def test3_duplicated_parameter_specification(self):
"""
This function will randomly choose a parameter in hyper_parameters and specify it as a parameter in the
model. Depending on the random error_number generated, the following is being done to the model parameter
and hyper-parameter:
error_number = 0: set model parameter to be a value in the hyper-parameter value list, should
generate error;
error_number = 1: set model parameter to be default value, should not generate error in this case;
error_number = 2: make sure model parameter is not set to default and choose a value not in the
hyper-parameter value list.
:return: None
"""
print("*******************************************************************************************")
print("test3_duplicated_parameter_specification for GLM " + self.family)
error_number = np.random.random_integers(0, 2, 1) # randomly choose an error
print("error_number is {0}".format(error_number[0]))
params_dict, error_hyper_params = \
pyunit_utils.generate_redundant_parameters(self.hyper_params, self.gridable_parameters,
self.gridable_defaults, error_number[0])
params_dict["family"] = self.family
params_dict["nfolds"] = self.nfolds
print("Your hyper-parameter dict is: ")
print(error_hyper_params)
print("Your model parameters are: ")
print(params_dict)
# copied from Eric to catch execution run errors and not quit
try:
grid_model = H2OGridSearch(H2OGeneralizedLinearEstimator(**params_dict),
hyper_params=error_hyper_params)
grid_model.train(x=self.x_indices, y=self.y_index, training_frame=self.training1_data)
# if error_number = 1, it is okay. Else it should fail.
if not (error_number[0] == 1):
self.test_failed += 1
self.test_failed_array[self.test_num] = 1
print("test3_duplicated_parameter_specification failed: Java error exception should have been "
"thrown but did not!")
else:
print("test3_duplicated_parameter_specification passed: Java error exception should not have "
"been thrown and did not!")
except:
if error_number[0] == 1:
self.test_failed += 1
self.test_failed_array[self.test_num] = 1
print("test3_duplicated_parameter_specification failed: Java error exception should not "
"have been thrown! ")
else:
print("test3_duplicated_parameter_specification passed: Java error exception should "
"have been thrown and did.")
def test3_duplicated_parameter_specification(self):
"""
This function will randomly choose a parameter in hyper_parameters and specify it as a parameter in the
model. Depending on the random error_number generated, the following is being done to the model parameter
and hyper-parameter:
error_number = 0: set model parameter to be a value in the hyper-parameter value list, should
generate error;
error_number = 1: set model parameter to be default value, should not generate error in this case;
error_number = 2: make sure model parameter is not set to default and choose a value not in the
hyper-parameter value list.
:return: None
"""
print("*******************************************************************************************")
print("test3_duplicated_parameter_specification for GLM " + self.family)
error_number = np.random.random_integers(0, 2, 1) # randomly choose an error
params_dict, error_hyper_params = \
pyunit_utils.generate_redundant_parameters(self.hyper_params, self.gridable_parameters,
self.gridable_defaults, error_number[0])
params_dict["family"] = self.family
params_dict["nfolds"] = self.nfolds
print("Your hyper-parameter dict is: ")
print(error_hyper_params)
print("Your model parameters are: ")
print(params_dict)
# copied from Eric to catch execution run errors and not quit
try:
if "max_runtime_secs" in list(params_dict): # need to set max_runtime_secs when calling train
max_runtime_secs = params_dict["max_runtime_secs"]
del params_dict["max_runtime_secs"]
grid_model = H2OGridSearch(H2OGeneralizedLinearEstimator(**params_dict),
hyper_params=error_hyper_params)
grid_model.train(x=self.x_indices, y=self.y_index, training_frame=self.training1_data,
max_runtime_secs=max_runtime_secs)
else:
grid_model = H2OGridSearch(H2OGeneralizedLinearEstimator(**params_dict),
hyper_params=error_hyper_params)
grid_model.train(x=self.x_indices, y=self.y_index, training_frame=self.training1_data)
# if error_number = 1, it is okay. Else it should fail.
if not (error_number[0] == 1):
self.test_failed += 1
self.test_failed_array[self.test_num] = 1
print("test3_duplicated_parameter_specification failed: Java error exception should have been "
"thrown but did not!")
else:
print("test3_duplicated_parameter_specification passed: Java error exception should not have "
"been thrown and did not!")
except:
if error_number[0] == 1:
self.test_failed += 1
self.test_failed_array[self.test_num] = 1
print("test3_duplicated_parameter_specification failed: Java error exception should not "
"have been thrown! ")
else:
print("test3_duplicated_parameter_specification passed: Java error exception should "
"have been thrown and did.")