ViewVC Help

View File | Revision Log | Show Annotations | Root Listing

root/cvsroot/UserCode/VHbb/python/evaluateMVA.py

Comparing UserCode/VHbb/python/evaluateMVA.py (file contents):
Revision 1.17 by peller, Tue Nov 27 09:00:27 2012 UTC vs.
Revision 1.25 by bortigno, Tue Feb 19 18:03:41 2013 UTC

#	Line 1 | Line 1
1		#!/usr/bin/env python
2	+	from __future__ import print_function
3		import sys
4		import os
5		import ROOT
5	–	from ROOT import TFile
6		from array import array
7		from math import sqrt
8		from copy import copy
#	Line 10 | Line 10 | from copy import copy
10		import warnings
11		warnings.filterwarnings( action='ignore', category=RuntimeWarning, message='creating converter.*' )
12		from optparse import OptionParser
13	–	from BetterConfigParser import BetterConfigParser
14	–	from samplesclass import sample
15	–	from mvainfos import mvainfo
13		import pickle
14	<	from progbar import progbar
18	<	from printcolor import printc
14	>
15
16		#CONFIGURE
17	<	print 'hello'
17	>	ROOT.gROOT.SetBatch(True)
18	>	print('hello')
19		#load config
23	–	#os.mkdir(path+'/sys')
20		argv = sys.argv
21		parser = OptionParser()
22		parser.add_option("-U", "--update", dest="update", default=0,
23		help="update infofile")
24		parser.add_option("-D", "--discr", dest="discr", default="",
25		help="discriminators to be added")
30	–	#parser.add_option("-I", "--inpath", dest="inpath", default="",
31	–	#                      help="path to samples")
32	–	#parser.add_option("-O", "--outpath", dest="outpath", default="",
33	–	#                      help="path where to store output samples")
26		parser.add_option("-S", "--samples", dest="names", default="",
27		help="samples you want to run on")
28		parser.add_option("-C", "--config", dest="config", default=[], action="append",
29		help="configuration file")
30		(opts, args) = parser.parse_args(argv)
31	+
32		if opts.config =="":
33		opts.config = "config"
34	+
35	+	#Import after configure to get help message
36	+	from myutils import BetterConfigParser, progbar, printc, ParseInfo, MvaEvaluator
37	+
38		config = BetterConfigParser()
42	–	#config.read('./config7TeV_ZZ')
39		config.read(opts.config)
40		anaTag = config.get("Analysis","tag")
41
42		#get locations:
43		Wdir=config.get('Directories','Wdir')
48	–	MVASubdir=config.get('Directories','MVAdir')
44		samplesinfo=config.get('Directories','samplesinfo')
45
46		#systematics
52	–	systematics=config.get('systematics','systematics')
53	–	systematics=systematics.split(' ')
54	–
55	–	#TreeVar Array
56	–	#MVA_Vars={}
57	–	#for systematic in systematics:
58	–	#    MVA_Vars[systematic]=config.get('treeVars',systematic)
59	–	#    MVA_Vars[systematic]=MVA_Vars[systematic].split(' ')
60	–
61	–	######################
62	–	#Evaluate multi: Must Have same treeVars!!!
