def getCleanedShape(self, sample):
subtract = copy.copy(self.histograms.keys())
subtract.remove(sample)
subtract.remove('data')
hist = hu.clean_control_region(self.histograms,
data_label='data',
subtract=subtract,
fix_to_zero=True)
return hist
def plot_fit_variable( histograms, fit_variable, variable, bin_range,
fit_variable_distribution, qcd_fit_variable_distribution,
title, save_path ):
global fit_variable_properties, b_tag_bin, save_as, b_tag_bin_ctl
mc_uncertainty = 0.10
prepare_histograms( histograms, rebin = fit_variable_properties[fit_variable]['rebin'], scale_factor = measurement_config.luminosity_scale )
histogram_properties = Histogram_properties()
histogram_properties.x_axis_title = fit_variable_properties[fit_variable]['x-title']
histogram_properties.y_axis_title = fit_variable_properties[fit_variable]['y-title']
histogram_properties.x_limits = [fit_variable_properties[fit_variable]['min'], fit_variable_properties[fit_variable]['max']]
histogram_lables = ['data', 'QCD', 'V+Jets', 'Single-Top', samples_latex['TTJet']]
histogram_colors = ['black', 'yellow', 'green', 'magenta', 'red']
# qcd_from_data = histograms['data'][qcd_fit_variable_distribution].Clone()
# clean against other processes
histograms_for_cleaning = {'data':histograms['data'][qcd_fit_variable_distribution],
'V+Jets':histograms['V+Jets'][qcd_fit_variable_distribution],
'SingleTop':histograms['SingleTop'][qcd_fit_variable_distribution],
'TTJet':histograms['TTJet'][qcd_fit_variable_distribution]}
qcd_from_data = clean_control_region( histograms_for_cleaning, subtract = ['TTJet', 'V+Jets', 'SingleTop'] )
histograms_to_draw = [histograms['data'][qcd_fit_variable_distribution],
histograms['QCD'][qcd_fit_variable_distribution],
histograms['V+Jets'][qcd_fit_variable_distribution],
histograms['SingleTop'][qcd_fit_variable_distribution],
histograms['TTJet'][qcd_fit_variable_distribution]]
histogram_properties.title = title + ', ' + b_tag_bins_latex[b_tag_bin_ctl]
histogram_properties.name = variable + '_' + bin_range + '_' + fit_variable + '_%s_QCDConversions' % b_tag_bin_ctl
make_data_mc_comparison_plot( histograms_to_draw, histogram_lables, histogram_colors,
histogram_properties,
save_folder = save_path + '/qcd/',
show_ratio = False,
save_as = save_as,
)
histograms_to_draw = [qcd_from_data,
histograms['QCD'][qcd_fit_variable_distribution],
]
histogram_properties.name = variable + '_' + bin_range + '_' + fit_variable + '_%s_QCDConversions_subtracted' % b_tag_bin_ctl
make_data_mc_comparison_plot( histograms_to_draw,
histogram_lables = ['data', 'QCD'],
histogram_colors = ['black', 'yellow'],
histogram_properties = histogram_properties,
save_folder = save_path + '/qcd/',
show_ratio = False,
save_as = save_as,
)
# scale QCD to predicted
n_qcd_predicted_mc = histograms['QCD'][fit_variable_distribution].Integral()
n_qcd_fit_variable_distribution = qcd_from_data.Integral()
if not n_qcd_fit_variable_distribution == 0:
qcd_from_data.Scale( 1.0 / n_qcd_fit_variable_distribution * n_qcd_predicted_mc )
histograms_to_draw = [histograms['data'][fit_variable_distribution], qcd_from_data,
histograms['V+Jets'][fit_variable_distribution],
histograms['SingleTop'][fit_variable_distribution], histograms['TTJet'][fit_variable_distribution]]
histogram_properties.title = title + ', ' + b_tag_bins_latex[b_tag_bin]
histogram_properties.name = variable + '_' + bin_range + '_' + fit_variable + '_' + b_tag_bin
make_data_mc_comparison_plot( histograms_to_draw,
histogram_lables,
histogram_colors,
histogram_properties,
save_folder = save_path,
show_ratio = False,
save_as = save_as,
)
histogram_properties.mc_error = mc_uncertainty
histogram_properties.mc_errors_label = '$\mathrm{t}\\bar{\mathrm{t}}$ uncertainty'
histogram_properties.name = variable + '_' + bin_range + '_' + fit_variable + '_' + b_tag_bin + '_templates'
# change histogram order for better visibility
histograms_to_draw = [histograms['TTJet'][fit_variable_distribution] + histograms['SingleTop'][fit_variable_distribution],
histograms['TTJet'][fit_variable_distribution],
histograms['SingleTop'][fit_variable_distribution],
histograms['V+Jets'][fit_variable_distribution],
qcd_from_data]
histogram_lables = ['QCD', 'V+Jets', 'Single-Top', samples_latex['TTJet'], samples_latex['TTJet'] + ' + ' + 'Single-Top']
histogram_lables.reverse()
# change QCD color to orange for better visibility
histogram_colors = ['orange', 'green', 'magenta', 'red', 'black']
histogram_colors.reverse()
make_shape_comparison_plot( shapes = histograms_to_draw,
names = histogram_lables,
colours = histogram_colors,
histogram_properties = histogram_properties,
fill_area = False,
alpha = 1,
save_folder = save_path,
save_as = save_as,
)
def compare_qcd_control_regions( variable = 'MET', met_type = 'patType1CorrectedPFMet', title = 'Untitled'):
''' Compares the templates from the control regions in different bins
of the current variable'''
global fit_variable_properties, b_tag_bin, save_as, b_tag_bin_ctl
variable_bins = variable_bins_ROOT[variable]
histogram_template = get_histogram_template( variable )
for fit_variable in electron_fit_variables:
all_hists = {}
inclusive_hist = None
if '_bl' in fit_variable:
b_tag_bin_ctl = '1orMoreBtag'
else:
b_tag_bin_ctl = '0orMoreBtag'
save_path = 'plots/fit_variables/%dTeV/%s/%s/' % (measurement_config.centre_of_mass_energy, variable, fit_variable)
make_folder_if_not_exists(save_path + '/qcd/')
max_bins = 3
for bin_range in variable_bins[0:max_bins]:
params = {'met_type': met_type, 'bin_range':bin_range, 'fit_variable':fit_variable, 'b_tag_bin':b_tag_bin, 'variable':variable}
