def bkg_predict(flavor):
return BackgroundFit.do_bkg_fit(data, mc, 100., flavor)
def run_mc_closure(flavor):
"""
Check MC closure for a range of MCT cuts. Writes a latex table containing the results
"""
# all mc except diboson and DY
test_mc = mc[((mc.mc_cat != 'WV') & (mc.mc_cat != 'ZZ') & (mc.mc_cat != 'DY'))]
diboson_mc = mc[((mc.mc_cat == 'WV') | (mc.mc_cat == 'ZZ'))]
dy_mc = mc[(mc.mc_cat == "DY")]
# double weights, since we're only using half for each
diboson_mc.weight *= 2
dy_mc.weight *= 2
# choose half of the diboson MC to use
template_indices = numpy.random.choice([True, False], size=diboson_mc.weight.count(), replace=True)
template_mc = diboson_mc[template_indices]
# add the rest to the test sample
test_mc = test_mc.append(diboson_mc[~template_indices])
# now do the same for DY
template_indices = numpy.random.choice([True, False], size=dy_mc.weight.count(), replace=True)
template_mc = template_mc.append(dy_mc[template_indices])
# add the rest to the test sample
test_mc = test_mc.append(dy_mc[~template_indices])
f = open("tables/table{}.tex".format(flavor), 'w')
f.write("""
\\begin{tabular}{|l|c|c|c|c|c|c|c|c|}
""")
table = defaultdict(lambda : defaultdict(lambda : defaultdict(dict)))
for cut in numpy.arange(70., 130., 10.) :
predictions = BackgroundFit.do_bkg_fit( test_mc, template_mc, cut, flavor, 0.07)
truth = mctruth.get_mc_truth(test_mc, cut, flavor)
for mctype in predictions['low'].keys() :
table['Low \mctp\ Region'][mctype]['prediction'][cut] = predictions['low'][mctype]
table['Low \mctp\ Region'][mctype]['truth'][cut] = truth['low'][mctype]
for mctype in predictions['high'].keys() :
table['High \mctp\ Region'][mctype]['prediction'][cut] = predictions['high'][mctype]
table['High \mctp\ Region'][mctype]['truth'][cut] = truth['high'][mctype]
mc_nicename = {"Top":"Top", "VV":"Diboson", "DY":"\PZz/\Pgg$^*$", "W":"Fake Lepton", "Total":"Total"}
mctypes=['Top', 'VV', 'DY', 'W', "Total"]
regions = table.keys()
regions.reverse()
for region in regions :
f.write("\hline\n \multicolumn{7}{|c|}{"+region+"}\\\\ \n \hline\n")
mccutstring = ""
for mctype in mctypes :
if mctype in table[region].keys():
cuts = table[region][mctype]['prediction'].keys()
cuts.sort()
if mccutstring == "" :
mccutstring += "\mctp\ cut "
for cut in cuts :
mccutstring += (" & "+str(cut))
mccutstring += "\\\\ \n \hline\n"
f.write(mccutstring)
truthline = mc_nicename[mctype]+" truth"
for cut in cuts :
truthline += " & ${:.2f}$".format(table[region][mctype]['truth'][cut])
truthline += "\\\\ \n\hline\n"
f.write(truthline)
predline = mc_nicename[mctype]+" prediction"
for cut in cuts :
predline += " & ${:.2f} \pm {:.2f}$".format(*table[region][mctype]['prediction'][cut])
predline += "\\\\ \n \hline\n"
f.write(predline)
f.write("""\\end{tabular}
""")
f.close()