63	–
64	–	#OUTpath=opts.outpath
65	–	#INpath=opts.inpath
47		INpath = config.get('Directories','MVAin')
48		OUTpath = config.get('Directories','MVAout')
49
50	<	infofile = open(samplesinfo,'r')
51	<	info = pickle.load(infofile)
71	<	infofile.close()
50	>	info = ParseInfo(samplesinfo,INpath)
51	>
52		arglist=opts.discr #RTight_blavla,bsbsb
53
54		namelistIN=opts.names
55		namelist=namelistIN.split(',')
56
57	<	doinfo=bool(int(opts.update))
57	>	#doinfo=bool(int(opts.update))
58
59		MVAlist=arglist.split(',')
80	–	MVAdir=config.get('Directories','vhbbpath')
60
61		#CONFIG
62		#factory
63		factoryname=config.get('factory','factoryname')
64	+
65	+	#load the namespace
66	+	VHbbNameSpace=config.get('VHbbNameSpace','library')
67	+	ROOT.gSystem.Load(VHbbNameSpace)
68	+
69		#MVA
86	–	#MVAnames=[]
87	–	#for MVA in MVAlist:
88	–	#    print MVA
89	–	#    MVAnames.append(config.get(MVA,'MVAname'))
90	–	#print Wdir+'/weights/'+factoryname+'_'+MVAname+'.info'
91	–	#MVAinfofiles=[]
70		MVAinfos=[]
71	+	MVAdir=config.get('Directories','vhbbpath')
72		for MVAname in MVAlist:
73		MVAinfofile = open(MVAdir+'/data/'+factoryname+'_'+MVAname+'.info','r')
74		MVAinfos.append(pickle.load(MVAinfofile))
75		MVAinfofile.close()
76
98	–	treeVarSet=MVAinfos[0].varset
99	–	#variables
100	–	#TreeVar Array
101	–	MVA_Vars={}
102	–	for systematic in systematics:
103	–	MVA_Vars[systematic]=config.get(treeVarSet,systematic)
104	–	MVA_Vars[systematic]=MVA_Vars[systematic].split(' ')
105	–	#Spectators:
106	–	#spectators=config.get(treeVarSet,'spectators')
107	–	#spectators=spectators.split(' ')
108	–	#progbar quatsch
77		longe=40
78		#Workdir
79		workdir=ROOT.gDirectory.GetPath()
112	–	#os.mkdir(Apath+'/MVAout')
80
81	<	#Book TMVA readers: MVAlist=["MMCC_bla","CC5050_bla"]
82	<	readers=[]
83	<	for MVA in MVAlist:
84	<	readers.append(ROOT.TMVA.Reader("!Color:!Silent"))
85	<
86	<	#define variables and specatators
120	<	MVA_var_buffer = []
121	<	MVA_var_buffer4 = []
122	<	for i in range(len( MVA_Vars['Nominal'])):
123	<	MVA_var_buffer.append(array( 'f', [ 0 ] ))
124	<	for reader in readers:
125	<	reader.AddVariable( MVA_Vars['Nominal'][i],MVA_var_buffer[i])
126	<	#MVA_spectator_buffer = []
127	<	#for i in range(len(spectators)):
128	<	#    MVA_spectator_buffer.append(array( 'f', [ 0 ] ))
129	<	#    for reader in readers:
130	<	#        reader.AddSpectator(spectators[i],MVA_spectator_buffer[i])
131	<	#Load raeder
132	<	for i in range(0,len(readers)):
133	<	readers[i].BookMVA(MVAinfos[i].MVAname,MVAdir+'/data/'+MVAinfos[i].getweightfile())
134	<	#--> Now the MVA is booked
135	<
136	<	#Apply samples
137	<	infofile = open(samplesinfo,'r')
138	<	Ainfo = pickle.load(infofile)
139	<	infofile.close()
81	>
82	>
83	>	theMVAs = []
84	>	for mva in MVAinfos:
85	>	theMVAs.append(MvaEvaluator(config,mva))
86	>
87
88		#eval
89	<	for job in Ainfo:
90	<	if eval(job.active):
91	<	if job.name in namelist:
92	<	#get trees:
93	<	print INpath+'/'+job.prefix+job.identifier+'.root'
94	<	input = TFile.Open(INpath+'/'+job.prefix+job.identifier+'.root','read')