fit_variable_distribution = histogram_template % params
qcd_fit_variable_distribution = fit_variable_distribution.replace( 'Ref selection', 'QCDConversions' )
qcd_fit_variable_distribution = qcd_fit_variable_distribution.replace( b_tag_bin, b_tag_bin_ctl )
# format: histograms['data'][qcd_fit_variable_distribution]
histograms = get_histograms_from_files( [qcd_fit_variable_distribution], histogram_files )
prepare_histograms( histograms, rebin = fit_variable_properties[fit_variable]['rebin'], scale_factor = measurement_config.luminosity_scale )
histograms_for_cleaning = {'data':histograms['data'][qcd_fit_variable_distribution],
'V+Jets':histograms['V+Jets'][qcd_fit_variable_distribution],
'SingleTop':histograms['SingleTop'][qcd_fit_variable_distribution],
'TTJet':histograms['TTJet'][qcd_fit_variable_distribution]}
qcd_from_data = clean_control_region( histograms_for_cleaning, subtract = ['TTJet', 'V+Jets', 'SingleTop'] )
# clean
all_hists[bin_range] = qcd_from_data
# create the inclusive distributions
inclusive_hist = deepcopy(all_hists[variable_bins[0]])
for bin_range in variable_bins[1:max_bins]:
inclusive_hist += all_hists[bin_range]
for bin_range in variable_bins[0:max_bins]:
if not all_hists[bin_range].Integral() == 0:
all_hists[bin_range].Scale(1/all_hists[bin_range].Integral())
# normalise all histograms
inclusive_hist.Scale(1/inclusive_hist.Integral())
# now compare inclusive to all bins
histogram_properties = Histogram_properties()
histogram_properties.x_axis_title = fit_variable_properties[fit_variable]['x-title']
histogram_properties.y_axis_title = fit_variable_properties[fit_variable]['y-title']
histogram_properties.y_axis_title = histogram_properties.y_axis_title.replace('Events', 'a.u.')
histogram_properties.x_limits = [fit_variable_properties[fit_variable]['min'], fit_variable_properties[fit_variable]['max']]
# histogram_properties.y_limits = [0, 0.5]
histogram_properties.title = title + ', ' + b_tag_bins_latex[b_tag_bin_ctl]
histogram_properties.name = variable + '_' + fit_variable + '_' + b_tag_bin_ctl + '_QCD_template_comparison'
measurements = {bin_range + ' GeV': histogram for bin_range, histogram in all_hists.iteritems()}
measurements = OrderedDict(sorted(measurements.items()))
compare_measurements(models = {'inclusive' : inclusive_hist},
measurements = measurements,
show_measurement_errors = True,
histogram_properties = histogram_properties,
save_folder = save_path + '/qcd/',
save_as = save_as)
def make_plot( channel, x_axis_title, y_axis_title,
signal_region_tree,
control_region_tree,
branchName,
name_prefix, x_limits, nBins,
use_qcd_data_region = False,
compare_qcd_signal_with_data_control = False,
y_limits = [],
y_max_scale = 1.3,
rebin = 1,
legend_location = ( 0.98, 0.78 ), cms_logo_location = 'right',
log_y = False,
legend_color = False,
ratio_y_limits = [0.3, 2.5],
normalise = False,
):
global output_folder, measurement_config, category, normalise_to_fit
global preliminary, norm_variable, sum_bins, b_tag_bin, histogram_files
controlToCompare = []
if 'electron' in channel :
controlToCompare = ['QCDConversions', 'QCD non iso e+jets']
elif 'muon' in channel :
controlToCompare = ['QCD iso > 0.3', 'QCD 0.12 < iso <= 0.3']
histogramsToCompare = {}
for qcd_data_region in controlToCompare:
print 'Doing ',qcd_data_region
# Input files, normalisations, tree/region names
title = title_template % ( measurement_config.new_luminosity, measurement_config.centre_of_mass_energy )
normalisation = None
weightBranchSignalRegion = 'EventWeight'
if 'electron' in channel:
histogram_files['data'] = measurement_config.data_file_electron_trees
histogram_files['QCD'] = measurement_config.electron_QCD_MC_category_templates_trees[category]
if normalise_to_fit:
normalisation = normalisations_electron[norm_variable]
# if use_qcd_data_region:
# qcd_data_region = 'QCDConversions'
# # qcd_data_region = 'QCD non iso e+jets'
if not 'QCD' in channel and not 'NPU' in branchName:
weightBranchSignalRegion += ' * ElectronEfficiencyCorrection'
if 'muon' in channel:
histogram_files['data'] = measurement_config.data_file_muon_trees
histogram_files['QCD'] = measurement_config.muon_QCD_MC_category_templates_trees[category]
if normalise_to_fit:
normalisation = normalisations_muon[norm_variable]
# if use_qcd_data_region:
# qcd_data_region = 'QCD iso > 0.3'
if not 'QCD' in channel and not 'NPU' in branchName:
weightBranchSignalRegion += ' * MuonEfficiencyCorrection'
if not "_NPUNoWeight" in name_prefix:
weightBranchSignalRegion += ' * PUWeight'
if not "_NBJetsNoWeight" in name_prefix:
weightBranchSignalRegion += ' * BJetWeight'
selection = '1'
if branchName == 'abs(lepton_eta)' :
selection = 'lepton_eta > -10'
else:
selection = '%s >= 0' % branchName
# if 'QCDConversions' in signal_region_tree:
# selection += '&& isTightElectron'
# print selection
histograms = get_histograms_from_trees( trees = [signal_region_tree, control_region_tree], branch = branchName, weightBranch = weightBranchSignalRegion, files = histogram_files, nBins = nBins, xMin = x_limits[0], xMax = x_limits[-1], selection = selection )
histograms_QCDControlRegion = None
if use_qcd_data_region:
qcd_control_region = signal_region_tree.replace( 'Ref selection', qcd_data_region )
histograms_QCDControlRegion = get_histograms_from_trees( trees = [qcd_control_region], branch = branchName, weightBranch = 'EventWeight', files = histogram_files, nBins = nBins, xMin = x_limits[0], xMax = x_limits[-1], selection = selection )
# Split histograms up into signal/control (?)
signal_region_hists = {}
control_region_hists = {}
for sample in histograms.keys():
signal_region_hists[sample] = histograms[sample][signal_region_tree]
if compare_qcd_signal_with_data_control:
if sample is 'data':
signal_region_hists[sample] = histograms[sample][control_region_tree]
elif sample is 'QCD' :
signal_region_hists[sample] = histograms[sample][signal_region_tree]
else:
del signal_region_hists[sample]
if use_qcd_data_region:
control_region_hists[sample] = histograms_QCDControlRegion[sample][qcd_control_region]
# Prepare histograms
if normalise_to_fit:
# only scale signal region to fit (results are invalid for control region)
prepare_histograms( signal_region_hists, rebin = rebin,
scale_factor = measurement_config.luminosity_scale,
normalisation = normalisation )
elif normalise_to_data:
totalMC = 0
for sample in signal_region_hists:
if sample is 'data' : continue
totalMC += signal_region_hists[sample].Integral()
newScale = signal_region_hists['data'].Integral() / totalMC
prepare_histograms( signal_region_hists, rebin = rebin,
scale_factor = newScale,
)
else:
print measurement_config.luminosity_scale
prepare_histograms( signal_region_hists, rebin = rebin,
scale_factor = measurement_config.luminosity_scale )
prepare_histograms( control_region_hists, rebin = rebin,
scale_factor = measurement_config.luminosity_scale )
# Use qcd from data control region or not
qcd_from_data = None
if use_qcd_data_region:
qcd_from_data = clean_control_region( control_region_hists,
subtract = ['TTJet', 'V+Jets', 'SingleTop'] )
# Normalise control region correctly
nBins = signal_region_hists['QCD'].GetNbinsX()
n, error = signal_region_hists['QCD'].integral(0,nBins+1,error=True)
n_qcd_predicted_mc_signal = ufloat( n, error)
n, error = control_region_hists['QCD'].integral(0,nBins+1,error=True)
n_qcd_predicted_mc_control = ufloat( n, error)
n, error = qcd_from_data.integral(0,nBins+1,error=True)
n_qcd_control_region = ufloat( n, error)
if not n_qcd_control_region == 0:
dataDrivenQCDScale = n_qcd_predicted_mc_signal / n_qcd_predicted_mc_control
print 'Overall scale : ',dataDrivenQCDScale
qcd_from_data.Scale( dataDrivenQCDScale.nominal_value )
signalToControlScale = n_qcd_predicted_mc_signal / n_qcd_control_region
dataToMCscale = n_qcd_control_region / n_qcd_predicted_mc_control
print "Signal to control :",signalToControlScale
print "QCD scale : ",dataToMCscale
else:
qcd_from_data = signal_region_hists['QCD']
# Which histograms to draw, and properties
histograms_to_draw = []
histogram_lables = []
histogram_colors = []
if compare_qcd_signal_with_data_control :
histograms_to_draw = [signal_region_hists['data'], qcd_from_data ]
histogram_lables = ['data', 'QCD']
histogram_colors = ['black', 'yellow']
else :
histograms_to_draw = [signal_region_hists['data'], qcd_from_data,
signal_region_hists['V+Jets'],
signal_region_hists['SingleTop'],
signal_region_hists['TTJet']]
histogram_lables = ['data', 'QCD', 'V+Jets', 'Single-Top', samples_latex['TTJet']]
histogram_colors = [colours['data'], colours['QCD'], colours['V+Jets'], colours['Single-Top'], colours['TTJet'] ]
print list(qcd_from_data.y())
histogramsToCompare[qcd_data_region] = qcd_from_data
print histogramsToCompare
histogram_properties = Histogram_properties()
histogram_properties.name = 'QCD_control_region_comparison_' + channel + '_' + branchName
histogram_properties.title = title
histogram_properties.x_axis_title = x_axis_title
histogram_properties.y_axis_title = y_axis_title
histogram_properties.x_limits = x_limits
histogram_properties.y_limits = y_limits
histogram_properties.mc_error = 0.0
histogram_properties.legend_location = ( 0.98, 0.78 )
histogram_properties.ratio_y_limits = ratio_y_limits
if 'electron' in channel:
make_control_region_comparison(histogramsToCompare['QCDConversions'], histogramsToCompare['QCD non iso e+jets'],
name_region_1='Conversions', name_region_2='Non Iso',
histogram_properties=histogram_properties, save_folder=output_folder)
elif 'muon' in channel:
make_control_region_comparison(histogramsToCompare['QCD iso > 0.3'], histogramsToCompare['QCD 0.12 < iso <= 0.3'],
name_region_1='QCD iso > 0.3', name_region_2='QCD 0.12 < iso <= 0.3',
histogram_properties=histogram_properties, save_folder=output_folder)
def get_qcd_histograms( input_files, variable, variable_bin, channel,
fit_variable_hist_name, rebin = 1 ):
'''
Retrieves the data-driven QCD template and normalises it to MC prediction.
It uses the inclusive template (across all variable bins) and removes other processes
before normalising the QCD template.
