def get_datadriven_fakerate(samples, cutloose, cuttight, var, dosf=False):
# Retrieve histograms
h_loose_data = get_histogram(samples, "data" , cutloose , var , dosf)
h_loose_top = get_histogram(samples, "top" , cutloose , var , dosf)
h_loose_w = get_histogram(samples, "w" , cutloose , var , dosf)
h_loose_dy = get_histogram(samples, "dy" , cutloose , var , dosf)
# Retrieve histograms
h_tight_data = get_histogram(samples, "data" , cuttight , var , dosf)
h_tight_top = get_histogram(samples, "top" , cuttight , var , dosf)
h_tight_w = get_histogram(samples, "w" , cuttight , var , dosf)
h_tight_dy = get_histogram(samples, "dy" , cuttight , var , dosf)
# Get total ewk
h_loose_ewk = p.get_total_hist([h_loose_top, h_loose_dy, h_loose_w])
h_tight_ewk = p.get_total_hist([h_tight_top, h_tight_dy, h_tight_w])
# Get data - ewk
h_loose_data.Add(h_loose_ewk, -1)
h_tight_data.Add(h_tight_ewk, -1)
# Compute fakerate
h_tight_data.Divide(h_loose_data)
return h_tight_data
def hlt_prescale():
cut = "TwoMuHLT17"
var = "Mll_Z"
# Retrieve histograms
h_cn_data = get_histogram(samples_cn, "data" , cut , var)
h_cn_top = get_histogram(samples_cn, "top" , cut , var)
h_cn_w = get_histogram(samples_cn, "w" , cut , var)
h_cn_dy = get_histogram(samples_cn, "dy" , cut , var)
h_cn_qcd = get_histogram(samples_cn, "qcd" , cut , var)
b = p.get_total_hist([h_cn_top, h_cn_dy, h_cn_w])
b.Rebin(b.GetNbinsX())
h_cn_data.Rebin(h_cn_data.GetNbinsX())
b.Divide(h_cn_data)
print cut, var, b.GetBinContent(1), b.GetBinError(1)
cut = "TwoElHLT17"
var = "Mll_Z"
# Retrieve histograms
h_cn_data = get_histogram(samples_cn, "data" , cut , var)
h_cn_top = get_histogram(samples_cn, "top" , cut , var)
h_cn_w = get_histogram(samples_cn, "w" , cut , var)
h_cn_dy = get_histogram(samples_cn, "dy" , cut , var)
h_cn_qcd = get_histogram(samples_cn, "qcd" , cut , var)
b = p.get_total_hist([h_cn_top, h_cn_dy, h_cn_w])
b.Rebin(b.GetNbinsX())
h_cn_data.Rebin(h_cn_data.GetNbinsX())
b.Divide(h_cn_data)
print cut, var, b.GetBinContent(1), b.GetBinError(1)
def get_datadriven_fakerate(channel, is3l, samples):
cut = ("OneElLoose" if channel == "e" else "OneMuLoose") if not is3l else ("OneEl3lLoose" if channel == "e" else "OneMu3lLoose")
histname = cut+"/lep_ptcorrcoarse_vs_etacoarse"
dataname = "/data/{}{}".format(channel, channel)
elsf = ewksf("OneElTightEWKCR", "e", samples) if not is3l else ewksf("OneEl3lTightEWKCR", "e", samples)
musf = ewksf("OneMuTightEWKCR", "m", samples) if not is3l else ewksf("OneMu3lTightEWKCR", "m", samples)
sf = -elsf if channel == "e" else -musf
bgs = [
samples.getHistogram("/top", histname).Clone("Top"),
samples.getHistogram("/Zonelep", histname).Clone("DY"),
samples.getHistogram("/Wonelep", histname).Clone("W"),
]
ddqcd = samples.getHistogram(dataname, histname).Clone("Data")
totalbkg = p.get_total_hist(bgs)
move_in_overflow(totalbkg)
move_in_overflow(ddqcd)
ddqcd.Add(totalbkg, sf)
bgstight = [
samples.getHistogram("/top", str(histname).replace("Loose","Tight")).Clone("Top"),
