OnlyTrain.py

from trainerChain import Trainer
from evaluaterRun2 import Evaluater
from evaluaterRun2 import Systematic
from evaluaterRun2 import Sample
from subprocess import call
from SampleScaleGetter import *

import sys
import json

print len(sys.argv)
if len(sys.argv)<6 or len(sys.argv)>6:
  print "usage:"
  print "python TrainAndEval.py CATEGORY NAME FINALBDTCONFIG_FILE INPUTPATH [test | train]"
  print "\n"
  print "CATEGORY = for example 43 or 62 or DB or so "
  print "NAME = name of outputfiles "
  print "FINALBDTCONFIG_FILE = BDT config file e.g. see example files "
  print "INPUTPATH = path to input trees prepared in ../1PrepareSamples. Has to contain directories called Category_*/* "
  print "[test | train] = do limit on test or trainings samples "
  print "\n output is out in ./output/"
  exit(0)
  
category=sys.argv[1]
name=sys.argv[2]
FinalBDTConfig=sys.argv[3]
TestOrTrainSwitch=sys.argv[5]
if TestOrTrainSwitch=="train" or TestOrTrainSwitch=="Train":
  doTestonTrain=True
elif TestOrTrainSwitch=="test" or TestOrTrainSwitch=="Test":
  doTestonTrain=False
else:
  doTestonTrain=None

AnaPath=sys.argv[4]
SubBDTConfig="NONE"

print category
print name
print FinalBDTConfig
print SubBDTConfig

f = open("output/TrainAndEval_"+name+".txt","w")

evaluationPath=AnaPath+"/Category_"+category+"/Even/"
trainingPath=AnaPath+"/Category_"+category+"/Odd/"


if doTestonTrain==True:
  print "!!!!!!!!!!DOING EVALUATION ON TRAINING SAMPLE!!!!!!!!"
  evaluationPath=trainingPath
elif doTestonTrain==None:
  print "!!!!!!!!!!DOING SPLITTING OF TRAINING SAMPLE!!!!!!!!"
  evaluationPath="NONE"
  
variablesFinal=[]

FBDTConfigFile=open(FinalBDTConfig,"r")
FBDTConfigList=list(FBDTConfigFile)
#splitlist=FBDTConfigList[0].split(" ")
for line in FBDTConfigList:
   if "variables" in line:
     line=line.split("=",1)
     #print line
     line=line[1].strip().replace("\'","\"")
     print type(line), line
     variablesFinalbuffer=json.loads(line)
     #print variablesFinal
   if "BDTOptions" in line:
     line=line.split("=",1)
     print line
     FinalBDTOptions=line[1]
     print FinalBDTOptions
buffername="output/FinalBDT_Config_"+category+"_"+name+".txt"
call(['cp', './'+FinalBDTConfig,buffername])

f.write("TrainingsPath: "+trainingPath+"\n")
f.write("EvalPath: "+evaluationPath+"\n\n")

for var in variablesFinalbuffer:
  variablesFinal.append(str(var))

f.write("FinalBDT Variables:\n")
f.write(str(variablesFinal)+"\n\n")

print "Training Final BDT"
if evaluationPath=="NONE":
  trainerFinal=Trainer(trainingPath+"ttHbb_*_nominal.root",trainingPath+"ttbar_*_nominal.root","","",FinalBDTOptions,variablesFinal,[],False,"weights/weights_Final_"+category+"_"+name+".xml")
else:
  trainerFinal=Trainer(trainingPath+"ttHbb_*_nominal.root",trainingPath+"ttbar_*_nominal.root",evaluationPath+"ttHbb_*_nominal.root",evaluationPath+"ttbar_*_nominal.root",FinalBDTOptions,variablesFinal,[],False,"weights/weights_Final_"+category+"_"+name+".xml")
trainerFinal.useTransformations(False)

trainerFinal.run(variablesFinal,FinalBDTOptions)

f.write("FinalBDT Parameters:\n")
f.write(FinalBDTOptions+"\n\n")
f.close()


if doTestonTrain==True:
  autotrainsuffix="train"
elif doTestonTrain==None:
  autotrainsuffix="split"
else:
  autotrainsuffix="test"

infname="autotrain.root"
outfname="output/training_"+category+"_"+name+"_"+autotrainsuffix+".root"
call(["cp",infname,outfname])
print "all done"

exit(0)

print "Writing FinalBDT output"
weightfileFinal=trainerFinal.weightfile
evaluaterFinal=Evaluater(weightfileFinal,variablesFinal,samplesTrainingFinal,[],[])
evaluaterFinal.WriteBDTVars("","_FinalBDT","Final")