'''
global electron_QCD_MC_file, muon_QCD_MC_file, analysis_type, \
electron_control_region, muon_control_region, b_tag_bin
control_region = ''
control_region_btag = '0btag'
if 'M_bl' in fit_variable_hist_name or 'angle_bl' in fit_variable_hist_name:
control_region_btag = '1orMoreBtag'
qcd_file = ''
samples = ['data', 'V+Jets', 'SingleTop', 'TTJet']
if channel == 'electron':
control_region = electron_control_region
qcd_file = electron_QCD_MC_file
if channel == 'muon':
control_region = muon_control_region
qcd_file = muon_QCD_MC_file
inclusive_control_region_hists = {}
for var_bin in variable_bins_ROOT[variable]:
hist_name = fit_variable_hist_name.replace( variable_bin, var_bin )
control_region_hist_name = hist_name.replace( 'Ref selection',
control_region )
for sample in samples:
if not inclusive_control_region_hists.has_key( sample ):
inclusive_control_region_hists[sample] = get_histogram(
input_files[sample],
control_region_hist_name,
control_region_btag,
)
else:
inclusive_control_region_hists[sample] += get_histogram(
input_files[sample],
control_region_hist_name,
control_region_btag,
)
for sample in samples:
inclusive_control_region_hists[sample].Rebin( rebin )
inclusive_control_region_hists['data'] = clean_control_region( inclusive_control_region_hists,
subtract = ['TTJet', 'V+Jets', 'SingleTop'] )
# now apply proper normalisation
QCD_normalisation_factor = 1
signal_region_mc = get_histogram( qcd_file, fit_variable_hist_name, b_tag_bin )
n_mc = signal_region_mc.Integral()
n_control = inclusive_control_region_hists['data'].Integral()
if not n_control == 0: # scale to MC prediction
if not n_mc == 0:
QCD_normalisation_factor = 1 / n_control * n_mc
else:
QCD_normalisation_factor = 1 / n_control
inclusive_control_region_hists['data'].Scale( QCD_normalisation_factor )
return inclusive_control_region_hists['data']
def compare_qcd_control_regions(variable='MET',
met_type='patType1CorrectedPFMet',
title='Untitled',
channel='electron'):
''' Compares the templates from the control regions in different bins
of the current variable'''
global fit_variable_properties, b_tag_bin, save_as, b_tag_bin_ctl
variable_bins = variable_bins_ROOT[variable]
histogram_template = get_histogram_template(variable)
for fit_variable in electron_fit_variables:
all_hists = {}
inclusive_hist = None
if '_bl' in fit_variable:
b_tag_bin_ctl = '1orMoreBtag'
else:
b_tag_bin_ctl = '0orMoreBtag'
save_path = 'plots/%dTeV/fit_variables/%s/%s/' % (
measurement_config.centre_of_mass_energy, variable, fit_variable)
make_folder_if_not_exists(save_path + '/qcd/')
max_bins = 3
for bin_range in variable_bins[0:max_bins]:
params = {
'met_type': met_type,
'bin_range': bin_range,
'fit_variable': fit_variable,
'b_tag_bin': b_tag_bin,
'variable': variable
}
fit_variable_distribution = histogram_template % params
qcd_fit_variable_distribution = fit_variable_distribution.replace(
'Ref selection', 'QCDConversions')
qcd_fit_variable_distribution = qcd_fit_variable_distribution.replace(
b_tag_bin, b_tag_bin_ctl)
# format: histograms['data'][qcd_fit_variable_distribution]
histograms = get_histograms_from_files(
[qcd_fit_variable_distribution], histogram_files)
prepare_histograms(
histograms,
rebin=fit_variable_properties[fit_variable]['rebin'],
scale_factor=measurement_config.luminosity_scale)
histograms_for_cleaning = {
'data': histograms['data'][qcd_fit_variable_distribution],
'V+Jets': histograms['V+Jets'][qcd_fit_variable_distribution],
'SingleTop':
histograms['SingleTop'][qcd_fit_variable_distribution],
'TTJet': histograms['TTJet'][qcd_fit_variable_distribution]
}
qcd_from_data = clean_control_region(
histograms_for_cleaning,
subtract=['TTJet', 'V+Jets', 'SingleTop'])
# clean
all_hists[bin_range] = qcd_from_data
# create the inclusive distributions
inclusive_hist = deepcopy(all_hists[variable_bins[0]])
for bin_range in variable_bins[1:max_bins]:
inclusive_hist += all_hists[bin_range]
for bin_range in variable_bins[0:max_bins]:
if not all_hists[bin_range].Integral() == 0:
all_hists[bin_range].Scale(1 / all_hists[bin_range].Integral())
# normalise all histograms
inclusive_hist.Scale(1 / inclusive_hist.Integral())
# now compare inclusive to all bins
histogram_properties = Histogram_properties()
histogram_properties.x_axis_title = fit_variable_properties[
fit_variable]['x-title']
histogram_properties.y_axis_title = fit_variable_properties[
fit_variable]['y-title']
histogram_properties.y_axis_title = histogram_properties.y_axis_title.replace(
'Events', 'a.u.')
histogram_properties.x_limits = [
fit_variable_properties[fit_variable]['min'],
fit_variable_properties[fit_variable]['max']
]
# histogram_properties.y_limits = [0, 0.5]
histogram_properties.title = title
histogram_properties.additional_text = channel_latex[
channel] + ', ' + b_tag_bins_latex[b_tag_bin_ctl]
histogram_properties.name = variable + '_' + fit_variable + '_' + b_tag_bin_ctl + '_QCD_template_comparison'
histogram_properties.y_max_scale = 1.5
measurements = {
bin_range + ' GeV': histogram
for bin_range, histogram in all_hists.iteritems()
}
measurements = OrderedDict(sorted(measurements.items()))
compare_measurements(models={'inclusive': inclusive_hist},
measurements=measurements,
show_measurement_errors=True,
histogram_properties=histogram_properties,
