def __init__(self,tree,name,weight_file,var_file,cutval):
""" A class to handle the decision of the BDT"""
TMVA.Tools.Instance()
self._reader = TMVA.Reader()
self._tree = tree
self._variables = {}
self._cutvalue = -1
self._bdtscore = -9999
self._name = name
self._weight_file = weight_file
self._var_file = var_file
self.SetReader(self._name,self._weight_file,self._var_file)
self.SetCutValue(cutval)
def __init__(self,
name,
weight_file,
variables,
cutval,
training='training'):
""" A class to handle the decision of the BDT"""
TMVA.Tools.Instance()
self._reader = TMVA.Reader()
self._vars = variables
self._vals = [array('f', [0.]) for i in range(0, len(variables))]
self._cutval = cutval
self._score = -9999
self._name = name
self._training = training
def __init__(self, **kwargs):
"""
This initialization function sets up the TMVA macro path, and
sets up variables for the TMVA reader. These variables are defined as
single entry Python arrays. One has to use these arrays as TMVA/ROOT works
by reference. This means that adding variables to the TMVA reader amounts
to adding pointers to the TMVA reader. We have to then iterate
(not very numpy) through our variable arrays, setting the reader variable arrays (and evaluating as we go).
kwargs:
- var_names: the names of the variables used in the BDT application
(default 10 variables)
- tmva_dir: the installation location of TMVA macros (/home/dean/software/tmva/TMVA-v4.2.0/test)
"""
var_names = kwargs.get('var_names', [
'peaks', 'mean_peaks', 'integral', 'integral_over_peaks', 'max',
'mean', 'max_over_mean', 'std_dev_peaks', 'entropy', 'ps_integral'
])
tmvadir = kwargs.get("tmva_dir",
"/home/dean/software/tmva/TMVA-v4.2.0/test")
#setting up TMVA macro path
try:
macro = os.path.join(tmvadir, "TMVAlogon.C")
loadmacro = os.path.join(tmvadir, "TMVAGui.C")
gROOT.SetMacroPath(tmvadir)
gROOT.Macro(macro)
gROOT.LoadMacro(loadmacro)
except:
print("Couldn't successfully establish TMVA macro path")
sys.exit()
self.reader = TMVA.Reader("!Color")
variables = []
for i in xrange(len(var_names)):
variables.append(array('f', [0]))
for name, var in zip(var_names, variables):
self.reader.AddVariable(name, var)
self.variables = variables
class OutputTree(TreeModel):
classID = FloatCol(default=-1111.)
var1 = FloatCol(default=-1111.)
var2 = FloatCol(default=-1111.)
var3 = FloatCol(default=-1111.)
weight = FloatCol(default=-1111.)
cls0 = FloatCol(default=-1111.)
cls1 = FloatCol(default=-1111.)
cls2 = FloatCol(default=-1111.)
outfile = root_open('tmva_example_multiple_backgrounds__applied.root',
'recreate')
output_tree = Tree('multiBkg', model=OutputTree)
r1 = TMVA.Reader("!Color:!Silent")
r2 = TMVA.Reader("!Color:!Silent")
# r3 = TMVA.Reader( "!Color:!Silent" )
r1.AddVariable('var1', output_tree.var1)
r1.AddVariable('var2', output_tree.var2)
r1.AddVariable('var3', output_tree.var3)
r2.AddVariable('var1', output_tree.var1)
r2.AddVariable('var2', output_tree.var2)
r2.AddVariable('var3', output_tree.var3)
# r3.AddVariable('var1', output_tree.var1)
# r3.AddVariable('var2', output_tree.var2)
# r3.AddVariable('var3', output_tree.var3)
Muon3_innerTrack_numberOfLostTrackerInnerHits = array('f', [0.])
Muon3_numberOfMatchedStations = array('f', [0.])
Muon3_calEnergy_had = array('f', [0.])
Muon3_innerTrack_numberofValidHits = array('f', [0.])
Muon3_calEnergy_em = array('f', [0.])
Muon3_innerTrack_numberOfLostTrackerOuterHits = array('f', [0.])
Muon3_innerTrack_normalizedChi2 = array('f', [0.])
Muon3_innerTrack_validFraction = array('f', [0.])
Muon3_innerTrack_pixelLayersWithMeasurement = array('f', [0.])
Muon3_ptErrOverPt = array('f', [0.])
fake = array('f', [0.])
MuonId = array('f', [-99.0])
readerMuonId = TMVA.Reader("!Color:!Silent")
readerMuonId.AddVariable("Muon_caloCompatibility", Muon3_caloCompatibility)
readerMuonId.AddVariable("Muon_segmentCompatibility",
Muon3_segmentCompatibility)
readerMuonId.AddVariable("Muon_innerTrack_numberOfLostTrackerHits",
Muon3_innerTrack_numberOfLostTrackerHits)
readerMuonId.AddVariable("Muon_innerTrack_numberOfLostTrackerInnerHits",
Muon3_innerTrack_numberOfLostTrackerInnerHits)
readerMuonId.AddVariable("Muon_numberOfMatchedStations",
Muon3_numberOfMatchedStations)
readerMuonId.AddVariable("Muon_calEnergy_had", Muon3_calEnergy_had)
readerMuonId.AddVariable("Muon_innerTrack_numberofValidHits",
Muon3_innerTrack_numberofValidHits)
readerMuonId.AddVariable("Muon_calEnergy_em", Muon3_calEnergy_em)
readerMuonId.AddVariable("Muon_innerTrack_numberOfLostTrackerOuterHits",
Muon3_innerTrack_numberOfLostTrackerOuterHits)
## This tutorial shows how to apply a trained model to new data (regression).
##
## \macro_code
##
## \date 2017
## \author TMVA Team
from ROOT import TMVA, TFile, TString
from array import array
from subprocess import call
from os.path import isfile
# Setup TMVA
TMVA.Tools.Instance()
TMVA.PyMethodBase.PyInitialize()
reader = TMVA.Reader("Color:!Silent")
# Load data
if not isfile('tmva_reg_example.root'):
call([
'curl', '-L', '-O', 'http://root.cern.ch/files/tmva_reg_example.root'
])
data = TFile.Open('tmva_reg_example.root')
tree = data.Get('TreeR')
branches = {}
for branch in tree.GetListOfBranches():
branchName = branch.GetName()
branches[branchName] = array('f', [-999])
tree.SetBranchAddress(branchName, branches[branchName])
from root_numpy.tmva import add_classification_events, evaluate_reader
from root_numpy import ROOT_VERSION
from ROOT import TMVA, TFile, TCut
reader = TMVA.Reader()
def arr():
return array('f', [0.])
reader2j.AddVariable("hmmpt", arr())
reader2j.AddVariable("hmmrap", arr())
reader2j.AddVariable("hmmthetacs", arr())
reader2j.AddVariable("hmmphics", arr())
reader2j.AddVariable("j1pt", arr())
reader2j.AddVariable("j1eta", arr())
reader2j.AddVariable("j2pt", arr())
reader2j.AddVariable("detajj", arr())
reader2j.AddVariable("dphijj", arr())
reader2j.AddVariable("mjj", arr())
reader2j.AddVariable("met", arr())
reader2j.AddVariable("zepen", arr())
reader2j.AddVariable("njets", arr())
reader2j.AddVariable("drmj", arr())
reader2j.AddVariable("m1ptOverMass", arr())
reader2j.AddVariable("m2ptOverMass", arr())
reader2j.AddVariable("m1eta", arr())
reader2j.AddVariable("m2eta", arr())
reader2j.AddSpectator("hmerr", arr())
reader2j.AddSpectator("weight", arr())
def main():
try:
# Retrive command line options
shortopts = "m:i:t:o:vh?"
longopts = [
"methods=", "inputfile=", "inputtrees=", "outputfile=", "verbose",
"help", "usage"
]
opts, args = getopt.getopt(sys.argv[1:], shortopts, longopts)
except getopt.GetoptError:
# Print help information and exit:
print "ERROR: unknown options in argument %s" % sys.argv[1:]
usage()
sys.exit(1)
infname = DEFAULT_INFNAME
treeNameSig = DEFAULT_TREESIG
treeNameBkg = DEFAULT_TREEBKG
outfname = DEFAULT_OUTFNAME
methods = DEFAULT_METHODS
verbose = False
for o, a in opts:
if o in ("-?", "-h", "--help", "--usage"):
usage()
sys.exit(0)
elif o in ("-m", "--methods"):
methods = a
elif o in ("-i", "--inputfile"):
infname = a
elif o in ("-o", "--outputfile"):
outfname = a
elif o in ("-t", "--inputtrees"):
a.strip()
trees = a.rsplit(' ')
trees.sort()
trees.reverse()
if len(trees) - trees.count('') != 2:
print "ERROR: need to give two trees (each one for signal and background)"
print trees
sys.exit(1)
treeNameSig = trees[0]
treeNameBkg = trees[1]
elif o in ("-v", "--verbose"):
verbose = True
# Print methods
mlist = methods.replace(' ', ',').split(',')
print "=== TMVApplication: use method(s)..."
for m in mlist:
if m.strip() != '':
print "=== - <%s>" % m.strip()
# Import ROOT classes
from ROOT import gSystem, gROOT, gApplication, TFile, TTree, TCut, TH1F, TStopwatch
# check ROOT version, give alarm if 5.18
if gROOT.GetVersionCode() >= 332288 and gROOT.GetVersionCode() < 332544:
print "*** You are running ROOT version 5.18, which has problems in PyROOT such that TMVA"
print "*** does not run properly (function calls with enums in the argument are ignored)."
print "*** Solution: either use CINT or a C++ compiled version (see TMVA/macros or TMVA/examples),"
print "*** or use another ROOT version (e.g., ROOT 5.19)."
sys.exit(1)
# Logon not automatically loaded through PyROOT (logon loads TMVA library) load also GUI
gROOT.SetMacroPath("../macros/")
gROOT.Macro("../macros/TMVAlogon.C")
# Import TMVA classes from ROOT
from ROOT import TMVA
# Create the Reader object
reader = TMVA.Reader("!Color")
# Create a set of variables and declare them to the reader
# - the variable names must corresponds in name and type to
# those given in the weight file(s) that you use
# what to do ???
var1 = array('f', [0])
var2 = array('f', [0])
var3 = array('f', [0])
var4 = array('f', [0])
reader.AddVariable("var1+var2", var1)
reader.AddVariable("var1-var2", var2)
reader.AddVariable("var3", var3)
reader.AddVariable("var4", var4)
# book the MVA methods
dir = "weights/"
prefix = "TMVAnalysis_"
for m in mlist:
reader.BookMVA(m + " method", dir + prefix + m + ".weights.txt")
#######################################################################
# For an example how to apply your own plugin method, please see
# TMVA/macros/TMVApplication.C
#######################################################################
# Book output histograms
nbin = 80
histList = []
for m in mlist:
histList.append(TH1F(m, m, nbin, -3, 3))
# Book example histogram for probability (the other methods would be done similarly)
if "Fisher" in mlist:
probHistFi = TH1F("PROBA_MVA_Fisher", "PROBA_MVA_Fisher", nbin, 0, 1)
rarityHistFi = TH1F("RARITY_MVA_Fisher", "RARITY_MVA_Fisher", nbin, 0,
1)
# Prepare input tree (this must be replaced by your data source)
# in this example, there is a toy tree with signal and one with background events
# we'll later on use only the "signal" events for the test in this example.
#
fname = "./tmva_example.root"
print "--- Accessing data file: %s" % fname
input = TFile.Open(fname)
if not input:
print "ERROR: could not open data file: %s" % fname
sys.exit(1)
#
# Prepare the analysis tree
# - here the variable names have to corresponds to your tree
# - you can use the same variables as above which is slightly faster,
# but of course you can use different ones and copy the values inside the event loop
#
print "--- Select signal sample"
theTree = input.Get("TreeS")
userVar1 = array('f', [0])
userVar2 = array('f', [0])
theTree.SetBranchAddress("var1", userVar1)
theTree.SetBranchAddress("var2", userVar2)
theTree.SetBranchAddress("var3", var3)
theTree.SetBranchAddress("var4", var4)
# Efficiency calculator for cut method
nSelCuts = 0
effS = 0.7
# Process the events
print "--- Processing: %i events" % theTree.GetEntries()
sw = TStopwatch()
sw.Start()
for ievt in range(theTree.GetEntries()):
if ievt % 1000 == 0:
print "--- ... Processing event: %i" % ievt
# Fill event in memory
theTree.GetEntry(ievt)
# Compute MVA input variables
var1[0] = userVar1[0] + userVar2[0]
var2[0] = userVar1[0] - userVar2[0]
# Return the MVAs and fill to histograms
if "CutsGA" in mlist:
passed = reader.EvaluateMVA("CutsGA method", effS)
if passed:
nSelCuts = nSelCuts + 1
# Fill histograms with MVA outputs
for h in histList:
h.Fill(reader.EvaluateMVA(h.GetName() + " method"))
# Retrieve probability instead of MVA output
if "Fisher" in mlist:
probHistFi.Fill(reader.GetProba("Fisher method"))
rarityHistFi.Fill(reader.GetRarity("Fisher method"))
# Get elapsed time
sw.Stop()
print "--- End of event loop: %s" % sw.Print()
# Return computed efficeincies
if "CutsGA" in mlist:
eff = float(nSelCuts) / theTree.GetEntries()
deff = math.sqrt(eff * (1.0 - eff) / theTree.GetEntries())
print "--- Signal efficiency for Cuts method : %.5g +- %.5g (required was: %.5g)" % (
eff, deff, effS)
# Test: retrieve cuts for particular signal efficiency
mcuts = reader.FindMVA("CutsGA method")
cutsMin = array('d', [0, 0, 0, 0])
cutsMax = array('d', [0, 0, 0, 0])
mcuts.GetCuts(0.7, cutsMin, cutsMax)
print "--- -------------------------------------------------------------"
print "--- Retrieve cut values for signal efficiency of 0.7 from Reader"
for ivar in range(4):
print "... Cut: %.5g < %s <= %.5g" % (
cutsMin[ivar], reader.GetVarName(ivar), cutsMax[ivar])
print "--- -------------------------------------------------------------"
#
# write histograms
#
target = TFile("TMVApp.root", "RECREATE")
for h in histList:
h.Write()
# Write also probability hists
if "Fisher" in mlist:
probHistFi.Write()
rarityHistFi.Write()
target.Close()
print "--- Created root file: \"TMVApp.root\" containing the MVA output histograms"
print "==> TMVApplication is done!"
