def makePrediction(self,
model,
testdatacollection,
outputDir,
ident='',
store_labels=False,
monkey_class=''):
import numpy as np
from root_numpy import array2root
import os
monkey_class_obj = None
if monkey_class:
module, classname = tuple(monkey_class.split(':'))
_temp = __import__(module, globals(), locals(), [classname], -1)
monkey_class_obj = getattr(_temp, classname)
outputDir = os.path.abspath(outputDir)
if len(ident) > 0:
ident = '_' + ident
self.__sourceroots = []
self.__predictroots = []
self.metrics = []
fullnumpyarray = np.array([])
for i in range(len(testdatacollection.samples)):
sample = testdatacollection.samples[i]
originroot = testdatacollection.originRoots[i]
outrootfilename = os.path.splitext(
os.path.basename(originroot))[0] + '_predict' + ident + '.root'
fullpath = testdatacollection.getSamplePath(sample)
if monkey_class_obj is not None:
testdatacollection.dataclass = monkey_class_obj()
td = testdatacollection.dataclass
td.readIn(fullpath)
if hasattr(td, 'customlabels'):
import copy
formatstring = copy.deepcopy(td.customlabels)
else:
truthclasses = td.getUsedTruth()
formatstring = []
if len(truthclasses) > 0 and len(truthclasses[0]) > 0:
formatstring = [
'prob_%s%s' % (i, ident) for i in truthclasses
]
regressionclasses = []
if hasattr(td, 'regressiontargetclasses'):
regressionclasses = td.regressiontargetclasses
#new implementation. Please check with the store_labels option, Mauro
formatstring.extend(
['reg_%s%s' % (i, ident) for i in regressionclasses])
features = td.x
labels = td.y
weights = td.w[0]
prediction = model.predict(features)
if self.use_only:
prediction = [prediction[i] for i in self.use_only]
if isinstance(prediction, list):
all_write = np.concatenate(prediction, axis=1)
else:
all_write = prediction
#if a prediction functor was set, it's output is also added to the final predictions
if hasattr(td, 'predictionFunctor') and hasattr(
td, 'predictionFunctorClasses'):
print(
'Extending the output with the configured prediction functor output'
)
formatstring.extend(getattr(td, 'predictionFunctorClasses'))
all_write = np.concatenate([
all_write,
[getattr(td, 'predictionFunctor')(p) for p in prediction]
],
axis=1)
if all_write.ndim == 2:
all_write = np.concatenate([all_write, weights], axis=1)
formatstring.append('weight')
if not all_write.shape[1] == len(formatstring):
print(formatstring, ' vs ', all_write.shape[1])
raise ValueError(
'Prediction output does not match with the provided targets!'
)
all_write = np.core.records.fromarrays(
np.transpose(all_write), names=','.join(formatstring))
array2root(all_write,
outputDir + '/' + outrootfilename,
"tree",
mode="recreate")
#self.metrics.append(metric)
self.__sourceroots.append(originroot)
self.__predictroots.append(outputDir + '/' + outrootfilename)
print(formatstring)
print('\ncreated prediction friend tree ' + outputDir + '/' +
outrootfilename + ' for ' + originroot)
if self.addnumpyoutput:
if len(fullnumpyarray):
fullnumpyarray = np.concatenate(
(fullnumpyarray, all_write))
else:
fullnumpyarray = np.array(all_write)
else:
c_storeTensor.store(
np.ascontiguousarray(all_write,
dtype=np.float32).ctypes.data,
list(np.shape(all_write)),
outputDir + '/' + outrootfilename)
self.__sourceroots.append(originroot)
self.__predictroots.append(outputDir + '/' + outrootfilename)
if self.addnumpyoutput:
if len(fullnumpyarray):
fullnumpyarray = np.concatenate(
(fullnumpyarray, all_write))
else:
fullnumpyarray = np.array(all_write)
if self.addnumpyoutput:
np.save(outputDir + '/' + 'allprediction.npy', fullnumpyarray)
def makePrediction_compat(self,
model,
testdatacollection,
outputDir,
ident='',
store_labels=False,
monkey_class=''):
deprwarning = '''
WARNING! makePrediction is being run in compatibilty mode. This mode will soon be deprecated.
The new prediction mode requires to define a function of the TrainData class that takes as input the prediction of the network \
and as output a list of formatted predictions and associated names.
