Belle II Software light-2406-ragdoll
builtin_splot.py
1#!/usr/bin/env python3
2
3
10
11import basf2_mva
12
13if __name__ == "__main__":
14 variables = ['p', 'pt', 'pz', 'phi',
15 'chiProb', 'dr', 'dz', 'dphi',
16 'daughter(0, dr)', 'daughter(1, dr)', 'daughter(0, dz)', 'daughter(1, dz)',
17 'daughter(0, dphi)', 'daughter(1, dphi)',
18 'daughter(0, chiProb)', 'daughter(1, chiProb)', 'daughter(2, chiProb)', 'daughter(2, M)',
19 'daughter(0, atcPIDBelle(3,2))', 'daughter(1, atcPIDBelle(3,2))',
20 'daughter(2, daughter(0, E))', 'daughter(2, daughter(1, E))',
21 'daughter(2, daughter(0, clusterLAT))', 'daughter(2, daughter(1, clusterLAT))',
22 'daughter(2, daughter(0, clusterHighestE))', 'daughter(2, daughter(1, clusterHighestE))',
23 'daughter(2, daughter(0, clusterNHits))', 'daughter(2, daughter(1, clusterNHits))',
24 'daughter(2, daughter(0, clusterE9E25))', 'daughter(2, daughter(1, clusterE9E25))',
25 'daughter(2, daughter(0, minC2TDist))', 'daughter(2, daughter(1, minC2TDist))',
26 # We do not use kinematic variables of the daughters
27 # 'daughter(0, p)', 'daughter(0, pz)', 'daughter(0, pt)', 'daughter(0, phi)',
28 # 'daughter(1, p)', 'daughter(1, pz)', 'daughter(1, pt)', 'daughter(1, phi)',
29 # 'daughter(2, p)', 'daughter(2, pz)', 'daughter(2, pt)', 'daughter(2, phi)',
30 # 'daughterInvM(1, 2)', 'daughterInvM(0, 1)', 'daughterInvM(0, 2)',
31 # 'daughterAngle(0, 1)', 'daughterAngle(0, 2)', 'daughterAngle(1, 2)',
32 'M',
33 ]
34
35 # Perform an sPlot training
36 general_options = basf2_mva.GeneralOptions()
37 general_options.m_datafiles = basf2_mva.vector("train_mc.root")
38 general_options.m_identifier = "MVAFull"
39 general_options.m_treename = "tree"
40 general_options.m_variables = basf2_mva.vector(*variables)
41 general_options.m_target_variable = "isSignal"
42
43 fastbdt_options = basf2_mva.FastBDTOptions()
44 # SPlot is more stable if one doesn't use the randRatio
45 # FastBDT has a special sPlot mode, but which isn't implemented yet in the mva package
46 # fastbdt_options.m_nTrees = 100
47 fastbdt_options.m_randRatio = 1.0
48 basf2_mva.teacher(general_options, fastbdt_options)
49
50 general_options.m_identifier = "MVAOrdinary"
51 general_options.m_variables = basf2_mva.vector(*variables[:-1])
52 basf2_mva.teacher(general_options, fastbdt_options)
53
54 meta_options = basf2_mva.MetaOptions()
55 meta_options.m_use_splot = True
56 meta_options.m_splot_variable = "M"
57 # SPlot training assumes that the datafile given to the general options contains only data
58 # It requires an additional file with MC information from which it can extract the distribution
59 # of the discriminating variable (in this case M).
60 # Here we use the same file
61 general_options.m_datafiles = basf2_mva.vector("train_data.root")
62 meta_options.m_splot_mc_files = basf2_mva.vector("train_mc.root")
63
64 # First we do an ordinary sPlot training
65 general_options.m_identifier = "MVASPlot"
66 meta_options.m_splot_combined = False
67 meta_options.m_splot_boosted = False
68 basf2_mva.teacher(general_options, fastbdt_options, meta_options)
69
70 # Now we combine the sPlot training with a PDF classifier for M, in one step
71 general_options.m_identifier = "MVASPlotCombined"
72 meta_options.m_splot_combined = True
73 meta_options.m_splot_boosted = False
74 basf2_mva.teacher(general_options, fastbdt_options, meta_options)
75
76 # Now we use a boosted sPlot training
77 general_options.m_identifier = "MVASPlotBoosted"
78 meta_options.m_splot_combined = False
79 meta_options.m_splot_boosted = True
80 basf2_mva.teacher(general_options, fastbdt_options, meta_options)
81
82 # And finally a boosted and combined training
83 general_options.m_identifier = "MVASPlotCombinedBoosted"
84 meta_options.m_splot_combined = True
85 meta_options.m_splot_boosted = True
86 basf2_mva.teacher(general_options, fastbdt_options, meta_options)
87
88 # Also do a training of only the pdf classifier
89 pdf_options = basf2_mva.PDFOptions()
90 general_options.m_method = 'PDF'
91 general_options.m_identifier = "MVAPdf"
92 general_options.m_variables = basf2_mva.vector('M')
93 basf2_mva.teacher(general_options, pdf_options)