def GetBJetPlusXHistograms(self):
this_analyses = ["trigbbh_CSVTM", "trigbbl_CSVTM", "trigbbll_CSVTM", "trigbbh_trigbbl_CSVTM"]
this_samples = ["BJetPlusX_2012", "BJetPlusX_2012BCD"]
for analysis in this_analyses:
for sample in this_samples:
name = analysis + "_" + sample
f = ROOT.TFile(analysis_config.get_b_histogram_filename(analysis, sample), "READ")
self._mjj_histograms_fine[name] = f.Get("BHistograms/h_pfjet_mjj")
self._mjj_histograms_fine[name].SetName("h_" + name + "_mjj_fine")
self._mjj_histograms_fine[name].SetDirectory(0)
self._mjj_histograms[name] = histogram_tools.rebin_histogram(self._mjj_histograms_fine[name], self._mass_bins, normalization_bin_width=1)
self._mjj_histograms[name].SetName("h_" + name + "_mjj")
self._mjj_histograms[name].SetDirectory(0)
self._mjj_histograms_csvorder[name] = f.Get("BHistograms/h_pfjet_mjj_csvorder")
self._mjj_histograms_csvorder[name].SetName("h_" + name + "_mjj_csvorder")
self._mjj_histograms_csvorder[name].SetDirectory(0)
self._mjj_histograms_csvorder[name] = histogram_tools.rebin_histogram(self._mjj_histograms_csvorder[name], self._mass_bins, normalization_bin_width=1)
self._mjj_histograms_vetothirdjet[name] = f.Get("BHistograms/h_pfjet_mjj_vetothirdjet")
self._mjj_histograms_vetothirdjet[name].SetName("h_" + name + "_mjj_vetothirdjet")
self._mjj_histograms_vetothirdjet[name].SetDirectory(0)
self._mjj_histograms_vetothirdjet[name] = histogram_tools.rebin_histogram(self._mjj_histograms_vetothirdjet[name], self._mass_bins, normalization_bin_width=1)
self._mjj_histograms_fine[name].Rebin(5)
self._analyses.append(name)
f.Close()
if "bbll" in analysis:
self._mjj_histograms[name].Scale(1.7)
self._mjj_histograms_fine[name].Scale(1.7)
def GetSingleMuHistograms(self):
#this_analyses = ["trigmu_highmass_CSVTM", "trigmu_lowmass_CSVTM", "trigmubbh_highmass_CSVTM", "trigmubbl_lowmass_CSVTM", "trigmubbll_lowmass_CSVTM", "trigmu24i_lowmass_CSVTM", "trigmu24ibbl_lowmass_CSVTM", "trigmu24ibbll_lowmass_CSVTM", "trigmu40_lowmass_CSVTM", "trigmu40bbl_lowmass_CSVTM", "trigmu40bbll_lowmass_CSVTM"]
this_analyses = ["trigmu24i_lowmass_CSVTM", "trigmu24ibbl_lowmass_CSVTM"]
this_analyses.extend(["trigmu24i_highmass_CSVTM", "trigmu24ibbh_highmass_CSVTM"])
for analysis in this_analyses:
print "Opening " + analysis_config.get_b_histogram_filename(analysis, "SingleMu_2012")
f = ROOT.TFile(analysis_config.get_b_histogram_filename(analysis, "SingleMu_2012"), "READ")
self._mjj_histograms_fine[analysis] = f.Get("BHistograms/h_pfjet_mjj")
self._mjj_histograms_fine[analysis].SetName("h_" + analysis + "_mjj_fine")
self._mjj_histograms_fine[analysis].SetDirectory(0)
self._mjj_histograms[analysis] = histogram_tools.rebin_histogram(self._mjj_histograms_fine[analysis], self._mass_bins, normalization_bin_width=1)
self._mjj_histograms[analysis].SetName("h_" + analysis + "_mjj")
self._mjj_histograms[analysis].SetDirectory(0)
if not "highmass" in analysis:
self._mjj_histograms_csvorder[analysis] = f.Get("BHistograms/h_pfjet_mjj_csvorder")
self._mjj_histograms_csvorder[analysis].SetName("h_" + analysis + "_mjj_csvorder")
self._mjj_histograms_csvorder[analysis].SetDirectory(0)
self._mjj_histograms_csvorder[analysis] = histogram_tools.rebin_histogram(self._mjj_histograms_csvorder[analysis], self._mass_bins, normalization_bin_width=1)
self._mjj_histograms_vetothirdjet[analysis] = f.Get("BHistograms/h_pfjet_mjj_vetothirdjet")
self._mjj_histograms_vetothirdjet[analysis].SetName("h_" + analysis + "_mjj_vetothirdjet")
self._mjj_histograms_vetothirdjet[analysis].SetDirectory(0)
self._mjj_histograms_vetothirdjet[analysis] = histogram_tools.rebin_histogram(self._mjj_histograms_vetothirdjet[analysis], self._mass_bins, normalization_bin_width=1)
if "bbll" in analysis:
self._mjj_histograms_fine[analysis].Scale(1.7) # Prescale for singlemu + 60/53. The prescale was not computer for these analyses.
self._mjj_histograms[analysis].Scale(1.7) # Prescale for singlemu + 60/53. The prescale was not computer for these analyses.
self._mjj_histograms_fine[analysis].Rebin(5)
self._analyses.append(analysis)
f.Close()
def data_jes_plot(new_sample, old_sample, analysis):
print "Making data JES plot for {}".format(analysis)
f_new = TFile(analysis_config.get_b_histogram_filename(analysis, new_sample), "READ")
h_new = f_new.Get("BHistograms/h_pfjet_mjj")
h_new.SetName("h_pfjet_mjj_newJEC")
h_new.SetDirectory(0)
h_new = histogram_tools.rebin_histogram(h_new, dijet_binning, normalization_bin_width=1.)
print "New integral = {}".format(h_new.Integral())
f_old = TFile(analysis_config.get_b_histogram_filename(analysis, old_sample).replace("EightTeeEeVeeBee/BHistograms", "EightTeeEeVeeBee/BHistograms/old/JEC2012"), "READ")
h_old = f_old.Get("BHistograms/h_pfjet_mjj")
h_old.SetName("h_pfjet_mjj_oldJEC")
h_old.SetDirectory(0)
h_old = histogram_tools.rebin_histogram(h_old, dijet_binning, normalization_bin_width=1.)
print "Old integral = {}".format(h_old.Integral())
c = TCanvas("c_jeccomparison_{}".format(analysis), "c_jeccomparison_{}".format(analysis), 800, 1200)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
top.SetLogy()
top.Draw()
top.cd()
h_new.SetMarkerStyle(20)
h_new.GetXaxis().SetTitleSize(0)
h_new.GetXaxis().SetLabelSize(0)
h_new.GetYaxis().SetTitle("Events / GeV")
h_new.Draw()
h_old.SetMarkerStyle(25)
h_old.Draw("same")
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetTopMargin(0.02)
bottom.SetBottomMargin(0.2)
bottom.Draw()
bottom.cd()
h_ratio = h_new.Clone()
h_ratio.Divide(h_old)
h_ratio.GetXaxis().SetTitle("m_{jj} [GeV]")
h_ratio.GetYaxis().SetTitle("New JEC / Old JEC")
h_ratio.SetMarkerStyle(21)
h_ratio.Draw()
c.cd()
c.SaveAs(analysis_config.figure_directory + "/" + c.GetName() + ".pdf")
f_new.Close()
f_old.Close()
def plot_all_masses(analysis, model, masses):
ratio_up = {}
ratio_down = {}
for mass in masses:
f = TFile(analysis_config.get_b_histogram_filename(analysis, analysis_config.simulation.get_signal_tag(model, mass, "FULLSIM")), "READ")
h_central = f.Get("BHistograms/h_pfjet_mjj")
h_central.SetDirectory(0)
h_central = histogram_tools.rebin_histogram(h_central, dijet_binning, normalization_bin_width=1.)
h_up = f.Get("BHistograms/h_pfjet_mjj_BTagOfflineSFUp")
h_up.SetDirectory(0)
h_up = histogram_tools.rebin_histogram(h_up, dijet_binning, normalization_bin_width=1.)
h_down = f.Get("BHistograms/h_pfjet_mjj_BTagOfflineSFDown")
h_down.SetDirectory(0)
h_down = histogram_tools.rebin_histogram(h_down, dijet_binning, normalization_bin_width=1.)
ratio_up[mass] = h_up.Clone()
ratio_up[mass].SetName("ratio_up_{}".format(mass))
ratio_up[mass].SetDirectory(0)
ratio_up[mass].Divide(h_central)
ratio_up[mass].SetMarkerStyle(26)
ratio_up[mass].Draw("p hist same")
ratio_down[mass] = h_down.Clone()
ratio_down[mass].SetName("ratio_down_{}".format(mass))
ratio_down[mass].SetDirectory(0)
ratio_down[mass].Divide(h_central)
ratio_down[mass].SetMarkerStyle(32)
ratio_down[mass].Draw("p hist same")
norm = h_central.Integral()
print "{}\t{}\t{}:{}".format(analysis, model, mass, h_up.Integral()/norm-1.)
f.Close()
c = TCanvas("c_offb_unc_{}_{}_all".format(analysis, model), "c_offb_unc_{}_{}_all".format(analysis, model), 800, 600)
frame = TH1D("frame", "frame", 100, 0., 1300.)
frame.SetMinimum(0.8)
frame.SetMaximum(1.2)
frame.GetXaxis().SetTitle("m_{jj}")
frame.GetYaxis().SetTitle("SF Variation / Central")
frame.Draw()
for style_counter, mass in enumerate(masses):
ratio_up[mass].SetMarkerStyle(20)
ratio_up[mass].SetMarkerColor(seaborn.GetColorRoot("cubehelixlarge", style_counter, len(masses)+2))
ratio_up[mass].SetLineColor(seaborn.GetColorRoot("cubehelixlarge", style_counter, len(masses)+2))
ratio_up[mass].Draw("hist same")
ratio_down[mass].SetMarkerStyle(20)
ratio_down[mass].SetMarkerColor(seaborn.GetColorRoot("cubehelixlarge", style_counter, len(masses)+2))
ratio_down[mass].SetLineColor(seaborn.GetColorRoot("cubehelixlarge", style_counter, len(masses)+2))
ratio_down[mass].Draw("hist same")
c.SaveAs(analysis_config.figure_directory + "/" + c.GetName() + ".pdf")
def GetJetHTHistogram(self):
for sr_name in ["highmass", "lowmass"]:
analyses = {}
HT_slices = []
for mass in xrange(200, 600, 50):
HT_slices.append("HT" + str(mass))
analyses["HT" + str(mass)] = "trigjetht" + str(mass)
if sr_name == "lowmass":
analyses["HT" + str(mass)] += "_eta1p7"
analyses["HT" + str(mass)] += "_CSVTM"
sample = "JetHT_2012BCD"
HT_slice_histograms = {}
for HT_slice in HT_slices:
f = TFile(analysis_config.get_b_histogram_filename(analyses[HT_slice], sample), "READ")
HT_slice_histograms[HT_slice] = f.Get("BHistograms/h_pfjet_mjj")
print "On file " + analysis_config.get_b_histogram_filename(analyses[HT_slice], sample)
HT_slice_histograms[HT_slice].SetName(HT_slice_histograms[HT_slice].GetName() + "_" + analyses[HT_slice])
HT_slice_histograms[HT_slice].SetDirectory(0)
f.Close()
HT_slices.append("HTUnprescaled")
unprescaled_analysis_name = "trigjetht"
if sr_name == "lowmass":
unprescaled_analysis_name += "_eta1p7"
unprescaled_analysis_name += "_CSVTM"
analyses["HTUnprescaled"] = unprescaled_analysis_name
f_unprescaled = TFile(analysis_config.get_b_histogram_filename(unprescaled_analysis_name, sample), "READ")
HT_slice_histograms["HTUnprescaled"] = f_unprescaled.Get("BHistograms/h_pfjet_mjj")
HT_slice_histograms["HTUnprescaled"].SetName(HT_slice_histograms["HTUnprescaled"].GetName() + "_" + analyses["HTUnprescaled"])
HT_slice_histograms["HTUnprescaled"].SetDirectory(0)
f_unprescaled.Close()
ranges = {
"HT200":[220, 386],
"HT250":[386, 489],
"HT300":[489, 526],
"HT350":[526, 606],
"HT400":[606, 649],
"HT450":[649, 740],
"HT500":[740, 788],
"HT550":[788, 890],
#"HT650":[800, 890],
"HTUnprescaled":[890, 2000]
}
self._analyses.append("JetHT")
self._mjj_histograms_fine["JetHT"] = self.FrankenHist(HT_slices, HT_slice_histograms, ranges)
self._mjj_histograms["JetHT"] = histogram_tools.rebin_histogram(self._mjj_histograms_fine["JetHT"], self._mass_bins, normalization_bin_width=1)
self._mjj_histograms_fine["JetHT"].Rebin(5)
this_num_hist = f_num.Get("BHistograms/h_pfjet_mjj")
#this_num_hist.Scale(19700.*qcd_cross_sections[qcd_sample] / input_nevents_num)
input_nevents_den = f_den.Get("BHistograms/h_input_nevents").Integral()
this_den_hist = f_den.Get("BHistograms/h_pfjet_mjj")
#this_den_hist.Scale(19700.*qcd_cross_sections[qcd_sample] / input_nevents_den)
if not sr in qcd_hists_num:
qcd_hists_num[sr] = this_num_hist.Clone()
qcd_hists_num[sr].SetDirectory(0)
qcd_hists_den[sr] = this_den_hist.Clone()
qcd_hists_den[sr].SetDirectory(0)
else:
qcd_hists_num[sr].Add(this_num_hist)
qcd_hists_den[sr].Add(this_den_hist)
f_num.Close()
f_den.Close()
qcd_hists_num[sr] = histogram_tools.rebin_histogram(
qcd_hists_num[sr], dijet_binning)
qcd_hists_den[sr] = histogram_tools.rebin_histogram(
qcd_hists_den[sr], dijet_binning)
for bin in xrange(1, qcd_hists_num[sr].GetNbinsX() + 1):
bin_center = qcd_hists_num[sr].GetXaxis().GetBinCenter(bin)
if sr == "lowmass":
if bin_center < 296 or bin_center > 1058:
qcd_hists_num[sr].SetBinContent(bin, 0)
qcd_hists_num[sr].SetBinError(bin, 0)
qcd_hists_den[sr].SetBinContent(bin, 0)
qcd_hists_den[sr].SetBinError(bin, 0)
elif sr == "highmass":
if bin_center < 526 or bin_center > 1607:
qcd_hists_num[sr].SetBinContent(bin, 0)
qcd_hists_num[sr].SetBinError(bin, 0)
qcd_hists_den[sr].SetBinContent(bin, 0)
def FrankenEfficiencyPlot(self, logy=True, binning=None, simple_rebin=None, save_tag="", x_range=None, ratio_range=None, legend_position="topright", numerator_legend=None, denominator_legend=None, ratio_title=None):
ht_slices = ["HT200","HT250","HT300","HT350","HT400","HT450","HT500","HT550","HTUnprescaled"] # HT650
if self._numerator_analysis == "trigjetht_eta1p7_CSVTM":
numerator_analyses = {
"HT200":"trigjetht200_eta1p7_CSVTM",
"HT250":"trigjetht250_eta1p7_CSVTM",
"HT300":"trigjetht300_eta1p7_CSVTM",
"HT350":"trigjetht350_eta1p7_CSVTM",
"HT400":"trigjetht400_eta1p7_CSVTM",
"HT450":"trigjetht450_eta1p7_CSVTM",
"HT500":"trigjetht500_eta1p7_CSVTM",
"HT550":"trigjetht550_eta1p7_CSVTM",
#"HT650":"trigjetht650_eta1p7_CSVTM",
"HTUnprescaled":"trigjetht_eta1p7_CSVTM"
}
elif self._numerator_analysis == "trigjetht_eta2p2_CSVTM":
numerator_analyses = {
"HT200":"trigjetht200_CSVTM",
"HT250":"trigjetht250_CSVTM",
"HT300":"trigjetht300_CSVTM",
"HT350":"trigjetht350_CSVTM",
"HT400":"trigjetht400_CSVTM",
"HT450":"trigjetht450_CSVTM",
"HT500":"trigjetht500_CSVTM",
"HT550":"trigjetht550_CSVTM",
#"HT650":"trigjetht650_CSVTM",
"HTUnprescaled":"trigjetht_CSVTM"
}
else:
print "[OfflineBTagPlots::FrankenEfficiencyPlot] ERROR : numerator analysis must be trigjetht_eta1p7_CSVTM or trigjetht_eta2p2_CSVTM"
sys.exit(1)
if self._denominator_analysis == "trigjetht_eta1p7":
denominator_analyses = {
"HT200":"trigjetht200_eta1p7",
"HT250":"trigjetht250_eta1p7",
"HT300":"trigjetht300_eta1p7",
"HT350":"trigjetht350_eta1p7",
"HT400":"trigjetht400_eta1p7",
"HT450":"trigjetht450_eta1p7",
"HT500":"trigjetht500_eta1p7",
"HT550":"trigjetht550_eta1p7",
#"HT650":"trigjetht650_eta1p7",
"HTUnprescaled":"trigjetht_eta1p7"
}
elif self._denominator_analysis == "trigjetht_eta2p2":
denominator_analyses = {
"HT200":"trigjetht200",
"HT250":"trigjetht250",
"HT300":"trigjetht300",
"HT350":"trigjetht350",
"HT400":"trigjetht400",
"HT450":"trigjetht450",
"HT500":"trigjetht500",
"HT550":"trigjetht550",
#"HT650":"trigjetht650",
"HTUnprescaled":"trigjetht"
}
else:
print "[OfflineBTagPlots::FrankenEfficiencyPlot] ERROR : denominator analysis must be trigjetht_eta1p7 or trigjetht_eta2p2"
sys.exit(1)
ht_ranges = {
"HT200":[220, 386],
"HT250":[386, 489],
"HT300":[489, 526],
"HT350":[526, 606],
"HT400":[606, 649],
"HT450":[649, 740],
"HT500":[740, 788],
"HT550":[788, 890],
#"HT650":[890, 2000],
"HTUnprescaled":[890, 2000]
}
numerator_histogram = None
denominator_histogram = None
for sample in self._samples:
print "[EfficiencyPlot] DEBUG : Sample " + sample
numerator_slice_histograms = {}
denominator_slice_histograms = {}
for slice_name in ht_slices:
print slice_name
print analysis_config.get_b_histogram_filename(numerator_analyses[slice_name], sample)
numerator_file = TFile(analysis_config.get_b_histogram_filename(numerator_analyses[slice_name], sample), "READ")
print analysis_config.get_b_histogram_filename(denominator_analyses[slice_name], sample)
denominator_file = TFile(analysis_config.get_b_histogram_filename(denominator_analyses[slice_name], sample), "READ")
# Check input nevents
num_nevents = numerator_file.Get("BHistograms/h_input_nevents").Integral()
den_nevents = denominator_file.Get("BHistograms/h_input_nevents").Integral()
if num_nevents != den_nevents:
# Allow tiny differences...?
if abs((num_nevents - den_nevents) / den_nevents) < 0.001:
print "[EfficiencyPlot] ERROR : Small inconsistency between number of events between numerator and denominator. I'm going to rescale away the difference, but you may want to fix this."