# In[11]:
#path = "/eos/uscms/store/user/corderom/Corrections2016Rereco/"
#phVals = ["EE","EB"]
phVals = ["EE"]
path = ""
file = {}
reader = {}
for ph in phVals:
file[ph] = "TMVAnalysis_BDT_"+ph+".weights.xml"
reader[ph] = TMVA.Reader()
# In[12]:
varName= {
"recoPhi" :["photonOnePhi"],
"r9" :["photonOneR9"],
"sieieFull5x5" :["photonOneSieie"],
"sieipFull5x5" :["photonOneSieip"],
"e2x2Full5x5/e5x5Full5x5":["photonOneE2x2","photonOneE5x5"],
"recoSCEta" :["photonOneEta"],
"rawE" :["photonOneScRawE"],
"scEtaWidth" :["photonOneScEtaWidth"],
"scPhiWidth" :["photonOneScPhiWidth"],
'var_Iso08MuMin', 'var_MaxMuonImpactAngle', 'var_MaxtrkKink',
'var_MuonglbkinkSum', 'var_MinMatchedStations', 'var_MaxdeltaMuZ'
]
varlist_help = ['var_mass12', 'var_mass13']
print "Initializing TMVA..."
# Setup TMVA
TMVA.Tools.Instance()
mvaReaderList = ['readerA', 'readerB', 'readerC', 'readerBC']
mvaReaders = {}
mvaWeights = {}
print "Setting up reader..."
for reader in mvaReaderList:
mvaReaders[reader] = TMVA.Reader('Color:!Silent')
# tmpstr = 'MyTMVAClassification_Test_vars_'+fileStr(reader)
tmpstr = args.weights_prefix + '_vars_' + fileStr(reader)
#--weights-prefix
mvaWeights[reader] = TString(
BASEDIR + '/' + tmpstr + '_BDT.weights.xml'
) # weights weights.xml file after training, place it to CommonFiles
#Vars2016_vars_
# Open input file and set branch addresses
rootfile = TFile(filename, 'READ')
rootfile.ls()
tree_bkg = rootfile.Get('TreeB')
tree_ds = rootfile.Get('TreeS_Ds')
def main():
usage = 'usage: %prog [options]'
parser = optparse.OptionParser(usage)
parser.add_option(
'-s',
'--suffix',
dest='input_suffix',
help='suffix used to identify inputs from network training',
default=None,
type='string')
(opt, args) = parser.parse_args()
'''if opt.json == None:
print 'input variable .json not defined!'
sys.exit(1)'''
if opt.input_suffix == None:
print 'Input files suffix not defined!'
sys.exit(1)
classifier_suffix = opt.input_suffix
classifier_parent_dir = 'BinaryClassifier_BDTG_%s' % (classifier_suffix)
# Setup TMVA
TMVA.Tools.Instance()
TMVA.PyMethodBase.PyInitialize()
reader = TMVA.Reader("Color:!Silent")
#ttH_filename = 'samples/2017MC/training_region/TTHnobb.root'
ttH_filename = 'samples/DiLepTR_ttH_bInclude.root'
data_ttH = TFile.Open(ttH_filename)
#data_ttH_tree = data_ttH.Get('syncTree')
data_ttH_tree = data_ttH.Get('BOOM')
print 'ttH file to evaluate: ', ttH_filename
if classifier_suffix == "ttHvsttJets":
#bckg_filename = 'samples/2017MC/testing_region/TTJets.root'
#bckg_filename = 'samples/2017MC/training_region/TTJets.root'
bckg_filename = 'samples/DiLepTR_ttJets_bInclude.root'
data_bckg = TFile.Open(bckg_filename)
#data_bckg_tree = data_bckg.Get('syncTree')
data_bckg_tree = data_bckg.Get('BOOM')
elif classifier_suffix == "ttHvsttV":
#bckg_filename = 'samples/2017MC/testing_region/ttV.root'
#bckg_filename = 'samples/2017MC/training_region/ttV.root'
bckg_filename = 'samples/DiLepTR_ttV_bInclude.root'
data_bckg = TFile.Open(bckg_filename)
#data_bckg_tree = data_bckg.Get('syncTree')
data_bckg_tree = data_bckg.Get('BOOM')
print 'ttJets file to evaluate: ', bckg_filename
keys_list = []
if classifier_suffix == "ttHvsttJets":
keys_list = [(
"max(abs(LepGood_eta[iLepFO_Recl[0]]),abs(LepGood_eta[iLepFO_Recl[1]]))",
"maxeta"), ("nJet25_Recl", "Jet_numLoose"),
("mindr_lep1_jet", "mindrlep1jet"),
("mindr_lep2_jet", "mindrlep2jet"),
("MT_met_lep1", "SR_InvarMassT"),
("max(-1.1,BDTv8_eventReco_mvaValue)", "hadTop_BDT")]
elif classifier_suffix == "ttHvsttV":
keys_list = [(
"max(abs(LepGood_eta[iLepFO_Recl[0]]),abs(LepGood_eta[iLepFO_Recl[1]]))",
"maxeta"), ("nJet25_Recl", "Jet_numLoose"),
("mindr_lep1_jet", "mindrlep1jet"),
("mindr_lep2_jet", "mindrlep2jet"),
("MT_met_lep1", "SR_InvarMassT"),
("LepGood_conePt[iLepFO_Recl[1]]", "corrptlep1"),
("LepGood_conePt[iLepFO_Recl[0]]", "corrptlep2"),
("max(-1.1,BDTv8_eventReco_Hj_score)", "Hj1_BDT")]
branches_ttree = {}
for reader_key, ttree_key in keys_list:
print 'Setting Branch address: ', ttree_key
if ttree_key == 'Jet_numLoose':
#branches_ttree[ttree_key] = array('I', [999])
branches_ttree[ttree_key] = array('d', [999])
else:
#branches_ttree[ttree_key] = array('f', [-999])
branches_ttree[ttree_key] = array('d', [-999])
if 'hadTop_BDT' in ttree_key:
keyname = 'hadTop_BDT'
data_ttH_tree.SetBranchAddress(str(keyname),
branches_ttree[ttree_key])
data_bckg_tree.SetBranchAddress(str(keyname),
branches_ttree[ttree_key])
elif 'Hj1_BDT' in ttree_key:
keyname = 'Hj1_BDT'
data_ttH_tree.SetBranchAddress(str(keyname),
branches_ttree[ttree_key])
data_bckg_tree.SetBranchAddress(str(keyname),
branches_ttree[ttree_key])
else:
data_ttH_tree.SetBranchAddress(str(ttree_key),
branches_ttree[ttree_key])
data_bckg_tree.SetBranchAddress(str(ttree_key),
branches_ttree[ttree_key])
# Keep track of event numbers for cross-checks.
event_number_branch = array('L', [999])
data_ttH_tree.SetBranchAddress('nEvent', event_number_branch)
branches_reader = {}
# Register names of inputs with reader. Together with the name give the address of the local variable that carries the updated input variables during event loop.
for reader_key, ttree_key in keys_list:
print 'Add variable name %s: ' % reader_key
branches_reader[reader_key] = array('f', [-999])
reader.AddVariable(str(reader_key), branches_reader[reader_key])
iLepFO_Recl0 = array('f', [-999])
reader.AddSpectator('iLepFO_Recl[0]', iLepFO_Recl0)
iLepFO_Recl1 = array('f', [-999])
reader.AddSpectator('iLepFO_Recl[1]', iLepFO_Recl1)
iLepFO_Recl2 = array('f', [-999])
reader.AddSpectator('iLepFO_Recl[2]', iLepFO_Recl2)
# Book methods
# First argument is user defined name. Doesn not have to be same as training name.
# True type of method and full configuration are read from the weights file specified in the second argument.
if classifier_suffix == "ttHvsttJets":
#mva_weights_dir = 'BinaryClassifier_BDTG_%s/weights/2lss_ttbar_withBDTv8_BDTG.weights.xml' % classifier_suffix
mva_weights_dir = '../../../FromMarco/2lss_ttbar_withBDTv8_BDTG.weights.xml'
elif classifier_suffix == "ttHvsttV":
#mva_weights_dir = 'BinaryClassifier_BDTG_%s/weights/2lss_ttV_withHj_BDTG.weights.xml' % classifier_suffix
mva_weights_dir = '../../../FromMarco/2lss_ttV_withHj_BDTG.weights.xml'
print 'using weights file: ', mva_weights_dir
reader.BookMVA('BDTG', TString(mva_weights_dir))
classifier_samples_dir = classifier_parent_dir + "/outputs"
classifier_plots_dir = classifier_parent_dir + "/plots"
if not os.path.exists(classifier_plots_dir):
os.makedirs(classifier_plots_dir)
if not os.path.exists(classifier_samples_dir):
os.makedirs(classifier_samples_dir)
# Define outputs: files to store histograms/ttree with results from application of classifiers and any histos/trees themselves.
output_file_name = '%s/Applied_%s.root' % (classifier_samples_dir,
classifier_parent_dir)
output_file = TFile.Open(output_file_name, 'RECREATE')
network_evaluation(data_ttH_tree, keys_list, 'ttH', branches_ttree,
branches_reader, reader)
if classifier_suffix == 'ttHvsttJets':
network_evaluation(data_bckg_tree, keys_list, 'ttjets', branches_ttree,
branches_reader, reader)
if classifier_suffix == 'ttHvsttV':
network_evaluation(data_bckg_tree, keys_list, 'ttV', branches_ttree,
branches_reader, reader)
output_file.Close()
foutname += str(emax) + '.root'
fout = TFile(foutname, 'recreate')
if sample == 'ttbar': is_ttbar = 1
else: is_ttbar = 0
# print 'emin and max', emin, emax
if emax == 0:
print 'emax == 0'
print "Using top mass constraint"
# if sample != 'data': b1 = Particle('b1',tkin,1)
# else: b1 = Particle('b1',tkin)
# if sample != 'data': b2 = Particle('b2',tkin,1)
# else: b2 = Particle('b2',tkin)
reader = TMVA.Reader("!Color:!Silent")
readerMisc = TMVA.Reader("!Color:!Silent")
readerQGL = TMVA.Reader("!Color:!Silent")
mytree = ManageTTree(tvars, tkin)
#usevar = ['top1_m','top2_m','w1_m','w2_m','deltaRl1l2','deltaRq1q2','deltaRb1b2','deltaRb1w1','deltaRb2w2','deltaPhiw1w2','deltaPhit1t2','q1b1_mass','p1b2_mass','deltaRb1w2','deltaRb2w1','mindeltaRb1q','simple_chi2','mindeltaRb2p', 'deltaEtal1l2', 'deltaEtaq1q2','jet_CSV[0]','jet_CSV[1]','jet_CSV[2]','jet_CSV[3]','jet_CSV[4]','jet_CSV[5]']
#usevar = ['top1_m','top2_m','w1_m','w2_m','deltaRl1l2','deltaRq1q2','deltaRb1b2','deltaRb1w1','deltaRb2w2','deltaPhit1t2','q1b1_mass','p1b2_mass','deltaRb1w2','deltaRb2w1','simple_chi2','mindeltaRb2p', 'deltaEtal1l2', 'deltaEtaq1q2','jet_CSV[0]','jet_CSV[1]','jet_CSV[2]','jet_CSV[3]','jet_CSV[4]','jet_CSV[5]']
usevar = [
'top1_m', 'top2_m', 'w1_m', 'w2_m', 'deltaRl1l2', 'deltaRq1q2',
'deltaRb1b2', 'deltaRb1w1', 'deltaRb2w2', 'deltaPhiw1w2',
'deltaPhit1t2', 'q1b1_mass', 'p1b2_mass', 'deltaRb1w2', 'deltaRb2w1',
'mindeltaRb1q', 'mindeltaRb2p', 'deltaEtal1l2', 'deltaEtaq1q2',
'jet_CSV[0]', 'jet_CSV[1]', 'jet_CSV[2]', 'jet_CSV[3]', 'jet_CSV[4]',
file2 = open('../mvavars2.txt')
file3 = open('../mvavars3.txt')
mvavars1 = []
mvavars2 = []
mvavars3 = []
for line in file1:
mvavars1.append(line.strip())
for line in file2:
mvavars2.append(line.strip())
for line in file3:
print line
mvavars3.append(line.strip())
reader3 = TMVA.Reader()
var3_ = []
for i, var3 in enumerate(mvavars1):
var3_.append(array.array('f',[0]))
reader3.AddVariable(var3,var3_[i])
reader3.BookMVA("MLP","weights/TMVAClassification_HZZ_2e2mu_MLP.weights.xml")
#weight weight ele1_pt ele1_eta ele1_phi ele1_charge ele1_trackIso ele1_EcalIso ele1_HcalIso ele1_X ele1_SIP ele2_pt ele2_eta ele2_phi ele2_charge ele2_trackIso ele2_EcalIso ele2_HcalIso ele2_X ele2_SIP ele3_pt ele3_eta ele3_phi ele3_charge ele3_trackIso ele3_EcalIso ele3_HcalIso ele3_X ele3_SIP ele4_pt ele4_eta ele4_phi ele4_charge ele4_trackIso ele4_EcalIso ele4_HcalIso ele4_X ele4_SIP
#worst_iso_X second_worst_iso_X worst_vertex second_worst_vertex mZ mZstar mbestH index channel sample end
def loop(hweights, sig, hs, c1):
#------- sig events ---------
for period, channel in itertools.product(periods, channels):
dijetDEta = array.array('f', [-999])
dijetDPhi = array.array('f', [-999])
llgJJDPhi = array.array('f', [-999])
jPhotonDRMin = array.array('f', [-999])
ptt = array.array('f', [-999])
jetOnePt = array.array('f', [-999])
jetTwoPt = array.array('f', [-999])
kin_bdt = array.array('f', [-999])
vbfPtBalance = array.array('f', [-999])
photonZepp = array.array('f', [-999])
kin_bdt_james = array.array('f', [-999])
vbfWeightsFile = 'trained_bdts/vbf_bdt_combined_ming-yan_current.xml'
vbf_reader = t.Reader("!Color:Silent")
vbf_reader.AddVariable('dijetDEta', dijetDEta)
vbf_reader.AddVariable('dijetDPhi', dijetDPhi)
vbf_reader.AddVariable('llgJJDPhi', llgJJDPhi)
vbf_reader.AddVariable('jPhotonDRMin', jPhotonDRMin)
vbf_reader.AddVariable('ptt', ptt)
vbf_reader.AddVariable('jetOnePt', jetOnePt)
vbf_reader.AddVariable('jetTwoPt', jetTwoPt)
vbf_reader.AddVariable('kin_bdt', kin_bdt)
vbf_reader.AddVariable('vbfPtBalance', vbfPtBalance)
vbf_reader.AddVariable('photonZepp', photonZepp)
vbf_reader.BookMVA('BDT method', vbfWeightsFile)
#vbfWeightsFileJames = 'trained_bdts/vbf_bdt_combined_james_current_half_signal_BDT.weights.xml'
## \macro_code
##
## \author Lailin XU
from ROOT import TMVA, TFile, TTree, TCut, TH1F, TCanvas, gROOT, TLegend
from subprocess import call
from os.path import isfile
from array import array
gROOT.SetStyle("ATLAS")
# Setup TMVA
TMVA.Tools.Instance()