samples.getHistogram("/Zonelep", str(histname).replace("Loose","Tight")).Clone("DY"),
samples.getHistogram("/Wonelep", str(histname).replace("Loose","Tight")).Clone("W"),
]
ddqcdtight = samples.getHistogram(dataname, str(histname).replace("Loose","Tight")).Clone("Data")
totalbkgtight = p.get_total_hist(bgstight)
move_in_overflow(totalbkgtight)
move_in_overflow(ddqcdtight)
ddqcdtight.Add(totalbkgtight, sf)
ddqcdtight.Divide(ddqcd)
#compute_fake_factor(ddqcdtight)
return ddqcdtight
def compute_nvtx_reweighting():
f = ROOT.TFile("nvtxreweight.root", "recreate")
cut = "OneMuTightEWKCR3NoNvtxRewgt"
var = "nvtx"
# Retrieve histograms
h_cn_data = get_histogram(samples_cn, "data" , cut , var)
h_cn_top = get_histogram(samples_cn, "top" , cut , var)
h_cn_w = get_histogram(samples_cn, "w" , cut , var)
h_cn_dy = get_histogram(samples_cn, "dy" , cut , var)
h_cn_qcd = get_histogram(samples_cn, "qcd" , cut , var)
b = p.get_total_hist([h_cn_top, h_cn_dy, h_cn_w])
# Get total integral to get the total offset
bi = b.Integral()
di = h_cn_data.Integral()
ratio = di / bi
# Compute reweighting
h_cn_data.Divide(b)
h_cn_data.Scale(1./ratio)
h_cn_data.SetName(cut+"_"+var)
h_cn_data.SetTitle(cut+"_"+var)
h_cn_data.Write()
cut = "OneElTightEWKCR3NoNvtxRewgt"
var = "nvtx"
# Retrieve histograms
h_cn_data = get_histogram(samples_cn, "data" , cut , var)
h_cn_top = get_histogram(samples_cn, "top" , cut , var)
h_cn_w = get_histogram(samples_cn, "w" , cut , var)
h_cn_dy = get_histogram(samples_cn, "dy" , cut , var)
h_cn_qcd = get_histogram(samples_cn, "qcd" , cut , var)
b = p.get_total_hist([h_cn_top, h_cn_dy, h_cn_w])
# Get total integral to get the total offset
bi = b.Integral()
di = h_cn_data.Integral()
ratio = di / bi
# Compute reweighting
h_cn_data.Divide(b)
h_cn_data.Scale(1./ratio)
h_cn_data.SetName(cut+"_"+var)
h_cn_data.SetTitle(cut+"_"+var)
h_cn_data.Write()
f.Close()
def ewksf(cut, channel, samples):
data = samples.getHistogram("/data/{}{}".format(channel, channel) , cut+"/MTOneLepFixed").Clone("Data")
bgs = []
bgs.append(samples.getHistogram("/top" , cut+"/MTOneLepFixed").Clone("Top"))
bgs.append(samples.getHistogram("/Zonelep" , cut+"/MTOneLepFixed").Clone("DY"))
bgs.append(samples.getHistogram("/Wonelep" , cut+"/MTOneLepFixed").Clone("W"))
totalbkg = p.get_total_hist(bgs)
data.Rebin(4)
totalbkg.Rebin(4)
data.Divide(totalbkg)
return data.GetBinContent(3)
def plot_stack(histname,
output_name,
systs=None,
options={},
plotfunc=p.plot_hist):
ismu = str(histname).find("Mu") != -1
is2d = str(histname).find("_vs_") != -1
# Options
alloptions = {
"ratio_range": [0.4, 1.6],
#"nbins": 30,
#"autobin": True,
"legend_scalex": 1.0,
"legend_scaley": 1.0,
"legend_smart": True,
"legend_alignment": "topleft",
"output_name": "{}/{}_stack.pdf".format(output_plot_dir, output_name),
"bkg_sort_method": "unsorted",
"divide_by_bin_width":
True if output_name.find("varbin") != -1 else False,
"yaxis_log": True if output_name.find("varbin") != -1 else False,
"yaxis_range": [1e3, 1e10] if output_name.find("varbin") != -1 else [],