Each list item will be assigned an own branch name in the output tree
def formatPrediction(self, predicted_list):
format_names = [name1, name2, ...]
out_pred = [predicted_list[1],predicted_list[0]] #e.g. if the order should be swapped
return out_pred, format_names
'''
print(deprwarning)
import numpy as np
from root_numpy import array2root
import os
monkey_class_obj = None
if monkey_class:
module, classname = tuple(monkey_class.split(':'))
_temp = __import__(module, globals(), locals(), [classname], -1)
monkey_class_obj = getattr(_temp, classname)
outputDir = os.path.abspath(outputDir)
if len(ident) > 0:
ident = '_' + ident
self.__sourceroots = []
self.__predictroots = []
self.metrics = []
fullnumpyarray = np.array([])
for i in range(len(testdatacollection.samples)):
sample = testdatacollection.samples[i]
originroot = testdatacollection.originRoots[i]
outrootfilename = os.path.splitext(
os.path.basename(originroot))[0] + '_predict' + ident + '.root'
fullpath = testdatacollection.getSamplePath(sample)
if monkey_class_obj is not None:
testdatacollection.dataclass = monkey_class_obj()
td = testdatacollection.dataclass
td.readIn(fullpath)
if hasattr(td, 'customlabels'):
import copy
formatstring = copy.deepcopy(td.customlabels)
else:
truthclasses = td.getUsedTruth()
formatstring = []
if len(truthclasses) > 0 and len(truthclasses[0]) > 0:
formatstring = [
'prob_%s%s' % (i, ident) for i in truthclasses
]
regressionclasses = []
if hasattr(td, 'regressiontargetclasses'):
regressionclasses = td.regressiontargetclasses
#new implementation. Please check with the store_labels option, Mauro
formatstring.extend(
['reg_%s%s' % (i, ident) for i in regressionclasses])
features = td.x
labels = td.y
weights = None
if td.w is not None and len(td.w):
weights = td.w[0]
prediction = model.predict(features)
if self.use_only:
prediction = [prediction[i] for i in self.use_only]
if isinstance(prediction, list):
all_write = np.concatenate(prediction, axis=1)
else:
all_write = prediction
#if a prediction functor was set and the corresponding predictionFunctorClasses
#its output is also added to the final predictions
#predictionFunctorClasses: an array of strings labelling extra variables to add to the tree
#predictionFunctor: receives the numpy array with the predictions for N events
# returns MxN numpy array
# where M=len(predictionFunctorClasses)
# and N=no. events
if hasattr(td, 'predictionFunctor') and hasattr(
td, 'predictionFunctorClasses'):
print(
'Extending the output with the configured prediction functor output'
)
formatstring.extend(getattr(td, 'predictionFunctorClasses'))
all_write = np.concatenate(
[all_write,
getattr(td, 'predictionFunctor')(prediction)],
axis=1)
if all_write.ndim == 2:
if weights is not None:
all_write = np.concatenate([all_write, weights], axis=1)
formatstring.append('weight')
if not all_write.shape[1] == len(formatstring):
print(formatstring, ' vs ', all_write.shape[1])
raise ValueError(
'Prediction output does not match with the provided targets!'
)
all_write = np.core.records.fromarrays(
np.transpose(all_write), names=','.join(formatstring))
array2root(all_write,
outputDir + '/' + outrootfilename,
"tree",
mode="recreate")
#self.metrics.append(metric)
self.__sourceroots.append(originroot)
self.__predictroots.append(outputDir + '/' + outrootfilename)
print(formatstring)
print('\ncreated prediction friend tree ' + outputDir + '/' +
outrootfilename + ' for ' + originroot)
if self.addnumpyoutput:
if len(fullnumpyarray):
fullnumpyarray = np.concatenate(
(fullnumpyarray, all_write))
else:
fullnumpyarray = np.array(all_write)
else:
raise Exception(
"tensor output not supported (anymore) in compatibilty prediction mode!"
