)/getXsec(sample) if getXsec(sample) else 0.
self.sampleTree = tchain
def __initializeSample(self, sample):
self.sample = sample
tchain = ROOT.TChain(self.treeName)
sampleDirectory = '{0}/{1}'.format(self.ntupleDirectory, sample)
summedWeights = 0.
for f in glob.glob('{0}/*.root'.format(sampleDirectory)):
tfile = ROOT.TFile.Open(f)
summedWeights += tfile.Get("summedWeights").GetBinContent(1)
tfile.Close()
tchain.Add(f)
self.sampleLumi = float(summedWeights) / getXsec(sample) if getXsec(
sample) else 0.
self.sampleTree = tchain
def __initializeNtuple(self):
tchain = ROOT.TChain(self.treeName)
if self.inputFileList: # reading from a passed list of inputfiles
allFiles = []
with open(self.inputFileList, 'r') as f:
for line in f.readlines():
allFiles += [line.strip()]
else: # reading from an input directory (all files in directory will be processed)
allFiles = glob.glob('{0}/*.root'.format(self.ntupleDirectory))
if len(allFiles) == 0:
logging.error('No files found for sample {0}'.format(self.sample))
summedWeights = 0.
for f in allFiles:
tfile = ROOT.TFile.Open(f)
summedWeights += tfile.Get("summedWeights").GetBinContent(1)
tfile.Close()
tchain.Add(f)
if not summedWeights and not isData(self.sample):
logging.warning('No events for sample {0}'.format(self.sample))
self.intLumi = float(getLumi())
self.xsec = getXsec(self.sample)
self.sampleLumi = float(summedWeights) / self.xsec if self.xsec else 0.
self.sampleTree = tchain
self.files = allFiles
self.initialized = True
logging.debug(
'Initialized {0}: summedWeights = {1}; xsec = {2}; sampleLumi = {3}; intLumi = {4}'
.format(self.sample, summedWeights, self.xsec, self.sampleLumi,
self.intLumi))
def __initializeNtuple(self):
tchain = ROOT.TChain(self.treeName)
if self.inputFileList: # reading from a passed list of inputfiles
allFiles = []
with open(self.inputFileList,'r') as f:
for line in f.readlines():
allFiles += [line.strip()]
else: # reading from an input directory (all files in directory will be processed)
#allFiles = glob.glob('{0}/*.root'.format(self.ntupleDirectory))
allFiles = []
for root, dirnames, fnames in os.walk(self.ntupleDirectory):
if 'failed' in root: continue
for fname in fnmatch.filter(fnames, '*.root'):
allFiles.append(os.path.join(root,fname))
if len(allFiles)==0: logging.error('No files found for sample {0}'.format(self.sample))
summedWeights = 0.
for f in allFiles:
tfile = ROOT.TFile.Open(f)
summedWeights += tfile.Get("summedWeights").GetBinContent(1)
tfile.Close()
tchain.Add(f)
if not summedWeights and not isData(self.sample): logging.warning('No events for sample {0}'.format(self.sample))
self.xsec = getXsec(self.sample)
self.sampleLumi = float(summedWeights)/self.xsec if self.xsec else 0.
self.sampleTree = tchain
self.files = allFiles
self.initialized = True
logging.debug('Initialized {0}: summedWeights = {1}; xsec = {2}; sampleLumi = {3}; intLumi = {4}'.format(self.sample,summedWeights,self.xsec,self.sampleLumi,self.intLumi))
def __initializeNtuple(self):
tchain = ROOT.TChain(self.treeName)
if self.inputFileList: # reading from a passed list of inputfiles
allFiles = []
with open(self.inputFileList,'r') as f:
for line in f.readlines():
allFiles += [line.strip()]
else: # reading from an input directory (all files in directory will be processed)
allFiles = glob.glob('{0}/*.root'.format(self.ntupleDirectory))
#elif os.path.isfile(self.ntuple): # reading a single root file
# allFiles = [self.ntuple]
if len(allFiles)==0: logging.error('No files found for sample {0}'.format(self.sample))
summedWeights = 0.
