def GetANNInput(self, verbose=1):
if (self.mode[:4] == 'Slow' or self.mode[:4] == 'Save'):
self.verbose = verbose
if (self.verbose == 1):
Utils.stdinfo("Setting up input vectors")
#Import Variables as an numoy array
for DISetup in self.ListAnaSetup:
if (self.verbose == 1):
print("Processing Sample: " + DISetup.Name)
ListOfVariables = DISetup.LVars[:]
if (self.mode[:4] == 'Fast' and DISetup.Name != 'NLO'):
ListOfVariables.append('Weight')
else:
ListOfVariables.extend(DISetup.WeightList)
Arr = self.GetArray(ListOfVariables, DISetup)
np.random.shuffle(Arr) #ensure randomize of event order
DISetup.Samples = self.MakeSplit(Arr, DISetup) #Split samples
if (self.verbose == 1):
self.Info(DISetup)
#Combine the different sets into a DataSet
if (self.verbose == 1):
Utils.stdinfo("finalising input Preparation")
ListSamples = [DISetup.Samples for DISetup in self.ListAnaSetup]
train = self.Finalise("train", ListSamples)
test = self.Finalise("test", ListSamples)
vali = self.Finalise("validation", ListSamples)
DataSet = DIDataSet(train, test, vali)
# Save the numpy arrays to load them faster later
if (self.mode[:4] == 'Save'):
if (self.verbose == 1):
Utils.stdinfo("Saving data for Fast mode")
self.SaveNpy('TrainSet' + self.mode[4:], train)
self.SaveNpy('TestSet' + self.mode[4:], test)
self.SaveNpy('ValiSet' + self.mode[4:], vali)
# If stored as numpy arrays load the arrays
elif (self.mode[:4] == 'Fast'):
train = self.GetSampleNpy('TrainSet' + self.mode[4:])
test = self.GetSampleNpy('TestSet' + self.mode[4:])
vali = self.GetSampleNpy('ValiSet' + self.mode[4:])
DataSet = DIDataSet(train, test, vali)
#Make Plots of the Variables
if (self.Plots != False):
PlotService.VarHists(DataSet,
key='All',
Norm=self.NormFlag,
Sig=self.Plots)
return DataSet
def GetANNInput(self):
# SLOW or SAVE Modes ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ #
if self.mode[:4] == 'Slow' or self.mode[:4] == 'Save':
Utils.stdinfo("\nSetting up input vectors!\n")
print(
"------------------------------------------------------------------------------------------------"
)
print(
" Sample Total Training Testing Validation Yield "
)
print(
"================================================================================================"
)
# Import Variables as a NumPy array
for DISetup in self.ListAnaSetup:
print(" " * 3 + DISetup.Name.ljust(15, " "), end="")
ListOfVariables = DISetup.LVars[:]
ListOfVariables.extend(DISetup.WeightList)
Arr = self.GetArray(ListOfVariables, DISetup) # Defined below
np.random.shuffle(
Arr
) # Ensures randomization of event order. Seed fixed in FNN.py[17]
DISetup.Samples = self.MakeSplit(Arr, DISetup) # Split samples
self.Info(
DISetup
) # Defined below: Displays the information about the sample sizes
# Combine the different sets into a DataSet
Utils.stdinfo("\nFinalising input preparation!\n")
ListSamples = [DISetup.Samples for DISetup in self.ListAnaSetup]
train = self.Finalise("train", ListSamples) # Defined below
test = self.Finalise("test", ListSamples)
vali = self.Finalise("validation", ListSamples)
DataSet = DIDataSet(train, test, vali)
# Save the NumPy arrays to load them faster later
if self.mode[:4] == 'Save':
Utils.stdinfo("Saving data for Fast mode")
self.SaveNpy('TrainSet' + self.mode[4:], train)
self.SaveNpy('TestSet' + self.mode[4:], test)
self.SaveNpy('ValiSet' + self.mode[4:], vali)
# FAST Mode ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ #
elif self.mode[:4] == 'Fast':
train = self.GetSampleNpy('TrainSet' + self.mode[4:])
test = self.GetSampleNpy('TestSet' + self.mode[4:])
vali = self.GetSampleNpy('ValiSet' + self.mode[4:])
DataSet = DIDataSet(train, test, vali)
# Make Plots of the Variables
if self.Plots != False:
PlotService.VarHists(DataSet,
key='All',
Norm=self.NormFlag,
Sig=self.Plots)
return DataSet