}
alloptions.update(options)
sigs = []
bgs = [
samples.getHistogram("/top", histname).Clone("Top") if not is2d else
p.flatten_th2(samples.getHistogram("/top", histname).Clone("Top")),
samples.getHistogram("/Zonelep", histname).Clone("DY") if not is2d else
p.flatten_th2(samples.getHistogram("/Zonelep", histname).Clone("DY")),
samples.getHistogram("/Wonelep", histname).Clone("W") if not is2d else
p.flatten_th2(samples.getHistogram("/Wonelep", histname).Clone("W")),
samples.getHistogram("/qcd/mu" if ismu else "/qcd/el",
histname).Clone("QCD")
if not is2d else p.flatten_th2(
samples.getHistogram("/qcd/mu" if ismu else "/qcd/el",
histname).Clone("QCD")),
]
dataname = "/data"
if histname.find("El") != -1: dataname = "/data/ee"
if histname.find("Mu") != -1: dataname = "/data/mm"
data = samples.getHistogram(
dataname, histname).Clone("Data") if not is2d else p.flatten_th2(
samples.getHistogram(dataname, histname).Clone("Data"))
if histname.find("HLT") != -1:
totalbkg = p.get_total_hist(bgs)
ratio = samples.getHistogram(
dataname, histname).Clone("Data") if not is2d else p.flatten_th2(
samples.getHistogram(dataname, histname).Clone("Data"))
ratio.Rebin(ratio.GetNbinsX())
totalbkg.Rebin(totalbkg.GetNbinsX())
totalbkg.Divide(ratio)
print histname
totalbkg.Print("all")
if (histname.find("OneMuTightEWKCR3NoNvtxRewgt/nvtx") != -1
or histname.find("OneElTightEWKCR3NoNvtxRewgt/nvtx") != -1
) and options["nbins"] != 5:
totalbkg = p.get_total_hist(bgs[:-1])
nvtxweight = samples.getHistogram(
dataname,
histname).Clone(output_name) if not is2d else p.flatten_th2(
samples.getHistogram(dataname, histname).Clone(output_name))
scale = nvtxweight.Integral() / totalbkg.Integral()
nvtxweight.Divide(totalbkg)
nvtxweight.Scale(1. / scale)
#nvtxweight.Print("all")
#f = ROOT.TFile("{}.root".format(output_name),"recreate")
#nvtxweight.Write()
#f.Close()
if options["nbins"] == 5 and histname.find("TightEWKCR/MT") != -1:
totalbkg = p.get_total_hist(bgs[:-1])
ratio = samples.getHistogram(
dataname, histname).Clone("Data") if not is2d else p.flatten_th2(
samples.getHistogram(dataname, histname).Clone("Data"))
ratio.Rebin(4)
totalbkg.Rebin(4)
ratio.Divide(totalbkg)
print "EWK NF", histname, ratio.GetBinContent(3), ratio.GetBinError(3)
colors = [2005, 2003, 2001, 920]
sf = musf if ismu else elsf
if histname.find("MTOneLep") == -1:
bgs[0].Scale(sf)
bgs[1].Scale(sf)
bgs[2].Scale(sf)
if histname.find("HLT") != -1:
bgs = bgs[:-1]
plotfunc(sigs=sigs,
bgs=bgs,
data=data,
colors=colors,
syst=systs,
options=alloptions)
def plot_datadriven_fakeratecomp(histname,
output_name,
systs=None,
options={},
plotfunc=p.plot_hist):
if str(histname).find("Loose") == -1: return
ismu = str(histname).find("Mu") != -1
is2d = str(histname).find("_vs_") != -1
# Options
alloptions = {
"ratio_range": [0.0, 2.0],
#"nbins": 30,
"autobin":
True,
"legend_scalex":
1.0,
"legend_scaley":
1.0,
"legend_smart":
True,
"legend_alignment":
"topleft",
"legend_datalabel":
"Data FR",
"output_name":