for f in allFiles:
tfile = ROOT.TFile.Open(f)
summedWeights += tfile.Get("summedWeights").GetBinContent(1)
tfile.Close()
tchain.Add(f)
if not summedWeights and not isData(self.sample): logging.warning('No events for sample {0}'.format(self.sample))
self.xsec = getXsec(self.sample)
if not self.xsec: logging.error('No xsec for sample {0}'.format(self.sample))
self.sampleLumi = float(summedWeights)/self.xsec if self.xsec else 0.
self.sampleTree = tchain
self.j += 1
#listname = 'selList{0}'.format(self.j)
#self.sampleTree.Draw('>>{0}'.format(listname),'1','entrylist')
#skim = ROOT.gDirectory.Get(listname)
#self.entryListMap['1'] = skim
self.files = allFiles
self.initialized = True
if not self.temp: self.fileHash = hashFile(*self.files)
if self.useProof: self.sampleTree.SetProof()
logging.debug('Initialized {0}: summedWeights = {1}; xsec = {2}; sampleLumi = {3}; intLumi = {4}'.format(self.sample,summedWeights,self.xsec,self.sampleLumi,self.intLumi))
def main(argv=None):
if argv is None:
argv = sys.argv[1:]
args = parse_command_line(argv)
if args.verbose and args.analysis:
table = PrettyTable(['Sample','xsec [pb]','entries','ratio neg.','lumi. [/pb]','eff. entries'])
else:
table = PrettyTable(['Sample','xsec [pb]'])
table.align = 'r'
table.align['Sample'] = 'l'
ntupleDir = getAnalysisNtupleDirectory(args.analysis,True) if args.verbose and args.analysis else getNtupleDirectory(version=args.version)
#Odd mix of local pathnames and xrootd access
# for sample in sorted(hdfs_ls_directory(ntupleDir)):
for sample in (glob.glob('/'.join([ntupleDir,'*']))):
name = os.path.basename(sample)
logging.info('Processing {0}'.format(name))
data = isData(name)
xsec = getXsec(name)
if args.verbose and args.analysis:
print sample
fnames = get_hdfs_root_files(sample)
# get total events, total weights
tree = ROOT.TChain(getTreeName(args.analysis))
summedWeights = 0.
for f in fnames:
tfile = ROOT.TFile.Open('/hdfs'+f)
summedWeights += tfile.Get("summedWeights").GetBinContent(1)
tfile.Close()
tree.Add('/hdfs'+f)
numEntries = tree.GetEntries(args.selection)
weightedEntries = 0.
negevents = 0.
seltree = tree.CopyTree(args.selection)
for row in seltree:
if data:
weightedEntries += 1.
else:
weightedEntries += row.genWeight
if row.genWeight<0.: negevents += 1
if data:
sampleLumi = getLumi()
else:
sampleLumi = float(summedWeights)/xsec if xsec else 0.
negratio = float(negevents)/numEntries if numEntries else 0.
effevents = weightedEntries*getLumi()/sampleLumi if sampleLumi else 0.
table.add_row([name,'{0:.6f}'.format(float(xsec)),numEntries,'{0:.3}'.format(float(negratio)),'{0:.6f}'.format(float(sampleLumi)),'{0:.3f}'.format(float(effevents))])
else:
table.add_row([name,xsec])
print table.get_string()
def main(argv=None):
if argv is None:
argv = sys.argv[1:]
args = parse_command_line(argv)
if args.verbose and args.analysis:
table = PrettyTable(['Sample','xsec [pb]','entries','ratio neg.','lumi. [/pb]','eff. entries'])
else:
table = PrettyTable(['Sample','xsec [pb]'])
table.align = 'r'
table.align['Sample'] = 'l'
ntupleDir = getAnalysisNtupleDirectory(args.analysis) if args.verbose and args.analysis else getNtupleDirectory(version=args.version)
for sample in sorted(glob.glob(os.path.join(ntupleDir,'*'))):
name = os.path.basename(sample)
logging.info('Processing {0}'.format(name))
data = isData(name)
xsec = getXsec(name)
if args.verbose and args.analysis:
fnames = get_hdfs_root_files(sample)
# get total events, total weights
tree = ROOT.TChain(getTreeName(args.analysis))
summedWeights = 0.