)
c_storeTensor.store(
np.ascontiguousarray(all_write,
dtype=np.float32).ctypes.data,
list(np.shape(all_write)),
outputDir + '/' + outrootfilename)
self.__sourceroots.append(originroot)
self.__predictroots.append(outputDir + '/' + outrootfilename)
if self.addnumpyoutput:
if len(fullnumpyarray):
fullnumpyarray = np.concatenate(
(fullnumpyarray, all_write))
else:
fullnumpyarray = np.array(all_write)
if self.addnumpyoutput:
np.save(outputDir + '/' + 'allprediction.npy', fullnumpyarray)
def makePrediction(self,
model,
testdatacollection,
outputDir,
ident='',
store_labels=False,
monkey_class='',
flatten_everything=True):
if not hasattr(testdatacollection.dataclass, 'formatPrediction'):
return self.makePrediction_compat(model, testdatacollection,
outputDir, ident, store_labels,
monkey_class)
if hasattr(testdatacollection.dataclass, 'customlabels'):
print(
'Warning: customlabels are replaced by the formatPrediction function and will be ignored'
)
import numpy as np
import os
monkey_class_obj = None
if monkey_class:
module, classname = tuple(monkey_class.split(':'))
_temp = __import__(module, globals(), locals(), [classname], -1)
monkey_class_obj = getattr(_temp, classname)
outputDir = os.path.abspath(outputDir)
if len(ident) > 0:
ident = '_' + ident
self.__sourceroots = []
self.__predictroots = []
self.metrics = []
for i in range(len(testdatacollection.samples)):
sample = testdatacollection.samples[i]
originroot = testdatacollection.originRoots[i]
outrootfilename = os.path.splitext(
os.path.basename(originroot))[0] + '_predict' + ident + '.root'
fullpath = testdatacollection.getSamplePath(sample)
if monkey_class_obj is not None:
testdatacollection.dataclass = monkey_class_obj()
td = testdatacollection.dataclass
td.readIn(fullpath)
features = td.x
#labels=td.y #not needed - maybe we can add this as option so all info is in the output trees (no friends)
weights = None
if len(td.w):
weights = td.w
prediction = model.predict(features)
if not isinstance(prediction, list):
prediction = [prediction]
prediction, formatstring = td.formatPrediction(prediction)
if len(formatstring) < len(prediction):
formatstring.extend([
'auto_pred_' + str(i)
for i in range(len(prediction) - len(formatstring))
])
if weights is not None:
prediction.append(weights)
formatstring.append('weight')
###### prediction functor needs to be adapted!!
#######NEW
if self.use_only:
prediction = [prediction[i] for i in self.use_only]
formatstring = [formatstring[i] for i in self.use_only]
cont_pred = [
np.ascontiguousarray(prediction[i], dtype=np.float32)
for i in range(len(prediction))
]
c_storeTensor.store([p.ctypes.data for p in cont_pred],
[a.shape for a in prediction],
outputDir + '/' + outrootfilename,
formatstring, flatten_everything)
self.__sourceroots.append(originroot)
self.__predictroots.append(outputDir + '/' + outrootfilename)
print('written prediction output ' + outputDir + '/' +
outrootfilename)
if self.addnumpyoutput:
print(
'addnumpyoutput: currently not supported (anymore), please request re-implementation'
)
def makePrediction(self, model, testdatacollection, outputDir,
ident='', store_labels = False, monkey_class=''):
import numpy as np
from root_numpy import array2root
import os
monkey_class_obj = None
if monkey_class:
module, classname = tuple(monkey_class.split(':'))
_temp = __import__(module, globals(), locals(), [classname], -1)
monkey_class_obj = getattr(_temp, classname)
outputDir=os.path.abspath(outputDir)
if len(ident)>0:
ident='_'+ident
self.__sourceroots=[]
self.__predictroots=[]
self.metrics=[]
fullnumpyarray=np.array([])
fullx = None
fulltest = None
fullpred = None
for i in range(len(testdatacollection.samples)):
sample=testdatacollection.samples[i]
originroot=testdatacollection.originRoots[i]
outrootfilename=os.path.splitext(os.path.basename(originroot))[0]+'_predict'+ident+'.root'
fullpath=testdatacollection.getSamplePath(sample)
if monkey_class_obj is not None:
testdatacollection.dataclass = monkey_class_obj()