# Reader. One reader for each application.
reader = TMVA.Reader("Color:!Silent")
reader_S = TMVA.Reader("Color:!Silent")
reader_B = TMVA.Reader("Color:!Silent")
# Inputs
# =============
# Load data
# An unknown sample
trfile = "Zp2TeV_ttbar.root"
data = TFile.Open(trfile)
tree = data.Get('tree')
# Known signal
trfile_S = "Zp1TeV_ttbar.root"
data_S = TFile.Open(trfile_S)
tree_S = data_S.Get('tree')
def beginLoop(self, setup):
super(TreeProducer, self).beginLoop(setup)
self.rootfile = TFile('/'.join([self.dirName, 'tree.root']),
'recreate')
self.tree = Tree('events', '')
self.tree.var('weight', float)
self.tree.var('nbjets', float)
bookParticle(self.tree, 'l')
bookMet(self.tree, 'met')
# fatjet stuff
for flavour in ['higgs', 'top']:
bookParticle(self.tree, '{}jet'.format(flavour))
bookParticle(self.tree, 'softDropped_{}jet'.format(flavour))
self.tree.var('{}jet_tau1'.format(flavour), float)
self.tree.var('{}jet_tau2'.format(flavour), float)
self.tree.var('{}jet_tau3'.format(flavour), float)
self.tree.var('{}jet_tau32'.format(flavour), float)
self.tree.var('{}jet_tau31'.format(flavour), float)
self.tree.var('{}jet_tau21'.format(flavour), float)
self.tree.var('{}jet_flow15'.format(flavour), float)
self.tree.var('{}jet_flow25'.format(flavour), float)
self.tree.var('{}jet_flow35'.format(flavour), float)
self.tree.var('{}jet_flow45'.format(flavour), float)
self.tree.var('{}jet_flow55'.format(flavour), float)
self.tree.var('{}jet_nbs'.format(flavour), float)
self.tree.var('{}jet_mbs'.format(flavour), float)
self.tree.var('{}jet_bdt_th'.format(flavour), float)
# for MVA
self.reader = TMVA.Reader()
self.bdt_tau1 = array.array('f', [0])
self.bdt_tau2 = array.array('f', [0])
self.bdt_tau3 = array.array('f', [0])
self.bdt_tau21 = array.array('f', [0])
self.bdt_tau31 = array.array('f', [0])
self.bdt_tau32 = array.array('f', [0])
self.bdt_flow15 = array.array('f', [0])
self.bdt_flow25 = array.array('f', [0])
self.bdt_flow35 = array.array('f', [0])
self.bdt_flow45 = array.array('f', [0])
self.bdt_flow55 = array.array('f', [0])
self.bdt_jet_m = array.array('f', [0])
self.bdt_softDropped_jet_m = array.array('f', [0])
self.bdt_jet_nbs = array.array('f', [0])
self.reader.AddVariable('jet_m', self.bdt_jet_m)
self.reader.AddVariable('softDropped_jet_m',
self.bdt_softDropped_jet_m)
self.reader.AddVariable('jet_tau1', self.bdt_tau1)
self.reader.AddVariable('jet_tau2', self.bdt_tau2)
self.reader.AddVariable('jet_tau3', self.bdt_tau3)
self.reader.AddVariable('jet_tau32', self.bdt_tau32)
self.reader.AddVariable('jet_tau31', self.bdt_tau31)
self.reader.AddVariable('jet_tau21', self.bdt_tau21)
self.reader.AddVariable('jet_flow15', self.bdt_flow15)
self.reader.AddVariable('jet_flow25', self.bdt_flow25)
self.reader.AddVariable('jet_flow35', self.bdt_flow35)
self.reader.AddVariable('jet_flow45', self.bdt_flow45)
self.reader.AddVariable('jet_flow55', self.bdt_flow55)
self.reader.AddVariable('jet_nbs', self.bdt_jet_nbs)
#path = "/afs/cern.ch/work/s/selvaggi/private/FCCSW/heppy/FCChhAnalyses/tth_boosted/"
path = "/eos/experiment/fcc/hh/analyses/Higgs/ttH/BDT/"
self.reader.BookMVA("BDT",
str(path) + "BDT_BDT_Higgs_vs_Top.weights.xml")
from array import array
from tqdm import tqdm
import math
from bdt_common import total_pot, variables, spectators, choose_shower
from bdt_common import is_active, is_fiducial, distance
from bdt_common import ELECTRON_MASS, PROTON_MASS, MAX_N_SHOWERS, MAX_N_TRACKS, N_UNI
from bdt_common import PROTON_THRESHOLD, ELECTRON_THRESHOLD
from proton_energy import length2energy
import time
import statistics
import sys
STORE_SYS = True
reader_reclass = TMVA.Reader(":".join(["!V", "!Silent", "Color"]))
angle = array("f", [0])
pca = array("f", [0])
res = array("f", [0])
open_angle = array("f", [0])
n_hits = array("f", [0])
ratio = array("f", [0])
reader_reclass.AddVariable("angle", angle)
reader_reclass.AddVariable("pca", pca)
reader_reclass.AddVariable("res", res)
reader_reclass.AddVariable("open_angle", open_angle)
reader_reclass.AddVariable("ratio", ratio)
reader_reclass.AddVariable("n_hits", n_hits)
reader_reclass.BookMVA(
zgLittleTheta = array.array('f', [-999])
zgBigTheta = array.array('f', [-999])
llgPtOverM = array.array('f', [-999])
leptonOneEta = array.array('f', [-999])
leptonTwoEta = array.array('f', [-999])
photonEta = array.array('f', [-999])
zgPhi = array.array('f', [-999])
photonMVA = array.array('f', [-999])
corrPhotonMVA = array.array('f', [-999])
photonERes = array.array('f', [-999])
lPhotonDRMin = array.array('f', [-999])
lPhotonDRMax = array.array('f', [-999])
kinWeightsFile = 'trained_bdts/kin_bdt_combined_ming-yan_current.xml'
kin_reader = t.Reader("!Color:Silent")
kin_reader.AddVariable('zgLittleTheta', zgLittleTheta)
kin_reader.AddVariable('zgBigTheta', zgBigTheta)
kin_reader.AddVariable('llgPtOverM', llgPtOverM)
kin_reader.AddVariable('leptonOneEta', leptonOneEta)
kin_reader.AddVariable('leptonTwoEta', leptonTwoEta)
kin_reader.AddVariable('photonEta', photonEta)
kin_reader.AddVariable('zgPhi', zgPhi)
kin_reader.AddVariable('photonMVA', corrPhotonMVA)
kin_reader.AddVariable('photonERes', photonERes)
kin_reader.AddVariable('lPhotonDRMin', lPhotonDRMin)
kin_reader.AddVariable('lPhotonDRMax', lPhotonDRMax)
kin_reader.BookMVA('BDT method', kinWeightsFile)
with open(filename) as f:
lines = f.read().splitlines()
emin = map(int, re.findall(r'\d+', lines[0]))[0]
emax = map(int, re.findall(r'\d+', lines[0]))[1]
foutname = 'Skim_'
foutname += sample + '_'
foutname += str(emax) + '.root'
fout = TFile(foutname, 'recreate')
if sample == 'ttbar': is_ttbar = 1
else: is_ttbar = 0
if emax == 0:
print 'emax == 0'
reader = TMVA.Reader("!Color:!Silent")
readerQCD = TMVA.Reader("!Color:!Silent")
mytree = ManageTTree(tvars, tkin)
usevar = [
'lp1_eta', 'lp2_eta', 'lq1_eta', 'lq2_eta', 'b1_eta', 'b2_eta',
'top1_m', 'top2_m', 'w1_m', 'w2_m', 'deltaRp1p2', 'deltaRq1q2',
'jet_MOverPt[0]', 'jet_MOverPt[1]', 'jet_MOverPt[2]', 'jet_MOverPt[3]',
'jet_MOverPt[4]', 'jet_MOverPt[5]', 'deltaRb1b2', 'deltaRb1w1',
'deltaRb2w2', 'deltaPhiw1w2', 'deltaPhit1t2', 'p1b1_mass', 'q1b2_mass',
'deltaRb1w2', 'deltaRb2w1', 'mindeltaRb1p', 'simple_chi2',
'mindeltaRb2q', 'deltaEtap1p2', 'deltaEtaq1q2', 'jet_CSV[0]',
'jet_CSV[1]', 'jet_CSV[2]', 'jet_CSV[3]', 'jet_CSV[4]', 'jet_CSV[5]'
]
useQCD = [
def tmva_model():
model_path = "dataset_dnn/weights/TMVAClassification_BDT.weights.xml"
print "loading TMVA........."
print "....\n...\n..\n."
reader = tmva.Reader()
#reloading all the features into the loader
#tmva requires that
le = len(y_test)
for col in columns:
reader.AddVariable(col,
np.zeros(le,
dtype='float32')) # accepts 32 bit only.
reader.BookMVA("BDT", model_path)
decision_value = []
prediction = []
for row in x_test:
a = root.std.vector(root.Double)()
for r in row:
a.push_back(r)
value = reader.EvaluateMVA(a, "BDT")
decision_value.append(value)
if value > 0:
prediction.append(1)
else:
prediction.append(0)
# print classification_report(y_test, prediction, target_names=["background", "signal"])
# print "Area under ROC curve: %.4f"%(roc_auc_score(y_test, prediction))
dval_arr = np.array(decision_value)
bg_scores = dval_arr[y_test == 0] # scores corresponding to background
sg_scores = dval_arr[y_test == 1] # scores corresponding to signal
plt.figure("fig-1")
plt.subplot(1, 2, 1)
#background histogram
h2 = plt.hist(bg_scores.ravel(),
color='r',
alpha=0.5,
range=(-0.7, 0.4),
bins=30)
#SIGNAL(tau-tau decay) histogram
h1 = plt.hist(sg_scores.ravel(),
color='b',
ec='b',
alpha=0.5,
range=(-.7, 0.4),
bins=30)
plt.xlabel("TMVA BDT output")
plt.ylabel("Events")
handles = [
Rectangle((0, 1), 1, 1, color='r', ec="k"),
Rectangle((0, 1), 1, 1, color='b', ec="k")
]
plt.legend(handles, ["background", "signal"])
AUC = roc_auc_score(y_test, prediction)
print "Area under ROC curve: %.4f" % (AUC)
plt.subplot(1, 2, 2)
#calculating false and true postive rate
fpr, tpr, thresholds = roc_curve(y_test, dval_arr)
draw_roc_curve(plt, fpr, tpr, AUC)
plt.show()
import ROOT, array
import ROOT, array
from ROOT import TFile, TH1F, TGraph, TCanvas, TLegend, TTree
from ROOT import TMVA, TMath
import sys
print sys.argv
process = sys.argv[1]
reader = TMVA.Reader("!V")
nJet = array.array('f', [0])
mindr_lep1_jet = array.array('f', [0])
mindr_lep2_jet = array.array('f', [0])
mindr_lep3_jet = array.array('f', [0])
avg_dr_jet = array.array('f', [0])
lep1_abs_eta = array.array('f', [0])
lep2_abs_eta = array.array('f', [0])
lep3_abs_eta = array.array('f', [0])
max_lep_eta = array.array('f', [0])
lep1_conePt = array.array('f', [0])
lep2_conePt = array.array('f', [0])
lep3_conePt = array.array('f', [0])
mindr_tau_jet = array.array('f', [0])
ptmiss = array.array('f', [0])
mT_lep1 = array.array('f', [0])
mT_lep2 = array.array('f', [0])
mT_lep3 = array.array('f', [0])
htmiss = array.array('f', [0])
dr_leps = array.array('f', [0])
tau_pt = array.array('f', [0])
tau_abs_eta = array.array('f', [0])
def main(weights, picklename, filename, treename='bTag_AntiKt2PV0TrackJets'):
'''
evaluate the tmva method after transforming input data into right format
Args:
-----
weights: .xml file out of mv2 training containing bdt parameters
picklename: name of the output pickle to store new mv2 values
filename: .root file with ntuples used to evaluate the tmva method
treename: (optional) name of the TTree to consider
Returns:
--------
status
Raises:
-------
nothing yet, but to be improved
'''
print 'Parsing XML file...'