for f in fnames:
tfile = ROOT.TFile.Open('/hdfs'+f)
summedWeights += tfile.Get("summedWeights").GetBinContent(1)
tfile.Close()
tree.Add('/hdfs'+f)
numEntries = tree.GetEntries(args.selection)
weightedEntries = 0.
negevents = 0.
seltree = tree.CopyTree(args.selection)
for row in seltree:
if data:
weightedEntries += 1.
else:
weightedEntries += row.genWeight
if row.genWeight<0.: negevents += 1
if data:
sampleLumi = getLumi()
else:
sampleLumi = float(summedWeights)/xsec if xsec else 0.
negratio = float(negevents)/numEntries if numEntries else 0.
effevents = weightedEntries*getLumi()/sampleLumi if sampleLumi else 0.
table.add_row([name,'{0:.6f}'.format(float(xsec)),numEntries,'{0:.3}'.format(float(negratio)),'{0:.6f}'.format(float(sampleLumi)),'{0:.3f}'.format(float(effevents))])
else:
table.add_row([name,xsec])
print table.get_string()
def __initializeNtuple(self):
tchain = ROOT.TChain(self.treeName)
if self.inputFileList: # reading from a passed list of inputfiles
allFiles = []
with open(self.inputFileList, 'r') as f:
for line in f.readlines():
allFiles += [line.strip()]
else: # reading from an input directory (all files in directory will be processed)
allFiles = glob.glob('{0}/*.root'.format(self.ntupleDirectory))
#elif os.path.isfile(self.ntuple): # reading a single root file
# allFiles = [self.ntuple]
if len(allFiles) == 0:
logging.error('No files found for sample {0}'.format(self.sample))
summedWeights = 0.
for f in allFiles:
tfile = ROOT.TFile.Open(f)
summedWeights += tfile.Get("summedWeights").GetBinContent(1)
tfile.Close()
tchain.Add(f)
if not summedWeights and not isData(self.sample):
logging.warning('No events for sample {0}'.format(self.sample))
self.intLumi = float(getLumi())
self.xsec = getXsec(self.sample)
if not self.xsec:
logging.error('No xsec for sample {0}'.format(self.sample))
self.sampleLumi = float(summedWeights) / self.xsec if self.xsec else 0.
self.sampleTree = tchain
self.j += 1
#listname = 'selList{0}'.format(self.j)
#self.sampleTree.Draw('>>{0}'.format(listname),'1','entrylist')
#skim = ROOT.gDirectory.Get(listname)
#self.entryListMap['1'] = skim
self.files = allFiles
self.initialized = True
if not self.temp: self.fileHash = hashFile(*self.files)
if self.useProof: self.sampleTree.SetProof()
logging.debug(
'Initialized {0}: summedWeights = {1}; xsec = {2}; sampleLumi = {3}; intLumi = {4}'
.format(self.sample, summedWeights, self.xsec, self.sampleLumi,
self.intLumi))
def __init__(self,**kwargs):
inputTreeName = kwargs.pop('inputTreeName','WZTree')
super(WZTrainer,self).__init__(**kwargs)
sampleDir = 'ntuples/WZ'
sampleMap = {
"dy10" : "DYJetsToLL_M-10to50_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"dy50" : "DYJetsToLL_M-50_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"ggzz2e2m" : "GluGluToContinToZZTo2e2mu_13TeV_MCFM701_pythia8",
"ggzz2e2t" : "GluGluToContinToZZTo2e2tau_13TeV_MCFM701_pythia8",
"ggzz2m2t" : "GluGluToContinToZZTo2mu2tau_13TeV_MCFM701_pythia8",
"ggzz4e" : "GluGluToContinToZZTo4e_13TeV_MCFM701_pythia8",
"ggzz4m" : "GluGluToContinToZZTo4mu_13TeV_MCFM701_pythia8",
"ggzz4t" : "GluGluToContinToZZTo4tau_13TeV_MCFM701_pythia8",
"tt" : "TTJets_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"ttw" : "TTWJetsToLNu_TuneCUETP8M1_13TeV-amcatnloFXFX-madspin-pythia8",
"w" : "WJetsToLNu_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"ww" : "WWTo2L2Nu_13TeV-powheg",
"wz3lnu" : "WZTo3LNu_TuneCUETP8M1_13TeV-powheg-pythia8",
"wz2l2q" : "WZTo2L2Q_13TeV_amcatnloFXFX_madspin_pythia8",
"wzz" : "WZZ_TuneCUETP8M1_13TeV-amcatnlo-pythia8",
"zg" : "ZGTo2LG_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"zz2l2n" : "ZZTo2L2Nu_13TeV_powheg_pythia8",
"zz2l2q" : "ZZTo2L2Q_13TeV_amcatnloFXFX_madspin_pythia8",
"zz4l" : "ZZTo4L_13TeV_powheg_pythia8",
#"tzq" : "tZq_ll_4f_13TeV-amcatnlo-pythia8_TuneCUETP8M1",
}
# get the trees
intLumis = {}
for s in sampleMap:
summedWeights = 0.