td=testdatacollection.dataclass
td.readIn(fullpath)
if hasattr(td, 'customlabels'):
import copy
formatstring=copy.deepcopy(td.customlabels)
else:
truthclasses=td.getUsedTruth()
formatstring=[]
if len(truthclasses)>0 and len(truthclasses[0])>0:
formatstring = ['prob_%s%s' % (i, ident) for i in truthclasses]
regressionclasses=[]
if hasattr(td, 'regressiontargetclasses'):
regressionclasses=td.regressiontargetclasses
#new implementation. Please check with the store_labels option, Mauro
formatstring.extend(['reg_%s%s' % (i, ident) for i in regressionclasses])
# add the truth to the friend tree
formatstring.extend(truthclasses)
features=td.x
labels=td.y
weights=td.w[0]
#print('x,y,w:')
#print(features)
#print(labels)
#print(weights)
prediction = model.predict(features)
if fulltest is not None:
fullx = np.concatenate([fullx, features[0]], axis=0)
fulltest = np.concatenate([fulltest, labels[0]], axis=0)
fullpred = np.concatenate([fullpred, prediction], axis=0)
else:
fullx = features[0]
fulltest = labels[0]
fullpred = prediction
if self.use_only:
prediction = [prediction[i] for i in self.use_only]
if isinstance(prediction, list):
all_write = np.concatenate(prediction, axis=1)
else:
all_write = prediction
if isinstance(labels,list):
all_labels = np.concatenate(labels, axis=1)
else:
all_labels = labels
all_write = np.concatenate([all_write, all_labels], axis=1)
#if a prediction functor was set and the corresponding predictionFunctorClasses
#its output is also added to the final predictions
#predictionFunctorClasses: an array of strings labelling extra variables to add to the tree
#predictionFunctor: receives the numpy array with the predictions for N events
# returns MxN numpy array
# where M=len(predictionFunctorClasses)
# and N=no. events
if hasattr(td, 'predictionFunctor') and hasattr(td,'predictionFunctorClasses'):
print('Extending the output with the configured prediction functor output')
formatstring.extend( getattr(td,'predictionFunctorClasses') )
all_write = np.concatenate([all_write,
getattr(td,'predictionFunctor')(prediction)],
axis=1)
if all_write.ndim == 2:
all_write = np.concatenate([all_write, weights], axis=1)
formatstring.append('weight')
if not all_write.shape[1] == len(formatstring):
print(formatstring, ' vs ', all_write.shape[1])
raise ValueError('Prediction output does not match with the provided targets!')
all_write = np.core.records.fromarrays(np.transpose(all_write), names= ','.join(formatstring))
array2root(all_write,outputDir+'/'+outrootfilename,"tree",mode="recreate")
#self.metrics.append(metric)
self.__sourceroots.append(originroot)
self.__predictroots.append(outputDir+'/'+outrootfilename)
print(formatstring)
print('\ncreated prediction friend tree '+outputDir+'/'+outrootfilename+ ' for '+originroot)
if self.addnumpyoutput:
if len(fullnumpyarray):
fullnumpyarray=np.concatenate((fullnumpyarray,all_write))
else:
fullnumpyarray=np.array(all_write)
else:
c_storeTensor.store(np.ascontiguousarray(all_write, dtype=np.float32).ctypes.data, list(np.shape(all_write)), outputDir+'/'+outrootfilename)
self.__sourceroots.append(originroot)
self.__predictroots.append(outputDir+'/'+outrootfilename)
if self.addnumpyoutput:
if len(fullnumpyarray):
fullnumpyarray=np.concatenate((fullnumpyarray,all_write))
else:
fullnumpyarray=np.array(all_write)
if self.addnumpyoutput:
np.save(outputDir+'/'+'allprediction.npy', fullnumpyarray)
# WARNING
# if you have to do this, you probably ahve aproblem before, need to debug
# but when you write a row with all 0s, things dont work correctly in the output files
# skip no truth
skip = np.all(fulltest==0, axis=1)
fullx = fullx[~skip]
fullpred = fullpred[~skip]
fulltest = fulltest[~skip]
# skip multi truth
skip = np.sum(fulltest, axis=1)>1
fullx = fullx[~skip]
fullpred = fullpred[~skip]
fulltest = fulltest[~skip]
make_confusion(td.getUsedTruth(),fulltest.argmax(axis=1),fullpred.argmax(axis=1),outputDir+'/confusion.png')
make_rocs(td.getUsedTruth(),fulltest,fullpred,outputDir+'/roc.png')