save_folder=save_path + '/qcd/',
save_as=save_as)
def plot_fit_variable(histograms,
fit_variable,
variable,
bin_range,
fit_variable_distribution,
qcd_fit_variable_distribution,
title,
save_path,
channel='electron'):
global fit_variable_properties, b_tag_bin, save_as, b_tag_bin_ctl
histograms_ = deepcopy(histograms)
mc_uncertainty = 0.10
prepare_histograms(histograms_,
rebin=fit_variable_properties[fit_variable]['rebin'],
scale_factor=measurement_config.luminosity_scale)
######################################
# plot the control regions as they are
######################################
histogram_properties = Histogram_properties()
histogram_properties.x_axis_title = fit_variable_properties[fit_variable][
'x-title']
histogram_properties.y_axis_title = fit_variable_properties[fit_variable][
'y-title']
histogram_properties.x_limits = [
fit_variable_properties[fit_variable]['min'],
fit_variable_properties[fit_variable]['max']
]
histogram_properties.y_max_scale = 2
histogram_lables = [
'data', 'QCD', 'V+Jets', 'Single-Top', samples_latex['TTJet']
]
histogram_colors = ['black', 'yellow', 'green', 'magenta', 'red']
# qcd_from_data = histograms_['data'][qcd_fit_variable_distribution].Clone()
# clean against other processes
histograms_for_cleaning = {
'data': histograms_['data'][qcd_fit_variable_distribution],
'V+Jets': histograms_['V+Jets'][qcd_fit_variable_distribution],
'SingleTop': histograms_['SingleTop'][qcd_fit_variable_distribution],
'TTJet': histograms_['TTJet'][qcd_fit_variable_distribution]
}
qcd_from_data = clean_control_region(
histograms_for_cleaning, subtract=['TTJet', 'V+Jets', 'SingleTop'])
histograms_to_draw = [
histograms_['data'][qcd_fit_variable_distribution],
histograms_['QCD'][qcd_fit_variable_distribution],
histograms_['V+Jets'][qcd_fit_variable_distribution],
histograms_['SingleTop'][qcd_fit_variable_distribution],
histograms_['TTJet'][qcd_fit_variable_distribution]
]
histogram_properties.title = title
histogram_properties.additional_text = channel_latex[
channel] + ', ' + b_tag_bins_latex[b_tag_bin_ctl]
histogram_properties.name = variable + '_' + bin_range + '_' + fit_variable + '_%s_QCDConversions' % b_tag_bin_ctl
make_data_mc_comparison_plot(
histograms_to_draw,
histogram_lables,
histogram_colors,
histogram_properties,
save_folder=save_path + '/qcd/',
show_ratio=False,
save_as=save_as,
)
######################################
# plot QCD against data control region with TTJet, SingleTop and V+Jets removed
######################################
histograms_to_draw = [
qcd_from_data,
histograms_['QCD'][qcd_fit_variable_distribution],
]
histogram_properties.y_max_scale = 1.5
histogram_properties.name = variable + '_' + bin_range + '_' + fit_variable + '_%s_QCDConversions_subtracted' % b_tag_bin_ctl
make_data_mc_comparison_plot(
histograms_to_draw,
histogram_lables=['data', 'QCD'],
histogram_colors=['black', 'yellow'],
histogram_properties=histogram_properties,
save_folder=save_path + '/qcd/',
show_ratio=False,
save_as=save_as,
)
######################################
# plot signal region
######################################
# scale QCD to predicted
n_qcd_predicted_mc = histograms_['QCD'][
fit_variable_distribution].Integral()
n_qcd_fit_variable_distribution = qcd_from_data.Integral()
if not n_qcd_fit_variable_distribution == 0:
qcd_from_data.Scale(1.0 / n_qcd_fit_variable_distribution *
n_qcd_predicted_mc)
histograms_to_draw = [
histograms_['data'][fit_variable_distribution], qcd_from_data,
histograms_['V+Jets'][fit_variable_distribution],
histograms_['SingleTop'][fit_variable_distribution],
histograms_['TTJet'][fit_variable_distribution]
]
histogram_properties.additional_text = channel_latex[
channel] + ', ' + b_tag_bins_latex[b_tag_bin]
histogram_properties.name = variable + '_' + bin_range + '_' + fit_variable + '_' + b_tag_bin
make_data_mc_comparison_plot(
histograms_to_draw,
histogram_lables,
histogram_colors,
histogram_properties,
save_folder=save_path,
show_ratio=False,
save_as=save_as,
)
######################################
# plot templates
######################################
histogram_properties.mc_error = mc_uncertainty
histogram_properties.mc_errors_label = '$\mathrm{t}\\bar{\mathrm{t}}$ uncertainty'
histogram_properties.name = variable + '_' + bin_range + '_' + fit_variable + '_' + b_tag_bin + '_templates'
histogram_properties.y_max_scale = 2
# change histogram order for better visibility
histograms_to_draw = [
histograms_['TTJet'][fit_variable_distribution] +
histograms_['SingleTop'][fit_variable_distribution],
histograms_['TTJet'][fit_variable_distribution],
histograms_['SingleTop'][fit_variable_distribution],
histograms_['V+Jets'][fit_variable_distribution], qcd_from_data
]
histogram_lables = [
'QCD', 'V+Jets', 'Single-Top', samples_latex['TTJet'],
samples_latex['TTJet'] + ' + ' + 'Single-Top'
]
histogram_lables.reverse()
# change QCD color to orange for better visibility
histogram_colors = ['orange', 'green', 'magenta', 'red', 'black']
histogram_colors.reverse()
# plot template
make_shape_comparison_plot(
shapes=histograms_to_draw,
names=histogram_lables,
colours=histogram_colors,
histogram_properties=histogram_properties,
fill_area=False,
alpha=1,
save_folder=save_path,
save_as=save_as,
)
def make_plot(
channel,
x_axis_title,
y_axis_title,