# -- Load XML file
tree = ET.parse(weights)
root = tree.getroot()
# -- Get list of variable names from XML file
var_list = [
var.attrib['Label']
for var in root.findall('Variables')[0].findall('Variable')
]
# -- Count the input variables that go into MV2:
n_vars = len(var_list)
print 'Loading .root file for evaluation...'
# -- Get ntuples:
df = pup.root2panda(
filename,
treename,
branches=[
'jet_pt', 'jet_eta', 'jet_phi', 'jet_m', 'jet_ip2d_pu',
'jet_ip2d_pc', 'jet_ip2d_pb', 'jet_ip3d_pu', 'jet_ip3d_pc',
'jet_ip3d_pb', 'jet_sv1_vtx_x', 'jet_sv1_vtx_y', 'jet_sv1_vtx_z',
'jet_sv1_ntrkv', 'jet_sv1_m', 'jet_sv1_efc', 'jet_sv1_n2t',
'jet_sv1_sig3d', 'jet_jf_n2t', 'jet_jf_ntrkAtVx', 'jet_jf_nvtx',
'jet_jf_nvtx1t', 'jet_jf_m', 'jet_jf_efc', 'jet_jf_sig3d',
'jet_jf_deta', 'jet_jf_dphi', 'PVx', 'PVy', 'PVz'
])
# -- Insert default values, calculate MV2 variables from the branches in df
df = transformVars(df)
# -- Map ntuple names to var_list
names_mapping = {
'pt': 'jet_pt',
'abs(eta)': 'abs(jet_eta)',
'ip2': 'jet_ip2',
'ip2_c': 'jet_ip2_c',
'ip2_cu': 'jet_ip2_cu',
'ip3': 'jet_ip3',
'ip3_c': 'jet_ip3_c',
'ip3_cu': 'jet_ip3_cu',
'sv1_ntkv': 'jet_sv1_ntrkv',
'sv1_mass': 'jet_sv1_m',
'sv1_efrc': 'jet_sv1_efc',
'sv1_n2t': 'jet_sv1_n2t',
'sv1_Lxy': 'jet_sv1_Lxy',
'sv1_L3d': 'jet_sv1_L3d',
'sv1_sig3': 'jet_sv1_sig3d',
'sv1_dR': 'jet_sv1_dR',
'jf_n2tv': 'jet_jf_n2t',
'jf_ntrkv': 'jet_jf_ntrkAtVx',
'jf_nvtx': 'jet_jf_nvtx',
'jf_nvtx1t': 'jet_jf_nvtx1t',
'jf_mass': 'jet_jf_m',
'jf_efrc': 'jet_jf_efc',
'jf_dR': 'jet_jf_dR',
'jf_sig3': 'jet_jf_sig3d'
}
print 'Initializing TMVA...'
# -- TMVA: Initialize reader, add empty variables and weights from training
reader = TMVA.Reader()
for n in range(n_vars):
reader.AddVariable(var_list[n], array('f', [0]))
reader.BookMVA('BDTG akt2', weights)
print 'Creating feature matrix...'
# -- Get features for each event and store them in X_test
X_buf = []
for event in df[[names_mapping[var] for var in var_list]].values:
X_buf.extend(
np.array([normalize_type(jet) for jet in event]).T.tolist())
X_test = np.array(X_buf)
print 'Evaluating!'
# -- TMVA: Evaluate!
twoclass_output = evaluate_reader(reader, 'BDTG akt2', X_test)
# -- Reshape the MV2 output into event-jet format
reorganized = match_shape(twoclass_output, df['jet_pt'])
import cPickle
print 'Saving new MV2 weights in {}'.format(picklename)
cPickle.dump(reorganized, open(picklename, 'wb'))
# -- Write the new branch to the tree (currently de-activated)
#add_branch(reorganized, filename, treename, 'jet_mv2c20_new')
print 'Done. Success!'
return 0
def main():
usage = 'usage: %prog [options]'
parser = optparse.OptionParser(usage)
parser.add_option(
'-s',
'--suffix',
dest='input_suffix',
help='suffix used to identify inputs from network training',
default=None,
type='string')
parser.add_option('-j',
'--json',
dest='json',
help='json file with list of variables',
default=None,
type='string')
parser.add_option('-i',
'--input',
dest='input_file',
help='input file',
default=None,
type='string')
(opt, args) = parser.parse_args()
jsonFile = open(opt.json, 'r')
if opt.json == None:
print 'input variable .json not defined!'
sys.exit(1)
if opt.input_suffix == None:
print 'Input files suffix not defined!'
sys.exit(1)
if opt.input_file == None:
print 'Input file not defined!'
sys.exit(1)
new_variable_list = json.load(jsonFile, encoding="utf-8").items()
n_input_vars = 0
for key, value in new_variable_list:
n_input_vars = n_input_vars + 1
input_file = opt.input_file
classifier_suffix = opt.input_suffix
classifier_parent_dir = '/afs/cern.ch/work/j/jthomasw/private/IHEP/ttHML/github/ttH_multilepton/DNN/Evaluation/V7-DNN_%s' % (
classifier_suffix)
# Setup TMVA
TMVA.Tools.Instance()
TMVA.PyMethodBase.PyInitialize()
reader = TMVA.Reader("Color:!Silent")
# Check files exist
if not isfile(input_file):
print 'No such input file: %s' % input_file
# Open files and load ttrees
data_file = TFile.Open(input_file)
data_tree = data_file.Get("syncTree")
branches_tree = {}
integer_branches_tree = {}
for key, value in new_variable_list:
if 'hadTop_BDT' in key:
branches_tree[key] = array('f', [-999])
keyname = 'hadTop_BDT'
data_tree.SetBranchAddress(str(keyname), branches_tree[key])
elif 'Hj1_BDT' in key:
branches_tree[key] = array('f', [-999])
keyname = 'Hj1_BDT'
data_tree.SetBranchAddress(str(keyname), branches_tree[key])
elif ('n_fakeablesel_mu' in key) or ('n_fakeablesel_ele'
in key) or ('Jet_numLoose'
in key):
integer_branches_tree[key] = array('I', [9999])
data_tree.SetBranchAddress(str(key), integer_branches_tree[key])
else:
branches_tree[key] = array('f', [-999])
data_tree.SetBranchAddress(str(key), branches_tree[key])
branches_reader = {}
# Register names of inputs with reader. Together with the name give the address of the local variable that carries the updated input variables during event loop.
for key, value in new_variable_list:
print 'Add variable name %s: ' % key
branches_reader[key] = array('f', [-999])
reader.AddVariable(str(key), branches_reader[key])
event_number = array('f', [-999])
reader.AddSpectator('nEvent', event_number)
# Book methods
# First argument is user defined name. Doesn not have to be same as training name.
# True type of method and full configuration are read from the weights file specified in the second argument.
mva_weights_dir = '%s/weights/Factory_V7-DNN_%s_DNN.weights.xml' % (
classifier_parent_dir, classifier_suffix)
print 'using weights file: ', mva_weights_dir
reader.BookMVA('DNN', TString(mva_weights_dir))
if '/2L/' in input_file:
analysis_region = '2L'
elif '/ttWctrl/' in input_file:
analysis_region = 'ttWctrl'
elif '/JESDownttWctrl/' in input_file:
analysis_region = 'JESDownttWctrl'
elif '/JESUpttWctrl/' in input_file:
analysis_region = 'JESUpttWctrl'
elif '/ClosTTWctrl/' in input_file:
analysis_region = 'ClosTTWctrl'
elif '/ttZctrl/' in input_file:
analysis_region = 'ttZctrl'
elif '/Clos2LSS/' in input_file:
analysis_region = 'Closure'
elif '/JESDown2L/' in input_file:
analysis_region = 'JESDown2L'
elif '/JESUp2L/' in input_file:
analysis_region = 'JESUp2L'
time_suffix = str(datetime.now(pytz.utc)).split(' ')
print time_suffix[0]
#classifier_samples_dir = classifier_parent_dir+"/outputs"
classifier_samples_dir = classifier_parent_dir + "/outputs-newbinning"
#classifier_plots_dir = classifier_parent_dir+"/plots"
classifier_plots_dir = classifier_parent_dir + "/plots-newbinning"
if not os.path.exists(classifier_plots_dir):
os.makedirs(classifier_plots_dir)
if not os.path.exists(classifier_samples_dir):
os.makedirs(classifier_samples_dir)
analysis_region_samples_dir = '%s/%s' % (classifier_samples_dir,
analysis_region)
analysis_region_plots_dir = '%s/%s' % (classifier_plots_dir,
analysis_region)
if not os.path.exists(analysis_region_plots_dir):
os.makedirs(analysis_region_plots_dir)
if not os.path.exists(analysis_region_samples_dir):
os.makedirs(analysis_region_samples_dir)
output_suffix = input_file[input_file.rindex('/') + 1:]
print 'output_suffix: ', output_suffix
# Define outputs: files to store histograms/ttree with results from application of classifiers and any histos/trees themselves.
output_file_name = '%s/Evaluated_%s_%s' % (
analysis_region_samples_dir, classifier_suffix, output_suffix)
output_file = TFile.Open(output_file_name, 'RECREATE')
output_tree = data_tree.CopyTree("")
output_tree.SetName("output_tree")
nEvents_check = output_tree.BuildIndex("nEvent", "run")
print 'Copied %s events from original tree' % (nEvents_check)
sample_nickname = ''
if 'THQ_htt_2L' in input_file:
sample_nickname = 'THQ_htt_2L'
if 'THQ_hzz_2L' in input_file:
sample_nickname = 'THQ_hzz_2L'
if 'THW_hww_2L' in input_file:
sample_nickname = 'THW_hww_2L'
if 'TTH_hmm_2L' in input_file:
sample_nickname = 'TTH_hmm_2L'
if 'TTH_htt_2L' in input_file:
sample_nickname = 'TTH_htt_2L'
if 'TTH_hzz_2L' in input_file:
sample_nickname = 'TTH_hzz_2L'
if 'THQ_hww_2L' in input_file:
sample_nickname = 'THQ_hww_2L'
if 'THW_htt_2L' in input_file:
sample_nickname = 'THW_htt_2L'
if 'THW_hzz_2L' in input_file:
sample_nickname = 'THW_hzz_2L'
if 'TTH_hot_2L' in input_file:
sample_nickname = 'TTH_hot_2L'
if 'TTH_hww_2L' in input_file:
sample_nickname = 'TTH_hww_2L'
if 'TTWW_2L' in input_file:
sample_nickname = 'TTWW_2L'
if 'TTW_2L' in input_file:
sample_nickname = 'TTW_2L'
if 'TTZ_2L' in input_file:
sample_nickname = 'TTZ_2L'
if 'Conv_2L' in input_file:
sample_nickname = 'Conv_2L'
if 'EWK_2L' in input_file:
sample_nickname = 'EWK_2L'
if 'Fakes_2L' in input_file:
sample_nickname = 'Fakes_2L'
if 'Flips_2L' in input_file:
sample_nickname = 'Flips_2L'
if 'Rares_2L' in input_file:
sample_nickname = 'Rares_2L'
if 'TT_Clos' in input_file:
sample_nickname = 'TT_Clos'
if 'Data' in input_file:
sample_nickname = 'Data'
# Evaluate network and use max node response to categorise event. Only maximum node response will be plotted per event meaning each event will only contribute in the maximum nodes response histogram.
network_evaluation(data_tree, new_variable_list, sample_nickname,
branches_tree, integer_branches_tree, branches_reader,
reader, True, output_tree)
output_file.Write()
gDirectory.Delete("syncTree;*")
output_file.Close()
print 'Job complete. Exiting.'
sys.exit(0)
def main():
usage = 'usage: %prog [options]'
parser = optparse.OptionParser(usage)
parser.add_option(
'-s',
'--suffix',
dest='input_suffix',
help='suffix used to identify inputs from network training',
default=None,
type='string')
parser.add_option('-j',
'--json',
dest='json',
help='json file with list of variables',
default=None,
type='string')
(opt, args) = parser.parse_args()
jsonFile = open(opt.json, 'r')
if opt.json == None:
print 'input variable .json not defined!'
sys.exit(1)
if opt.input_suffix == None:
print 'Input files suffix not defined!'