for f in glob.glob('{0}/{1}/*.root'.format(sampleDir, sampleMap[s])):
tfile = ROOT.TFile.Open(f)
hist = tfile.Get('summedWeights')
summedWeights += hist.GetBinContent(1)
tfile.Close()
intLumis[s] = float(summedWeights)/getXsec(sampleMap[s])
sigTrees = {}
for sig in ['wz3lnu']:
sigTrees[sig] = ROOT.TChain('WZTree')
for f in glob.glob('{0}/{1}/*.root'.format(sampleDir,sampleMap[sig])):
sigTrees[sig].Add(f)
bgTrees = {}
for bg in ['dy10','dy50','ggzz2e2m','ggzz2m2t','ggzz4e','ggzz4m','ggzz4t','tt','ttw','wzz','zg','zz2l2n','zz2l2q','zz4l']:
bgTrees[bg] = ROOT.TChain('WZTree')
for f in glob.glob('{0}/{1}/*.root'.format(sampleDir,sampleMap[bg])):
bgTrees[bg].Add(f)
intLumi = getLumi()
# add to factory
for sig in sigTrees:
self.factory.AddSignalTree(sigTrees[sig],intLumi/intLumis[sig])
for bg in bgTrees:
self.factory.AddBackgroundTree(bgTrees[bg],intLumi/intLumis[bg])
# per event weight
weight = 'genWeight'
self.factory.SetWeightExpression(weight)
# variables
self.factory.AddVariable('z1_pt','F')
self.factory.AddVariable('z2_pt','F')
self.factory.AddVariable('w1_pt','F')
self.factory.AddVariable('z_mass','F')
self.factory.AddVariable('met_pt','F')
self.factory.AddVariable('numBjetsTight30','I')
# preselection cut
passCut = ROOT.TCut('z1_passMedium==1 && z2_passMedium==1 && w1_passTight==1')
self.factory.PrepareTrainingAndTestTree(
passCut,
":".join(
[
"nTrain_Signal=0",
"nTrain_Background=0",
"SplitMode=Random",
"NormMode=NumEvents",
"!V"
]
)
)
# options:
# H : display help
# V : turn on verbosity
# IgnoreNegWeightsInTraining : ignore events with negative weights for training, keep for testing
# book method
method = self.factory.BookMethod(
ROOT.TMVA.Types.kBDT,
"BDT",
":".join(
[
"NTrees=850",
"MaxDepth=3",
"BoostType=AdaBoost",
"AdaBoostBeta=0.5",
"SeparationType=GiniIndex",
"nCuts=20",
"PruneMethod=NoPruning",
]
)
)
def __init__(self, **kwargs):
inputTreeName = kwargs.pop('inputTreeName', 'WZTree')
super(WZTrainer, self).__init__(**kwargs)
sampleDir = 'ntuples/WZ'
sampleMap = {
"dy10":
"DYJetsToLL_M-10to50_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"dy50": "DYJetsToLL_M-50_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"ggzz2e2m": "GluGluToContinToZZTo2e2mu_13TeV_MCFM701_pythia8",
"ggzz2e2t": "GluGluToContinToZZTo2e2tau_13TeV_MCFM701_pythia8",
"ggzz2m2t": "GluGluToContinToZZTo2mu2tau_13TeV_MCFM701_pythia8",
"ggzz4e": "GluGluToContinToZZTo4e_13TeV_MCFM701_pythia8",
"ggzz4m": "GluGluToContinToZZTo4mu_13TeV_MCFM701_pythia8",
"ggzz4t": "GluGluToContinToZZTo4tau_13TeV_MCFM701_pythia8",
"tt": "TTJets_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"ttw":