signal_region_tree,
control_region_tree,
branchName,
name_prefix,
x_limits,
nBins,
use_qcd_data_region=False,
compare_qcd_signal_with_data_control=False,
y_limits=[],
y_max_scale=1.3,
rebin=1,
legend_location=(0.98, 0.78),
cms_logo_location='right',
log_y=False,
legend_color=False,
ratio_y_limits=[0.3, 2.5],
normalise=False,
):
global output_folder, measurement_config, category, normalise_to_fit
global preliminary, norm_variable, sum_bins, b_tag_bin, histogram_files
controlToCompare = []
if 'electron' in channel:
controlToCompare = ['QCDConversions', 'QCD non iso e+jets']
elif 'muon' in channel:
controlToCompare = ['QCD iso > 0.3', 'QCD 0.12 < iso <= 0.3']
histogramsToCompare = {}
for qcd_data_region in controlToCompare:
print 'Doing ', qcd_data_region
# Input files, normalisations, tree/region names
title = title_template % (measurement_config.new_luminosity,
measurement_config.centre_of_mass_energy)
normalisation = None
weightBranchSignalRegion = 'EventWeight'
if 'electron' in channel:
histogram_files[
'data'] = measurement_config.data_file_electron_trees
histogram_files[
'QCD'] = measurement_config.electron_QCD_MC_category_templates_trees[
category]
if normalise_to_fit:
normalisation = normalisations_electron[norm_variable]
# if use_qcd_data_region:
# qcd_data_region = 'QCDConversions'
# # qcd_data_region = 'QCD non iso e+jets'
if not 'QCD' in channel and not 'NPU' in branchName:
weightBranchSignalRegion += ' * ElectronEfficiencyCorrection'
if 'muon' in channel:
histogram_files['data'] = measurement_config.data_file_muon_trees
histogram_files[
'QCD'] = measurement_config.muon_QCD_MC_category_templates_trees[
category]
if normalise_to_fit:
normalisation = normalisations_muon[norm_variable]
# if use_qcd_data_region:
# qcd_data_region = 'QCD iso > 0.3'
if not 'QCD' in channel and not 'NPU' in branchName:
weightBranchSignalRegion += ' * MuonEfficiencyCorrection'
if not "_NPUNoWeight" in name_prefix:
weightBranchSignalRegion += ' * PUWeight'
if not "_NBJetsNoWeight" in name_prefix:
weightBranchSignalRegion += ' * BJetWeight'
selection = '1'
if branchName == 'abs(lepton_eta)':
selection = 'lepton_eta > -10'
else:
selection = '%s >= 0' % branchName
# if 'QCDConversions' in signal_region_tree:
# selection += '&& isTightElectron'
# print selection
histograms = get_histograms_from_trees(
trees=[signal_region_tree, control_region_tree],
branch=branchName,
weightBranch=weightBranchSignalRegion,
files=histogram_files,
nBins=nBins,
xMin=x_limits[0],
xMax=x_limits[-1],
selection=selection)
histograms_QCDControlRegion = None
if use_qcd_data_region:
qcd_control_region = signal_region_tree.replace(
'Ref selection', qcd_data_region)
histograms_QCDControlRegion = get_histograms_from_trees(
trees=[qcd_control_region],
branch=branchName,
weightBranch='EventWeight',
files=histogram_files,
nBins=nBins,
xMin=x_limits[0],
xMax=x_limits[-1],
selection=selection)
# Split histograms up into signal/control (?)
signal_region_hists = {}
control_region_hists = {}
for sample in histograms.keys():
signal_region_hists[sample] = histograms[sample][
signal_region_tree]
if compare_qcd_signal_with_data_control:
if sample is 'data':
signal_region_hists[sample] = histograms[sample][
control_region_tree]
elif sample is 'QCD':
signal_region_hists[sample] = histograms[sample][
signal_region_tree]
else:
del signal_region_hists[sample]
if use_qcd_data_region:
control_region_hists[sample] = histograms_QCDControlRegion[
sample][qcd_control_region]
# Prepare histograms
if normalise_to_fit:
# only scale signal region to fit (results are invalid for control region)
prepare_histograms(
signal_region_hists,
rebin=rebin,
scale_factor=measurement_config.luminosity_scale,
normalisation=normalisation)
elif normalise_to_data:
totalMC = 0
for sample in signal_region_hists:
if sample is 'data': continue
totalMC += signal_region_hists[sample].Integral()
newScale = signal_region_hists['data'].Integral() / totalMC
prepare_histograms(
signal_region_hists,
rebin=rebin,
scale_factor=newScale,
)
else:
print measurement_config.luminosity_scale
prepare_histograms(
signal_region_hists,
rebin=rebin,
scale_factor=measurement_config.luminosity_scale)
prepare_histograms(
control_region_hists,
rebin=rebin,
scale_factor=measurement_config.luminosity_scale)
# Use qcd from data control region or not
qcd_from_data = None
if use_qcd_data_region:
qcd_from_data = clean_control_region(
control_region_hists,
subtract=['TTJet', 'V+Jets', 'SingleTop'])
# Normalise control region correctly
nBins = signal_region_hists['QCD'].GetNbinsX()
n, error = signal_region_hists['QCD'].integral(0,
nBins + 1,
error=True)
n_qcd_predicted_mc_signal = ufloat(n, error)
n, error = control_region_hists['QCD'].integral(0,
nBins + 1,
error=True)
n_qcd_predicted_mc_control = ufloat(n, error)
n, error = qcd_from_data.integral(0, nBins + 1, error=True)
n_qcd_control_region = ufloat(n, error)
if not n_qcd_control_region == 0:
dataDrivenQCDScale = n_qcd_predicted_mc_signal / n_qcd_predicted_mc_control
print 'Overall scale : ', dataDrivenQCDScale
qcd_from_data.Scale(dataDrivenQCDScale.nominal_value)
signalToControlScale = n_qcd_predicted_mc_signal / n_qcd_control_region
dataToMCscale = n_qcd_control_region / n_qcd_predicted_mc_control
print "Signal to control :", signalToControlScale