sys.exit(1)
new_variable_list = json.load(jsonFile, encoding='utf-8').items()
# Setup TMVA
TMVA.Tools.Instance()
TMVA.PyMethodBase.PyInitialize()
reader = TMVA.Reader("Color:!Silent")
input_file_ttH = 'samples/2017_updated_MC/2LSS/TTH_2LSS.root'
input_file_ttV = 'samples/2017_updated_MC/2LSS/ttV.root'
#input_file_ttJets = 'samples/2017_updated_MC/2LSS/TTJets.root'
input_file_ttJets = 'samples/2017_updated_MC/2LSS/Fakes_2LSS.root'
# Check files exist
if not isfile(input_file_ttH):
print 'No such input file: %s' % input_file_ttH
if not isfile(input_file_ttJets):
print 'No such input file: %s' % input_file_ttJets
if not isfile(input_file_ttV):
print 'No such input file: %s' % input_file_ttV
# Open files and load ttrees
data_ttH = TFile.Open(input_file_ttH)
data_ttV = TFile.Open(input_file_ttV)
data_ttJets = TFile.Open(input_file_ttJets)
data_ttH_tree = data_ttH.Get('syncTree')
data_ttV_tree = data_ttV.Get('syncTree')
data_ttJets_tree = data_ttJets.Get('syncTree')
branches_tree = {}
integer_branches_tree = {}
for key, value in new_variable_list:
#branches_tree[key] = array('d', [-999])
branches_tree[key] = array('f', [-999])
if 'hadTop_BDT' in key:
keyname = 'hadTop_BDT'
data_ttH_tree.SetBranchAddress(str(keyname), branches_tree[key])
data_ttV_tree.SetBranchAddress(str(keyname), branches_tree[key])
data_ttJets_tree.SetBranchAddress(str(keyname), branches_tree[key])
elif 'Hj1_BDT' in key:
keyname = 'Hj1_BDT'
data_ttH_tree.SetBranchAddress(str(keyname), branches_tree[key])
data_ttV_tree.SetBranchAddress(str(keyname), branches_tree[key])
data_ttJets_tree.SetBranchAddress(str(keyname), branches_tree[key])
elif ('n_fakeablesel_mu' in key) or ('n_fakeablesel_ele'
in key) or ('Jet_numLoose'
in key):
integer_branches_tree[key] = array('I', [9999])
data_ttH_tree.SetBranchAddress(str(key),
integer_branches_tree[key])
data_ttH_tree.SetBranchAddress(str(key),
integer_branches_tree[key])
data_ttJets_tree.SetBranchAddress(str(key),
integer_branches_tree[key])
else:
data_ttH_tree.SetBranchAddress(str(key), branches_tree[key])
data_ttV_tree.SetBranchAddress(str(key), branches_tree[key])
data_ttJets_tree.SetBranchAddress(str(key), branches_tree[key])
branches_reader = {}
# Register names of inputs with reader. Together with the name give the address of the local variable that carries the updated input variables during event loop.
for key, value in new_variable_list:
print 'Add variable name %s: ' % key
branches_reader[key] = array('f', [-999])
reader.AddVariable(str(key), branches_reader[key])
event_number = array('f', [-999])
reader.AddSpectator('nEvent', event_number)
num_inputs = 0
for key, value in new_variable_list:
num_inputs = num_inputs + 1
classifier_suffix = opt.input_suffix
# Book methods
# First argument is user defined name. Doesn not have to be same as training name.
# True type of method and full configuration are read from the weights file specified in the second argument.
mva_weights_dir = 'MultiClass_DNN_%sVars_%s/weights/Factory_MultiClass_DNN_%sVars_%s_DNN.weights.xml' % (
str(num_inputs), classifier_suffix, str(num_inputs), classifier_suffix)
print 'using weights file: ', mva_weights_dir
reader.BookMVA('DNN', TString(mva_weights_dir))
classifier_samples_dir = 'MultiClass_DNN_%sVars_%s/outputs' % (
str(num_inputs), classifier_suffix)
classifier_plots_dir = 'MultiClass_DNN_%sVars_%s/plots' % (
str(num_inputs), classifier_suffix)
if not os.path.exists(classifier_plots_dir):
os.makedirs(classifier_plots_dir)
if not os.path.exists(classifier_samples_dir):
os.makedirs(classifier_samples_dir)
# Define outputs: files to store histograms/ttree with results from application of classifiers and any histos/trees themselves.
output_file_name = '%s/Applied_MultiClass_DNN_%sVars_%s.root' % (
classifier_samples_dir, str(num_inputs), classifier_suffix)
output_file = TFile.Open(output_file_name, 'RECREATE')
# Evaluate network and make plots of the response on each of the nodes to each of the simulated samples.
#network_evaluation(data_ttH_tree, new_variable_list, 'ttH', branches_tree, branches_reader, reader, False)
#network_evaluation(data_ttV_tree, new_variable_list, 'ttV', branches_tree, branches_reader, reader, False)
#network_evaluation(data_ttJets_tree, new_variable_list, 'ttJets', branches_tree, branches_reader, reader, False)
# Evaluate network and use max node response to categorise event. Only maximum node response will be plotted per event meaning each event will only contribute in the maximum nodes response histogram.
network_evaluation(data_ttH_tree, new_variable_list, 'ttH', branches_tree,
integer_branches_tree, branches_reader, reader, True)
network_evaluation(data_ttV_tree, new_variable_list, 'ttV', branches_tree,
integer_branches_tree, branches_reader, reader, True)
network_evaluation(data_ttJets_tree, new_variable_list, 'ttJets',
branches_tree, integer_branches_tree, branches_reader,
reader, True)
output_file.Close()
def main():
Muon1_cLP = array('f', [-1.0])
Muon1_cLM = array('f', [-1.0])
Muon1_staRelChi2 = array('f', [-1.0])
Muon1_trkRelChi2 = array('f', [-1.0])
Muon1_glbdEP = array('f', [-1.0])
Muon1_trkKink = array('f', [-1.0])
Muon1_glbKink = array('f', [-1.0])
Muon1_glbTrkP = array('f', [-1.0])
Muon1_nTVH = array('f', [-1.0])
Muon1_nVPH = array('f', [-1.0])
Muon1_vMHC = array('f', [-1.0])
Muon1_nMS = array('f', [-1.0])
Muon1_segComp = array('f', [-1.0])
Muon1_tIpOnOut = array('f', [-1.0])
Muon1_glbNChi2 = array('f', [-1.0])
Muon1_inner_nChi2 = array('f', [-1.0])
Muon1_outer_nChi2 = array('f', [-1.0])
Muon1_innner_VF = array('f', [-1.0])
mu1_eta = array('f', [-1.0])
mu1_pt = array('f', [-1.0])
mu1_phi = array('f', [-1.0])
mu1_SoftMVA = array('f', [-1.0])
Muon2_cLP = array('f', [-1.0])
Muon2_cLM = array('f', [-1.0])
Muon2_staRelChi2 = array('f', [-1.0])
Muon2_trkRelChi2 = array('f', [-1.0])
Muon2_glbdEP = array('f', [-1.0])
Muon2_trkKink = array('f', [-1.0])
Muon2_glbKink = array('f', [-1.0])
Muon2_glbTrkP = array('f', [-1.0])
Muon2_nTVH = array('f', [-1.0])
Muon2_nVPH = array('f', [-1.0])
Muon2_vMHC = array('f', [-1.0])
Muon2_nMS = array('f', [-1.0])
Muon2_segComp = array('f', [-1.0])
Muon2_tIpOnOut = array('f', [-1.0])
Muon2_glbNChi2 = array('f', [-1.0])
Muon2_inner_nChi2 = array('f', [-1.0])
Muon2_outer_nChi2 = array('f', [-1.0])
Muon2_innner_VF = array('f', [-1.0])
mu2_eta = array('f', [-1.0])
mu2_pt = array('f', [-1.0])
mu2_phi = array('f', [-1.0])
mu2_SoftMVA = array('f', [-1.0])
Muon3_cLP = array('f', [-1.0])
Muon3_cLM = array('f', [-1.0])
Muon3_staRelChi2 = array('f', [-1.0])
Muon3_trkRelChi2 = array('f', [-1.0])
Muon3_glbdEP = array('f', [-1.0])
Muon3_trkKink = array('f', [-1.0])
Muon3_glbKink = array('f', [-1.0])
Muon3_glbTrkP = array('f', [-1.0])
Muon3_nTVH = array('f', [-1.0])
Muon3_nVPH = array('f', [-1.0])
Muon3_vMHC = array('f', [-1.0])
Muon3_nMS = array('f', [-1.0])
Muon3_segComp = array('f', [-1.0])
Muon3_tIpOnOut = array('f', [-1.0])
Muon3_glbNChi2 = array('f', [-1.0])
Muon3_inner_nChi2 = array('f', [-1.0])
Muon3_outer_nChi2 = array('f', [-1.0])
Muon3_innner_VF = array('f', [-1.0])
mu3_eta = array('f', [-1.0])
mu3_pt = array('f', [-1.0])
mu3_phi = array('f', [-1.0])
mu3_SoftMVA = array('f', [-1.0])
var_dnnSegCompMuMin = array('f', [-1.0])
# Muon Id 1
reader_Muon1Id_barrel = TMVA.Reader("!Color:!Silent")
reader_Muon1Id_barrel.AddVariable(
"mu_combinedQuality_chi2LocalMomentum>250?250:mu_combinedQuality_chi2LocalMomentum",
Muon1_cLM)
reader_Muon1Id_barrel.AddVariable(
"mu_combinedQuality_chi2LocalPosition>50?50:mu_combinedQuality_chi2LocalPosition",
Muon1_cLP)
reader_Muon1Id_barrel.AddVariable(
"mu_combinedQuality_staRelChi2>50?50:mu_combinedQuality_staRelChi2",
Muon1_staRelChi2)
reader_Muon1Id_barrel.AddVariable(
"mu_combinedQuality_trkRelChi2>50?50:mu_combinedQuality_trkRelChi2",
Muon1_trkRelChi2)
reader_Muon1Id_barrel.AddVariable("mu_combinedQuality_globalDeltaEtaPhi",
Muon1_glbdEP)
reader_Muon1Id_barrel.AddVariable("log(mu_combinedQuality_trkKink)",
Muon1_trkKink)
reader_Muon1Id_barrel.AddVariable("log(mu_combinedQuality_glbKink)",
Muon1_glbKink)
reader_Muon1Id_barrel.AddVariable(
"mu_combinedQuality_glbTrackProbability>150?150:mu_combinedQuality_glbTrackProbability",
Muon1_glbTrkP)
reader_Muon1Id_barrel.AddVariable("mu_Numberofvalidpixelhits", Muon1_nVPH)
reader_Muon1Id_barrel.AddVariable("mu_trackerLayersWithMeasurement",
Muon1_nTVH)
reader_Muon1Id_barrel.AddVariable("mu_validMuonHitComb", Muon1_vMHC)
reader_Muon1Id_barrel.AddVariable("mu_numberOfMatchedStations", Muon1_nMS)
reader_Muon1Id_barrel.AddVariable("mu_segmentCompatibility", Muon1_segComp)
reader_Muon1Id_barrel.AddVariable("mu_timeAtIpInOutErr", Muon1_tIpOnOut)
reader_Muon1Id_barrel.AddVariable("mu_GLnormChi2>50?50:mu_GLnormChi2",
Muon1_glbNChi2)
reader_Muon1Id_barrel.AddVariable(
"mu_innerTrack_normalizedChi2>50?50:mu_innerTrack_normalizedChi2",
Muon1_inner_nChi2)
reader_Muon1Id_barrel.AddVariable(
"mu_outerTrack_normalizedChi2>80?80:mu_outerTrack_normalizedChi2",
Muon1_outer_nChi2)
reader_Muon1Id_barrel.AddVariable("mu_innerTrack_validFraction",
Muon1_innner_VF)
reader_Muon1Id_barrel.AddSpectator("mu_eta", mu1_eta)
reader_Muon1Id_barrel.AddSpectator("mu_pt", mu1_pt)
reader_Muon1Id_barrel.AddSpectator("mu_phi", mu1_phi)
reader_Muon1Id_barrel.AddSpectator("mu_SoftMVA", mu1_SoftMVA)
reader_Muon1Id_barrel.BookMVA(
"BDT", basedir + "weights/weights_barrel/TMVA_new_BDT.weights.xml")
# weights weights.xml file after training, place it to CommonFiles
reader_Muon1Id_endcap = TMVA.Reader("!Color:!Silent")
reader_Muon1Id_endcap.AddVariable(
"mu_combinedQuality_chi2LocalMomentum>250?250:mu_combinedQuality_chi2LocalMomentum",
Muon1_cLM)
reader_Muon1Id_endcap.AddVariable(
"mu_combinedQuality_chi2LocalPosition>50?50:mu_combinedQuality_chi2LocalPosition",
Muon1_cLP)
reader_Muon1Id_endcap.AddVariable(
"mu_combinedQuality_staRelChi2>50?50:mu_combinedQuality_staRelChi2",
Muon1_staRelChi2)
reader_Muon1Id_endcap.AddVariable(
"mu_combinedQuality_trkRelChi2>50?50:mu_combinedQuality_trkRelChi2",
Muon1_trkRelChi2)
reader_Muon1Id_endcap.AddVariable("mu_combinedQuality_globalDeltaEtaPhi",
Muon1_glbdEP)
reader_Muon1Id_endcap.AddVariable("log(mu_combinedQuality_trkKink)",
Muon1_trkKink)
reader_Muon1Id_endcap.AddVariable("log(mu_combinedQuality_glbKink)",
Muon1_glbKink)
reader_Muon1Id_endcap.AddVariable(
"mu_combinedQuality_glbTrackProbability>150?150:mu_combinedQuality_glbTrackProbability",
Muon1_glbTrkP)
reader_Muon1Id_endcap.AddVariable("mu_Numberofvalidpixelhits", Muon1_nVPH)
reader_Muon1Id_endcap.AddVariable("mu_trackerLayersWithMeasurement",
Muon1_nTVH)
reader_Muon1Id_endcap.AddVariable("mu_validMuonHitComb", Muon1_vMHC)