numerator_normalization = den_nevents / num_nevents
denominator_normalization = 1.
else:
print "[EfficiencyPlot] ERROR : Inconsistent number of events between numerator and denominator. Results would be wrong, so I'm aborting."
print "[EfficiencyPlot] ERROR : \tNumerator = " + str(numerator_file.Get("BHistograms/h_input_nevents").Integral())
print "[EfficiencyPlot] ERROR : \tDenominator = " + str(denominator_file.Get("BHistograms/h_input_nevents").Integral())
sys.exit(1)
else:
numerator_normalization = 1.
denominator_normalization = 1.
numerator_slice_histograms[slice_name] = numerator_file.Get("BHistograms/h_pfjet_mjj")
numerator_slice_histograms[slice_name].SetName("h_pfjet_mjj_num_" + slice_name + "_" + sample)
numerator_slice_histograms[slice_name].SetDirectory(0)
numerator_slice_histograms[slice_name].Scale(numerator_normalization)
denominator_slice_histograms[slice_name] = denominator_file.Get("BHistograms/h_pfjet_mjj")
denominator_slice_histograms[slice_name].SetName("h_pfjet_mjj_den_" + slice_name + "_" + sample)
denominator_slice_histograms[slice_name].SetDirectory(0)
denominator_slice_histograms[slice_name].Scale(denominator_normalization)
numerator_file.Close()
denominator_file.Close()
# Make frankenhist
this_numerator_histogram = self.FrankenHist(ht_slices, numerator_slice_histograms, ht_ranges)
this_denominator_histogram = self.FrankenHist(ht_slices, denominator_slice_histograms, ht_ranges)
if not numerator_histogram:
numerator_histogram = this_numerator_histogram.Clone()
numerator_histogram.SetName(numerator_histogram.GetName() + save_tag + "_num_" + str(time.time()))
denominator_histogram = this_denominator_histogram.Clone()
denominator_histogram.SetName(denominator_histogram.GetName() + save_tag + "_num_" + str(time.time()))
else:
numerator_histogram.Add(this_numerator_histogram)
denominator_histogram.Add(this_denominator_histogram)
# Rebin
if binning:
numerator_histogram = histogram_tools.rebin_histogram(numerator_histogram, binning)
denominator_histogram = histogram_tools.rebin_histogram(denominator_histogram, binning)
elif simple_rebin:
numerator_histogram.Rebin(simple_rebin)
denominator_histogram.Rebin(simple_rebin)
cname = "c_offline_btag_eff_mjj"
if logy:
cname += "_log"
c = TCanvas(cname, "Offline b-tag #epsilon", 800, 1000)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
if logy:
top.SetLogy()
top.Draw()
top.cd()
frame_top = numerator_histogram.Clone()
frame_top.Reset()
if x_range:
frame_top.GetXaxis().SetRangeUser(x_range[0], x_range[1])
if logy:
y_min = 0.1
y_max = max(numerator_histogram.GetMaximum(), denominator_histogram.GetMaximum()) * 10.
else:
y_min = 0.
y_max = max(numerator_histogram.GetMaximum(), denominator_histogram.GetMaximum()) * 1.5
frame_top.SetMinimum(y_min)
frame_top.SetMaximum(y_max)
frame_top.GetXaxis().SetLabelSize(0)
frame_top.GetXaxis().SetTitleSize(0)
if binning:
#frame_top.GetYaxis().SetTitle("Events / 1 GeV")
frame_top.GetYaxis().SetTitle("Events")
else:
frame_top.GetYaxis().SetTitle("Events")
frame_top.Draw("axis")
print "numerator integral = " + str(numerator_histogram.Integral())
print "denominator integral = " + str(denominator_histogram.Integral())
numerator_histogram.SetMarkerStyle(20)
numerator_histogram.SetMarkerColor(seaborn.GetColorRoot("default", 0))
numerator_histogram.SetLineColor(seaborn.GetColorRoot("default", 0))
numerator_histogram.Draw("same")
denominator_histogram.SetMarkerStyle(24)
denominator_histogram.SetMarkerColor(seaborn.GetColorRoot("default", 2))
denominator_histogram.SetLineColor(seaborn.GetColorRoot("default", 2))
denominator_histogram.Draw("same")
if legend_position == "topright":
l = TLegend(0.6, 0.6, 0.85, 0.8)
elif legend_position == "bottomright":
l = TLegend(0.6, 0.2, 0.85, 0.4)
l.SetFillColor(0)
l.SetBorderSize(0)
if numerator_legend:
l.AddEntry(numerator_histogram, numerator_legend)
else:
l.AddEntry(numerator_histogram, "CSVT+CSVM")
if denominator_legend:
l.AddEntry(denominator_histogram, denominator_legend)
else:
l.AddEntry(denominator_histogram, "No CSV")
l.Draw()
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetTopMargin(0.01)
bottom.SetBottomMargin(0.2)
bottom.Draw()
bottom.cd()
ratio_histogram = numerator_histogram.Clone()
ratio_histogram.Reset()
if x_range:
ratio_histogram.GetXaxis().SetRangeUser(x_range[0], x_range[1])
ratio_histogram.SetName(numerator_histogram.GetName() + "_ratio_" + save_tag + str(time.time()))
ratio_histogram.SetDirectory(0)
for bin in xrange(1, numerator_histogram.GetNbinsX() + 1):
# Undo bin normalization
if numerator_histogram.GetBinError(bin) > 0 and denominator_histogram.GetBinError(bin) > 0:
num_unnormalized = (numerator_histogram.GetBinContent(bin))**2 / (numerator_histogram.GetBinError(bin))**2
den_unnormalized = (denominator_histogram.GetBinContent(bin))**2 / (denominator_histogram.GetBinError(bin))**2
num = numerator_histogram.GetBinContent(bin)
den = denominator_histogram.GetBinContent(bin)
ratio = 1. * num_unnormalized / den_unnormalized
ratio_err = sqrt(ratio * (1. - ratio) / den_unnormalized)
#ratio_err = max(sqrt(ratio * (1. - ratio) / den), 1./den)
else:
ratio = 0.
ratio_err = 0.
ratio_histogram.SetBinContent(bin, ratio)
ratio_histogram.SetBinError(bin, ratio_err)
ratio_histogram.SetMarkerSize(1)
ratio_histogram.SetMarkerColor(kBlack)
ratio_histogram.SetLineColor(kBlack)
ratio_histogram.SetLineWidth(2)
ratio_histogram.GetXaxis().SetTitle("m_{jj} [GeV]")
if ratio_title:
ratio_histogram.GetYaxis().SetTitle(ratio_title)
else:
ratio_histogram.GetYaxis().SetTitle("Offline 2#timesb-tag efficiency")
if ratio_range:
ratio_histogram.SetMinimum(ratio_range[0])
ratio_histogram.SetMaximum(ratio_range[1])
else:
ratio_histogram.SetMinimum(0.)
ratio_histogram.SetMaximum(1.)
ratio_histogram.Draw()
c.cd()
c.SaveAs(analysis_config.figure_directory + "/OfflineBTag/" + c.GetName() + save_tag + ".pdf")
ROOT.SetOwnership(c, False)
ROOT.SetOwnership(top, False)
ROOT.SetOwnership(bottom, False)
def EfficiencyPlot(self, var="mjj", logy=True, binning=None, simple_rebin=None, save_tag="", sample_xsecs=None, x_range=None, ratio_range=None, legend_position="topright", numerator_legend=None, denominator_legend=None, ratio_title=None, prescaled=False, fit=False, fit_range=None):
numerator_histogram = None
denominator_histogram = None
for sample in self._samples:
if sample_xsecs:
xsec = sample_xsecs[sample]
numerator_normalization = 19710. * xsec / self._input_files["numerator"][sample].Get("BHistograms/h_sample_nevents").Integral()
denominator_normalization = 19710. * xsec / self._input_files["denominator"][sample].Get("BHistograms/h_sample_nevents").Integral()
if numerator_normalization != denominator_normalization:
print "[EfficiencyPlot] WARNING : Numerator and denominator normalizations do not match! Continuing, but investigate this. Numerator nevents = " + str(self._input_files["numerator"][sample].Get("BHistograms/h_sample_nevents").Integral()) + ", denominator nevents = " + str(self._input_files["denominator"][sample].Get("BHistograms/h_sample_nevents").Integral())
else:
# Check input nevents
if self._input_files["numerator"][sample].Get("BHistograms/h_sample_nevents"):
num_nevents = self._input_files["numerator"][sample].Get("BHistograms/h_sample_nevents").Integral()
den_nevents = self._input_files["denominator"][sample].Get("BHistograms/h_sample_nevents").Integral()
else:
num_nevents = self._input_files["numerator"][sample].Get("BHistograms/h_input_nevents").Integral()
den_nevents = self._input_files["denominator"][sample].Get("BHistograms/h_input_nevents").Integral()
if num_nevents != den_nevents:
# Allow tiny differences...?
if abs((num_nevents - den_nevents) / den_nevents) < 0.001:
print "[EfficiencyPlot] ERROR : Small inconsistency between number of events between numerator and denominator. I'm going to rescale away the difference, but you may want to fix this."
numerator_normalization = den_nevents / num_nevents
denominator_normalization = 1.
else:
print "[EfficiencyPlot] ERROR : Inconsistent number of events between numerator and denominator. Results would be wrong, so I'm aborting."
print "[EfficiencyPlot] ERROR : \tNumerator = " + str(self._input_files["numerator"][sample].Get("BHistograms/h_sample_nevents").Integral())
print "[EfficiencyPlot] ERROR : \tDenominator = " + str(self._input_files["denominator"][sample].Get("BHistograms/h_sample_nevents").Integral())
sys.exit(1)
else:
numerator_normalization = 1.
denominator_normalization = 1.
if numerator_histogram:
numerator_histogram.Add(self._input_files["numerator"][sample].Get("BHistograms/h_pfjet_" + var), numerator_normalization)
else:
numerator_histogram = self._input_files["numerator"][sample].Get("BHistograms/h_pfjet_" + var).Clone()
numerator_histogram.Scale(numerator_normalization)
numerator_histogram.SetDirectory(0)
numerator_histogram.SetName(numerator_histogram.GetName() + save_tag + "_num" + str(time.time()))
if denominator_histogram:
denominator_histogram.Add(self._input_files["denominator"][sample].Get("BHistograms/h_pfjet_" + var), denominator_normalization)
else:
denominator_histogram = self._input_files["denominator"][sample].Get("BHistograms/h_pfjet_" + var).Clone()
denominator_histogram.Scale(denominator_normalization)
denominator_histogram.SetDirectory(0)
denominator_histogram.SetName(denominator_histogram.GetName() + save_tag + "_den" + str(time.time()))
# Rebin
numerator_histogram_fine = numerator_histogram.Clone()
denominator_histogram_fine = denominator_histogram.Clone()
if binning:
numerator_histogram = histogram_tools.rebin_histogram(numerator_histogram, binning)
denominator_histogram = histogram_tools.rebin_histogram(denominator_histogram, binning)
elif simple_rebin:
numerator_histogram.Rebin(simple_rebin)
denominator_histogram.Rebin(simple_rebin)
cname = "c_offline_btag_eff_" + var
if logy:
cname += "_log"
c = TCanvas(cname, "Offline b-tag #epsilon", 800, 1000)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
if logy:
top.SetLogy()
top.Draw()
top.cd()
frame_top = numerator_histogram.Clone()
frame_top.Reset()
if x_range:
frame_top.GetXaxis().SetRangeUser(x_range[0], x_range[1])
if logy:
y_min = 0.1
y_max = max(numerator_histogram.GetMaximum(), denominator_histogram.GetMaximum()) * 10.
else:
y_min = 0.
y_max = max(numerator_histogram.GetMaximum(), denominator_histogram.GetMaximum()) * 1.5
frame_top.SetMinimum(y_min)
frame_top.SetMaximum(y_max)
frame_top.GetXaxis().SetLabelSize(0)
frame_top.GetXaxis().SetTitleSize(0)
if binning:
#frame_top.GetYaxis().SetTitle("Events / 1 GeV")
frame_top.GetYaxis().SetTitle("Events")
else:
frame_top.GetYaxis().SetTitle("Events")
frame_top.Draw("axis")
print "numerator integral = " + str(numerator_histogram.Integral())
print "denominator integral = " + str(denominator_histogram.Integral())
numerator_histogram.SetMarkerStyle(20)
numerator_histogram.SetMarkerColor(seaborn.GetColorRoot("default", 0))
numerator_histogram.SetLineColor(seaborn.GetColorRoot("default", 0))
numerator_histogram.Draw("same")
denominator_histogram.SetMarkerStyle(24)
denominator_histogram.SetMarkerColor(seaborn.GetColorRoot("default", 2))
denominator_histogram.SetLineColor(seaborn.GetColorRoot("default", 2))
denominator_histogram.Draw("same")
if legend_position == "topright":
l = TLegend(0.6, 0.6, 0.85, 0.8)
elif legend_position == "bottomright":
l = TLegend(0.6, 0.2, 0.85, 0.4)
l.SetFillColor(0)
l.SetBorderSize(0)
if numerator_legend:
l.AddEntry(numerator_histogram, numerator_legend)
else:
l.AddEntry(numerator_histogram, "CSVT+CSVM")
if denominator_legend:
l.AddEntry(denominator_histogram, denominator_legend)
else:
l.AddEntry(denominator_histogram, "No CSV")
l.Draw()
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetTopMargin(0.01)
bottom.SetBottomMargin(0.2)
bottom.Draw()
bottom.cd()
ratio_histogram = numerator_histogram.Clone()
ratio_histogram.Reset()
if x_range:
ratio_histogram.GetXaxis().SetRangeUser(x_range[0], x_range[1])
ratio_histogram.SetName(numerator_histogram.GetName() + "_ratio_" + save_tag + str(time.time()))
ratio_histogram.SetDirectory(0)
for bin in xrange(1, numerator_histogram.GetNbinsX() + 1):
# Undo bin normalization
if numerator_histogram.GetBinError(bin) > 0 and denominator_histogram.GetBinError(bin) > 0:
if prescaled:
num = (numerator_histogram.GetBinContent(bin))**2 / (numerator_histogram.GetBinError(bin))**2
den = (denominator_histogram.GetBinContent(bin))**2 / (denominator_histogram.GetBinError(bin))**2
else:
num = numerator_histogram.GetBinContent(bin)
den = denominator_histogram.GetBinContent(bin)
ratio = 1. * num / den
ratio_err = sqrt(ratio * (1. - ratio) / den)
#ratio_err = max(sqrt(ratio * (1. - ratio) / den), 1./den)
else:
ratio = 0.
ratio_err = 0.
ratio_histogram.SetBinContent(bin, ratio)
ratio_histogram.SetBinError(bin, ratio_err)
ratio_histogram.SetMarkerSize(1)
ratio_histogram.SetMarkerColor(kBlack)
ratio_histogram.SetLineColor(kBlack)
ratio_histogram.SetLineWidth(2)
if "mjj" in var:
ratio_histogram.GetXaxis().SetTitle("m_{jj} [GeV]")
elif "pt" in var:
ratio_histogram.GetXaxis().SetTitle("p_{T} [GeV]")
if ratio_title:
ratio_histogram.GetYaxis().SetTitle(ratio_title)
else:
ratio_histogram.GetYaxis().SetTitle("Offline 2#timesb-tag efficiency")
if ratio_range:
ratio_histogram.SetMinimum(ratio_range[0])
ratio_histogram.SetMaximum(ratio_range[1])
else:
ratio_histogram.SetMinimum(0.)