"{}/{}_fakeratecomp.pdf".format(output_plot_dir, output_name),
"bkg_sort_method":
"unsorted",
"draw_points":
True,
}
alloptions.update(options)
sigs = []
# Loose
bgs = [
samples.getHistogram("/top", histname).Clone("Top") if not is2d else
p.flatten_th2(samples.getHistogram("/top", histname).Clone("Top")),
samples.getHistogram("/Zonelep", histname).Clone("DY") if not is2d else
p.flatten_th2(samples.getHistogram("/Zonelep", histname).Clone("DY")),
samples.getHistogram("/Wonelep", histname).Clone("W") if not is2d else
p.flatten_th2(samples.getHistogram("/Wonelep", histname).Clone("W")),
#samples.getHistogram("/qcd/mu" if ismu else "/qcd/el", histname).Clone("QCD") if not is2d else p.flatten_th2(samples.getHistogram("/qcd/mu" if ismu else "/qcd/el", histname).Clone("QCD")),
]
dataname = "/data"
if histname.find("El") != -1: dataname = "/data/ee"
if histname.find("Mu") != -1: dataname = "/data/mm"
ddqcd = samples.getHistogram(
dataname, histname).Clone("Data") if not is2d else p.flatten_th2(
samples.getHistogram(dataname, histname).Clone("Data"))
qcd = samples.getHistogram(
"/qcd/mu" if ismu else "/qcd/el",
histname).Clone("QCD") if not is2d else p.flatten_th2(
samples.getHistogram("/qcd/mu" if ismu else "/qcd/el",
histname).Clone("QCD"))
totalbkg = p.get_total_hist(bgs)
# tight
bgstight = [
samples.getHistogram(
"/top",
str(histname).replace(
"Loose", "Tight")).Clone("Top") if not is2d else p.flatten_th2(
samples.getHistogram(
"/top",
str(histname).replace("Loose", "Tight")).Clone("Top")),
samples.getHistogram(
"/Zonelep",
str(histname).replace(
"Loose", "Tight")).Clone("DY") if not is2d else p.flatten_th2(
samples.getHistogram(
"/Zonelep",
str(histname).replace("Loose", "Tight")).Clone("DY")),
samples.getHistogram(
"/Wonelep",
str(histname).replace(
"Loose", "Tight")).Clone("W") if not is2d else p.flatten_th2(
samples.getHistogram(
"/Wonelep",
str(histname).replace("Loose", "Tight")).Clone("W")),
#samples.getHistogram("/qcd/mu" if ismu else "/qcd/el", str(histname).replace("Loose","Tight")).Clone("QCD") if not is2d else p.flatten_th2(samples.getHistogram("/qcd/mu" if ismu else "/qcd/el", str(histname).replace("Loose","Tight")).Clone("QCD")),
]
dataname = "/data"
if str(histname).replace("Loose", "Tight").find("El") != -1:
dataname = "/data/ee"
if str(histname).replace("Loose", "Tight").find("Mu") != -1:
dataname = "/data/mm"
ddqcdtight = samples.getHistogram(
dataname,
str(histname).replace(
"Loose", "Tight")).Clone("Data") if not is2d else p.flatten_th2(
samples.getHistogram(dataname,
str(histname).replace(
"Loose", "Tight")).Clone("Data"))
qcdtight = samples.getHistogram(
"/qcd/mu" if ismu else "/qcd/el",
str(histname).replace(
"Loose", "Tight")).Clone("QCD") if not is2d else p.flatten_th2(
samples.getHistogram("/qcd/mu" if ismu else "/qcd/el",
str(histname).replace(
"Loose", "Tight")).Clone("QCD"))
totalbkgtight = p.get_total_hist(bgstight)
# Get data fakerate
p.remove_overflow(totalbkg)
p.remove_overflow(totalbkgtight)
p.remove_overflow(ddqcd)