reader_Muon1Id_endcap.AddVariable("mu_numberOfMatchedStations", Muon1_nMS)
reader_Muon1Id_endcap.AddVariable("mu_segmentCompatibility", Muon1_segComp)
reader_Muon1Id_endcap.AddVariable("mu_timeAtIpInOutErr", Muon1_tIpOnOut)
reader_Muon1Id_endcap.AddVariable("mu_GLnormChi2>50?50:mu_GLnormChi2",
Muon1_glbNChi2)
reader_Muon1Id_endcap.AddVariable(
"mu_innerTrack_normalizedChi2>50?50:mu_innerTrack_normalizedChi2",
Muon1_inner_nChi2)
reader_Muon1Id_endcap.AddVariable(
"mu_outerTrack_normalizedChi2>80?80:mu_outerTrack_normalizedChi2",
Muon1_outer_nChi2)
reader_Muon1Id_endcap.AddVariable("mu_innerTrack_validFraction",
Muon1_innner_VF)
reader_Muon1Id_endcap.AddSpectator("mu_eta", mu1_eta)
reader_Muon1Id_endcap.AddSpectator("mu_pt", mu1_pt)
reader_Muon1Id_endcap.AddSpectator("mu_phi", mu1_phi)
reader_Muon1Id_endcap.AddSpectator("mu_SoftMVA", mu1_SoftMVA)
reader_Muon1Id_endcap.BookMVA(
"BDT", basedir + "weights/weights_endcap/TMVA_new_BDT.weights.xml")
# weights weights.xml file after training, place it to CommonFiles
# (MuonId 2)
reader_Muon2Id_barrel = TMVA.Reader("!Color:!Silent")
reader_Muon2Id_barrel.AddVariable(
"mu_combinedQuality_chi2LocalMomentum>250?250:mu_combinedQuality_chi2LocalMomentum",
Muon2_cLM)
reader_Muon2Id_barrel.AddVariable(
"mu_combinedQuality_chi2LocalPosition>50?50:mu_combinedQuality_chi2LocalPosition",
Muon2_cLP)
reader_Muon2Id_barrel.AddVariable(
"mu_combinedQuality_staRelChi2>50?50:mu_combinedQuality_staRelChi2",
Muon2_staRelChi2)
reader_Muon2Id_barrel.AddVariable(
"mu_combinedQuality_trkRelChi2>50?50:mu_combinedQuality_trkRelChi2",
Muon2_trkRelChi2)
reader_Muon2Id_barrel.AddVariable("mu_combinedQuality_globalDeltaEtaPhi",
Muon2_glbdEP)
reader_Muon2Id_barrel.AddVariable("log(mu_combinedQuality_trkKink)",
Muon2_trkKink)
reader_Muon2Id_barrel.AddVariable("log(mu_combinedQuality_glbKink)",
Muon2_glbKink)
reader_Muon2Id_barrel.AddVariable(
"mu_combinedQuality_glbTrackProbability>150?150:mu_combinedQuality_glbTrackProbability",
Muon2_glbTrkP)
reader_Muon2Id_barrel.AddVariable("mu_Numberofvalidpixelhits", Muon2_nVPH)
reader_Muon2Id_barrel.AddVariable("mu_trackerLayersWithMeasurement",
Muon2_nTVH)
reader_Muon2Id_barrel.AddVariable("mu_validMuonHitComb", Muon2_vMHC)
reader_Muon2Id_barrel.AddVariable("mu_numberOfMatchedStations", Muon2_nMS)
reader_Muon2Id_barrel.AddVariable("mu_segmentCompatibility", Muon2_segComp)
reader_Muon2Id_barrel.AddVariable("mu_timeAtIpInOutErr", Muon2_tIpOnOut)
reader_Muon2Id_barrel.AddVariable("mu_GLnormChi2>50?50:mu_GLnormChi2",
Muon2_glbNChi2)
reader_Muon2Id_barrel.AddVariable(
"mu_innerTrack_normalizedChi2>50?50:mu_innerTrack_normalizedChi2",
Muon2_inner_nChi2)
reader_Muon2Id_barrel.AddVariable(
"mu_outerTrack_normalizedChi2>80?80:mu_outerTrack_normalizedChi2",
Muon2_outer_nChi2)
reader_Muon2Id_barrel.AddVariable("mu_innerTrack_validFraction",
Muon2_innner_VF)
reader_Muon2Id_barrel.AddSpectator("mu_eta", mu2_eta)
reader_Muon2Id_barrel.AddSpectator("mu_pt", mu2_pt)
reader_Muon2Id_barrel.AddSpectator("mu_phi", mu2_phi)
reader_Muon2Id_barrel.AddSpectator("mu_SoftMVA", mu2_SoftMVA)
reader_Muon2Id_barrel.BookMVA(
"BDT", basedir + "/weights/weights_barrel/TMVA_new_BDT.weights.xml")
# weights weights.xml file after training, place it to CommonFiles
reader_Muon2Id_endcap = TMVA.Reader("!Color:!Silent")
reader_Muon2Id_endcap.AddVariable(
"mu_combinedQuality_chi2LocalMomentum>250?250:mu_combinedQuality_chi2LocalMomentum",
Muon2_cLM)
reader_Muon2Id_endcap.AddVariable(
"mu_combinedQuality_chi2LocalPosition>50?50:mu_combinedQuality_chi2LocalPosition",
Muon2_cLP)
reader_Muon2Id_endcap.AddVariable(
"mu_combinedQuality_staRelChi2>50?50:mu_combinedQuality_staRelChi2",
Muon2_staRelChi2)
reader_Muon2Id_endcap.AddVariable(
"mu_combinedQuality_trkRelChi2>50?50:mu_combinedQuality_trkRelChi2",
Muon2_trkRelChi2)
reader_Muon2Id_endcap.AddVariable("mu_combinedQuality_globalDeltaEtaPhi",
Muon2_glbdEP)
reader_Muon2Id_endcap.AddVariable("log(mu_combinedQuality_trkKink)",
Muon2_trkKink)
reader_Muon2Id_endcap.AddVariable("log(mu_combinedQuality_glbKink)",
Muon2_glbKink)
reader_Muon2Id_endcap.AddVariable(
"mu_combinedQuality_glbTrackProbability>150?150:mu_combinedQuality_glbTrackProbability",
Muon2_glbTrkP)
reader_Muon2Id_endcap.AddVariable("mu_Numberofvalidpixelhits", Muon2_nVPH)
reader_Muon2Id_endcap.AddVariable("mu_trackerLayersWithMeasurement",
Muon2_nTVH)
reader_Muon2Id_endcap.AddVariable("mu_validMuonHitComb", Muon2_vMHC)
reader_Muon2Id_endcap.AddVariable("mu_numberOfMatchedStations", Muon2_nMS)
reader_Muon2Id_endcap.AddVariable("mu_segmentCompatibility", Muon2_segComp)
reader_Muon2Id_endcap.AddVariable("mu_timeAtIpInOutErr", Muon2_tIpOnOut)
reader_Muon2Id_endcap.AddVariable("mu_GLnormChi2>50?50:mu_GLnormChi2",
Muon2_glbNChi2)
reader_Muon2Id_endcap.AddVariable(
"mu_innerTrack_normalizedChi2>50?50:mu_innerTrack_normalizedChi2",
Muon2_inner_nChi2)
reader_Muon2Id_endcap.AddVariable(
"mu_outerTrack_normalizedChi2>80?80:mu_outerTrack_normalizedChi2",
Muon2_outer_nChi2)
reader_Muon2Id_endcap.AddVariable("mu_innerTrack_validFraction",
Muon2_innner_VF)
reader_Muon2Id_endcap.AddSpectator("mu_eta", mu2_eta)
reader_Muon2Id_endcap.AddSpectator("mu_pt", mu2_pt)
reader_Muon2Id_endcap.AddSpectator("mu_phi", mu2_phi)
reader_Muon2Id_endcap.AddSpectator("mu_SoftMVA", mu2_SoftMVA)
reader_Muon2Id_endcap.BookMVA(
"BDT", basedir + "/weights/weights_endcap/TMVA_new_BDT.weights.xml")
# weights weights.xml file after training, place it to CommonFiles
# (MuonId 3)
reader_Muon3Id_barrel = TMVA.Reader("!Color:!Silent")
reader_Muon3Id_barrel.AddVariable(
"mu_combinedQuality_chi2LocalMomentum>250?250:mu_combinedQuality_chi2LocalMomentum",
Muon3_cLM)
reader_Muon3Id_barrel.AddVariable(
"mu_combinedQuality_chi2LocalPosition>50?50:mu_combinedQuality_chi2LocalPosition",
Muon3_cLP)
reader_Muon3Id_barrel.AddVariable(
"mu_combinedQuality_staRelChi2>50?50:mu_combinedQuality_staRelChi2",
Muon3_staRelChi2)
reader_Muon3Id_barrel.AddVariable(
"mu_combinedQuality_trkRelChi2>50?50:mu_combinedQuality_trkRelChi2",
Muon3_trkRelChi2)
reader_Muon3Id_barrel.AddVariable("mu_combinedQuality_globalDeltaEtaPhi",
Muon3_glbdEP)
reader_Muon3Id_barrel.AddVariable("log(mu_combinedQuality_trkKink)",
Muon3_trkKink)
reader_Muon3Id_barrel.AddVariable("log(mu_combinedQuality_glbKink)",
Muon3_glbKink)
reader_Muon3Id_barrel.AddVariable(
"mu_combinedQuality_glbTrackProbability>150?150:mu_combinedQuality_glbTrackProbability",
Muon3_glbTrkP)
reader_Muon3Id_barrel.AddVariable("mu_Numberofvalidpixelhits", Muon3_nVPH)
reader_Muon3Id_barrel.AddVariable("mu_trackerLayersWithMeasurement",
Muon3_nTVH)
reader_Muon3Id_barrel.AddVariable("mu_validMuonHitComb", Muon3_vMHC)
reader_Muon3Id_barrel.AddVariable("mu_numberOfMatchedStations", Muon3_nMS)
reader_Muon3Id_barrel.AddVariable("mu_segmentCompatibility", Muon3_segComp)
reader_Muon3Id_barrel.AddVariable("mu_timeAtIpInOutErr", Muon3_tIpOnOut)
reader_Muon3Id_barrel.AddVariable("mu_GLnormChi2>50?50:mu_GLnormChi2",
Muon3_glbNChi2)
reader_Muon3Id_barrel.AddVariable(
"mu_innerTrack_normalizedChi2>50?50:mu_innerTrack_normalizedChi2",
Muon3_inner_nChi2)
reader_Muon3Id_barrel.AddVariable(
"mu_outerTrack_normalizedChi2>80?80:mu_outerTrack_normalizedChi2",
Muon3_outer_nChi2)
reader_Muon3Id_barrel.AddVariable("mu_innerTrack_validFraction",
Muon3_innner_VF)
reader_Muon3Id_barrel.AddSpectator("mu_eta", mu2_eta)
reader_Muon3Id_barrel.AddSpectator("mu_pt", mu2_pt)
reader_Muon3Id_barrel.AddSpectator("mu_phi", mu2_phi)
reader_Muon3Id_barrel.AddSpectator("mu_SoftMVA", mu2_SoftMVA)
reader_Muon3Id_barrel.BookMVA(
"BDT", basedir + "/weights/weights_barrel/TMVA_new_BDT.weights.xml")
# weights weights.xml file after training, place it to CommonFiles
reader_Muon3Id_endcap = TMVA.Reader("!Color:!Silent")
reader_Muon3Id_endcap.AddVariable(
"mu_combinedQuality_chi2LocalMomentum>250?250:mu_combinedQuality_chi2LocalMomentum",
Muon3_cLM)
reader_Muon3Id_endcap.AddVariable(
"mu_combinedQuality_chi2LocalPosition>50?50:mu_combinedQuality_chi2LocalPosition",
Muon3_cLP)
reader_Muon3Id_endcap.AddVariable(
"mu_combinedQuality_staRelChi2>50?50:mu_combinedQuality_staRelChi2",
Muon3_staRelChi2)
reader_Muon3Id_endcap.AddVariable(
"mu_combinedQuality_trkRelChi2>50?50:mu_combinedQuality_trkRelChi2",
Muon3_trkRelChi2)
reader_Muon3Id_endcap.AddVariable("mu_combinedQuality_globalDeltaEtaPhi",
Muon3_glbdEP)
reader_Muon3Id_endcap.AddVariable("log(mu_combinedQuality_trkKink)",
Muon3_trkKink)
reader_Muon3Id_endcap.AddVariable("log(mu_combinedQuality_glbKink)",
Muon3_glbKink)
reader_Muon3Id_endcap.AddVariable(
"mu_combinedQuality_glbTrackProbability>150?150:mu_combinedQuality_glbTrackProbability",
Muon3_glbTrkP)
reader_Muon3Id_endcap.AddVariable("mu_Numberofvalidpixelhits", Muon3_nVPH)
reader_Muon3Id_endcap.AddVariable("mu_trackerLayersWithMeasurement",
Muon3_nTVH)
reader_Muon3Id_endcap.AddVariable("mu_validMuonHitComb", Muon3_vMHC)
reader_Muon3Id_endcap.AddVariable("mu_numberOfMatchedStations", Muon3_nMS)
reader_Muon3Id_endcap.AddVariable("mu_segmentCompatibility", Muon3_segComp)
reader_Muon3Id_endcap.AddVariable("mu_timeAtIpInOutErr", Muon3_tIpOnOut)
reader_Muon3Id_endcap.AddVariable("mu_GLnormChi2>50?50:mu_GLnormChi2",
Muon3_glbNChi2)
reader_Muon3Id_endcap.AddVariable(
"mu_innerTrack_normalizedChi2>50?50:mu_innerTrack_normalizedChi2",
Muon3_inner_nChi2)
reader_Muon3Id_endcap.AddVariable(
"mu_outerTrack_normalizedChi2>80?80:mu_outerTrack_normalizedChi2",
Muon3_outer_nChi2)
reader_Muon3Id_endcap.AddVariable("mu_innerTrack_validFraction",
Muon3_innner_VF)
reader_Muon3Id_endcap.AddSpectator("mu_eta", mu3_eta)
reader_Muon3Id_endcap.AddSpectator("mu_pt", mu3_pt)
reader_Muon3Id_endcap.AddSpectator("mu_phi", mu3_phi)
reader_Muon3Id_endcap.AddSpectator("mu_SoftMVA", mu3_SoftMVA)
reader_Muon3Id_endcap.BookMVA(
"BDT", basedir + "/weights/weights_endcap/TMVA_new_BDT.weights.xml")
# weights weights.xml file after training, place it to CommonFiles
file_ = ROOT.TFile(
"/eos/user/b/bjoshi/RunIITau23Mu/AnalysisTrees/T3MSelectionTreeInput_combined_with_keras_scores.root",
"READ")
old_tree_bkg = file_.Get('TreeB')
old_tree_ds = file_.Get('TreeS_Ds')
old_tree_bu = file_.Get('TreeS_Bu')
old_tree_bd = file_.Get('TreeS_Bd')
new_file = ROOT.TFile(
"/eos/user/b/bjoshi/RunIITau23Mu/AnalysisTrees/T3MSelectionTreeInput_combined_globalMuonId.root",
"RECREATE")
new_tree_bkg = old_tree_bkg.CloneTree(0)
new_tree_ds = old_tree_ds.CloneTree(0)
new_tree_bu = old_tree_bu.CloneTree(0)
new_tree_bd = old_tree_bd.CloneTree(0)
old_tree_list = [old_tree_bkg, old_tree_ds, old_tree_bu, old_tree_bd]
new_tree_list = [new_tree_bkg, new_tree_ds, new_tree_bu, new_tree_bd]
globalMuon1Id = [
array('f', [-99.]),
array('f', [-99.]),
array('f', [-99.]),
array('f', [-99.])