print "QCD scale : ", dataToMCscale
else:
qcd_from_data = signal_region_hists['QCD']
# Which histograms to draw, and properties
histograms_to_draw = []
histogram_lables = []
histogram_colors = []
if compare_qcd_signal_with_data_control:
histograms_to_draw = [signal_region_hists['data'], qcd_from_data]
histogram_lables = ['data', 'QCD']
histogram_colors = ['black', 'yellow']
else:
histograms_to_draw = [
signal_region_hists['data'], qcd_from_data,
signal_region_hists['V+Jets'],
signal_region_hists['SingleTop'], signal_region_hists['TTJet']
]
histogram_lables = [
'data', 'QCD', 'V+Jets', 'Single-Top', samples_latex['TTJet']
]
histogram_colors = [
colours['data'], colours['QCD'], colours['V+Jets'],
colours['Single-Top'], colours['TTJet']
]
print list(qcd_from_data.y())
histogramsToCompare[qcd_data_region] = qcd_from_data
print histogramsToCompare
histogram_properties = Histogram_properties()
histogram_properties.name = 'QCD_control_region_comparison_' + channel + '_' + branchName
histogram_properties.title = title
histogram_properties.x_axis_title = x_axis_title
histogram_properties.y_axis_title = y_axis_title
histogram_properties.x_limits = x_limits
histogram_properties.y_limits = y_limits
histogram_properties.mc_error = 0.0
histogram_properties.legend_location = (0.98, 0.78)
histogram_properties.ratio_y_limits = ratio_y_limits
if 'electron' in channel:
make_control_region_comparison(
histogramsToCompare['QCDConversions'],
histogramsToCompare['QCD non iso e+jets'],
name_region_1='Conversions',
name_region_2='Non Iso',
histogram_properties=histogram_properties,
save_folder=output_folder)
elif 'muon' in channel:
make_control_region_comparison(
histogramsToCompare['QCD iso > 0.3'],
histogramsToCompare['QCD 0.12 < iso <= 0.3'],
name_region_1='QCD iso > 0.3',
name_region_2='QCD 0.12 < iso <= 0.3',
histogram_properties=histogram_properties,
save_folder=output_folder)
def background_subtraction(self, histograms):
ttjet_hist = clean_control_region(histograms,
subtract=['QCD', 'V+Jets', 'SingleTop'])
self.normalisation[
'TTJet'] = hist_to_value_error_tuplelist(ttjet_hist)
def compare_QCD_control_regions_to_MC():
config = XSectionConfig(13)
ctrl_e1 = 'TTbar_plus_X_analysis/EPlusJets/QCDConversions/FitVariables'
ctrl_e2 = 'TTbar_plus_X_analysis/EPlusJets/QCD non iso e+jets/FitVariables'
mc_e = 'TTbar_plus_X_analysis/EPlusJets/Ref selection/FitVariables'
data_file_e = config.data_file_electron_trees
ttbar_file = config.ttbar_category_templates_trees['central']
vjets_file = config.VJets_category_templates_trees['central']
singleTop_file = config.SingleTop_category_templates_trees['central']
qcd_file_e = config.electron_QCD_MC_tree_file
ctrl_mu1 = 'TTbar_plus_X_analysis/MuPlusJets/QCD iso > 0.3/FitVariables'
ctrl_mu2 = 'TTbar_plus_X_analysis/MuPlusJets/QCD 0.12 < iso <= 0.3/FitVariables'
mc_mu = 'TTbar_plus_X_analysis/MuPlusJets/Ref selection/FitVariables'
data_file_mu = config.data_file_muon_trees
qcd_file_mu = config.muon_QCD_MC_tree_file
weight_branches_electron = [
"EventWeight", "PUWeight", "BJetWeight", "ElectronEfficiencyCorrection"
]
weight_branches_mu = [
"EventWeight", "PUWeight", "BJetWeight", "MuonEfficiencyCorrection"
]
variables = [
'MET', 'HT', 'ST', 'NJets', 'lepton_pt', 'abs_lepton_eta', 'WPT'
]
# variables = ['abs_lepton_eta']
for variable in variables:
branch = variable
selection = '{0} >= 0'.format(branch)
if variable == 'abs_lepton_eta':
branch = 'abs(lepton_eta)'
selection = 'lepton_eta >= -3'
for channel in ['electron', 'muon']:
data_file = data_file_e
qcd_file = qcd_file_e
ctrl1 = ctrl_e1
ctrl2 = ctrl_e2
mc = mc_e
weight_branches = weight_branches_electron
if channel == 'muon':
data_file = data_file_mu
qcd_file = qcd_file_mu
ctrl1 = ctrl_mu1
ctrl2 = ctrl_mu2
mc = mc_mu
weight_branches = weight_branches_mu
inputs = {
'branch': branch,
'weight_branches': weight_branches,
'tree': ctrl1,
'bin_edges': bin_edges_vis[variable],
'selection': selection,
}
hs_ctrl1 = {
'data':
get_histogram_from_tree(input_file=data_file, **inputs),
'TTJet':
get_histogram_from_tree(input_file=ttbar_file, **inputs),
'VJets':
get_histogram_from_tree(input_file=vjets_file, **inputs),
'SingleTop':
get_histogram_from_tree(input_file=singleTop_file, **inputs),
'QCD':
get_histogram_from_tree(input_file=qcd_file, **inputs),
}
inputs['tree'] = ctrl2
hs_ctrl2 = {
'data':
get_histogram_from_tree(input_file=data_file, **inputs),
'TTJet':
get_histogram_from_tree(input_file=ttbar_file, **inputs),
'VJets':
get_histogram_from_tree(input_file=vjets_file, **inputs),
'SingleTop':
get_histogram_from_tree(input_file=singleTop_file, **inputs),
'QCD':
get_histogram_from_tree(input_file=qcd_file, **inputs),
}
inputs['tree'] = mc
h_qcd = get_histogram_from_tree(input_file=qcd_file, **inputs)
h_ctrl1 = clean_control_region(
hs_ctrl1,
data_label='data',
subtract=['TTJet', 'VJets', 'SingleTop'],
fix_to_zero=True)
h_ctrl2 = clean_control_region(
hs_ctrl2,
data_label='data',
subtract=['TTJet', 'VJets', 'SingleTop'],
fix_to_zero=True)
n_qcd_ctrl1 = hs_ctrl1['QCD'].integral()
n_qcd_ctrl2 = hs_ctrl2['QCD'].integral()
n_data1 = h_ctrl1.integral()
n_data2 = h_ctrl2.integral()
n_qcd_sg = h_qcd.integral()
ratio_ctrl1 = n_data1 / n_qcd_ctrl1
ratio_ctrl2 = n_data2 / n_qcd_ctrl2
qcd_estimate_ctrl1 = n_qcd_sg * ratio_ctrl1