ratio_histogram.SetMaximum(1.)
ratio_histogram.Draw()
c.cd()
c.SaveAs(analysis_config.figure_directory + "/OfflineBTag/" + c.GetName() + save_tag + ".pdf")
ROOT.SetOwnership(c, False)
ROOT.SetOwnership(top, False)
ROOT.SetOwnership(bottom, False)
# Fit the ratio
if fit:
# Make finely binned ratio
#ratio_histogram_fine = numerator_histogram_fine.Clone()
#ratio_histogram_fine.Reset()
#if x_range:
# ratio_histogram_fine.GetXaxis().SetRangeUser(x_range[0], x_range[1])
#ratio_histogram_fine.SetName(numerator_histogram_fine.GetName() + "_ratio_fine_" + save_tag + str(time.time()))
#ratio_histogram_fine.SetDirectory(0)
#for bin in xrange(1, numerator_histogram_fine.GetNbinsX() + 1):
# # Undo bin normalization
# if numerator_histogram_fine.GetBinError(bin) > 0 and denominator_histogram_fine.GetBinError(bin) > 0:
# if prescaled:
# num = (numerator_histogram_fine.GetBinContent(bin))**2 / (numerator_histogram_fine.GetBinError(bin))**2
# den = (denominator_histogram_fine.GetBinContent(bin))**2 / (denominator_histogram_fine.GetBinError(bin))**2
# else:
# num = numerator_histogram_fine.GetBinContent(bin)
# den = denominator_histogram_fine.GetBinContent(bin)
# ratio = 1. * num / den
# ratio_err = sqrt(ratio * (1. - ratio) / den)
# #ratio_err = max(sqrt(ratio * (1. - ratio) / den), 1./den)
# else:
# ratio = 0.
# ratio_err = 0.
# ratio_histogram_fine.SetBinContent(bin, ratio)
# ratio_histogram_fine.SetBinError(bin, ratio_err)
print "Fitting " + save_tag + " with some functions"
fitter = TBinomialEfficiencyFitter(numerator_histogram, denominator_histogram)
fits = {}
print "\nFitting quadratic"
fits["quadratic"] = TF1("quadratic", "[0]+[1]*x+[2]*x*x", fit_range[0], fit_range[1])
fits["quadratic"].SetParameter(0, 0.01)
fits["quadratic"].SetParameter(1, -1.e-5)
fits["quadratic"].SetParLimits(1, -1.e5, 1.e5)
fits["quadratic"].SetParameter(2, 1.e-11)
fits["quadratic"].SetParLimits(2, -1.e5, 1.e5)
ratio_histogram.Fit(fits["quadratic"], "R0")
print "\nFitting quadexp"
fits["quadexp"] = TF1("quadexp", "[0]+[1]*x+[2]*x*x+ [3]*exp([4]*x)", fit_range[0], fit_range[1])
fits["quadexp"].SetParameter(0, 3.97724e-03)
fits["quadexp"].SetParameter(1, -3.69019e-06)
fits["quadexp"].SetParameter(2, 1.37598e-09)
fits["quadexp"].SetParameter(3, 2.95236e+00)
fits["quadexp"].SetParameter(4, -1.54774e-02)
fitter.Fit(fits["quadexp"], "R0")
print "\nFitting linexp"
fits["linexp"] = TF1("linexp", "[0]+[1]*x + [2]*exp([3]*x)", fit_range[0], fit_range[1])
fits["linexp"].SetParameter(0, 3.97724e-03)
fits["linexp"].SetParameter(1, -3.69019e-06)
fits["linexp"].SetParameter(2, 2.95236e+00)
fits["linexp"].SetParameter(3, -1.54774e-02)
fitter.Fit(fits["linexp"], "R0")
# Print chi2/ndfs
for fit_name, fit_function in fits.iteritems():
print "chi^2/ndf (p) for " + fit_name + " = " + str(fit_function.GetChisquare()) + "/" + str(fit_function.GetNDF()) + " (" + str(TMath.Prob(fit_function.GetChisquare(), fit_function.GetNDF()))
pull_histograms = {}
for fit_name, fit_function in fits.iteritems():
print "[debug] Making pull histogram for " + fit_name
pull_histograms[fit_name] = ratio_histogram.Clone()
pull_histograms[fit_name].SetDirectory(0)
pull_histograms[fit_name].SetName("ratio_fit_pulls_" + fit_name)
pull_histograms[fit_name].Reset()
for bin in xrange(1, pull_histograms[fit_name].GetNbinsX() + 1):
bin_center = pull_histograms[fit_name].GetXaxis().GetBinCenter(bin)
if (bin_center < fit_range[0]) or (bin_center > fit_range[1]):
#print "Pull hist : skipping bin with center " + str(bin_center)
pull_histograms[fit_name].SetBinContent(bin, 0.)
elif ratio_histogram.GetBinError(bin):
#print "[debug] Bin " + str(bin) + " / pull = " + str(round(ratio_histogram.GetBinContent(bin), 2)) + "-" + str(round(fit_function.Eval(bin_center), 2)) + "/" + str(round(ratio_histogram.GetBinError(bin), 2)) + " = " + str(round((ratio_histogram.GetBinContent(bin) - fit_function.Eval(bin_center)) / ratio_histogram.GetBinError(bin), 2))
pull_histograms[fit_name].SetBinContent(bin, (ratio_histogram.GetBinContent(bin) - fit_function.Eval(bin_center)) / ratio_histogram.GetBinError(bin))
else:
pull_histograms[fit_name].SetBinContent(bin, 0.)
c_eff_fit = TCanvas("c_offline_btag_fit", "c_offline_btag_fit", 800, 1000)
top_eff_fit = TPad("top_fit", "top_fit", 0., 0.5, 1., 1.)
top_eff_fit.SetBottomMargin(0.02)
top_eff_fit.Draw()
top_eff_fit.cd()
ratio_histogram.GetXaxis().SetTitleSize(0)
ratio_histogram.GetXaxis().SetLabelSize(0)
ratio_histogram.GetYaxis().SetTitle("Offline b-tag efficiency")
ratio_histogram.Draw()
fits["quadratic"].SetLineColor(seaborn.GetColorRoot("dark", 2))
fits["quadratic"].SetLineWidth(2)
pull_histograms["quadratic"].SetLineColor(seaborn.GetColorRoot("dark", 2))
fits["quadratic"].Draw("same")
fits["linexp"].SetLineColor(seaborn.GetColorRoot("dark", 4))
pull_histograms["linexp"].SetLineColor(seaborn.GetColorRoot("dark", 4))
fits["linexp"].SetLineWidth(2)
fits["linexp"].Draw("same")
l_eff_fit = TLegend(0.6, 0.6, 0.88, 0.8)
l_eff_fit.SetBorderSize(0)
l_eff_fit.SetFillColor(0)
l_eff_fit.AddEntry(ratio_histogram, "Data", "pl")
l_eff_fit.AddEntry(fits["quadratic"], "Fit (quadratic)", "l")
l_eff_fit.AddEntry(fits["linexp"], "Fit (linear+exp)", "l")
l_eff_fit.Draw()
c_eff_fit.cd()
bottom_eff_fit = TPad("bottom_fit", "bottom_fit", 0., 0., 1., 0.5)
bottom_eff_fit.SetBottomMargin(0.2)
bottom_eff_fit.SetTopMargin(0.02)
bottom_eff_fit.Draw()
bottom_eff_fit.cd()
if fit_range[0] <= 296:
default_fit = "quadratic"
else:
default_fit = "linexp"
pull_histograms[default_fit].SetLineWidth(2)
pull_histograms[default_fit].GetXaxis().SetTitle("m_{jj} [GeV]")
pull_histograms[default_fit].GetYaxis().SetTitle
if "mjj" in var:
pull_histograms[default_fit].GetXaxis().SetTitle("m_{jj} [GeV]")
elif "pt" in var:
pull_histograms[default_fit].GetXaxis().SetTitle("p_{T} [GeV]")
pull_histograms[default_fit].GetYaxis().SetTitle("(fit - data) / (#sigma(data))")
pull_histograms[default_fit].SetMinimum(-10.)
pull_histograms[default_fit].SetMaximum(10.)
pull_histograms[default_fit].Draw("hist")
#pull_histograms["quadratic"].SetLineWidth(2)
#pull_histograms["quadratic"].SetLineColor(seaborn.GetColorRoot("dark", 2))
#pull_histograms["quadratic"].Draw("hist same")
c_eff_fit.cd()
c_eff_fit.SaveAs(analysis_config.figure_directory + "/OfflineBTag/" + c_eff_fit.GetName() + save_tag + ".pdf")
ROOT.SetOwnership(c_eff_fit, False)
ROOT.SetOwnership(top_eff_fit, False)
ROOT.SetOwnership(bottom_eff_fit, False)
f = TFile(analysis_config.get_b_histogram_filename(analysis, qcd_sample), "READ")
input_nevents = f.Get("BHistograms/h_input_nevents").Integral()
this_h_mjj = f.Get("BHistograms/h_pfjet_mjj").Clone()
this_h_mjj.Scale(19700.*qcd_cross_sections[qcd_sample] / input_nevents)
this_h_mjj_truthbb = f.Get("BHistograms/h_pfjet_mjj_truthbb").Clone()
this_h_mjj_truthbb.Scale(19700.*qcd_cross_sections[qcd_sample] / input_nevents)
if not h_mjj:
h_mjj = this_h_mjj.Clone()
h_mjj.SetDirectory(0)
h_mjj_truthbb = this_h_mjj_truthbb.Clone()
h_mjj_truthbb.SetDirectory(0)
else:
h_mjj.Add(this_h_mjj)
h_mjj_truthbb.Add(this_h_mjj_truthbb)
f.Close()
h_mjj = histogram_tools.rebin_histogram(h_mjj, dijet_binning, normalization_bin_width=1.)
h_mjj_truthbb = histogram_tools.rebin_histogram(h_mjj_truthbb, dijet_binning, normalization_bin_width=1.)
c = TCanvas("c_purity_{}".format(analysis), "c_purity_{}".format(analysis), 800, 1000)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
top.SetLogy()
top.Draw()
top.cd()
h_mjj.SetLineColor(1)
h_mjj.SetLineWidth(2)
h_mjj.GetXaxis().SetTitleSize(0)
h_mjj.GetXaxis().SetLabelSize(0)
h_mjj.GetYaxis().SetTitle("Events / GeV")
h_mjj.GetXaxis().SetRangeUser(0., 2000.)
h_mjj.Draw("hist")
h_mjj_truthbb.SetLineColor(seaborn.GetColorRoot("cubehelixlarge", 2, 20))
this_num_hist = f_num.Get("BHistograms/h_pfjet_mjj")
#this_num_hist.Scale(19700.*qcd_cross_sections[qcd_sample] / input_nevents_num)
input_nevents_den = f_den.Get("BHistograms/h_input_nevents").Integral()
this_den_hist = f_den.Get("BHistograms/h_pfjet_mjj")
#this_den_hist.Scale(19700.*qcd_cross_sections[qcd_sample] / input_nevents_den)
if not sr in qcd_hists_num:
qcd_hists_num[sr] = this_num_hist.Clone()
qcd_hists_num[sr].SetDirectory(0)
qcd_hists_den[sr] = this_den_hist.Clone()
qcd_hists_den[sr].SetDirectory(0)
else:
qcd_hists_num[sr].Add(this_num_hist)
qcd_hists_den[sr].Add(this_den_hist)
f_num.Close()
f_den.Close()
qcd_hists_num[sr] = histogram_tools.rebin_histogram(qcd_hists_num[sr], dijet_binning)
qcd_hists_den[sr] = histogram_tools.rebin_histogram(qcd_hists_den[sr], dijet_binning)
for bin in xrange(1, qcd_hists_num[sr].GetNbinsX()+1):
bin_center = qcd_hists_num[sr].GetXaxis().GetBinCenter(bin)
if sr == "lowmass":
if bin_center < 296 or bin_center > 1058:
qcd_hists_num[sr].SetBinContent(bin, 0)
qcd_hists_num[sr].SetBinError(bin, 0)
qcd_hists_den[sr].SetBinContent(bin, 0)
qcd_hists_den[sr].SetBinError(bin, 0)
elif sr == "highmass":
if bin_center < 526 or bin_center > 1607:
qcd_hists_num[sr].SetBinContent(bin, 0)
qcd_hists_num[sr].SetBinError(bin, 0)
qcd_hists_den[sr].SetBinContent(bin, 0)
qcd_hists_den[sr].SetBinError(bin, 0)
def plot_all_masses(analysis, model, masses):
ratio_up = {}
ratio_down = {}
for mass in masses:
f = TFile(
analysis_config.get_b_histogram_filename(
analysis,
analysis_config.simulation.get_signal_tag(
model, mass, "FULLSIM")), "READ")
h_central = f.Get("BHistograms/h_pfjet_mjj")
h_central.SetDirectory(0)
h_central = histogram_tools.rebin_histogram(h_central,
dijet_binning,
normalization_bin_width=1.)
h_up = f.Get("BHistograms/h_pfjet_mjj_BTagOfflineSFUp")
h_up.SetDirectory(0)
h_up = histogram_tools.rebin_histogram(h_up,
dijet_binning,
normalization_bin_width=1.)
h_down = f.Get("BHistograms/h_pfjet_mjj_BTagOfflineSFDown")
h_down.SetDirectory(0)
h_down = histogram_tools.rebin_histogram(h_down,
dijet_binning,
normalization_bin_width=1.)
ratio_up[mass] = h_up.Clone()
ratio_up[mass].SetName("ratio_up_{}".format(mass))
ratio_up[mass].SetDirectory(0)
ratio_up[mass].Divide(h_central)
ratio_up[mass].SetMarkerStyle(26)
ratio_up[mass].Draw("p hist same")
ratio_down[mass] = h_down.Clone()
ratio_down[mass].SetName("ratio_down_{}".format(mass))
ratio_down[mass].SetDirectory(0)
ratio_down[mass].Divide(h_central)
ratio_down[mass].SetMarkerStyle(32)
ratio_down[mass].Draw("p hist same")
norm = h_central.Integral()
print "{}\t{}\t{}:{}".format(analysis, model, mass,
h_up.Integral() / norm - 1.)
f.Close()
c = TCanvas("c_offb_unc_{}_{}_all".format(analysis, model),
"c_offb_unc_{}_{}_all".format(analysis, model), 800, 600)
frame = TH1D("frame", "frame", 100, 0., 1300.)
