def load_datasets(self):
''' load train and validation dataset '''
self.train_data = data_frame.DataFrame(
self.in_path+"_train.h5", output_label = self.class_label,
phi_padding = self.phi_padding )
self.val_data = data_frame.DataFrame(
self.in_path+"_val.h5", output_label = self.class_label,
phi_padding = self.phi_padding )
self.num_classes = self.train_data.num_classes
def load_datasets(self):
''' load train and validation dataset '''
self.train_data = data_frame.DataFrame(
self.in_path + "_train.h5",
output_label=self.class_label,
variables=self.input_variables,
n_particles=self.n_particles,
normed_inputs=self.normed_inputs)
self.val_data = data_frame.DataFrame(self.in_path + "_val.h5",
output_label=self.class_label,
variables=self.input_variables,
n_particles=self.n_particles,
normed_inputs=self.normed_inputs)
self.num_classes = self.train_data.num_classes
def eval_model(self):
# loading test examples
self.test_data = data_frame.DataFrame(
self.in_path+"_test.h5", output_label = self.class_label,
phi_padding = self.phi_padding )
self.target_names = [self.test_data.inverted_label_dict[i] for i in range(
self.test_data.min_jets, self.test_data.max_jets+1)]
self.test_eval = self.model.evaluate(
self.test_data.X, self.test_data.one_hot)
print("test loss: {}".format( self.test_eval[0] ))
for im, metric in enumerate(self.eval_metrics):
print("test {}: {}".format( metric, self.test_eval[im+1] ))
self.history = self.trained_model.history
self.predicted_vector = self.model.predict( self.test_data.X )
self.predicted_classes = np.argmax( self.predicted_vector, axis = 1)
self.predicted_classes = np.array([self.test_data.max_jets if j >= self.test_data.max_jets \
else self.test_data.min_jets if j <= self.test_data.min_jets \
else j for j in self.predicted_classes])
self.confusion_matrix = confusion_matrix(
self.test_data.Y, self.predicted_classes )
def _load_datasets(self):
''' load data set '''
return data_frame.DataFrame(input_samples=self.input_samples,
event_category=self.event_category,
train_variables=self.train_variables,
test_percentage=self.test_percentage,
norm_variables=True,
additional_cut=self.additional_cut)
def _load_datasets(self):
''' load dataset '''
return data_frame.DataFrame(path_to_input_files=self.in_path,
classes=self.event_classes,
event_category=self.event_category,
train_variables=self.train_variables,
prenet_targets=self.prenet_targets,
test_percentage=self.test_percentage,
norm_variables=True,
additional_cut=self.additional_cut)
def _load_datasets(self):
''' load data set '''
return data_frame.DataFrame(
input_samples = self.input_samples,
input_features = self.inputs,
target_features = self.targets,
feature_scaling = self.feature_scaling,
test_percentage = self.test_percentage,
val_percentage = self.val_percentage)
def _load_datasets(self, shuffle_seed, balanceSamples):
''' load data set '''
return data_frame.DataFrame(input_samples=self.input_samples,
event_category=self.category_cutString,
train_variables=self.train_variables,
test_percentage=self.test_percentage,
norm_variables=self.norm_variables,
shuffleSeed=shuffle_seed,
balanceSamples=balanceSamples,
evenSel=self.evenSel,
addSampleSuffix=self.addSampleSuffix)
def eval_model(self):
self.test_data = data_frame.DataFrame(self.in_path + "_test.h5")
self.test_eval = self.model.evaluate(self.test_data.X,
self.test_data.X)
print("test loss: {}".format(self.test_eval[0]))
for im, metric in enumerate(self.eval_metrics):
print("test {}: {}".format(metric, self.test_eval[im + 1]))
self.history = self.trained_model.history
self.test_encoded_images = self.encoder.predict(self.test_data.X)
self.test_decoded_images = self.model.predict(self.test_data.X)
def eval_model(self):
# loading test examples
self.test_data = data_frame.DataFrame(
self.in_path+"_test.h5", output_label = self.class_label, one_hot = False,
phi_padding = self.phi_padding )
self.test_eval = self.model.evaluate(
self.test_data.X, self.test_data.Y)
print("test loss: {}".format( self.test_eval[0] ))
for im, metric in enumerate(self.eval_metrics):
print("test {}: {}".format( metric, self.test_eval[im+1] ))
self.history = self.trained_model.history
self.predicted_vector = self.model.predict( self.test_data.X )
# correct predictons (kind of arbitrary)
# everything so far only implemented for a single output neuron with integer targets
self.predicted_classes = [int(val+0.5) for val in self.predicted_vector[:,0]]
self.confusion_matrix = confusion_matrix(
self.test_data.Y, self.predicted_classes)
def load_datasets(self):
''' load train and validation dataset '''
self.train_data = data_frame.DataFrame(self.in_path + "_train.h5")
self.val_data = data_frame.DataFrame(self.in_path + "_val.h5")