95	<	print OUTpath+'/'+job.prefix+job.identifier+'.root'
96	<	outfile = TFile.Open(OUTpath+'/'+job.prefix+job.identifier+'.root','recreate')
97	<	input.cd()
98	<	obj = ROOT.TObject
99	<	for key in ROOT.gDirectory.GetListOfKeys():
100	<	input.cd()
101	<	obj = key.ReadObj()
102	<	#print obj.GetName()
103	<	if obj.GetName() == job.tree:
104	<	continue
105	<	outfile.cd()
106	<	#print key.GetName()
107	<	obj.Write(key.GetName())
108	<	tree = input.Get(job.tree)
109	<	nEntries = tree.GetEntries()
110	<	outfile.cd()
111	<	newtree = tree.CloneTree(0)
112	<	#input.Close()
113	<
114	<	#MCs:
115	<	if job.type != 'DATA':
116	<	MVA_formulas={}
117	<	MVA_formulas4={}
118	<	for systematic in systematics:
119	<	#print '\t\t - ' + systematic
120	<	MVA_formulas[systematic]=[]
121	<	MVA_formulas4[systematic]=[]
122	<	#create TTreeFormulas
123	<	for j in range(len( MVA_Vars['Nominal'])):
124	<	MVA_formulas[systematic].append(ROOT.TTreeFormula("MVA_formula%s_%s"%(j,systematic),MVA_Vars[systematic][j],tree))
125	<	MVA_formulas4[systematic].append(ROOT.TTreeFormula("MVA_formula4%s_%s"%(j,systematic),MVA_Vars['Nominal'][j]+'+('+MVA_Vars[systematic][j]+'-'+MVA_Vars['Nominal'][j]+')*4',tree))#HERE change
126	<	outfile.cd()
127	<	#Setup Branches
128	<	MVAbranches=[]
129	<	MVAbranches4=[]
130	<	for i in range(0,len(readers)):
131	<	MVAbranches.append(array('f',[0]*11))
132	<	MVAbranches4.append(array('f',[0]*11))
186	<	newtree.Branch(MVAinfos[i].MVAname,MVAbranches[i],'nominal:JER_up:JER_down:JES_up:JES_down:beff_up:beff_down:bmis_up:bmis_down:beff1_up:beff1_down/F')
187	<	newtree.Branch(MVAinfos[i].MVAname+'_4',MVAbranches4[i],'nominal:JER_up:JER_down:JES_up:JES_down:beff_up:beff_down:bmis_up:bmis_down:beff1_up:beff1_down/F')
188	<	print '\n--> ' + job.name +':'
189	<	#progbar setup
190	<	if nEntries >= longe:
191	<	step=int(nEntries/longe)
192	<	long=longe
193	<	else:
194	<	long=nEntries
195	<	step = 1
196	<	bar=progbar(long)
197	<	#Fill event by event:
198	<	for entry in range(0,nEntries):
199	<	if entry % step == 0:
200	<	bar.move()
201	<	#load entry
202	<	tree.GetEntry(entry)
203	<	for systematic in systematics:
204	<	for j in range(len( MVA_Vars['Nominal'])):
205	<	MVA_var_buffer[j][0] = MVA_formulas[systematic][j].EvalInstance()
206	<
207	<	for j in range(0,len(readers)):
208	<	MVAbranches[j][systematics.index(systematic)] = readers[j].EvaluateMVA(MVAinfos[j].MVAname)
209	<
210	<	for j in range(len( MVA_Vars['Nominal'])):
211	<	MVA_var_buffer[j][0] = MVA_formulas4[systematic][j].EvalInstance()
89	>
90	>	samples = info.get_samples(namelist)
91	>	print(samples)
92	>	for job in samples:
93	>	#get trees:
94	>	print(INpath+'/'+job.prefix+job.identifier+'.root')
95	>	input = ROOT.TFile.Open(INpath+'/'+job.prefix+job.identifier+'.root','read')
96	>	print(OUTpath+'/'+job.prefix+job.identifier+'.root')
97	>	outfile = ROOT.TFile.Open(OUTpath+'/'+job.prefix+job.identifier+'.root','recreate')
98	>	input.cd()
99	>	obj = ROOT.TObject
100	>	for key in ROOT.gDirectory.GetListOfKeys():
101	>	input.cd()