"TTWJetsToLNu_TuneCUETP8M1_13TeV-amcatnloFXFX-madspin-pythia8",
"w": "WJetsToLNu_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"ww": "WWTo2L2Nu_13TeV-powheg",
"wz3lnu": "WZTo3LNu_TuneCUETP8M1_13TeV-powheg-pythia8",
"wz2l2q": "WZTo2L2Q_13TeV_amcatnloFXFX_madspin_pythia8",
"wzz": "WZZ_TuneCUETP8M1_13TeV-amcatnlo-pythia8",
"zg": "ZGTo2LG_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8",
"zz2l2n": "ZZTo2L2Nu_13TeV_powheg_pythia8",
"zz2l2q": "ZZTo2L2Q_13TeV_amcatnloFXFX_madspin_pythia8",
"zz4l": "ZZTo4L_13TeV_powheg_pythia8",
#"tzq" : "tZq_ll_4f_13TeV-amcatnlo-pythia8_TuneCUETP8M1",
}
# get the trees
intLumis = {}
for s in sampleMap:
summedWeights = 0.
for f in glob.glob('{0}/{1}/*.root'.format(sampleDir,
sampleMap[s])):
tfile = ROOT.TFile.Open(f)
hist = tfile.Get('summedWeights')
summedWeights += hist.GetBinContent(1)
tfile.Close()
intLumis[s] = float(summedWeights) / getXsec(sampleMap[s])
sigTrees = {}
for sig in ['wz3lnu']:
sigTrees[sig] = ROOT.TChain('WZTree')
for f in glob.glob('{0}/{1}/*.root'.format(sampleDir,
sampleMap[sig])):
sigTrees[sig].Add(f)
bgTrees = {}
for bg in [
'dy10', 'dy50', 'ggzz2e2m', 'ggzz2m2t', 'ggzz4e', 'ggzz4m',
'ggzz4t', 'tt', 'ttw', 'wzz', 'zg', 'zz2l2n', 'zz2l2q', 'zz4l'
]:
bgTrees[bg] = ROOT.TChain('WZTree')
for f in glob.glob('{0}/{1}/*.root'.format(sampleDir,
sampleMap[bg])):
bgTrees[bg].Add(f)
intLumi = getLumi()
# add to factory
for sig in sigTrees:
self.factory.AddSignalTree(sigTrees[sig], intLumi / intLumis[sig])
for bg in bgTrees:
self.factory.AddBackgroundTree(bgTrees[bg], intLumi / intLumis[bg])
# per event weight
weight = 'genWeight'
self.factory.SetWeightExpression(weight)
# variables
self.factory.AddVariable('z1_pt', 'F')
self.factory.AddVariable('z2_pt', 'F')
self.factory.AddVariable('w1_pt', 'F')
self.factory.AddVariable('z_mass', 'F')
self.factory.AddVariable('met_pt', 'F')
self.factory.AddVariable('numBjetsTight30', 'I')
# preselection cut
passCut = ROOT.TCut(
'z1_passMedium==1 && z2_passMedium==1 && w1_passTight==1')
self.factory.PrepareTrainingAndTestTree(
passCut, ":".join([
"nTrain_Signal=0", "nTrain_Background=0", "SplitMode=Random",
"NormMode=NumEvents", "!V"
]))
# options:
# H : display help
# V : turn on verbosity
# IgnoreNegWeightsInTraining : ignore events with negative weights for training, keep for testing
# book method
method = self.factory.BookMethod(
ROOT.TMVA.Types.kBDT, "BDT", ":".join([
"NTrees=850",
"MaxDepth=3",
"BoostType=AdaBoost",
"AdaBoostBeta=0.5",
"SeparationType=GiniIndex",
"nCuts=20",
"PruneMethod=NoPruning",
]))