p.remove_overflow(ddqcdtight)
sf = -musf if ismu else -elsf
ddqcd.Add(totalbkg, sf)
ddqcdtight.Add(totalbkgtight, sf)
ddqcdtight.Divide(ddqcd)
# Get QCD fakerate
p.remove_overflow(qcd)
p.remove_overflow(qcdtight)
qcdtight.Divide(qcd)
bgs = [qcdtight.Clone("QCD FR")]
data = ddqcdtight
# Compute data driven fake rate
colors = [2]
plotfunc(sigs=sigs,
bgs=bgs,
data=data,
colors=colors,
syst=systs,
options=alloptions)
def plot_closure(cut, var, options={}, doprintclosure=False):
# Sanity check that the name of the cut contains "Predict"
# If not, then quietly exit the function without doing anything
ispredict = cut.find("Predict") != -1
if not ispredict:
print "ERROR - requested plot_closure on", cut, var, "but did not find Predict in name"
return
# Sanity check that the name of the cut contains "Loose"
# If not, then quietly exit the function without doing anything
isloose = cut.find("Loose") != -1
if not isloose:
print "ERROR - requested plot_closure on", cut, var, "but did not find Loose in name"
return
# Setting variables to be used for option setting
output_name = cut + "_" + var
isvarbin = output_name.find("varbin") != -1
divide_by_bin_width = isvarbin
yaxis_log = isvarbin
yaxis_range = [1e3, 1e10] if isvarbin else []
# Options
alloptions= {
"ratio_range":[0.0,2.0],
"nbins": 9,
"legend_scalex": 1.4,
"legend_scaley": 1.0,
"legend_smart": True,
"legend_alignment": "topleft",
"output_name": "plots/{}_closure.pdf".format(output_name),
"bkg_sort_method": "unsorted",
"divide_by_bin_width": divide_by_bin_width,
"yaxis_log": yaxis_log,
"yaxis_range": yaxis_range,
"legend_datalabel": "t#bar{t} + W prediction",
"xaxis_ndivisions": 101,
}
# Whatever the option passed through the argument can override anything
alloptions.update(options)
# Colors of the MC histograms
colors = [ 2005, 2001 ] # in order of top, Z, W, QCD
# Retrieve histograms (loose means "estimation", i.e. apply the "tight" selection. I KNOW IT'S A TERRIBLE NAMING....)
h_loose_cn_top = get_histogram(samples_cn, "top" , cut , var)
h_loose_cn_w = get_histogram(samples_cn, "w" , cut , var)
# Retrieve histograms (tight means "prediction", i.e. loose!tight * fake-factor)
h_tight_cn_top = get_histogram(samples_cn, "top" , cut.replace("Loose", "Tight") , var)
h_tight_cn_w = get_histogram(samples_cn, "w" , cut.replace("Loose", "Tight") , var)
# Form the "prediction" which is the sum of 'tight'
h_tight_cn_top.Add(h_tight_cn_w)
# Name the histograms
h_loose_cn_w.SetName("W estimation")
h_loose_cn_top.SetName("t#bar{t} estimation")
# Arrange the histograms
sigs = []
bgs = [ h_loose_cn_top, h_loose_cn_w ]
data = h_tight_cn_top
# if do print closure then print the ratio
if doprintclosure:
prd = p.get_total_hist(bgs)
est = h_tight_cn_top.Clone("estimate")
prd_num = E(prd.GetBinContent(1), prd.GetBinError(1))
est_num = E(est.GetBinContent(1), est.GetBinError(1))
print est_num, prd_num, est_num / prd_num
# Plot!
p.plot_hist(
sigs = sigs,
bgs = bgs,
data = data,
colors = colors,
options=alloptions)