frame.SetMinimum(0.8)
frame.SetMaximum(1.2)
frame.GetXaxis().SetTitle("m_{jj}")
frame.GetYaxis().SetTitle("SF Variation / Central")
frame.Draw()
for style_counter, mass in enumerate(masses):
ratio_up[mass].SetMarkerStyle(20)
ratio_up[mass].SetMarkerColor(
seaborn.GetColorRoot("cubehelixlarge", style_counter,
len(masses) + 2))
ratio_up[mass].SetLineColor(
seaborn.GetColorRoot("cubehelixlarge", style_counter,
len(masses) + 2))
ratio_up[mass].Draw("hist same")
ratio_down[mass].SetMarkerStyle(20)
ratio_down[mass].SetMarkerColor(
seaborn.GetColorRoot("cubehelixlarge", style_counter,
len(masses) + 2))
ratio_down[mass].SetLineColor(
seaborn.GetColorRoot("cubehelixlarge", style_counter,
len(masses) + 2))
ratio_down[mass].Draw("hist same")
c.SaveAs(analysis_config.figure_directory + "/" + c.GetName() + ".pdf")
variables = ["mjj", "pt1", "pt2", "pt_btag1", "pt_btag2"]
histograms = {}
for analysis in analyses:
histograms[analysis] = {}
for variable in variables:
histograms[analysis][variable] = files[analysis].Get(
"BHistograms/h_pfjet_" + variable)
histograms[analysis][variable].SetName("h_" + analysis + "_" +
variable)
histograms[analysis][variable].SetDirectory(0)
if analysis == "trigmubbll_lowmass":
histograms[analysis][variable].Scale(1.7) # Prescale
if variable == "mjj":
print "(rebinning)...",
histograms[analysis][variable] = histogram_tools.rebin_histogram(
histograms[analysis][variable],
mass_bins,
normalization_bin_width=1)
print "(done)"
elif "pt" in variable:
histograms[analysis][variable].Rebin(25)
print "Making efficiency histograms"
efficiency_pairs = {
"highmass": ["trigmubbh_highmass_CSVTM", "trigmu_highmass_CSVTM"],
"lowmass": ["trigmubbl_lowmass_CSVTM", "trigmu_lowmass_CSVTM"],
"llowmass": ["trigmubbll_lowmass_CSVTM", "trigmu_lowmass_CSVTM"]
}
efficiency_titles = {
"highmass": "160/120",
"lowmass": "80/70",
"llowmass": "60/53"
def FrankenEfficiencyPlot(self,
logy=True,
binning=None,
simple_rebin=None,
save_tag="",
x_range=None,
ratio_range=None,
legend_position="topright",
numerator_legend=None,
denominator_legend=None,
ratio_title=None):
ht_slices = [
"HT200", "HT250", "HT300", "HT350", "HT400", "HT450", "HT500",
"HT550", "HTUnprescaled"
] # HT650
if self._numerator_analysis == "trigjetht_eta1p7_CSVTM":
numerator_analyses = {
"HT200": "trigjetht200_eta1p7_CSVTM",
"HT250": "trigjetht250_eta1p7_CSVTM",
"HT300": "trigjetht300_eta1p7_CSVTM",
"HT350": "trigjetht350_eta1p7_CSVTM",
"HT400": "trigjetht400_eta1p7_CSVTM",
"HT450": "trigjetht450_eta1p7_CSVTM",
"HT500": "trigjetht500_eta1p7_CSVTM",
"HT550": "trigjetht550_eta1p7_CSVTM",
#"HT650":"trigjetht650_eta1p7_CSVTM",
"HTUnprescaled": "trigjetht_eta1p7_CSVTM"
}
elif self._numerator_analysis == "trigjetht_eta2p2_CSVTM":
numerator_analyses = {
"HT200": "trigjetht200_CSVTM",
"HT250": "trigjetht250_CSVTM",
"HT300": "trigjetht300_CSVTM",
"HT350": "trigjetht350_CSVTM",
"HT400": "trigjetht400_CSVTM",
"HT450": "trigjetht450_CSVTM",
"HT500": "trigjetht500_CSVTM",
"HT550": "trigjetht550_CSVTM",
#"HT650":"trigjetht650_CSVTM",
"HTUnprescaled": "trigjetht_CSVTM"
}
else:
print "[OfflineBTagPlots::FrankenEfficiencyPlot] ERROR : numerator analysis must be trigjetht_eta1p7_CSVTM or trigjetht_eta2p2_CSVTM"
sys.exit(1)
if self._denominator_analysis == "trigjetht_eta1p7":
denominator_analyses = {
"HT200": "trigjetht200_eta1p7",
"HT250": "trigjetht250_eta1p7",
"HT300": "trigjetht300_eta1p7",
"HT350": "trigjetht350_eta1p7",
"HT400": "trigjetht400_eta1p7",
"HT450": "trigjetht450_eta1p7",
"HT500": "trigjetht500_eta1p7",
"HT550": "trigjetht550_eta1p7",
#"HT650":"trigjetht650_eta1p7",
"HTUnprescaled": "trigjetht_eta1p7"
}
elif self._denominator_analysis == "trigjetht_eta2p2":
denominator_analyses = {
"HT200": "trigjetht200",
"HT250": "trigjetht250",
"HT300": "trigjetht300",
"HT350": "trigjetht350",
"HT400": "trigjetht400",
"HT450": "trigjetht450",
"HT500": "trigjetht500",
"HT550": "trigjetht550",
#"HT650":"trigjetht650",
"HTUnprescaled": "trigjetht"
}
else:
print "[OfflineBTagPlots::FrankenEfficiencyPlot] ERROR : denominator analysis must be trigjetht_eta1p7 or trigjetht_eta2p2"
sys.exit(1)
ht_ranges = {
"HT200": [220, 386],
"HT250": [386, 489],
"HT300": [489, 526],
"HT350": [526, 606],
"HT400": [606, 649],
"HT450": [649, 740],
"HT500": [740, 788],
"HT550": [788, 890],
#"HT650":[890, 2000],
"HTUnprescaled": [890, 2000]
}
numerator_histogram = None
denominator_histogram = None
for sample in self._samples:
print "[EfficiencyPlot] DEBUG : Sample " + sample
numerator_slice_histograms = {}
denominator_slice_histograms = {}
for slice_name in ht_slices:
print slice_name
print analysis_config.get_b_histogram_filename(
numerator_analyses[slice_name], sample)
numerator_file = TFile(
analysis_config.get_b_histogram_filename(
numerator_analyses[slice_name], sample), "READ")
print analysis_config.get_b_histogram_filename(
denominator_analyses[slice_name], sample)
denominator_file = TFile(
analysis_config.get_b_histogram_filename(
denominator_analyses[slice_name], sample), "READ")
# Check input nevents
num_nevents = numerator_file.Get(
"BHistograms/h_input_nevents").Integral()
den_nevents = denominator_file.Get(
"BHistograms/h_input_nevents").Integral()
if num_nevents != den_nevents:
# Allow tiny differences...?
if abs((num_nevents - den_nevents) / den_nevents) < 0.001:
print "[EfficiencyPlot] ERROR : Small inconsistency between number of events between numerator and denominator. I'm going to rescale away the difference, but you may want to fix this."
numerator_normalization = den_nevents / num_nevents
denominator_normalization = 1.
else:
print "[EfficiencyPlot] ERROR : Inconsistent number of events between numerator and denominator. Results would be wrong, so I'm aborting."
print "[EfficiencyPlot] ERROR : \tNumerator = " + str(
numerator_file.Get(
"BHistograms/h_input_nevents").Integral())
print "[EfficiencyPlot] ERROR : \tDenominator = " + str(
denominator_file.Get(
"BHistograms/h_input_nevents").Integral())
sys.exit(1)
else:
numerator_normalization = 1.
denominator_normalization = 1.
numerator_slice_histograms[slice_name] = numerator_file.Get(
"BHistograms/h_pfjet_mjj")
numerator_slice_histograms[slice_name].SetName(
"h_pfjet_mjj_num_" + slice_name + "_" + sample)
numerator_slice_histograms[slice_name].SetDirectory(0)
numerator_slice_histograms[slice_name].Scale(
numerator_normalization)
denominator_slice_histograms[
slice_name] = denominator_file.Get(
"BHistograms/h_pfjet_mjj")
denominator_slice_histograms[slice_name].SetName(
"h_pfjet_mjj_den_" + slice_name + "_" + sample)
denominator_slice_histograms[slice_name].SetDirectory(0)
denominator_slice_histograms[slice_name].Scale(
denominator_normalization)
numerator_file.Close()
denominator_file.Close()
# Make frankenhist
this_numerator_histogram = self.FrankenHist(
ht_slices, numerator_slice_histograms, ht_ranges)
this_denominator_histogram = self.FrankenHist(
ht_slices, denominator_slice_histograms, ht_ranges)
if not numerator_histogram:
numerator_histogram = this_numerator_histogram.Clone()
numerator_histogram.SetName(numerator_histogram.GetName() +
save_tag + "_num_" +
str(time.time()))
denominator_histogram = this_denominator_histogram.Clone()
denominator_histogram.SetName(denominator_histogram.GetName() +
save_tag + "_num_" +
str(time.time()))
else:
numerator_histogram.Add(this_numerator_histogram)
denominator_histogram.Add(this_denominator_histogram)
# Rebin
if binning:
numerator_histogram = histogram_tools.rebin_histogram(
numerator_histogram, binning)
denominator_histogram = histogram_tools.rebin_histogram(
denominator_histogram, binning)
elif simple_rebin:
numerator_histogram.Rebin(simple_rebin)
denominator_histogram.Rebin(simple_rebin)
cname = "c_offline_btag_eff_mjj"
if logy:
cname += "_log"
c = TCanvas(cname, "Offline b-tag #epsilon", 800, 1000)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
if logy:
top.SetLogy()
top.Draw()
top.cd()
frame_top = numerator_histogram.Clone()
frame_top.Reset()
if x_range:
frame_top.GetXaxis().SetRangeUser(x_range[0], x_range[1])
if logy:
y_min = 0.1
y_max = max(numerator_histogram.GetMaximum(),
denominator_histogram.GetMaximum()) * 10.
else:
y_min = 0.
y_max = max(numerator_histogram.GetMaximum(),
denominator_histogram.GetMaximum()) * 1.5
frame_top.SetMinimum(y_min)
frame_top.SetMaximum(y_max)
frame_top.GetXaxis().SetLabelSize(0)
frame_top.GetXaxis().SetTitleSize(0)
if binning:
#frame_top.GetYaxis().SetTitle("Events / 1 GeV")
frame_top.GetYaxis().SetTitle("Events")
else:
frame_top.GetYaxis().SetTitle("Events")
frame_top.Draw("axis")
print "numerator integral = " + str(numerator_histogram.Integral())
print "denominator integral = " + str(denominator_histogram.Integral())
numerator_histogram.SetMarkerStyle(20)
numerator_histogram.SetMarkerColor(seaborn.GetColorRoot("default", 0))
numerator_histogram.SetLineColor(seaborn.GetColorRoot("default", 0))
numerator_histogram.Draw("same")
denominator_histogram.SetMarkerStyle(24)
denominator_histogram.SetMarkerColor(seaborn.GetColorRoot(
"default", 2))
denominator_histogram.SetLineColor(seaborn.GetColorRoot("default", 2))
denominator_histogram.Draw("same")
if legend_position == "topright":
l = TLegend(0.6, 0.6, 0.85, 0.8)
elif legend_position == "bottomright":
l = TLegend(0.6, 0.2, 0.85, 0.4)
l.SetFillColor(0)
l.SetBorderSize(0)
if numerator_legend:
l.AddEntry(numerator_histogram, numerator_legend)
else:
l.AddEntry(numerator_histogram, "CSVT+CSVM")
if denominator_legend:
l.AddEntry(denominator_histogram, denominator_legend)
else:
l.AddEntry(denominator_histogram, "No CSV")
l.Draw()
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetTopMargin(0.01)
bottom.SetBottomMargin(0.2)
bottom.Draw()
bottom.cd()
ratio_histogram = numerator_histogram.Clone()
ratio_histogram.Reset()
if x_range:
ratio_histogram.GetXaxis().SetRangeUser(x_range[0], x_range[1])
ratio_histogram.SetName(numerator_histogram.GetName() + "_ratio_" +
save_tag + str(time.time()))
ratio_histogram.SetDirectory(0)
for bin in xrange(1, numerator_histogram.GetNbinsX() + 1):
# Undo bin normalization
if numerator_histogram.GetBinError(
bin) > 0 and denominator_histogram.GetBinError(bin) > 0:
num_unnormalized = (numerator_histogram.GetBinContent(
bin))**2 / (numerator_histogram.GetBinError(bin))**2
den_unnormalized = (denominator_histogram.GetBinContent(
bin))**2 / (denominator_histogram.GetBinError(bin))**2
num = numerator_histogram.GetBinContent(bin)
den = denominator_histogram.GetBinContent(bin)
ratio = 1. * num_unnormalized / den_unnormalized
ratio_err = sqrt(ratio * (1. - ratio) / den_unnormalized)
#ratio_err = max(sqrt(ratio * (1. - ratio) / den), 1./den)
else:
ratio = 0.
ratio_err = 0.
ratio_histogram.SetBinContent(bin, ratio)
ratio_histogram.SetBinError(bin, ratio_err)
ratio_histogram.SetMarkerSize(1)
ratio_histogram.SetMarkerColor(kBlack)
ratio_histogram.SetLineColor(kBlack)
ratio_histogram.SetLineWidth(2)
ratio_histogram.GetXaxis().SetTitle("m_{jj} [GeV]")
if ratio_title:
ratio_histogram.GetYaxis().SetTitle(ratio_title)
else:
ratio_histogram.GetYaxis().SetTitle(
"Offline 2#timesb-tag efficiency")
if ratio_range:
ratio_histogram.SetMinimum(ratio_range[0])
ratio_histogram.SetMaximum(ratio_range[1])
else:
ratio_histogram.SetMinimum(0.)
ratio_histogram.SetMaximum(1.)
ratio_histogram.Draw()
c.cd()
c.SaveAs(analysis_config.figure_directory + "/OfflineBTag/" +
c.GetName() + save_tag + ".pdf")
ROOT.SetOwnership(c, False)
ROOT.SetOwnership(top, False)
ROOT.SetOwnership(bottom, False)
def EfficiencyPlot(self,
var="mjj",
logy=True,
binning=None,
simple_rebin=None,
save_tag="",
sample_xsecs=None,
x_range=None,
ratio_range=None,
legend_position="topright",
numerator_legend=None,
denominator_legend=None,
ratio_title=None,
prescaled=False,
fit=False,
fit_range=None):
numerator_histogram = None
denominator_histogram = None
for sample in self._samples:
if sample_xsecs:
xsec = sample_xsecs[sample]
numerator_normalization = 19710. * xsec / self._input_files[
"numerator"][sample].Get(
"BHistograms/h_sample_nevents").Integral()
denominator_normalization = 19710. * xsec / self._input_files[
"denominator"][sample].Get(
"BHistograms/h_sample_nevents").Integral()
if numerator_normalization != denominator_normalization:
print "[EfficiencyPlot] WARNING : Numerator and denominator normalizations do not match! Continuing, but investigate this. Numerator nevents = " + str(
self._input_files["numerator"]
[sample].Get("BHistograms/h_sample_nevents").Integral(
)) + ", denominator nevents = " + str(
self._input_files["denominator"][sample].Get(
"BHistograms/h_sample_nevents").Integral())
else:
# Check input nevents
if self._input_files["numerator"][sample].Get(
"BHistograms/h_sample_nevents"):
num_nevents = self._input_files["numerator"][sample].Get(
"BHistograms/h_sample_nevents").Integral()
den_nevents = self._input_files["denominator"][sample].Get(
"BHistograms/h_sample_nevents").Integral()
else:
num_nevents = self._input_files["numerator"][sample].Get(
"BHistograms/h_input_nevents").Integral()
den_nevents = self._input_files["denominator"][sample].Get(
"BHistograms/h_input_nevents").Integral()
if num_nevents != den_nevents:
# Allow tiny differences...?
if abs((num_nevents - den_nevents) / den_nevents) < 0.001:
print "[EfficiencyPlot] ERROR : Small inconsistency between number of events between numerator and denominator. I'm going to rescale away the difference, but you may want to fix this."
numerator_normalization = den_nevents / num_nevents
denominator_normalization = 1.
else:
print "[EfficiencyPlot] ERROR : Inconsistent number of events between numerator and denominator. Results would be wrong, so I'm aborting."
print "[EfficiencyPlot] ERROR : \tNumerator = " + str(
self._input_files["numerator"][sample].Get(
"BHistograms/h_sample_nevents").Integral())
print "[EfficiencyPlot] ERROR : \tDenominator = " + str(
self._input_files["denominator"][sample].Get(
"BHistograms/h_sample_nevents").Integral())
sys.exit(1)
else:
numerator_normalization = 1.
denominator_normalization = 1.
if numerator_histogram:
numerator_histogram.Add(
self._input_files["numerator"][sample].Get(
"BHistograms/h_pfjet_" + var), numerator_normalization)
else:
numerator_histogram = self._input_files["numerator"][
sample].Get("BHistograms/h_pfjet_" + var).Clone()
numerator_histogram.Scale(numerator_normalization)
numerator_histogram.SetDirectory(0)
numerator_histogram.SetName(numerator_histogram.GetName() +
save_tag + "_num" +
str(time.time()))
if denominator_histogram:
denominator_histogram.Add(
self._input_files["denominator"][sample].Get(
"BHistograms/h_pfjet_" + var),
denominator_normalization)
else:
denominator_histogram = self._input_files["denominator"][
sample].Get("BHistograms/h_pfjet_" + var).Clone()
denominator_histogram.Scale(denominator_normalization)
denominator_histogram.SetDirectory(0)
denominator_histogram.SetName(denominator_histogram.GetName() +
save_tag + "_den" +
str(time.time()))
# Rebin
numerator_histogram_fine = numerator_histogram.Clone()
denominator_histogram_fine = denominator_histogram.Clone()
if binning:
numerator_histogram = histogram_tools.rebin_histogram(
numerator_histogram, binning)
denominator_histogram = histogram_tools.rebin_histogram(
denominator_histogram, binning)
elif simple_rebin:
numerator_histogram.Rebin(simple_rebin)
denominator_histogram.Rebin(simple_rebin)
cname = "c_offline_btag_eff_" + var
if logy:
cname += "_log"
c = TCanvas(cname, "Offline b-tag #epsilon", 800, 1000)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
if logy:
top.SetLogy()
top.Draw()
top.cd()
frame_top = numerator_histogram.Clone()
frame_top.Reset()
if x_range:
frame_top.GetXaxis().SetRangeUser(x_range[0], x_range[1])
if logy:
y_min = 0.1
y_max = max(numerator_histogram.GetMaximum(),
denominator_histogram.GetMaximum()) * 10.
else:
y_min = 0.
y_max = max(numerator_histogram.GetMaximum(),
denominator_histogram.GetMaximum()) * 1.5
frame_top.SetMinimum(y_min)
frame_top.SetMaximum(y_max)
frame_top.GetXaxis().SetLabelSize(0)
frame_top.GetXaxis().SetTitleSize(0)
if binning:
#frame_top.GetYaxis().SetTitle("Events / 1 GeV")
frame_top.GetYaxis().SetTitle("Events")
else:
frame_top.GetYaxis().SetTitle("Events")
frame_top.Draw("axis")
print "numerator integral = " + str(numerator_histogram.Integral())
print "denominator integral = " + str(denominator_histogram.Integral())
numerator_histogram.SetMarkerStyle(20)
numerator_histogram.SetMarkerColor(seaborn.GetColorRoot("default", 0))
numerator_histogram.SetLineColor(seaborn.GetColorRoot("default", 0))
numerator_histogram.Draw("same")
denominator_histogram.SetMarkerStyle(24)
denominator_histogram.SetMarkerColor(seaborn.GetColorRoot(
"default", 2))
denominator_histogram.SetLineColor(seaborn.GetColorRoot("default", 2))
denominator_histogram.Draw("same")
if legend_position == "topright":
l = TLegend(0.6, 0.6, 0.85, 0.8)
elif legend_position == "bottomright":
l = TLegend(0.6, 0.2, 0.85, 0.4)
l.SetFillColor(0)
l.SetBorderSize(0)
if numerator_legend:
l.AddEntry(numerator_histogram, numerator_legend)
else:
l.AddEntry(numerator_histogram, "CSVT+CSVM")
if denominator_legend:
l.AddEntry(denominator_histogram, denominator_legend)
else:
l.AddEntry(denominator_histogram, "No CSV")
l.Draw()
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetTopMargin(0.01)
bottom.SetBottomMargin(0.2)
bottom.Draw()
bottom.cd()
ratio_histogram = numerator_histogram.Clone()
ratio_histogram.Reset()
if x_range:
ratio_histogram.GetXaxis().SetRangeUser(x_range[0], x_range[1])
ratio_histogram.SetName(numerator_histogram.GetName() + "_ratio_" +
save_tag + str(time.time()))
ratio_histogram.SetDirectory(0)
for bin in xrange(1, numerator_histogram.GetNbinsX() + 1):
# Undo bin normalization
if numerator_histogram.GetBinError(
bin) > 0 and denominator_histogram.GetBinError(bin) > 0:
if prescaled:
num = (numerator_histogram.GetBinContent(bin))**2 / (
numerator_histogram.GetBinError(bin))**2
den = (denominator_histogram.GetBinContent(bin))**2 / (
denominator_histogram.GetBinError(bin))**2
else:
num = numerator_histogram.GetBinContent(bin)
den = denominator_histogram.GetBinContent(bin)
ratio = 1. * num / den
ratio_err = sqrt(ratio * (1. - ratio) / den)
#ratio_err = max(sqrt(ratio * (1. - ratio) / den), 1./den)
else:
ratio = 0.