]
globalMuon2Id = [
array('f', [-99.]),
array('f', [-99.]),
array('f', [-99.]),
array('f', [-99.])
]
globalMuon3Id = [
array('f', [-99.]),
array('f', [-99.]),
array('f', [-99.]),
array('f', [-99.])
]
var_dnnSegCompMuMin = [
array('f', [-99.]),
array('f', [-99.]),
array('f', [-99.]),
array('f', [-99.])
]
for i in xrange(4):
new_tree_list[i].Branch("globalMuon1Id", globalMuon1Id[i],
"globalMuon1Id/F")
new_tree_list[i].Branch("globalMuon2Id", globalMuon2Id[i],
"globalMuon2Id/F")
new_tree_list[i].Branch("globalMuon3Id", globalMuon3Id[i],
"globalMuon3Id/F")
new_tree_list[i].Branch("var_dnnSegCompMuMin", var_dnnSegCompMuMin[i],
"var_dnnSegCompMuMin/F")
for i, tree in enumerate(old_tree_list):
nentries = tree.GetEntriesFast()
print tree.GetName()
for j in xrange(nentries):
if (j % 1000 == 0): print "Processing ", j, "/", nentries, " ..."
tree.GetEntry(j)
muon1_score = -99.0
muon2_score = -99.0
muon3_score = -99.0
Muon1_cLP[0] = tree.Muon1_combinedQuality_chi2LocalPosition
Muon1_cLM[0] = tree.Muon1_combinedQuality_chi2LocalMomentum
Muon1_staRelChi2[0] = tree.Muon1_combinedQuality_staRelChi2
Muon1_trkRelChi2[0] = tree.Muon1_combinedQuality_trkRelChi2
Muon1_glbdEP[0] = tree.Muon1_combinedQuality_globalDeltaEtaPhi
Muon1_trkKink[0] = np.log(0.01 +
tree.Muon1_combinedQuality_trkKink)
Muon1_glbKink[0] = np.log(0.01 +
tree.Muon1_combinedQuality_glbKink)
Muon1_glbTrkP[0] = tree.Muon1_combinedQuality_glbTrackProbability
Muon1_nTVH[0] = tree.Muon1_trackerLayersWithMeasurement
Muon1_nVPH[0] = tree.Muon1_numberofValidPixelHits
Muon1_vMHC[0] = tree.Muon1_vmuonhitcomb_reco
Muon1_nMS[0] = tree.Muon1_numberOfMatchedStations
Muon1_segComp[0] = tree.Muon1_segmentCompatibility
Muon1_tIpOnOut[0] = tree.var_Muon1_timeAtIpInOutErr
Muon1_glbNChi2[0] = tree.Muon1_normChi2
Muon1_inner_nChi2[0] = tree.Muon1_innerTrack_normalizedChi2
Muon1_outer_nChi2[0] = tree.Muon1_outerTrack_normalizedChi2
Muon1_innner_VF[0] = tree.Muon1_innerTrack_validFraction
Muon2_cLM[0] = tree.Muon2_combinedQuality_chi2LocalMomentum
Muon2_cLP[0] = tree.Muon2_combinedQuality_chi2LocalPosition
Muon2_staRelChi2[0] = tree.Muon2_combinedQuality_staRelChi2
Muon2_trkRelChi2[0] = tree.Muon2_combinedQuality_trkRelChi2
Muon2_glbdEP[0] = tree.Muon2_combinedQuality_globalDeltaEtaPhi
Muon2_trkKink[0] = np.log(0.01 +
tree.Muon2_combinedQuality_trkKink)
Muon2_glbKink[0] = np.log(0.01 +
tree.Muon2_combinedQuality_glbKink)
Muon2_glbTrkP[0] = tree.Muon2_combinedQuality_glbTrackProbability
Muon2_nTVH[0] = tree.Muon2_trackerLayersWithMeasurement
Muon2_nVPH[0] = tree.Muon2_numberofValidPixelHits
Muon2_vMHC[0] = tree.Muon2_vmuonhitcomb_reco
Muon2_nMS[0] = tree.Muon2_numberOfMatchedStations
Muon2_segComp[0] = tree.Muon2_segmentCompatibility
Muon2_tIpOnOut[0] = tree.var_Muon2_timeAtIpInOutErr
Muon2_glbNChi2[0] = tree.Muon2_normChi2
Muon2_inner_nChi2[0] = tree.Muon2_innerTrack_normalizedChi2
Muon2_outer_nChi2[0] = tree.Muon2_outerTrack_normalizedChi2
Muon2_innner_VF[0] = tree.Muon2_innerTrack_validFraction
Muon3_cLM[0] = tree.Muon3_combinedQuality_chi2LocalMomentum
Muon3_cLP[0] = tree.Muon3_combinedQuality_chi2LocalPosition
Muon3_staRelChi2[0] = tree.Muon3_combinedQuality_staRelChi2
Muon3_trkRelChi2[0] = tree.Muon3_combinedQuality_trkRelChi2
Muon3_glbdEP[0] = tree.Muon3_combinedQuality_globalDeltaEtaPhi
Muon3_trkKink[0] = np.log(0.01 +
tree.Muon3_combinedQuality_trkKink)
Muon3_glbKink[0] = np.log(0.01 +
tree.Muon3_combinedQuality_glbKink)
Muon3_glbTrkP[0] = tree.Muon3_combinedQuality_glbTrackProbability
Muon3_nTVH[0] = tree.Muon3_trackerLayersWithMeasurement
Muon3_nVPH[0] = tree.Muon3_numberofValidPixelHits
Muon3_vMHC[0] = tree.Muon3_vmuonhitcomb_reco
Muon3_nMS[0] = tree.Muon3_numberOfMatchedStations
Muon3_segComp[0] = tree.Muon3_segmentCompatibility
Muon3_tIpOnOut[0] = tree.var_Muon3_timeAtIpInOutErr
Muon3_glbNChi2[0] = tree.Muon3_normChi2
Muon3_inner_nChi2[0] = tree.Muon3_innerTrack_normalizedChi2
Muon3_outer_nChi2[0] = tree.Muon3_outerTrack_normalizedChi2
Muon3_innner_VF[0] = tree.Muon3_innerTrack_validFraction
if (abs(tree.var_Muon1_Eta) < 1.2):
muon1_score = reader_Muon1Id_barrel.EvaluateMVA("BDT")
else:
muon1_score = reader_Muon1Id_endcap.EvaluateMVA("BDT")
if (abs(tree.var_Muon2_Eta) < 1.2):
muon2_score = reader_Muon2Id_barrel.EvaluateMVA("BDT")
else:
muon2_score = reader_Muon2Id_endcap.EvaluateMVA("BDT")
if (abs(tree.var_Muon3_Eta) < 1.2):
muon3_score = reader_Muon3Id_barrel.EvaluateMVA("BDT")
else:
muon3_score = reader_Muon3Id_endcap.EvaluateMVA("BDT")
globalMuon1Id[i][0] = muon1_score
globalMuon2Id[i][0] = muon2_score
if (tree.threeGlobal == 1): globalMuon3Id[i][0] = muon3_score
else: globalMuon3Id[i][0] = -99.0
var_dnnSegCompMuMin[i][0] = min(tree.muon1_seg_comp_dnn,
tree.muon2_seg_comp_dnn,
tree.muon3_seg_comp_dnn)
new_tree_list[i].Fill()
new_file.cd()
new_tree_bkg.Write()
new_tree_ds.Write()
new_tree_bu.Write()
new_tree_bd.Write()
new_file.Close()
file_.Close()
def FillBDTscore(self):
phVals = ["EE", "EB"]
path = "/home/jcordero/CMS/JYCMCMS/SMP_ZG/python/Corrections/"
file = {}
for ph in phVals:
file[ph] = "TMVAnalysis_BDT_" + ph + ".weights.xml"
reader[ph] = TMVA.Reader()
###########################################
##
## Adding the variables to the MVA object
## &
## Fill in the variables from the Data
##
phVar = {}
varName = {
"recoPhi": ["photonOnePhi"],
"r9": ["photonOneR9"],
"sieieFull5x5": ["photonOneSieie"],
"sieipFull5x5": ["photonOneSieip"],
"e2x2Full5x5/e5x5Full5x5": ["photonOneE2x2", "photonOneE5x5"],
"recoSCEta": ["photonOneEta"],
"rawE": ["photonOneScRawE"],
"scEtaWidth": ["photonOneScEtaWidth"],
"scPhiWidth": ["photonOneScPhiWidth"],
"esEn/rawE": ["photonOnePreShowerE", "photonOneScRawE"],
"esRR": ["photonOneSrr"],
"rho": ["Rho"],
}
var = {}
for ph in ["EE", "EB"]:
var[ph] = {}
for vn in varName.keys():
var[ph][vn] = array.array('f', [0])
reader[ph].AddVariable(vn, var[ph][vn])
if ph == "EE": # this is just to fill in once
if len(varName[vn]) > 1:
phVar[vn] = ZG.GetWithCuts(
varName[vn][0]) / ZG.GetWithCuts(varName[vn][1])
else:
phVar[vn] = ZG.GetWithCuts(varName[vn][0])
reader[ph].BookMVA("BDT", path + file[ph])
#######################
##
## Fill the BDT
##
ShowerShapeBDT = []
for i in range(len(self.GetWithCuts('cuts'))):
if np.abs(phVar["recoSCEta"][i]) > 1.48:
ph = "EE"
else:
ph = "EB"
for vn in varName:
var[ph][vn] = phVar[vn]
ShowerShapeBDT.append(reader[ph].EvaluateMVA("BDT"))
#######################################
##
## Add the BDT score to the DataFrame
##
sefl.df["ShowerShapeBDT"] = ShowerShapeBDT
def main():
try:
# Retrive command line options
shortopts = "m:i:o:d:vh?"
longopts = [
"methods=", "inputfile=", "outputfile=", "datatype=", "verbose",
"help", "usage"
]
opts, args = getopt.getopt(sys.argv[1:], shortopts, longopts)
except getopt.GetoptError:
# Print help information and exit:
print "ERROR: unknown options in argument %s" % sys.argv[1:]
usage()
sys.exit(1)
treeNameSig = DEFAULT_TREESIG
treeNameBkg = DEFAULT_TREEBKG
methods = DEFAULT_METHODS
directory = DEFAULT_DATA
verbose = False
for o, a in opts:
if o in ("-?", "-h", "--help", "--usage"):
usage()
sys.exit(0)
elif o in ("-m", "--methods"):
methods = a
elif o in ("-d", "--datatype"):
directory = a
elif o in ("-v", "--verbose"):
verbose = True
# Print methods
#take leading and trailing white space out
methods = methods.strip(" ")
mlist = methods.replace(' ', ',').split(',')
print "=== TMVApplication: use method(s)..."
for m in mlist:
if m.strip() != '':
print "=== - <%s>" % m.strip()
# Import ROOT classes
from ROOT import gSystem, gROOT, gApplication, TFile, TTree, TCut, TH1F, TStopwatch
print("ROOT classes successfully imported!\n") # DCS 17/06/2016
# check ROOT version, give alarm if 5.18
if gROOT.GetVersionCode() >= 332288 and gROOT.GetVersionCode() < 332544:
print "*** You are running ROOT version 5.18, which has problems in PyROOT such that TMVA"
print "*** does not run properly (function calls with enums in the argument are ignored)."
print "*** Solution: either use CINT or a C++ compiled version (see TMVA/macros or TMVA/examples),"
print "*** or use another ROOT version (e.g., ROOT 5.19)."