qcd_estimate_ctrl2 = n_qcd_sg * ratio_ctrl2
h_ctrl1.Scale(qcd_estimate_ctrl1 / n_data1)
h_ctrl2.Scale(qcd_estimate_ctrl2 / n_data2)
properties = Histogram_properties()
properties.name = 'compare_qcd_control_regions_to_mc_{0}_{1}_channel'.format(
variable, channel)
properties.title = 'Comparison of QCD control regions ({0} channel)'.format(
channel)
properties.path = 'plots'
properties.has_ratio = False
properties.xerr = True
properties.x_limits = (bin_edges_vis[variable][0],
bin_edges_vis[variable][-1])
properties.x_axis_title = variables_latex[variable]
properties.y_axis_title = 'number of QCD events'
histograms = {
'control region 1': h_ctrl1,
'control region 2': h_ctrl2,
'MC prediction': h_qcd
}
diff = absolute(h_ctrl1 - h_ctrl2)
lower = h_ctrl1 - diff
upper = h_ctrl1 + diff
err_e = ErrorBand('uncertainty', lower, upper)
plot_e = Plot(histograms, properties)
plot_e.draw_method = 'errorbar'
plot_e.add_error_band(err_e)
compare_histograms(plot_e)
def compare_QCD_control_regions_to_MC():
config = XSectionConfig(13)
ctrl_e1 = 'TTbar_plus_X_analysis/EPlusJets/QCDConversions/FitVariables'
ctrl_e2 = 'TTbar_plus_X_analysis/EPlusJets/QCD non iso e+jets/FitVariables'
mc_e = 'TTbar_plus_X_analysis/EPlusJets/Ref selection/FitVariables'
data_file_e = config.data_file_electron_trees
ttbar_file = config.ttbar_category_templates_trees['central']
vjets_file = config.VJets_category_templates_trees['central']
singleTop_file = config.SingleTop_category_templates_trees['central']
qcd_file_e = config.electron_QCD_MC_tree_file
ctrl_mu1 = 'TTbar_plus_X_analysis/MuPlusJets/QCD iso > 0.3/FitVariables'
ctrl_mu2 = 'TTbar_plus_X_analysis/MuPlusJets/QCD 0.12 < iso <= 0.3/FitVariables'
mc_mu = 'TTbar_plus_X_analysis/MuPlusJets/Ref selection/FitVariables'
data_file_mu = config.data_file_muon_trees
qcd_file_mu = config.muon_QCD_MC_tree_file
weight_branches_electron = [
"EventWeight",
"PUWeight",
"BJetWeight",
"ElectronEfficiencyCorrection"
]
weight_branches_mu = [
"EventWeight",
"PUWeight",
"BJetWeight",
"MuonEfficiencyCorrection"
]
variables = ['MET', 'HT', 'ST', 'NJets',
'lepton_pt', 'abs_lepton_eta', 'WPT']
# variables = ['abs_lepton_eta']
for variable in variables:
branch = variable
selection = '{0} >= 0'.format(branch)
if variable == 'abs_lepton_eta':
branch = 'abs(lepton_eta)'
selection = 'lepton_eta >= -3'
for channel in ['electron', 'muon']:
data_file = data_file_e
qcd_file = qcd_file_e
ctrl1 = ctrl_e1
ctrl2 = ctrl_e2
mc = mc_e
weight_branches = weight_branches_electron
if channel == 'muon':
data_file = data_file_mu
qcd_file = qcd_file_mu
ctrl1 = ctrl_mu1
ctrl2 = ctrl_mu2
mc = mc_mu
weight_branches = weight_branches_mu
inputs = {
'branch': branch,
'weight_branches': weight_branches,
'tree': ctrl1,
'bin_edges': bin_edges_vis[variable],
'selection': selection,
}
hs_ctrl1 = {
'data': get_histogram_from_tree(input_file=data_file, **inputs),
'TTJet': get_histogram_from_tree(input_file=ttbar_file, **inputs),
'VJets': get_histogram_from_tree(input_file=vjets_file, **inputs),
'SingleTop': get_histogram_from_tree(input_file=singleTop_file, **inputs),
'QCD': get_histogram_from_tree(input_file=qcd_file, **inputs),
}
inputs['tree'] = ctrl2
hs_ctrl2 = {
'data': get_histogram_from_tree(input_file=data_file, **inputs),
'TTJet': get_histogram_from_tree(input_file=ttbar_file, **inputs),
'VJets': get_histogram_from_tree(input_file=vjets_file, **inputs),
'SingleTop': get_histogram_from_tree(input_file=singleTop_file, **inputs),
'QCD': get_histogram_from_tree(input_file=qcd_file, **inputs),
}
inputs['tree'] = mc
h_qcd = get_histogram_from_tree(input_file=qcd_file, **inputs)
h_ctrl1 = clean_control_region(
hs_ctrl1,
data_label='data',
subtract=['TTJet', 'VJets', 'SingleTop'],
fix_to_zero=True)
h_ctrl2 = clean_control_region(
hs_ctrl2,
data_label='data',
subtract=['TTJet', 'VJets', 'SingleTop'],
fix_to_zero=True)
n_qcd_ctrl1 = hs_ctrl1['QCD'].integral()
n_qcd_ctrl2 = hs_ctrl2['QCD'].integral()
n_data1 = h_ctrl1.integral()
n_data2 = h_ctrl2.integral()
n_qcd_sg = h_qcd.integral()
ratio_ctrl1 = n_data1 / n_qcd_ctrl1
ratio_ctrl2 = n_data2 / n_qcd_ctrl2
qcd_estimate_ctrl1 = n_qcd_sg * ratio_ctrl1
qcd_estimate_ctrl2 = n_qcd_sg * ratio_ctrl2
h_ctrl1.Scale(qcd_estimate_ctrl1 / n_data1)
h_ctrl2.Scale(qcd_estimate_ctrl2 / n_data2)
properties = Histogram_properties()
properties.name = 'compare_qcd_control_regions_to_mc_{0}_{1}_channel'.format(
variable, channel)
properties.title = 'Comparison of QCD control regions ({0} channel)'.format(
channel)
properties.path = 'plots'
properties.has_ratio = False
properties.xerr = True
properties.x_limits = (
bin_edges_vis[variable][0], bin_edges_vis[variable][-1])
properties.x_axis_title = variables_latex[variable]
properties.y_axis_title = 'number of QCD events'
histograms = {'control region 1': h_ctrl1,
'control region 2': h_ctrl2,
'MC prediction': h_qcd}
diff = absolute(h_ctrl1 - h_ctrl2)
lower = h_ctrl1 - diff
upper = h_ctrl1 + diff
err_e = ErrorBand('uncertainty', lower, upper)
plot_e = Plot(histograms, properties)
plot_e.draw_method = 'errorbar'
plot_e.add_error_band(err_e)
compare_histograms(plot_e)
def background_subtraction(self, histograms):
ttjet_hist = clean_control_region(
histograms, subtract=['QCD', 'V+Jets', 'SingleTop'])
self.normalisation['TTJet'] = hist_to_value_error_tuplelist(ttjet_hist)