ratio_err = 0.
ratio_histogram.SetBinContent(bin, ratio)
ratio_histogram.SetBinError(bin, ratio_err)
ratio_histogram.SetMarkerSize(1)
ratio_histogram.SetMarkerColor(kBlack)
ratio_histogram.SetLineColor(kBlack)
ratio_histogram.SetLineWidth(2)
if "mjj" in var:
ratio_histogram.GetXaxis().SetTitle("m_{jj} [GeV]")
elif "pt" in var:
ratio_histogram.GetXaxis().SetTitle("p_{T} [GeV]")
if ratio_title:
ratio_histogram.GetYaxis().SetTitle(ratio_title)
else:
ratio_histogram.GetYaxis().SetTitle(
"Offline 2#timesb-tag efficiency")
if ratio_range:
ratio_histogram.SetMinimum(ratio_range[0])
ratio_histogram.SetMaximum(ratio_range[1])
else:
ratio_histogram.SetMinimum(0.)
ratio_histogram.SetMaximum(1.)
ratio_histogram.Draw()
c.cd()
c.SaveAs(analysis_config.figure_directory + "/OfflineBTag/" +
c.GetName() + save_tag + ".pdf")
ROOT.SetOwnership(c, False)
ROOT.SetOwnership(top, False)
ROOT.SetOwnership(bottom, False)
# Fit the ratio
if fit:
# Make finely binned ratio
#ratio_histogram_fine = numerator_histogram_fine.Clone()
#ratio_histogram_fine.Reset()
#if x_range:
# ratio_histogram_fine.GetXaxis().SetRangeUser(x_range[0], x_range[1])
#ratio_histogram_fine.SetName(numerator_histogram_fine.GetName() + "_ratio_fine_" + save_tag + str(time.time()))
#ratio_histogram_fine.SetDirectory(0)
#for bin in xrange(1, numerator_histogram_fine.GetNbinsX() + 1):
# # Undo bin normalization
# if numerator_histogram_fine.GetBinError(bin) > 0 and denominator_histogram_fine.GetBinError(bin) > 0:
# if prescaled:
# num = (numerator_histogram_fine.GetBinContent(bin))**2 / (numerator_histogram_fine.GetBinError(bin))**2
# den = (denominator_histogram_fine.GetBinContent(bin))**2 / (denominator_histogram_fine.GetBinError(bin))**2
# else:
# num = numerator_histogram_fine.GetBinContent(bin)
# den = denominator_histogram_fine.GetBinContent(bin)
# ratio = 1. * num / den
# ratio_err = sqrt(ratio * (1. - ratio) / den)
# #ratio_err = max(sqrt(ratio * (1. - ratio) / den), 1./den)
# else:
# ratio = 0.
# ratio_err = 0.
# ratio_histogram_fine.SetBinContent(bin, ratio)
# ratio_histogram_fine.SetBinError(bin, ratio_err)
print "Fitting " + save_tag + " with some functions"
fitter = TBinomialEfficiencyFitter(numerator_histogram,
denominator_histogram)
fits = {}
print "\nFitting quadratic"
fits["quadratic"] = TF1("quadratic", "[0]+[1]*x+[2]*x*x",
fit_range[0], fit_range[1])
fits["quadratic"].SetParameter(0, 0.01)
fits["quadratic"].SetParameter(1, -1.e-5)
fits["quadratic"].SetParLimits(1, -1.e5, 1.e5)
fits["quadratic"].SetParameter(2, 1.e-11)
fits["quadratic"].SetParLimits(2, -1.e5, 1.e5)
ratio_histogram.Fit(fits["quadratic"], "R0")
print "\nFitting quadexp"
fits["quadexp"] = TF1("quadexp",
"[0]+[1]*x+[2]*x*x+ [3]*exp([4]*x)",
fit_range[0], fit_range[1])
fits["quadexp"].SetParameter(0, 3.97724e-03)
fits["quadexp"].SetParameter(1, -3.69019e-06)
fits["quadexp"].SetParameter(2, 1.37598e-09)
fits["quadexp"].SetParameter(3, 2.95236e+00)
fits["quadexp"].SetParameter(4, -1.54774e-02)
fitter.Fit(fits["quadexp"], "R0")
print "\nFitting linexp"
fits["linexp"] = TF1("linexp", "[0]+[1]*x + [2]*exp([3]*x)",
fit_range[0], fit_range[1])
fits["linexp"].SetParameter(0, 3.97724e-03)
fits["linexp"].SetParameter(1, -3.69019e-06)
fits["linexp"].SetParameter(2, 2.95236e+00)
fits["linexp"].SetParameter(3, -1.54774e-02)
fitter.Fit(fits["linexp"], "R0")
# Print chi2/ndfs
for fit_name, fit_function in fits.iteritems():
print "chi^2/ndf (p) for " + fit_name + " = " + str(
fit_function.GetChisquare()) + "/" + str(
fit_function.GetNDF()) + " (" + str(
TMath.Prob(fit_function.GetChisquare(),
fit_function.GetNDF()))
pull_histograms = {}
for fit_name, fit_function in fits.iteritems():
print "[debug] Making pull histogram for " + fit_name
pull_histograms[fit_name] = ratio_histogram.Clone()
pull_histograms[fit_name].SetDirectory(0)
pull_histograms[fit_name].SetName("ratio_fit_pulls_" +
fit_name)
pull_histograms[fit_name].Reset()
for bin in xrange(1,
pull_histograms[fit_name].GetNbinsX() + 1):
bin_center = pull_histograms[fit_name].GetXaxis(
).GetBinCenter(bin)
if (bin_center < fit_range[0]) or (bin_center >
fit_range[1]):
#print "Pull hist : skipping bin with center " + str(bin_center)
pull_histograms[fit_name].SetBinContent(bin, 0.)
elif ratio_histogram.GetBinError(bin):
#print "[debug] Bin " + str(bin) + " / pull = " + str(round(ratio_histogram.GetBinContent(bin), 2)) + "-" + str(round(fit_function.Eval(bin_center), 2)) + "/" + str(round(ratio_histogram.GetBinError(bin), 2)) + " = " + str(round((ratio_histogram.GetBinContent(bin) - fit_function.Eval(bin_center)) / ratio_histogram.GetBinError(bin), 2))
pull_histograms[fit_name].SetBinContent(
bin, (ratio_histogram.GetBinContent(bin) -
fit_function.Eval(bin_center)) /
ratio_histogram.GetBinError(bin))
else:
pull_histograms[fit_name].SetBinContent(bin, 0.)
c_eff_fit = TCanvas("c_offline_btag_fit", "c_offline_btag_fit",
800, 1000)
top_eff_fit = TPad("top_fit", "top_fit", 0., 0.5, 1., 1.)
top_eff_fit.SetBottomMargin(0.02)
top_eff_fit.Draw()
top_eff_fit.cd()
ratio_histogram.GetXaxis().SetTitleSize(0)
ratio_histogram.GetXaxis().SetLabelSize(0)
ratio_histogram.GetYaxis().SetTitle("Offline b-tag efficiency")
ratio_histogram.Draw()
fits["quadratic"].SetLineColor(seaborn.GetColorRoot("dark", 2))
fits["quadratic"].SetLineWidth(2)
pull_histograms["quadratic"].SetLineColor(
seaborn.GetColorRoot("dark", 2))
fits["quadratic"].Draw("same")
fits["linexp"].SetLineColor(seaborn.GetColorRoot("dark", 4))
pull_histograms["linexp"].SetLineColor(
seaborn.GetColorRoot("dark", 4))
fits["linexp"].SetLineWidth(2)
fits["linexp"].Draw("same")
l_eff_fit = TLegend(0.6, 0.6, 0.88, 0.8)
l_eff_fit.SetBorderSize(0)
l_eff_fit.SetFillColor(0)
l_eff_fit.AddEntry(ratio_histogram, "Data", "pl")
l_eff_fit.AddEntry(fits["quadratic"], "Fit (quadratic)", "l")
l_eff_fit.AddEntry(fits["linexp"], "Fit (linear+exp)", "l")
l_eff_fit.Draw()
c_eff_fit.cd()
bottom_eff_fit = TPad("bottom_fit", "bottom_fit", 0., 0., 1., 0.5)
bottom_eff_fit.SetBottomMargin(0.2)
bottom_eff_fit.SetTopMargin(0.02)
bottom_eff_fit.Draw()
bottom_eff_fit.cd()
if fit_range[0] <= 296:
default_fit = "quadratic"
else:
default_fit = "linexp"
pull_histograms[default_fit].SetLineWidth(2)
pull_histograms[default_fit].GetXaxis().SetTitle("m_{jj} [GeV]")
pull_histograms[default_fit].GetYaxis().SetTitle
if "mjj" in var:
pull_histograms[default_fit].GetXaxis().SetTitle(
"m_{jj} [GeV]")
elif "pt" in var:
pull_histograms[default_fit].GetXaxis().SetTitle("p_{T} [GeV]")
pull_histograms[default_fit].GetYaxis().SetTitle(
"(fit - data) / (#sigma(data))")
pull_histograms[default_fit].SetMinimum(-10.)
pull_histograms[default_fit].SetMaximum(10.)
pull_histograms[default_fit].Draw("hist")
#pull_histograms["quadratic"].SetLineWidth(2)
#pull_histograms["quadratic"].SetLineColor(seaborn.GetColorRoot("dark", 2))
#pull_histograms["quadratic"].Draw("hist same")
c_eff_fit.cd()
c_eff_fit.SaveAs(analysis_config.figure_directory +
"/OfflineBTag/" + c_eff_fit.GetName() + save_tag +
".pdf")
ROOT.SetOwnership(c_eff_fit, False)
ROOT.SetOwnership(top_eff_fit, False)
ROOT.SetOwnership(bottom_eff_fit, False)
denominator_legend="Without offline b-tag",
ratio_title="Offline b-tag efficiency)")
#for ht_slice in [200, 250, 300, 350, 400, 450, 500, 550, 650]:
# if sr == "lowmass":
# numerator_analysis = "trigjetht" + str(ht_slice) + "_eta1p7_CSVTM"
# denominator_analysis = "trigjetht" + str(ht_slice) + "_eta1p7"
# else:
# numerator_analysis = "trigjetht" + str(ht_slice) + "_CSVTM"
# denominator_analysis = "trigjetht" + str(ht_slice)
# plotter = OfflineBTagPlots(numerator_analysis, denominator_analysis, ["JetHT_2012BCD"])
# plotter.EfficiencyPlot(var="mjj", logy=True, binning=dijet_binning, save_tag="_CSVTM_" + sr + "_jetht" + str(ht_slice), ratio_range=[0.,0.01], x_range=[0., 2000.], legend_position="topright", numerator_legend="With offline b-tag", denominator_legend="Without offline b-tag", ratio_title="Offline b-tag efficiency", prescaled=True)
if args.do_singlejet:
for analysis in ["trigbbl_CSVTM", "trigbbh_CSVTM"]:
f = TFile(
analysis_config.get_b_histogram_filename(
analysis, "BJetPlusX_2012"), "READ")
h_mjj = histogram_tools.rebin_histogram(
f.Get("BHistograms/h_pfjet_mjj"), dijet_binning)
h_mjj_btagcorr = histogram_tools.rebin_histogram(
f.Get("BHistograms/h_pfjet_mjj_btagcorr"), dijet_binning)
EfficiencyPlot(h_mjj,
h_mjj_btagcorr,
name_num="No correction",
name_den="b tag correction",
logy=True,
save_directory=analysis_config.figure_directory +
"/OfflineBTag",
save_tag="from_singlejet_" + analysis)
def rooplot(self, save_tag, fit_functions, background_workspaces, fitted_signal_workspaces=None, expected_signal_workspaces=None, log=False, x_range=None, data_binning=None, normalization_bin_width=1, draw_chi2ndf=False):
print "RooPlotting " + save_tag
c = TCanvas("c_" + save_tag, "c_" + save_tag, 800, 1200)
l = TLegend(0.55, 0.6, 0.88, 0.88)
l.SetFillColor(0)
l.SetBorderSize(0)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.03)
top.Draw()
if log:
top.SetLogy()
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetTopMargin(0.02)
bottom.SetBottomMargin(0.2)
bottom.Draw()
ROOT.SetOwnership(c, False)
ROOT.SetOwnership(top, False)
ROOT.SetOwnership(bottom, False)
top.cd()
# Make data histogram
data_hist = self.data_histogram_.Clone()
if data_binning:
data_hist = histogram_tools.rebin_histogram(data_hist, data_binning, normalization_bin_width=normalization_bin_width)
roobinning = RooBinning(len(data_binning) - 1, data_binning, "data_binning")
else:
data_hist = data_hist_raw
bins_array = array('d', [])
for bin in xrange(1, data_hist.GetNbinsX() + 1):
bins_array.append(data_hist.GetXaxis().GetBinLowEdge(bin))
bins_array.append(data_hist.GetXaxis().GetBinUpEdge(data_hist.GetNbinsX()))
roobinning = RooBinning(len(bins_array) - 1, bins_array, "data_binning")
if x_range:
x_min = x_range[0]
x_max = x_range[1]
else:
x_min = self.mjj_.GetMin()
x_max = self.mjj_.GetMax()
frame_top = TH1D("frame_top", "frame_top", 100, x_min, x_max)
if log:
frame_top.SetMaximum(data_hist.GetMaximum() * 50.)
frame_top.SetMinimum(0.1)
else:
frame_top.SetMaximum(data_hist.GetMaximum() * 1.3)
frame_top.SetMinimum(0.)
frame_top.GetYaxis().SetTitle("Events / " + str(normalization_bin_width) + " GeV")
frame_top.GetXaxis().SetTitleSize(0)
frame_top.GetXaxis().SetLabelSize(0)
#if log:
# frame_top.SetMaximum(y_max * 10.)
# frame_top.SetMinimum(y_min / 100.)
#else:
# frame_top.SetMaximum(y_max * 1.3)
# frame_top.SetMinimum(0.)
frame_top.Draw()
#data_rdh.plotOn(frame_top, RooFit.Name("Data"))
data_hist.SetMarkerStyle(20)
data_hist.SetMarkerColor(1)
data_hist.SetMarkerSize(1)
data_hist.Draw("p same")
l.AddEntry(data_hist, "Data", "pl")
background_histograms = {}
background_histograms_fitted_range = {}
chi2s = {}
chi2ndfs = {}
mc_stack = THStack("mc_stack", "")
style_counter = 0
for fit_function in fit_functions:
f_workspace = TFile(background_workspaces[fit_function], "READ")
workspace = f_workspace.Get("w")
# MC histograms
background_histograms[fit_function + "_MC"] = data_hist.Clone()
background_histograms[fit_function + "_MC"].Reset()
background_histograms[fit_function + "_MC"].SetName(fit_function + "_MC")
if len(self.backgrounds_):
for mc_background in self.backgrounds_:
print "Trying to get " + "h_pfjet_mjj_" + mc_background
f_workspace.ls()
this_mc_histogram = f_workspace.Get("h_pfjet_mjj_" + mc_background)
this_mc_histogram.SetDirectory(0)
print "Fitted normalization is scaled from initial by " + str(workspace.var(background + "_norm_backgroundonly").getVal() / workspace.var(background + "_initial_norm_backgroundonly").getVal())
this_mc_histogram.Scale(workspace.var(background + "_norm_backgroundonly").getVal() / workspace.var(background + "_initial_norm_backgroundonly").getVal())
this_mc_histogram = histogram_tools.rebin_histogram(this_mc_histogram, data_binning, normalization_bin_width=normalization_bin_width)
background_histograms[fit_function + "_MC"].Add(this_mc_histogram)
background_histograms[fit_function + "_MC"].SetLineColor(seaborn.GetColorRoot("default", style_counter))
background_histograms[fit_function + "_MC"].SetLineStyle(2)
background_histograms[fit_function + "_MC"].SetLineWidth(2)
background_histograms[fit_function + "_MC"].SetMarkerStyle(20)
background_histograms[fit_function + "_MC"].SetMarkerSize(0)
background_histograms[fit_function + "_MC"].SetDirectory(0)
background_histograms[fit_function + "_MC"].Draw("hist same")
l.AddEntry(background_histograms[fit_function + "_MC"], "MC " + fit_function, "l")
# Make histogram from fitted background
fitresult_name = "fitresult_background_data_roohistogram"
fitresult = workspace.genobj(fitresult_name)
fitresult.Print()
fit = self.make_background_tf1_from_roofitresult(fit_function, fitresult, mjj_range=x_range)
# Normalize fit
scale_factor = workspace.var("fit_norm_backgroundonly").getVal() / fit.Integral(self.mjj_.getMin(), self.mjj_.getMax())
fit.SetParameter(0, fit.GetParameter(0) * scale_factor)
fit.SetParError(0, fit.GetParError(0) * scale_factor)
background_histograms[fit_function] = data_hist.Clone()
background_histograms[fit_function].Reset()
background_histograms[fit_function].SetName(fit_function)
for bin in xrange(1, background_histograms[fit_function].GetNbinsX() + 1):
low_edge = background_histograms[fit_function].GetXaxis().GetBinLowEdge(bin)
up_edge = background_histograms[fit_function].GetXaxis().GetBinUpEdge(bin)
background_histograms[fit_function].SetBinContent(bin, fit.Integral(low_edge, up_edge) / (up_edge - low_edge) * normalization_bin_width)
chi2s[fit_function] = 0.
chi2ndfs[fit_function] = 0.
ndf = -1 * fit.GetNpar()
for bin in xrange(1, self.data_histogram_.GetNbinsX() + 1):
low_edge = self.data_histogram_.GetXaxis().GetBinLowEdge(bin)
up_edge = self.data_histogram_.GetXaxis().GetBinUpEdge(bin)
if up_edge < self.mjj_.getMin() or low_edge > self.mjj_.getMax():
continue
if self.data_histogram_.GetBinError(bin):
chi2s[fit_function] += ((self.data_histogram_.GetBinContent(bin) - fit.Integral(low_edge, up_edge)) / self.data_histogram_.GetBinError(bin))**2
ndf += 1
chi2ndfs[fit_function] = chi2s[fit_function] / ndf
print "#chi^2/NDF(" + fit_function + ") = " + str(round(chi2s[fit_function], 3)) + "/" + str(ndf) + " = " + str(round(chi2ndfs[fit_function], 3)) + " / p = " + str(TMath.Prob(chi2s[fit_function], ndf))
# Add in all the MC background
background_histograms[fit_function].Add(background_histograms[fit_function + "_MC"])
# Draw total background histogram as a colored line
background_histograms[fit_function].SetLineColor(seaborn.GetColorRoot("default", style_counter))
background_histograms[fit_function].SetLineWidth(1)
background_histograms[fit_function].SetLineStyle(1)
background_histograms[fit_function].Draw("hist same")
if draw_chi2ndf:
legend_entry = "Total background " + fit_function + "(#chi^{2}/NDF=" + str(round(chi2ndfs[fit_function], 2)) + ")"
else:
legend_entry = "Total background " + fit_function
l.AddEntry(background_histograms[fit_function], legend_entry, "l")
background_histograms[fit_function].SetDirectory(0)
f_workspace.Close()
style_counter += 1
l.Draw()
# Pull histogram
c.cd()
bottom.cd()
pull_histograms = {}
pull_histograms_fitted_range = {}
for fit_function in fit_functions:
pull_histograms[fit_function] = data_hist.Clone()
pull_histograms[fit_function].Reset()
for bin in xrange(1, data_hist.GetNbinsX() + 1):
if data_hist.GetBinError(bin) > 0:
pull = (data_hist.GetBinContent(bin) - background_histograms[fit_function].GetBinContent(bin)) / (data_hist.GetBinError(bin))
else:
pull = 0.
#print "[debug] Pull = " + str(pull)
pull_histograms[fit_function].SetBinContent(bin, pull)
pull_histograms[fit_function].SetBinError(bin, 0.)
pull_histograms_fitted_range[fit_function] = pull_histograms[fit_function].Clone()
for bin in xrange(1, pull_histograms_fitted_range[fit_function].GetNbinsX() + 1):
bin_center = pull_histograms_fitted_range[fit_function].GetXaxis().GetBinCenter(bin)
if bin_center < self.mjj_.getMin() or bin_center > self.mjj_.getMax():
pull_histograms_fitted_range[fit_function].SetBinContent(bin, 0)
pull_histograms_fitted_range[fit_function].SetBinError(bin, 0)
#pull_histogram = frame_top.pullHist("Data", "B Fit")
frame_bottom = TH1D("frame_bottom", "frame_bottom", 100, x_min, x_max)
frame_bottom.SetMinimum(-5.)
frame_bottom.SetMaximum(5.)
#pull_histogram.plotOn(frame_bottom, RooFit.Name(fit_pdf_name))
frame_bottom.GetXaxis().SetTitle("m_{jj} [GeV]")
frame_bottom.GetYaxis().SetTitle("#frac{Data - Fit}{#sigma(Data)}")
frame_bottom.Draw()
style_counter = 0
for fit_function in fit_functions:
pull_histograms[fit_function].SetLineColor(seaborn.GetColorRoot("default", style_counter))
pull_histograms[fit_function].SetLineWidth(2)
pull_histograms[fit_function].SetLineStyle(2)
pull_histograms[fit_function].Draw("hist same")
pull_histograms_fitted_range[fit_function].SetLineColor(seaborn.GetColorRoot("dark", style_counter))
pull_histograms_fitted_range[fit_function].SetLineWidth(2)
pull_histograms_fitted_range[fit_function].SetLineStyle(1)
pull_histograms_fitted_range[fit_function].Draw("hist same")
style_counter += 1
#pull_histogram.GetXaxis().SetTitle("m_{jj} [GeV]")
#pull_histogram.GetYaxis().SetTitle("#frac{Data - Fit}{#sigma(Fit)}")
#pull_histogram.Draw("same")
c.cd()
c.SaveAs("/uscms/home/dryu/Dijets/data/EightTeeEeVeeBee/Results/figures/c_" + save_tag + ".pdf")
def mc_jes_plot(analysis, model, mass):
f_new = TFile(analysis_config.get_b_histogram_filename(analysis, analysis_config.simulation.get_signal_tag(model, mass, "FULLSIM")), "READ")
h_new = f_new.Get("BHistograms/h_pfjet_mjj")
h_new.SetName("h_pfjet_mjj_newJEC")
h_new.SetDirectory(0)
h_new = histogram_tools.rebin_histogram(h_new, dijet_binning, normalization_bin_width=1.)
f_old = TFile(analysis_config.get_b_histogram_filename(analysis, analysis_config.simulation.get_signal_tag(model, mass, "FULLSIM")).replace("EightTeeEeVeeBee/BHistograms", "EightTeeEeVeeBee/BHistograms/old/JEC2012"), "READ")
h_old = f_old.Get("BHistograms/h_pfjet_mjj")
h_old.SetName("h_pfjet_mjj_oldJEC")
h_old.SetDirectory(0)
h_old = histogram_tools.rebin_histogram(h_old, dijet_binning, normalization_bin_width=1.)
c = TCanvas("c_jeccomparison_{}_{}_{}".format(analysis, model, mass), "c_jeccomparison_{}_{}_{}".format(analysis, model, mass), 800, 1200)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
#top.SetLogy()
top.Draw()
top.cd()
frame_top = TH1D("frame_top", "frame_top", 100, 0., mass*2)
frame_top.GetXaxis().SetTitleSize(0)
frame_top.GetXaxis().SetLabelSize(0)
frame_top.GetYaxis().SetTitle("Events / GeV")
frame_top.SetMinimum(0.01)
frame_top.SetMaximum(h_new.GetMaximum() * 1.3)
frame_top.Draw()
h_new.SetMarkerStyle(20)
h_new.GetXaxis().SetTitleSize(0)
h_new.GetXaxis().SetLabelSize(0)
h_new.GetYaxis().SetTitle("Events / GeV")
h_new.Draw("same")
h_old.SetMarkerStyle(25)
h_old.Draw("same")
l = TLegend(0.6, 0.6, 0.88, 0.8)
l.SetFillColor(0)
l.SetBorderSize(0)
l.AddEntry(h_new, "New JEC")
l.AddEntry(h_old, "Old JEC")
l.Draw()
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetTopMargin(0.02)
bottom.SetBottomMargin(0.2)
bottom.Draw()
bottom.cd()
frame_bottom = TH1D("frame_bottom", "frame_bottom", 100, 0., mass*2)
frame_bottom.GetXaxis().SetTitle("m_{jj} [GeV]")
frame_bottom.GetYaxis().SetTitle("New JEC / Old JEC")
frame_bottom.SetMinimum(0.5)
frame_bottom.SetMaximum(1.5)
frame_bottom.Draw()
h_ratio = h_new.Clone()
h_ratio.Divide(h_old)
h_ratio.SetMarkerStyle(21)
h_ratio.Draw("same")
c.cd()
c.SaveAs(analysis_config.figure_directory + "/" + c.GetName() + ".pdf")
f_new.Close()
f_old.Close()
from array import array
mass_bins = array("d", [1, 3, 6, 10, 16, 23, 31, 40, 50, 61, 74, 88, 103, 119, 137, 156, 176, 197, 220, 244, 270, 296, 325, 354, 386, 419, 453, 489, 526, 565, 606, 649, 693, 740, 788, 838, 890, 944, 1000, 1058, 1118, 1181, 1246, 1313, 1383, 1455, 1530, 1607, 1687, 1770, 1856, 1945, 2037, 2132, 2231, 2332, 2438, 2546, 2659, 2775, 2895, 3019, 3147, 3279, 3416, 3558, 3704, 3854, 4010, 4171, 4337, 4509, 4686, 4869, 5058, 5253, 5455, 5663, 5877, 6099, 6328, 6564, 6808, 7060, 7320, 7589, 7866, 8000])
variables = ["mjj", "pt1", "pt2", "pt_btag1", "pt_btag2"]
histograms = {}
for analysis in analyses:
histograms[analysis] = {}
for variable in variables:
histograms[analysis][variable] = files[analysis].Get("BHistograms/h_pfjet_" + variable)
histograms[analysis][variable].SetName("h_" + analysis + "_" + variable)
histograms[analysis][variable].SetDirectory(0)
if analysis == "trigmubbll_lowmass":
histograms[analysis][variable].Scale(1.7) # Prescale
if variable == "mjj":
print "(rebinning)...",
histograms[analysis][variable] = histogram_tools.rebin_histogram(histograms[analysis][variable], mass_bins, normalization_bin_width=1)
print "(done)"
elif "pt" in variable:
histograms[analysis][variable].Rebin(25)
print "Making efficiency histograms"
efficiency_pairs = {
"highmass":["trigmubbh_highmass_CSVTM", "trigmu_highmass_CSVTM"],
"lowmass":["trigmubbl_lowmass_CSVTM", "trigmu_lowmass_CSVTM"],
"llowmass":["trigmubbll_lowmass_CSVTM", "trigmu_lowmass_CSVTM"]
}
efficiency_titles = {"highmass":"160/120", "lowmass":"80/70", "llowmass":"60/53"}
efficiency_colors = {"highmass":"red", "lowmass":"green", "llowmass":"blue"}
efficiency_histograms = {}
for efficiency_name, efficiency_analyses in efficiency_pairs.iteritems():
def plot(analysis, model, mass):
f = TFile(
analysis_config.get_b_histogram_filename(
analysis,
analysis_config.simulation.get_signal_tag(model, mass, "FULLSIM")),
"READ")
h_central = f.Get("BHistograms/h_pfjet_mjj")
h_central.SetDirectory(0)
h_central = histogram_tools.rebin_histogram(h_central,
dijet_binning,
normalization_bin_width=1.)
h_up = f.Get("BHistograms/h_pfjet_mjj_BTagOfflineSFUp")
h_up.SetDirectory(0)
h_up = histogram_tools.rebin_histogram(h_up,
dijet_binning,
normalization_bin_width=1.)
h_down = f.Get("BHistograms/h_pfjet_mjj_BTagOfflineSFDown")
h_down.SetDirectory(0)
h_down = histogram_tools.rebin_histogram(h_down,
dijet_binning,
normalization_bin_width=1.)
c = TCanvas("c_offb_unc_{}_{}{}".format(analysis, model, mass),
"c_offb_unc_{}_{}{}".format(analysis, model, mass), 800, 1000)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
#top.SetLogy()
top.Draw()
top.cd()
frame_top = TH1D("frame_top", "frame_top", 100, 0., mass * 2)
frame_top.GetXaxis().SetTitleSize(0)
frame_top.GetXaxis().SetLabelSize(0)
frame_top.GetYaxis().SetTitle("Events / GeV")
frame_top.SetMinimum(0.01)
frame_top.SetMaximum(h_central.GetMaximum() * 1.3)
frame_top.Draw()
h_central.SetMarkerStyle(20)
h_central.Draw("same")
h_up.SetMarkerStyle(26)
h_up.Draw("same")
h_down.SetMarkerStyle(32)
h_down.Draw("same")
l = TLegend(0.6, 0.6, 0.88, 0.8)
l.SetFillColor(0)
l.SetBorderSize(0)
l.AddEntry(h_central, "Central", "p")
l.AddEntry(h_up, "SF unc up", "p")
l.AddEntry(h_down, "SF unc down", "p")
l.Draw()
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetTopMargin(0.02)
bottom.SetBottomMargin(0.2)
bottom.Draw()
bottom.cd()
frame_bottom = TH1D("frame_bottom", "frame_bottom", 100, 0., mass * 2)
frame_bottom.GetXaxis().SetTitle("m_{jj} [GeV]")
frame_bottom.GetYaxis().SetTitle("Unc / Central")
frame_bottom.SetMinimum(0.5)
frame_bottom.SetMaximum(1.5)
frame_bottom.Draw()
ratio_up = h_up.Clone()
ratio_up.Divide(h_central)
ratio_up.SetMarkerStyle(26)
ratio_up.Draw("p hist same")
ratio_down = h_down.Clone()
ratio_down.Divide(h_central)
ratio_down.SetMarkerStyle(32)
ratio_down.Draw("p hist same")
avg_up = TLine(0.,
h_up.Integral() / h_central.Integral(), mass * 2,
h_up.Integral() / h_central.Integral())
avg_up.SetLineStyle(2)
avg_up.SetLineColor(seaborn.GetColorRoot("pastel", 2))
avg_up.Draw("same")
avg_down = TLine(0.,
h_down.Integral() / h_central.Integral(), mass * 2,
h_down.Integral() / h_central.Integral())
avg_down.SetLineStyle(2)
avg_down.SetLineColor(seaborn.GetColorRoot("pastel", 2))
avg_down.Draw("same")
c.cd()
c.SaveAs(analysis_config.figure_directory + "/" + c.GetName() + ".pdf")
f.Close()
def draw(self,
binning=None,
logy=False,
draw_ratio=False,
draw_pull=False,
x_range=None,
y_range=None,
ratio_range=None,
pull_range=None,
pull_dataerrors=False,
save_tag="",
cms_label=None,
x_title=None,
y_title=None,
ratio_title=None,
color_scheme="cubehelixhuge",
complex_rebinning=None,
simple_rebinning=None,
legend_position="topright",
lumi_string=None,
save_directory=None):
print "[debug] x_range = ",
print x_range
if complex_rebinning:
self._data_histogram = histogram_tools.rebin_histogram(
self._data_histogram,
complex_rebinning,
normalization_bin_width=1.)
self._mc_sum = histogram_tools.rebin_histogram(
self._mc_sum, complex_rebinning, normalization_bin_width=1.)
for mc_name in self._mc_names:
self._mc_histograms[mc_name] = histogram_tools.rebin_histogram(
self._mc_histograms[mc_name],
complex_rebinning,
normalization_bin_width=1.)
self._mc_histograms[mc_name].SetDirectory(0)
for signal_name in self._signal_names:
self._signal_histograms[
signal_name] = histogram_tools.rebin_histogram(
self._signal_histograms[signal_name],
complex_rebinning,
normalization_bin_width=1.)
self._signal_histograms[signal_name].SetDirectory(0)
elif simple_rebinning:
self._data_histogram.Rebin(simple_rebinning)
for mc_name in self._mc_names:
self._mc_histograms[mc_name].Rebin(simple_rebinning)
for signal_name in self._signal_names:
self._signal_histograms[signal_name].Rebin(simple_rebinning)
if draw_ratio or draw_pull:
self._canvas = TCanvas("c_" + save_tag, "c_" + save_tag, 800, 800)
self._top = TPad("top_" + save_tag, "top_" + save_tag, 0.,
self._bottom_fraction, 1., 1.)
self._top.SetTopMargin(0.07)
self._top.SetBottomMargin(0.03)
self._top.Draw()
self._canvas.cd()
self._bottom = TPad("bottom_" + save_tag, "bottom_" + save_tag, 0.,
0., 1., self._bottom_fraction)
self._bottom.SetTopMargin(0.04)
self._bottom.SetBottomMargin(0.28)
self._bottom.Draw()
self._canvas.cd()
else:
self._canvas = TCanvas("c_" + save_tag, "c_" + save_tag, 800, 600)
if legend_position == "topright":
self._legend = TLegend(0.5, 0.53, 0.92, 0.87)
elif legend_position == "topleft":
self._legend = TLegend(0.2, 0.6, 0.45, 0.8)
elif legend_position == "bottomright":
self._legend = TLegend(0.6, 0.2, 0.85, 0.4)
elif legend_position == "bottomleft":
self._legend = TLegend(0.2, 0.2, 0.45, 0.4)
self._legend.SetFillColor(0)
self._legend.SetBorderSize(0)
if draw_ratio or draw_pull:
self._top.cd()
self._frame_top = self._data_histogram.Clone()
self._frame_top.Reset()
if logy:
if draw_ratio or draw_pull:
self._top.SetLogy()
else:
self._canvas.SetLogy()
if x_range:
self._frame_top.GetXaxis().SetRangeUser(x_range[0], x_range[1])
if y_range:
y_min = y_range[0]
y_max = y_range[1]
else:
if logy:
y_min = 0.1
y_max = max(self._data_histogram.GetMaximum(),
self._mc_sum.GetMaximum()) * 50.
else:
y_min = 0.
y_max = max(self._data_histogram.GetMaximum(),
self._mc_sum.GetMaximum()) * 1.3
self._frame_top.SetMinimum(y_min)
self._frame_top.SetMaximum(y_max)
if draw_ratio or draw_pull:
self._frame_top.GetXaxis().SetLabelSize(0)
self._frame_top.GetXaxis().SetTitleSize(0)
else:
if x_title:
self._frame_top.GetXaxis().SetTitle(x_title)
if y_title:
self._frame_top.GetYaxis().SetTitle(y_title)
self._frame_top.GetYaxis().SetTitleSize(0.06)
self._frame_top.GetYaxis().SetTitleOffset(0.95)
self._frame_top.GetYaxis().SetLabelSize(0.035 /
(1. - self._bottom_fraction))
self._frame_top.Draw("axis")
# MC: sort histograms by integral and stack
for tf1_name in self._tf1_names:
if tf1_name in self._tf1_colors:
self._tf1s[tf1_name].SetLineColor(self._tf1_colors[tf1_name])
self._tf1s[tf1_name].SetLineWidth(2)
self._tf1s[tf1_name].SetLineStyle(1)
self._tf1s[tf1_name].Draw("same")
if not self._draw_tf1only:
mc_names_sorted = sorted(
self._mc_names,
key=lambda x: self._mc_histograms[x].Integral())
self._mc_stack = THStack("mc_stack", "mc_stack")
style_counter = 0
for mc_name in mc_names_sorted:
if mc_name in self._mc_colors:
self._mc_histograms[mc_name].SetFillColor(
self._mc_colors[mc_name])
else:
self._mc_histograms[mc_name].SetFillColor(
seaborn.GetColorRoot(color_scheme, style_counter,
len(mc_names_sorted)))
self._mc_histograms[mc_name].SetLineColor(kBlack)
self._mc_histograms[mc_name].SetLineWidth(1)
self._mc_histograms[mc_name].SetFillStyle(1001)
self._mc_histograms[mc_name].SetMarkerStyle(20)
self._mc_histograms[mc_name].SetMarkerSize(0)
self._mc_stack.Add(self._mc_histograms[mc_name])
self._mc_stack.Draw("hist same")
ROOT.SetOwnership(self._mc_stack, False)
for mc_name in mc_names_sorted:
ROOT.SetOwnership(self._mc_histograms[mc_name], False)
for signal_name in self._signal_names:
if self._stack_signals:
self._signal_histograms[signal_name].Add(self._mc_sum)
if signal_name in self._signal_styles:
self._signal_histograms[signal_name].SetLineStyle(
self._signal_styles[signal_name])
if signal_name in self._signal_colors:
self._signal_histograms[signal_name].SetLineColor(
self._signal_colors[signal_name])
self._signal_histograms[signal_name].SetLineWidth(3)
self._signal_histograms[signal_name].Draw("hist same ][")
self._data_histogram.SetMarkerStyle(20)
self._data_histogram.SetMarkerSize(1)
self._data_histogram.SetMarkerColor(kBlack)
self._data_histogram.Draw("same")
print "[debug] Data integral = " + str(
self._data_histogram.Integral()) + " / MC integral = " + str(
self._mc_sum.Integral())
self._frame_top.Draw("axis same")
self._legend.AddEntry(self._data_histogram, self._data_name, "ep")
for tf1_name in self._tf1_names:
self._legend.AddEntry(self._tf1s[tf1_name], tf1_name, "l")
if not self._draw_tf1only:
for mc_name in reversed(mc_names_sorted):
self._legend.AddEntry(self._mc_histograms[mc_name], mc_name,
"lf")
for signal_name in self._signal_names:
self._legend.AddEntry(self._signal_histograms[signal_name],
signal_name, "l")
self._legend.Draw()
if cms_label != None:
if draw_ratio or draw_pull:
cms_label_offset = 1. - self._bottom_fraction
else:
cms_label_offset = 1.
Root.CMSLabel(0.22, 0.8, cms_label, 1, 0.65, cms_label_offset)
if lumi_string:
Root.myText(0.74, 0.95, kBlack, lumi_string, 0.57)
if draw_ratio:
self._canvas.cd()
self._bottom.cd()
self._frame_bottom = self._data_histogram.Clone()
self._frame_bottom.Reset()
if ratio_range:
self._frame_bottom.SetMinimum(ratio_range[0])
self._frame_bottom.SetMaximum(ratio_range[1])
else:
self._frame_bottom.SetMinimum(0.)
self._frame_bottom.SetMaximum(2.)
if x_range:
self._frame_bottom.GetXaxis().SetRangeUser(
x_range[0], x_range[1])
if x_title:
self._frame_bottom.GetXaxis().SetTitle(x_title)
self._frame_bottom.GetXaxis().SetTitleSize(0.04 /
self._bottom_fraction)
self._frame_bottom.GetXaxis().SetTitleOffset(1.15)
self._frame_bottom.GetXaxis().SetLabelOffset(0.005 * 4.4)
self._frame_bottom.GetYaxis().SetTitleSize(0.04 /
self._bottom_fraction)
self._frame_bottom.GetYaxis().SetLabelSize(0.04 /
self._bottom_fraction)
self._frame_bottom.GetYaxis().SetTitleOffset(0.6)
self._frame_bottom.GetYaxis().SetLabelOffset(0.005 * 2.2)
self._frame_bottom.GetYaxis().SetTitle("Data / Background")
self._frame_bottom.Draw("axis")
self._ratio_histogram = self._data_histogram.Clone()
self._ratio_histogram.Reset()
for bin in xrange(1, self._data_histogram.GetNbinsX() + 1):
num = self._data_histogram.GetBinContent(bin)
num_err = self._data_histogram.GetBinError(bin)
den = self._mc_sum.GetBinContent(bin)
den_err = self._mc_sum.GetBinError(bin)
if num > 0 and den > 0:
ratio = num / den
ratio_err = ratio * sqrt((num_err / num)**2 +
(den_err / den)**2)
else:
ratio = 0.
ratio_err = 0.
self._ratio_histogram.SetBinContent(bin, ratio)
self._ratio_histogram.SetBinError(bin, ratio_err)
print "[debug] Set ratio(mjj=" + str(
self._data_histogram.GetXaxis().GetBinCenter(
bin)) + ") = " + str(ratio)
self._ratio_histogram.SetMarkerStyle(24)
self._ratio_histogram.SetMarkerSize(1)
self._ratio_histogram.SetMarkerColor(kBlack)
self._ratio_histogram.SetLineWidth(1)
self._ratio_histogram.SetLineColor(kBlack)
self._ratio_histogram.Draw("same")
self._frame_bottom.Draw("axis same")
elif draw_pull:
self._canvas.cd()
self._bottom.cd()
self._frame_bottom = self._data_histogram.Clone()
self._frame_bottom.Reset()
if pull_range:
self._frame_bottom.SetMinimum(pull_range[0])
self._frame_bottom.SetMaximum(pull_range[1])
else:
self._frame_bottom.SetMinimum(-3.)
self._frame_bottom.SetMaximum(3.)
if x_range:
self._frame_bottom.GetXaxis().SetRangeUser(
x_range[0], x_range[1])
if pull_dataerrors:
self._frame_bottom.GetYaxis().SetTitle(
"#frac{(Data - Bkgd)}{#sqrt{Data}}")
else:
self._frame_bottom.GetYaxis().SetTitle(
"#frac{(Data - Bkgd)}{#sqrt{Bkgd}}")
if x_title:
self._frame_bottom.GetXaxis().SetTitle(x_title)
self._frame_bottom.GetXaxis().SetTitleSize(0.035 /
self._bottom_fraction)
self._frame_bottom.GetXaxis().SetLabelSize(0.035 /
self._bottom_fraction)
self._frame_bottom.GetXaxis().SetTitleOffset(1.07)
self._frame_bottom.GetXaxis().SetLabelOffset(0.005 * 4.2)
self._frame_bottom.GetYaxis().SetTitleSize(0.035 /
self._bottom_fraction)
self._frame_bottom.GetYaxis().SetLabelSize(0.035 /
self._bottom_fraction)
self._frame_bottom.GetYaxis().SetTitleOffset(0.55)
self._frame_bottom.GetYaxis().SetLabelOffset(0.005 * 2.2)
self._frame_bottom.Draw("axis")
self._pull_histogram = self._data_histogram.Clone()
self._pull_histogram.Reset()
self._systematic_histogram = self._data_histogram.Clone()
self._systematic_histogram.Reset()
for bin in xrange(1, self._data_histogram.GetNbinsX() + 1):
data = self._data_histogram.GetBinContent(bin)
data_err = self._data_histogram.GetBinError(bin)
background = self._mc_sum.GetBinContent(bin)
background_err = self._mc_sum.GetBinError(bin)
if complex_rebinning:
# Bins are scaled by bin width. Need to undo before calculating background uncertainty.
bin_width = self._data_histogram.GetXaxis().GetBinWidth(
bin)
data = data * bin_width
data_err = data_err * bin_width
background = background * bin_width
background_err = background_err * bin_width
if background > 0:
if pull_dataerrors:
pull = (data - background) / data_err
else:
pull = (data - background) / sqrt(background)
syst = background_err / sqrt(background)
else:
pull = 0.
syst = 0.
self._pull_histogram.SetBinContent(bin, pull)
self._pull_histogram.SetBinError(bin, 0.)
self._systematic_histogram.SetBinContent(bin, 0.)
self._systematic_histogram.SetBinError(bin, syst)
self._systematic_histogram.SetFillColor(17)
self._systematic_histogram.SetFillStyle(1001)
self._systematic_histogram.SetMarkerStyle(20)
self._systematic_histogram.SetMarkerSize(0)
#self._systematic_histogram.Draw("e2 hist same")
self._pull_histogram.SetMarkerStyle(24)
self._pull_histogram.SetMarkerSize(1)
self._pull_histogram.SetMarkerColor(kBlack)
self._pull_histogram.SetLineStyle(1)
self._pull_histogram.SetLineWidth(2)
self._pull_histogram.SetLineColor(1)
self._pull_histogram.SetFillStyle(1001)
self._pull_histogram.SetFillColor(
seaborn.GetColorRoot("default", 2))
#self._pull_histogram.SetLineWidth(1)
#self._pull_histogram.SetLineColor(kBlack)
self._pull_histogram.Draw("hist same")
self._frame_bottom.Draw("axis same")
self._signal_pulls = {}
for signal_name in self._signal_names:
if self._signal_plot_pull[signal_name]:
self._signal_pulls[signal_name] = self._signal_histograms[
signal_name].Clone()
for bin in xrange(
1,
self._signal_histograms[signal_name].GetNbinsX() +
1):
signal_counts = self._signal_histograms[
signal_name].GetBinContent(bin)
signal_err = self._signal_histograms[
signal_name].GetBinError(bin)
background = self._mc_sum.GetBinContent(bin)
background_err = self._mc_sum.GetBinError(bin)
if complex_rebinning:
# Bins are scaled by bin width. Need to undo before calculating background uncertainty.
bin_width = self._data_histogram.GetXaxis(
).GetBinWidth(bin)
signal_counts = signal_counts * bin_width
signal_err = signal_err * bin_width
background = background * bin_width
background_err = background_err * bin_width
if background > 0:
if pull_dataerrors:
pull = signal_counts / signal_err
else:
pull = signal_counts / sqrt(background)
else:
pull = 0.
self._signal_pulls[signal_name].SetBinContent(
bin, pull)
self._signal_pulls[signal_name].SetBinError(bin, 0.)
self._signal_pulls[signal_name].Draw("hist same")
self._canvas.cd()
if draw_pull or draw_ratio:
ROOT.SetOwnership(self._canvas, False)
ROOT.SetOwnership(self._top, False)
ROOT.SetOwnership(self._bottom, False)
if save_directory:
self._canvas.cd()
self._canvas.SaveAs(save_directory + "/" + self._canvas.GetName() +
".png")
self._canvas.SaveAs(save_directory + "/" + self._canvas.GetName() +
".pdf")
self._canvas.SaveAs(save_directory + "/" + self._canvas.GetName() +
".eps")
return self._canvas
def plot(analysis, model, mass):
f = TFile(analysis_config.get_b_histogram_filename(analysis, analysis_config.simulation.get_signal_tag(model, mass, "FULLSIM")), "READ")
h_central = f.Get("BHistograms/h_pfjet_mjj")
h_central.SetDirectory(0)
h_central = histogram_tools.rebin_histogram(h_central, dijet_binning, normalization_bin_width=1.)
h_up = f.Get("BHistograms/h_pfjet_mjj_BTagOfflineSFUp")
h_up.SetDirectory(0)
h_up = histogram_tools.rebin_histogram(h_up, dijet_binning, normalization_bin_width=1.)
h_down = f.Get("BHistograms/h_pfjet_mjj_BTagOfflineSFDown")
h_down.SetDirectory(0)
h_down = histogram_tools.rebin_histogram(h_down, dijet_binning, normalization_bin_width=1.)
c = TCanvas("c_offb_unc_{}_{}{}".format(analysis, model, mass), "c_offb_unc_{}_{}{}".format(analysis, model, mass), 800, 1000)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
#top.SetLogy()
top.Draw()
top.cd()
frame_top = TH1D("frame_top", "frame_top", 100, 0., mass*2)
frame_top.GetXaxis().SetTitleSize(0)
frame_top.GetXaxis().SetLabelSize(0)
frame_top.GetYaxis().SetTitle("Events / GeV")
frame_top.SetMinimum(0.01)
frame_top.SetMaximum(h_central.GetMaximum() * 1.3)
frame_top.Draw()
h_central.SetMarkerStyle(20)
h_central.Draw("same")
h_up.SetMarkerStyle(26)
h_up.Draw("same")
h_down.SetMarkerStyle(32)
h_down.Draw("same")
l = TLegend(0.6, 0.6, 0.88, 0.8)
l.SetFillColor(0)
l.SetBorderSize(0)
l.AddEntry(h_central, "Central", "p")
l.AddEntry(h_up, "SF unc up", "p")
l.AddEntry(h_down, "SF unc down", "p")
l.Draw()
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetTopMargin(0.02)
bottom.SetBottomMargin(0.2)
bottom.Draw()
bottom.cd()
frame_bottom = TH1D("frame_bottom", "frame_bottom", 100, 0., mass*2)
frame_bottom.GetXaxis().SetTitle("m_{jj} [GeV]")
frame_bottom.GetYaxis().SetTitle("Unc / Central")
frame_bottom.SetMinimum(0.5)
frame_bottom.SetMaximum(1.5)
frame_bottom.Draw()
ratio_up = h_up.Clone()
ratio_up.Divide(h_central)
ratio_up.SetMarkerStyle(26)
ratio_up.Draw("p hist same")
ratio_down = h_down.Clone()
ratio_down.Divide(h_central)
ratio_down.SetMarkerStyle(32)
ratio_down.Draw("p hist same")
avg_up = TLine(0., h_up.Integral() / h_central.Integral(), mass*2, h_up.Integral() / h_central.Integral())
avg_up.SetLineStyle(2)
avg_up.SetLineColor(seaborn.GetColorRoot("pastel", 2))
avg_up.Draw("same")
avg_down = TLine(0., h_down.Integral() / h_central.Integral(), mass*2, h_down.Integral() / h_central.Integral())
avg_down.SetLineStyle(2)
avg_down.SetLineColor(seaborn.GetColorRoot("pastel", 2))
avg_down.Draw("same")
c.cd()
c.SaveAs(analysis_config.figure_directory + "/" + c.GetName() + ".pdf")
f.Close()
plotter = OfflineBTagPlots(numerator_analysis, denominator_analysis, ["SingleMu_2012"])
plotter.EfficiencyPlot(var="mjj", logy=True, binning=dijet_binning, save_tag="_" + wp + "_" + sr + "_onbtag_singlemu", ratio_range=[0.,0.1], x_range=[0., 2000.], legend_position="topright", numerator_legend="With online b-tag", denominator_legend="Without online b-tag", ratio_title="Online b-tag efficiency (no offline CSV)")
if args.do_jetht:
for sr in ["lowmass", "highmass"]:
if sr == "lowmass":
numerator_analysis = "trigjetht_eta1p7_CSVTM"
denominator_analysis = "trigjetht_eta1p7"
else:
numerator_analysis = "trigjetht_eta2p2_CSVTM"
denominator_analysis = "trigjetht_eta2p2"
plotter = OfflineBTagPlots(numerator_analysis, denominator_analysis, ["JetHT_2012BCD"])
plotter.FrankenEfficiencyPlot(logy=True, binning=dijet_binning, save_tag="_JetHT_CSVTM_" + sr, ratio_range=[0.,0.01], x_range=[0., 2000.], legend_position="topright", numerator_legend="With offline b-tag", denominator_legend="Without offline b-tag", ratio_title="Offline b-tag efficiency)")
#for ht_slice in [200, 250, 300, 350, 400, 450, 500, 550, 650]:
# if sr == "lowmass":
# numerator_analysis = "trigjetht" + str(ht_slice) + "_eta1p7_CSVTM"
# denominator_analysis = "trigjetht" + str(ht_slice) + "_eta1p7"
# else:
# numerator_analysis = "trigjetht" + str(ht_slice) + "_CSVTM"
# denominator_analysis = "trigjetht" + str(ht_slice)
# plotter = OfflineBTagPlots(numerator_analysis, denominator_analysis, ["JetHT_2012BCD"])
# plotter.EfficiencyPlot(var="mjj", logy=True, binning=dijet_binning, save_tag="_CSVTM_" + sr + "_jetht" + str(ht_slice), ratio_range=[0.,0.01], x_range=[0., 2000.], legend_position="topright", numerator_legend="With offline b-tag", denominator_legend="Without offline b-tag", ratio_title="Offline b-tag efficiency", prescaled=True)
if args.do_singlejet:
for analysis in ["trigbbl_CSVTM", "trigbbh_CSVTM"]:
f = TFile(analysis_config.get_b_histogram_filename(analysis, "BJetPlusX_2012"), "READ")
h_mjj = histogram_tools.rebin_histogram(f.Get("BHistograms/h_pfjet_mjj"), dijet_binning)
h_mjj_btagcorr = histogram_tools.rebin_histogram(f.Get("BHistograms/h_pfjet_mjj_btagcorr"), dijet_binning)
EfficiencyPlot(h_mjj, h_mjj_btagcorr, name_num="No correction", name_den="b tag correction", logy=True, save_directory=analysis_config.figure_directory + "/OfflineBTag", save_tag="from_singlejet_"+analysis)
this_h_mjj.Scale(19700. * qcd_cross_sections[qcd_sample] /
input_nevents)
this_h_mjj_truthbb = f.Get("BHistograms/h_pfjet_mjj_truthbb").Clone()
this_h_mjj_truthbb.Scale(19700. * qcd_cross_sections[qcd_sample] /
input_nevents)
if not h_mjj:
h_mjj = this_h_mjj.Clone()
h_mjj.SetDirectory(0)
h_mjj_truthbb = this_h_mjj_truthbb.Clone()
h_mjj_truthbb.SetDirectory(0)
else:
h_mjj.Add(this_h_mjj)
h_mjj_truthbb.Add(this_h_mjj_truthbb)
f.Close()
h_mjj = histogram_tools.rebin_histogram(h_mjj,
dijet_binning,
normalization_bin_width=1.)