sys.exit(1)
# Logon not automatically loaded through PyROOT (logon loads TMVA library) load also GUI
tmvadir = "/home/dean/software/tmva/TMVA-v4.2.0/test"
macro = os.path.join(tmvadir, "TMVAlogon.C")
loadmacro = os.path.join(tmvadir, "TMVAGui.C")
gROOT.SetMacroPath(tmvadir)
gROOT.Macro(macro)
gROOT.LoadMacro(loadmacro)
print("ROOT macro path loaded correctly!\n")
# Import TMVA classes from ROOT
from ROOT import TMVA
# Create the Reader object
reader = TMVA.Reader("!Color")
var1 = array('f', [0])
var2 = array('f', [0])
var3 = array('f', [0])
var4 = array('f', [0])
var5 = array('f', [0])
var6 = array('f', [0])
var7 = array('f', [0])
var8 = array('f', [0])
var9 = array('f', [0])
var10 = array('f', [0])
variables = [var1, var2, var3, var4, var5, var6, var7, var8, var9, var10]
var_names = [
'peaks', 'mean_peaks', 'integral', 'integral_over_peaks', 'max',
'mean', 'max_over_mean', 'std_dev_peaks', 'entropy', 'ps_integral'
]
#variables = [var1, var2, var3, var4]
#var_names = ['var1', 'var2', 'var3', 'var4']
for name, var in zip(var_names, variables):
reader.AddVariable(name, var)
print("Variables successfully loaded!\n")
#reader.AddVariable("Nclusters.value", var1)
#reader.AddVariable("(TMath::Log10(eventinfo_ALLOfflinePulseSeriesReco.tot_charge))*1000/eventinfo_ALLOfflinePulseSeriesReco.length" ,var2)
#reader.AddVariable("MDCOGLaunches.value*1000./eventinfo_ALLOfflinePulseSeriesReco.length",var3)
#reader.AddVariable("Nclusters.value*1000./eventinfo_ALLOfflinePulseSeriesReco.length" ,var4)
#reader.AddVariable("NSMT8TRIGGER.value/eventinfo_ALLOfflinePulseSeriesReco.nstrings",var5)
#reader.AddVariable("MedianCluster.value",var6)
# book the MVA methods
#dir = "weights/"
#prefix = "TMVAClassification_"
#
#for m in mlist:
# print( m + " method", dir + prefix + m + ".weights.xml")
# reader.BookMVA( m + " method", dir + prefix + m + ".weights.xml" )
weight_dir = "/home/dean/capstone/TMVA-classifier/weights/"
weights = [f for f in os.listdir(weight_dir) if ".xml" in f]
for i, f in enumerate(weights):
reader.BookMVA("BDT_{}".format(i),
os.path.join(weight_dir, f)) #only care about BDT
# reader.BookMVA("BDT","weights/TMVAClassification_BDT.weights.xml")
#######################################################################
# For an example how to apply your own plugin method, please see
# TMVA/macros/TMVApplication.C
#######################################################################
# Book output histograms
nbin = 100
histList = []
for m in mlist:
histList.append(TH1F(m, m, nbin, -3, 3))
for h in histList:
h.Fill(reader.EvaluateMVA(h.GetName() + " method"))
# Book example histogram for probability (the other methods would be done similarly)
if "Fisher" in mlist:
probHistFi = TH1F("PROBA_MVA_Fisher", "PROBA_MVA_Fisher", nbin, 0, 1)
rarityHistFi = TH1F("RARITY_MVA_Fisher", "RARITY_MVA_Fisher", nbin, 0,
1)
filelist = glob(directory + "/" + "Level4b*.hdf")
print 30 * "#"
print "the filelist, ", filelist
print 30 * "--"
for file in filelist:
try:
startfile = tables.openFile(file, "a")
#DELETE BDTs if they exist
for var in startfile.root._v_children.keys():
if re.match("BDT_", var):
startfile.removeNode("/", var)
startfile.removeNode("/__I3Index__", var)
#NOW CLONE THE NODE
for name in histList:
startfile.copyNode("/__I3Index__/StdDCOGLaunches",
"/__I3Index__", str(name.GetName()))
startfile.copyNode("/StdDCOGLaunches", "/",
str(name.GetName()))
startfile.close()
h5 = tables.openFile(file, 'r')
mcog_over_t = numpy.divide(h5.root.MDCOGLaunches.cols.value[:],\
h5.root.eventinfo_ALLOfflinePulseSeriesReco.cols.length[:]/1000.)
q_over_t = numpy.divide(numpy.log10(h5.root.eventinfo_ALLOfflinePulseSeriesReco.cols.tot_charge[:]),\
h5.root.eventinfo_ALLOfflinePulseSeriesReco.cols.length[:]/1000.)
ncluster_over_t = numpy.divide(h5.root.Nclusters.cols.value[:],\
h5.root.eventinfo_ALLOfflinePulseSeriesReco.cols.length[:]/1000.)
nsmt8_over_string = numpy.divide(h5.root.NSMT8TRIGGER.cols.value[:],\
h5.root.eventinfo_ALLOfflinePulseSeriesReco.cols.nstrings[:])
s1 = array('f', h5.root.Nclusters.cols.value[:])
s2 = array('f', q_over_t)
s3 = array('f', mcog_over_t[:])
s4 = array('f', ncluster_over_t[:])
s5 = array('f', nsmt8_over_string[:])
s6 = array('f', h5.root.MedianCluster.cols.value[:])
h5.close()
result = numpy.zeros((len(histList), len(s1)),
numpy.dtype([('Classifier', numpy.double)]))
for ievt in range(len(s1)):
#if ievt%1000 == 0:
# print "--- ... Processing event: %i" % ievt
# Fill event in memory
# Compute MVA input variables
var1[0] = s1[ievt]
var2[0] = s2[ievt]
var3[0] = s3[ievt]
var4[0] = s4[ievt]
var5[0] = s5[ievt]
var6[0] = s6[ievt]
# Fill histograms with MVA outputs
for j, h in enumerate(histList):
h.Fill(reader.EvaluateMVA(h.GetName() + " method"))
result[j][ievt]["Classifier"] = reader.EvaluateMVA(
h.GetName() + " method")
endfile = tables.openFile(file, 'a')
for k, name in enumerate(histList):
modifiedNode = endfile.getNode("/", str(name.GetName()))
modifiedNode.cols.value[ievt] = result[k][ievt][
"Classifier"]
endfile.close()
print time.strftime('Elapsed time - %H:%M:%S',
time.gmtime(time.clock()))
#sanity check of the mva values writen in the hdf files
#ifile=tables.openFile(file,'r')
# if len(ifile.root.BDT_400_20.cols.BDT) != len(ifile.root.MPEFit.cols.Zenith):
# ifile.close()
# print "Something wrong with file: ", k, j+1
#exit()
#os.system("rm "+"/data/icecube01/users/redlpete/IC59L2/TableIOL3/H5FilesIncludingScores/H5%0.2d%0.2d.hd5"%(k,j+1))
# ifile.close()
except ImportError as exce:
print "file does not exist", k, j + 1
print exce
exit()
ifile = tables.openFile("test.h5", mode='a')
class Score(IsDescription):
score = Float64Col()
group = ifile.createGroup("/", 'Background', 'Scoreinfo')
table = ifile.createTable(group, 'score', Score, "Example")
particle = table.row
for n in range(len(result)):
particle['score'] = result[n]
particle.append()
print "--- End of event loop: %s" % sw.Print()
target = TFile("TMVApp1.root", "RECREATE")
for h in histList:
h.Write()
target.Close()
print "--- Created root file: \"TMVApp.root\" containing the MVA output histograms"
print "==> TMVApplication is done!"
def ApplyAddCuts(InputFolder, OutputFolder, Campaign):
ChannelList = ChannelList_MC16a
if Campaign == "MC16d":
ChannelList = ChannelList_MC16d
if Campaign == "MC16e":
ChannelList = ChannelList_MC16e
TreeList = ["nominal"]
if "SYSLJ" in InputFolder:
TreeList = SystTreeList
InputFiles = glob.glob(InputFolder + "/*root*")
for InputFile in InputFiles:
OutputFile = InputFile.replace(InputFolder, OutputFolder)
if os.path.exists(OutputFile):
continue
fF = TFile(InputFile, "READ")
fF_out = TFile(OutputFile, "RECREATE")
# save sumWeights tree to new file
fT_weight = fF.Get("sumWeights")
fT_weight_out = fT_weight.CloneTree()
fT_weight_out.Write()
# save all bookkeeping histograms to new file
for Channel in ChannelList:
DIR = TDirectory()
fF.GetObject(Channel, DIR)
DIR.ReadAll()
fF_out.mkdir(Channel)
fF_out.cd(Channel)
DIR_new = TDirectory()
DIR.GetList().Write()
DIR_new.Write()
fF_out.cd()
for TreeName in TreeList:
print(
"---------------------------------------------------------------------------> Running over tree = ",
TreeName)
if TreeName == "nominal_SYST":
continue
fT = fF.Get(TreeName)
entries = fT.GetEntries()
reader1 = TMVA.Reader()
reader2 = TMVA.Reader()
var_LL_1 = array('f', [0])
reader1.AddVariable("klfitter_logLikelihood[0]", var_LL_1)
var_EvtProb_1 = array('f', [0])
reader1.AddVariable("klfitter_eventProbability[0]", var_EvtProb_1)
var_WhadPt_1 = array('f', [0])
reader1.AddVariable("klf_orig_Whad_pt", var_WhadPt_1)
var_WlepPt_1 = array('f', [0])
reader1.AddVariable("klf_orig_Wlep_pt", var_WlepPt_1)
var_ThadPt_1 = array('f', [0])
reader1.AddVariable("klf_orig_tophad_pt", var_ThadPt_1)
var_TlepPt_1 = array('f', [0])
reader1.AddVariable("klf_orig_toplep_pt", var_TlepPt_1)
var_j_n_1 = array('f', [0])
reader1.AddVariable("tma_njets", var_j_n_1)
var_met_1 = array('f', [0])
reader1.AddVariable("tma_met", var_met_1)
var_TTbarPt_1 = array('f', [0])
reader1.AddVariable("klf_orig_ttbar_pt", var_TTbarPt_1)
var_mwt_1 = array('f', [0])
reader1.AddVariable("tma_mtw", var_mwt_1)
var_DRwjets_1 = array('f', [0])
reader1.AddVariable("klf_orig_dR_qq_W", var_DRwjets_1)
var_DRbjets_1 = array('f', [0])
reader1.AddVariable("klf_orig_dR_bb", var_DRbjets_1)
var_LL_2 = array('f', [0])
reader2.AddVariable("klfitter_logLikelihood[0]", var_LL_2)
var_EvtProb_2 = array('f', [0])
reader2.AddVariable("klfitter_eventProbability[0]", var_EvtProb_2)
var_WhadPt_2 = array('f', [0])
reader2.AddVariable("klf_orig_Whad_pt", var_WhadPt_2)
var_WlepPt_2 = array('f', [0])
reader2.AddVariable("klf_orig_Wlep_pt", var_WlepPt_2)
var_ThadPt_2 = array('f', [0])
reader2.AddVariable("klf_orig_tophad_pt", var_ThadPt_2)
var_TlepPt_2 = array('f', [0])
reader2.AddVariable("klf_orig_toplep_pt", var_TlepPt_2)
var_j_n_2 = array('f', [0])
reader2.AddVariable("tma_njets", var_j_n_2)
var_met_2 = array('f', [0])
reader2.AddVariable("tma_met", var_met_2)
var_TTbarPt_2 = array('f', [0])
reader2.AddVariable("klf_orig_ttbar_pt", var_TTbarPt_2)
var_mwt_2 = array('f', [0])
reader2.AddVariable("tma_mtw", var_mwt_2)
var_DRwjets_2 = array('f', [0])
reader2.AddVariable("klf_orig_dR_qq_W", var_DRwjets_2)
var_DRbjets_2 = array('f', [0])
reader2.AddVariable("klf_orig_dR_bb", var_DRbjets_2)
reader1.BookMVA(
"BDT",
"TrainingSteffenChristmas/weights_noWindowCuts/MassTraining_BDT.weights.xml"
)
reader2.BookMVA(
"BDT",
"TrainingSteffenChristmas/weights_withWindowCuts/MassTraining_BDT.weights.xml"
)
fT_out = fT.CloneTree(0)
d_noWC = array('f', [0.])
d_withWC = array('f', [0.])
fT_out.Branch("bdtOutput_8TeVlike_noWC", d_noWC,
'bdtOutput_8TeVlike_noWC/F')
fT_out.Branch("bdtOutput_8TeVlike_withWC", d_withWC,
'bdtOutput_8TeVlike_withWC/F')
for i in range(0, entries):
fT.GetEntry(i)
var_LL_1[0] = fT.klfitter_logLikelihood[0]
var_EvtProb_1[0] = fT.klfitter_eventProbability[0]
var_WhadPt_1[0] = fT.klf_orig_Whad_pt
var_WlepPt_1[0] = fT.klf_orig_Wlep_pt
var_ThadPt_1[0] = fT.klf_orig_tophad_pt
var_TlepPt_1[0] = fT.klf_orig_toplep_pt
var_j_n_1[0] = fT.tma_njets
var_met_1[0] = fT.tma_met
var_TTbarPt_1[0] = fT.klf_orig_ttbar_pt
var_mwt_1[0] = fT.tma_mtw
var_DRwjets_1[0] = fT.klf_orig_dR_qq_W
var_DRbjets_1[0] = fT.klf_orig_dR_bb
var_LL_2[0] = fT.klfitter_logLikelihood[0]
var_EvtProb_2[0] = fT.klfitter_eventProbability[0]
var_WhadPt_2[0] = fT.klf_orig_Whad_pt
var_WlepPt_2[0] = fT.klf_orig_Wlep_pt
var_ThadPt_2[0] = fT.klf_orig_tophad_pt
var_TlepPt_2[0] = fT.klf_orig_toplep_pt
var_j_n_2[0] = fT.tma_njets
var_met_2[0] = fT.tma_met
var_TTbarPt_2[0] = fT.klf_orig_ttbar_pt
var_mwt_2[0] = fT.tma_mtw
var_DRwjets_2[0] = fT.klf_orig_dR_qq_W
var_DRbjets_2[0] = fT.klf_orig_dR_bb
d_noWC[0] = reader1.EvaluateMVA("BDT")
d_withWC[0] = reader2.EvaluateMVA("BDT")
#print d_noWC[0]," ",d_withWC[0]
fT_out.Fill()
fT_out.Write()
fF_out.Close()