102	>	obj = key.ReadObj()
103	>	#print obj.GetName()
104	>	if obj.GetName() == job.tree:
105	>	continue
106	>	outfile.cd()
107	>	#print key.GetName()
108	>	obj.Write(key.GetName())
109	>	tree = input.Get(job.tree)
110	>	nEntries = tree.GetEntries()
111	>	outfile.cd()
112	>	newtree = tree.CopyTree() #hard skim to get faster
113	>	input.Close()
114	>
115	>	#Set branch adress for all vars
116	>	for i in range(0,len(theMVAs)):
117	>	theMVAs[i].setVariables(newtree,job)
118	>	outfile.cd()
119	>	#Setup Branches
120	>	mvaVals=[]
121	>	for i in range(0,len(theMVAs)):
122	>	if job.type == 'Data':
123	>	mvaVals.append(array('f',[0]))
124	>	newtree.Branch(MVAinfos[i].MVAname,mvaVals[i],'nominal/F')
125	>	else:
126	>	mvaVals.append(array('f',[0]*11))
127	>	newtree.Branch(theMVAs[i].MVAname,mvaVals[i],'nominal:JER_up:JER_down:JES_up:JES_down:beff_up:beff_down:bmis_up:bmis_down:beff1_up:beff1_down/F')
128	>	MVA_formulas_Nominal = []
129	>	print('\n--> ' + job.name +':')
130	>	#Fill event by event:
131	>	for entry in range(0,nEntries):
132	>	newtree.GetEntry(entry)
133
134	<	for j in range(0,len(readers)):
135	<	MVAbranches4[j][systematics.index(systematic)] = readers[j].EvaluateMVA(MVAinfos[j].MVAname)
136	<	#Fill:
137	<	newtree.Fill()
138	<	newtree.AutoSave()
139	<	outfile.Close()
134	>	for i in range(0,len(theMVAs)):
135	>	theMVAs[i].evaluate(mvaVals[i],job)
136	>	#Fill:
137	>	newtree.Fill()
138	>	newtree.AutoSave()
139	>	outfile.Close()
140
141	<	#DATA:
221	<	if job.type == 'DATA':
222	<	#MVA Formulas
223	<	MVA_formulas_Nominal = []
224	<	#create TTreeFormulas
225	<	for j in range(len( MVA_Vars['Nominal'])):
226	<	MVA_formulas_Nominal.append(ROOT.TTreeFormula("MVA_formula%s_Nominal"%j, MVA_Vars['Nominal'][j],tree))
227	<	outfile.cd()
228	<	MVAbranches=[]
229	<	for i in range(0,len(readers)):
230	<	MVAbranches.append(array('f',[0]))
231	<	newtree.Branch(MVAinfos[i].MVAname,MVAbranches[i],'nominal/F')
232	<	newtree.Branch(MVAinfos[i].MVAname+'_4',MVAbranches[i],'nominal/F')
233	<	#progbar
234	<	print '\n--> ' + job.name +':'
235	<	if nEntries >= longe:
236	<	step=int(nEntries/longe)
237	<	long=longe
238	<	else:
239	<	long=nEntries
240	<	step = 1
241	<	bar=progbar(long)
242	<	#Fill event by event:
243	<	for entry in range(0,nEntries):
244	<	if entry % step == 0:
245	<	bar.move()
246	<	#load entry
247	<	tree.GetEntry(entry)
248	<	#nominal:
249	<	for j in range(len( MVA_Vars['Nominal'])):
250	<	MVA_var_buffer[j][0] = MVA_formulas_Nominal[j].EvalInstance()
251	<
252	<	for j in range(0,len(readers)):
253	<	MVAbranches[j][0]= readers[j].EvaluateMVA(MVAinfos[j].MVAname)
254	<	newtree.Fill()
255	<	newtree.AutoSave()
256	<	outfile.Close()
257	<
258	<	print '\n'
259	<
260	<	#Update Info:
261	<	if doinfo:
262	<	for job in Ainfo:
263	<	for MVAinfo in MVAinfos:
264	<	job.addcomment('Added MVA %s'%MVAinfo.MVAname)
265	<	job.addpath(MVAdir)
266	<	infofile = open(samplesinfo,'w')
267	<	pickle.dump(Ainfo,infofile)
268	<	infofile.close()
269	<
270	<
141	>	print('\n')

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines