def test_dataset_with_custom_feature_hyperparameters(self):
with ScratchDir("."):
os.environ["BEEP_PROCESSING_DIR"] = os.getcwd()
os.makedirs(os.path.join(os.getcwd(), "data-share", "raw", "parameters"))
parameter_files = os.listdir(os.path.join(TEST_FILE_DIR, "data-share", "raw", "parameters"))
for file in parameter_files:
shutil.copy(os.path.join(TEST_FILE_DIR, "data-share", "raw", "parameters", file),
os.path.join(os.getcwd(), "data-share", "raw", "parameters"))
hyperparameter_dict = {'RPTdQdVFeatures': [
{'test_time_filter_sec': 1000000, 'cycle_index_filter': 6,
'diag_ref': 0, 'diag_nr': 1, 'charge_y_n': 0, 'rpt_type': 'rpt_0.2C', 'plotting_y_n': 0},
{'test_time_filter_sec': 1000000, 'cycle_index_filter': 6,
'diag_ref': 0, 'diag_nr': 1, 'charge_y_n': 0, 'rpt_type': 'rpt_1C', 'plotting_y_n': 0},
{'test_time_filter_sec': 1000000, 'cycle_index_filter': 6,
'diag_ref': 0, 'diag_nr': 1, 'charge_y_n': 0, 'rpt_type': 'rpt_2C', 'plotting_y_n': 0}],
'HPPCResistanceVoltageFeatures': [
FEATURE_HYPERPARAMS['HPPCResistanceVoltageFeatures']],
'DiagnosticSummaryStats': [FEATURE_HYPERPARAMS['DiagnosticSummaryStats']]
}
dataset = BeepDataset.from_processed_cycler_runs('test_dataset',
project_list=None,
processed_run_list=[DIAGNOSTIC_PROCESSED,
FASTCHARGE_PROCESSED],
feature_class_list=FEATURIZER_CLASSES,
processed_dir=TEST_FILE_DIR,
hyperparameter_dict=hyperparameter_dict,
feature_dir='data-share/features')
self.assertEqual(dataset.name, 'test_dataset')
self.assertEqual(dataset.data.shape, (1, 159))
self.assertEqual(dataset.data.seq_num.iloc[0], 240)
self.assertIsNone(dataset.X_test)
self.assertEqual(dataset.missing.shape, (5, 2))
self.assertEqual(dataset.missing.filename.iloc[0],
os.path.split(FASTCHARGE_PROCESSED)[1])
def test_from_features(self):
dataset = BeepDataset.from_features('test_dataset', ['PreDiag'], FEATURIZER_CLASSES,
feature_dir=os.path.join(TEST_FILE_DIR, 'data-share/features'))
self.assertEqual(dataset.name, 'test_dataset')
self.assertEqual(dataset.data.shape, (2, 56))
#from pdb import set_trace; set_trace()
self.assertListEqual(list(dataset.data.seq_num), [196, 197])
self.assertIsNone(dataset.X_test)
self.assertSetEqual(set(dataset.feature_sets.keys()), {'RPTdQdVFeatures', 'DiagnosticSummaryStats'})
self.assertEqual(dataset.missing.feature_class.iloc[0], 'HPPCResistanceVoltageFeatures')
def test_serialization(self):
with ScratchDir("."):
os.environ["BEEP_PROCESSING_DIR"] = os.getcwd()
dataset = BeepDataset.from_features('test_dataset', ['PreDiag'], FEATURIZER_CLASSES,
feature_dir=os.path.join(TEST_FILE_DIR, 'data-share/features'))
dumpfn(dataset, 'temp_dataset.json')
dataset = loadfn('temp_dataset.json')
self.assertEqual(dataset.name, 'test_dataset')
self.assertEqual(dataset.data.shape, (2, 56))
# from pdb import set_trace; set_trace()
self.assertListEqual(list(dataset.data.seq_num), [196, 197])
self.assertIsNone(dataset.X_test)
self.assertSetEqual(set(dataset.feature_sets.keys()), {'RPTdQdVFeatures', 'DiagnosticSummaryStats'})
self.assertEqual(dataset.missing.feature_class.iloc[0], 'HPPCResistanceVoltageFeatures')
self.assertIsInstance(dataset.filenames, list)
os.environ["BEEP_PROCESSING_DIR"] = os.getcwd()
dataset2 = BeepDataset.from_features('test_dataset', ['PreDiag'], [RPTdQdVFeatures],
feature_dir=os.path.join(TEST_FILE_DIR, 'data-share/features'))
dumpfn(dataset2, "temp_dataset_2.json")
dataset2 = loadfn('temp_dataset_2.json')
self.assertEqual(dataset2.missing.columns.to_list(), ["filename", "feature_class"])
def test_train_test_split(self):
dataset = BeepDataset.from_features('test_dataset', ['PreDiag'],
FEATURIZER_CLASSES,
feature_dir=os.path.join(
TEST_FILE_DIR, 'data-share',
'features'))
predictors = dataset.feature_sets['RPTdQdVFeatures'][0:3] + \
dataset.feature_sets['DiagnosticSummaryStats'][-3:]
X_train, X_test, y_train, y_test = \
dataset.generate_train_test_split(predictors=predictors,
outcomes=dataset.feature_sets['RPTdQdVFeatures'][-1],
test_size=0.5, seed=123,
parameters_path=os.path.join(TEST_FILE_DIR,
'data-share',
'raw',
'parameters'))
self.assertEqual(dataset.data.shape, (2, 55))
self.assertEqual(dataset.X_test.shape, (1, 6))
self.assertEqual(dataset.X_train.shape, (1, 6))
parameter_dict = {
'PreDiag_000197': {
'project_name': 'PreDiag',
'seq_num': 197,
'template': 'diagnosticV3.000',
'charge_constant_current_1': 0.2,
'charge_percent_limit_1': 30,
'charge_constant_current_2': 0.2,
'charge_cutoff_voltage': 3.7,
'charge_constant_voltage_time': 30,
'charge_rest_time': 5,
'discharge_constant_current': 0.2,
'discharge_cutoff_voltage': 3.5,
'discharge_rest_time': 15,
'cell_temperature_nominal': 25,
'cell_type': 'Tesla_Model3_21700',
'capacity_nominal': 4.84,
'diagnostic_type': 'HPPC+RPT',
'diagnostic_parameter_set': 'Tesla21700',
'diagnostic_start_cycle': 30,
'diagnostic_interval': 100
}
}
self.assertDictEqual(dataset.train_cells_parameter_dict,
parameter_dict)
def test_from_processed_cycler_run_list(self):
with ScratchDir("."):
os.environ["BEEP_PROCESSING_DIR"] = os.getcwd()
dataset = BeepDataset.from_processed_cycler_runs(
'test_dataset',
project_list=None,
processed_run_list=[
DIAGNOSTIC_PROCESSED, FASTCHARGE_PROCESSED
],
feature_class_list=FEATURIZER_CLASSES,
processed_dir=TEST_FILE_DIR,
feature_dir='data-share/features')
self.assertEqual(dataset.name, 'test_dataset')
self.assertEqual(dataset.data.shape, (1, 118))
self.assertIsNone(dataset.X_test)
self.assertEqual(dataset.missing.shape, (3, 2))
self.assertEqual(dataset.missing.filename.iloc[0],
os.path.split(FASTCHARGE_PROCESSED)[1])
def test_from_processed_cycler_run_list(self):
with ScratchDir("."):
os.environ["BEEP_PROCESSING_DIR"] = os.getcwd()
os.makedirs(os.path.join(os.getcwd(), "data-share", "raw", "parameters"))
parameter_files = os.listdir(os.path.join(TEST_FILE_DIR, "data-share", "raw", "parameters"))
for file in parameter_files:
shutil.copy(os.path.join(TEST_FILE_DIR, "data-share", "raw", "parameters", file),
os.path.join(os.getcwd(), "data-share", "raw", "parameters"))
dataset = BeepDataset.from_processed_cycler_runs('test_dataset',
project_list=None,
processed_run_list=[DIAGNOSTIC_PROCESSED,
FASTCHARGE_PROCESSED],
feature_class_list=FEATURIZER_CLASSES,
processed_dir=TEST_FILE_DIR,
feature_dir='data-share/features')
self.assertEqual(dataset.name, 'test_dataset')
self.assertEqual(dataset.data.shape, (1, 143))
self.assertEqual(dataset.data.seq_num.iloc[0], 240)
self.assertIsNone(dataset.X_test)
self.assertEqual(dataset.missing.shape, (3, 2))
self.assertEqual(dataset.missing.filename.iloc[0],
os.path.split(FASTCHARGE_PROCESSED)[1])