h_mjj_truthbb = histogram_tools.rebin_histogram(h_mjj_truthbb,
dijet_binning,
normalization_bin_width=1.)
c = TCanvas("c_purity_{}".format(analysis), "c_purity_{}".format(analysis),
800, 1000)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.02)
top.SetLogy()
top.Draw()
top.cd()
h_mjj.SetLineColor(1)
h_mjj.SetLineWidth(2)
h_mjj.GetXaxis().SetTitleSize(0)
h_mjj.GetXaxis().SetLabelSize(0)
def rooplot(save_tag,
fit_functions,
background_workspace,
fitted_signal_workspaces=None,
expected_signal_workspaces=None,
log=False,
x_range=None,
data_binning=None,
normalization_bin_width=1,
draw_chi2ndf=False,
draw_chi2prob=False,
data_histogram=None,
trigger_correction=None,
draw_trigeff=False):
print "Making plot " + save_tag
c = TCanvas("c_" + save_tag, "c_" + save_tag, 800, 1200)
l = TLegend(0.65, 0.55, 0.88, 0.8)
l.SetFillColor(0)
l.SetBorderSize(0)
top = TPad("top", "top", 0., 0.5, 1., 1.)
top.SetBottomMargin(0.03)
top.SetLeftMargin(0.15)
top.Draw()
if log:
top.SetLogy()
c.cd()
bottom = TPad("bottom", "bottom", 0., 0., 1., 0.5)
bottom.SetLeftMargin(0.15)
bottom.SetTopMargin(0.02)
bottom.SetBottomMargin(0.2)
bottom.Draw()
ROOT.SetOwnership(c, False)
ROOT.SetOwnership(top, False)
ROOT.SetOwnership(bottom, False)
top.cd()
background_histograms = {}
background_histograms_fitted_range = {}
pull_histograms = {}
pull_histograms_fitted_range = {}
chi2s = {}
ndfs = {}
chi2ndfs = {}
style_counter = 0
first = True
print "[rooplot] INFO : Opening workspace file {}".format(
background_workspace)
f_workspace = TFile(background_workspace, "READ")
workspace = f_workspace.Get("w")
workspace.Print()
for fit_function in fit_functions:
mjj = workspace.var("mjj")
# First time: draw frame and data histogram
if first:
first = False
if not data_histogram:
fine_bins_array = array('d', range(int(x_min), int(x_max + 1)))
if data_binning:
roobinning = RooBinning(
len(data_binning) - 1, data_binning, "data_binning")
else:
roobinning = RooBinning(
len(fine_bins_array) - 1, fine_bins_array,
"data_binning")
fine_roobinning = RooBinning(
len(fine_bins_array) - 1, fine_bins_array,
"data_fine_binning")
data_histogram = workspace.data("data_obs").createHistogram(
"data", mjj, RooFit.Binning(roobinning))
data_histogram.SetDirectory(0)
data_histogram_unrebinned = workspace.data(
"data_obs").createHistogram(
"data_fine", mjj, RooFit.Binning(fine_roobinning))
data_histogram_unrebinned.SetDirectory(0)
else:
data_histogram_unrebinned = data_histogram.Clone()
data_histogram_unrebinned.SetDirectory(0)
data_histogram = histogram_tools.rebin_histogram(
data_histogram, mass_bins, normalization_bin_width=1)
if x_range:
x_min = x_range[0]
x_max = x_range[1]
else:
x_min = mjj.GetMin()
x_max = mjj.GetMax()
frame_top = TH1D("frame_top", "frame_top", 100, x_min, x_max)
frame_top.SetDirectory(0)
if log:
frame_top.SetMaximum(data_histogram.GetMaximum() * 50.)
frame_top.SetMinimum(0.1)
else:
frame_top.SetMaximum(data_histogram.GetMaximum() * 1.3)
frame_top.SetMinimum(0.)
if normalization_bin_width == 1:
frame_top.GetYaxis().SetTitle("Events / GeV")
else:
frame_top.GetYaxis().SetTitle(
"Events / " + str(int(normalization_bin_width)) + " GeV")
frame_top.GetYaxis().SetTitleSize(0.06)
frame_top.GetYaxis().SetTitleOffset(1)
frame_top.GetYaxis().SetLabelSize(0.06)
frame_top.GetXaxis().SetTitleSize(0)
frame_top.GetXaxis().SetLabelSize(0)
frame_top.Draw()
data_histogram.SetMarkerStyle(20)
data_histogram.SetMarkerColor(1)
data_histogram.SetMarkerSize(1)
data_histogram.Draw("p same")
l.AddEntry(data_histogram, "Data", "pl")
# Make histogram from fitted background
fitresult_name = "fitresult_model_" + fit_function + "_rooDatahist"
fitresult = workspace.genobj(fitresult_name)
fitresult.Print()
fit = make_background_tf1_from_roofitresult(fit_function,
fitresult,
mjj_range=x_range,
trigger_correction=None)
print "[debug] Printing fit function"
fit.Print()
# Normalize fit
scale_factor = workspace.var("background_" + fit_function +
"_norm").getVal() / fit.Integral(
mjj.getMin(), mjj.getMax())
fit.SetParameter(0, fit.GetParameter(0) * scale_factor)
fit.SetParError(0, fit.GetParError(0) * scale_factor)
background_histograms[fit_function] = data_histogram.Clone()
background_histograms[fit_function].Reset()
background_histograms[fit_function].SetDirectory(0)
background_histograms[fit_function].SetName(fit_function)
for bin in xrange(1,
background_histograms[fit_function].GetNbinsX() + 1):
low_edge = background_histograms[fit_function].GetXaxis(
).GetBinLowEdge(bin)
up_edge = background_histograms[fit_function].GetXaxis(
).GetBinUpEdge(bin)
background_histograms[fit_function].SetBinContent(
bin,
fit.Integral(low_edge, up_edge) / (up_edge - low_edge) *
normalization_bin_width)
#if draw_trigeff:
# fit_notrigcorr = make_background_tf1_from_roofitresult(fit_function, fitresult, mjj_range=x_range, trigger_correction=None)
# scale_factor = workspace.var("background_" + fit_function + "_norm").getVal() / fit_notrigcorr.Integral(mjj.getMin(), mjj.getMax())
# fit_notrigcorr.SetParameter(0, fit_notrigcorr.GetParameter(0) * scale_factor)
# fit_notrigcorr.SetParError(0, fit_notrigcorr.GetParError(0) * scale_factor)
# background_histograms[fit_function + "_notrigcorr"] = data_histogram.Clone()
# background_histograms[fit_function + "_notrigcorr"].Reset()
# background_histograms[fit_function + "_notrigcorr"].SetDirectory(0)
# background_histograms[fit_function + "_notrigcorr"].SetName(fit_function)
# for bin in xrange(1, background_histograms[fit_function + "_notrigcorr"].GetNbinsX() + 1):
# low_edge = background_histograms[fit_function + "_notrigcorr"].GetXaxis().GetBinLowEdge(bin)
# up_edge = background_histograms[fit_function + "_notrigcorr"].GetXaxis().GetBinUpEdge(bin)
# background_histograms[fit_function + "_notrigcorr"].SetBinContent(bin, fit_notrigcorr.Integral(low_edge, up_edge) / (up_edge - low_edge) * normalization_bin_width)
chi2s[fit_function] = 0.
chi2ndfs[fit_function] = 0.
ndfs[fit_function] = -1 * fit.GetNpar()
for bin in xrange(1, data_histogram_unrebinned.GetNbinsX() + 1):
low_edge = data_histogram_unrebinned.GetXaxis().GetBinLowEdge(bin)
up_edge = data_histogram_unrebinned.GetXaxis().GetBinUpEdge(bin)
if up_edge <= mjj.getMin() or low_edge >= mjj.getMax():
continue
# Data errors
#this_chi2 = ((data_histogram_unrebinned.GetBinContent(bin) - fit.Integral(low_edge, up_edge)) / data_histogram_unrebinned.GetBinError(bin))**2
# Fit errors
this_fit = fit.Integral(low_edge, up_edge)
this_chi2 = (
(data_histogram_unrebinned.GetBinContent(bin) - this_fit) /
(this_fit**0.5))**2
chi2s[fit_function] += this_chi2
ndfs[fit_function] += 1
#print "[debug] Bin {}, chi2 = (({} - {})/{})**2=\t{}".format(bin, data_histogram_unrebinned.GetBinContent(bin), fit.Integral(low_edge, up_edge), this_fit**0.5, this_chi2)
chi2ndfs[fit_function] = chi2s[fit_function] / ndfs[fit_function]
print "#chi^2/NDF(" + fit_function + ") = " + str(
round(chi2s[fit_function], 3)) + "/" + str(
ndfs[fit_function]) + " = " + str(
round(chi2ndfs[fit_function], 3)) + " / p = " + str(
TMath.Prob(chi2s[fit_function], ndfs[fit_function]))
# Draw total background histogram as a colored line
background_histograms[fit_function].SetLineColor(
seaborn.GetColorRoot("default", style_counter))
background_histograms[fit_function].SetLineWidth(1)
background_histograms[fit_function].SetLineStyle(1)
background_histograms[fit_function].Draw("hist same")
if draw_chi2ndf:
legend_entry = fit_function + " (#chi^{2}/NDF=" + str(
round(chi2ndfs[fit_function], 2)) + ")"
elif draw_chi2prob:
legend_entry = fit_function + " (p=" + str(
round(ROOT.TMath.Prob(chi2s[fit_function], ndfs[fit_function]),
2)) + ")"
else:
legend_entry = fit_function
l.AddEntry(background_histograms[fit_function], legend_entry, "l")
background_histograms[fit_function].SetDirectory(0)
if draw_trigeff:
background_histograms[fit_function + "_notrigcorr"].SetLineColor(
seaborn.GetColorRoot("pastel", style_counter))
background_histograms[fit_function + "_notrigcorr"].SetLineWidth(1)
background_histograms[fit_function + "_notrigcorr"].SetLineStyle(3)
background_histograms[fit_function +
"_notrigcorr"].Draw("hist same")
legend_entry = fit_function + ", no trig corr"
l.AddEntry(background_histograms[fit_function + "_notrigcorr"],
legend_entry, "l")
background_histograms[fit_function + "_notrigcorr"].SetDirectory(0)
pull_histograms[fit_function] = data_histogram.Clone()
pull_histograms[fit_function].Reset()
pull_histograms[fit_function].SetDirectory(0)
for bin in xrange(1, data_histogram.GetNbinsX() + 1):
if data_histogram.GetBinError(bin) > 0:
pull = (data_histogram.GetBinContent(bin) -
background_histograms[fit_function].GetBinContent(bin)
) / (data_histogram.GetBinError(bin))
else:
pull = 0.
#print "[debug] Pull = " + str(pull)
pull_histograms[fit_function].SetBinContent(bin, pull)
pull_histograms[fit_function].SetBinError(bin, 0.)
if draw_trigeff:
pull_histograms[fit_function +
"_notrigcorr"] = data_histogram.Clone()
pull_histograms[fit_function + "_notrigcorr"].Reset()
pull_histograms[fit_function + "_notrigcorr"].SetDirectory(0)
for bin in xrange(1, data_histogram.GetNbinsX() + 1):
if data_histogram.GetBinError(bin) > 0:
pull = (
data_histogram.GetBinContent(bin) -
background_histograms[fit_function + "_notrigcorr"].
GetBinContent(bin)) / (data_histogram.GetBinError(bin))
else:
pull = 0.
#print "[debug] Pull = " + str(pull)
pull_histograms[fit_function + "_notrigcorr"].SetBinContent(
bin, pull)
pull_histograms[fit_function + "_notrigcorr"].SetBinError(
bin, 0.)
pull_histograms_fitted_range[fit_function] = pull_histograms[
fit_function].Clone()
pull_histograms_fitted_range[fit_function].SetDirectory(0)
for bin in xrange(
1, pull_histograms_fitted_range[fit_function].GetNbinsX() + 1):
bin_center = pull_histograms_fitted_range[fit_function].GetXaxis(
).GetBinCenter(bin)
if bin_center < mjj.getMin() or bin_center > mjj.getMax():
pull_histograms_fitted_range[fit_function].SetBinContent(
bin, 0)
pull_histograms_fitted_range[fit_function].SetBinError(bin, 0)
style_counter += 1
f_workspace.Close()
l.Draw()
Root.CMSLabelTwoPane(0.65, 0.84, "Internal", 1, 0.6)
# Pull histogram
c.cd()
bottom.cd()
#pull_histogram = frame_top.pullHist("Data", "B Fit")
frame_bottom = TH1D("frame_bottom", "frame_bottom", 100, x_min, x_max)
frame_bottom.SetMinimum(-5.)
frame_bottom.SetMaximum(5.)
#pull_histogram.plotOn(frame_bottom, RooFit.Name(fit_pdf_name))
frame_bottom.GetXaxis().SetNdivisions(505)
frame_bottom.GetXaxis().SetTitle("m_{jj} [GeV]")
frame_bottom.GetXaxis().SetLabelSize(0.06)
frame_bottom.GetXaxis().SetTitleSize(0.06)
frame_bottom.GetXaxis().SetTitleOffset(1.0)
frame_bottom.GetYaxis().SetTitle("#frac{Data - Fit}{#sigma(Data)}")
frame_bottom.GetYaxis().SetLabelSize(0.06)
frame_bottom.GetYaxis().SetTitleSize(0.06)
frame_bottom.GetYaxis().SetTitleOffset(1.0)
frame_bottom.Draw()
style_counter = 0
for fit_function in fit_functions:
pull_histograms[fit_function].SetLineColor(
seaborn.GetColorRoot("default", style_counter))
pull_histograms[fit_function].SetLineWidth(2)
pull_histograms[fit_function].SetLineStyle(2)
pull_histograms[fit_function].Draw("hist same")
if draw_trigeff:
pull_histograms[fit_function].SetLineColor(
seaborn.GetColorRoot("pastel", style_counter))
pull_histograms[fit_function].SetLineWidth(1)
pull_histograms[fit_function].SetLineStyle(3)
pull_histograms[fit_function].Draw("hist same")
pull_histograms_fitted_range[fit_function].SetLineColor(
seaborn.GetColorRoot("dark", style_counter))
pull_histograms_fitted_range[fit_function].SetLineWidth(2)
pull_histograms_fitted_range[fit_function].SetLineStyle(1)
pull_histograms_fitted_range[fit_function].Draw("hist same")
style_counter += 1
#pull_histogram.GetXaxis().SetTitle("m_{jj} [GeV]")
#pull_histogram.GetYaxis().SetTitle("#frac{Data - Fit}{#sigma(Fit)}")
#pull_histogram.Draw("same")
c.cd()
c.SaveAs(
"/uscms/home/dryu/Dijets/data/EightTeeEeVeeBee/Results/figures/c_" +
save_tag + ".pdf")