def test_read_csv():
# TODO: Is there a reason to replace " " in the activity name with "_"?
event_stream_test = read_csv("demo/Detail_Supplier_IW-Frozen.csv")
assert isinstance(event_stream_test, obj.EventStream)
assert len(event_stream_test) == 5000
first_event = event_stream_test[0]
assert first_event["case:concept:name"] == "Case 496"
assert first_event["concept:name"] == "Process Creation"
assert isinstance(first_event["time:timestamp"], datetime)
assert first_event["time:timestamp"] == datetime(2010, 9, 21, 9, 0, 13)
second_event = event_stream_test[1]
assert second_event["case:concept:name"] == "Case 12186"
assert second_event["concept:name"] == "Process Creation"
assert isinstance(second_event["time:timestamp"], datetime)
assert second_event["time:timestamp"] == datetime(2010, 9, 21, 9, 0, 21)
second_to_last_event = event_stream_test[-2]
assert second_to_last_event["case:concept:name"] == "Case 8848"
assert second_to_last_event["concept:name"] == "ME Fabrication Checker"
assert isinstance(second_to_last_event["time:timestamp"], datetime)
assert second_to_last_event["time:timestamp"] == datetime(
2011, 4, 27, 8, 0, 59)
last_event = event_stream_test[-1]
assert last_event["case:concept:name"] == "Case 10634"
assert last_event["concept:name"] == "ME Fabrication Checker"
assert isinstance(last_event["time:timestamp"], datetime)
assert last_event["time:timestamp"] == datetime(2011, 4, 27, 9, 0, 0)
def test_run(self, process):
dir_path = 'demo'
filename = 'Detail_Supplier_IW-Frozen.csv'
stream = read_csv(f'{dir_path}/{filename}')
process.run(stream)
assert process.initialized
assert path.isfile(f'./metrics/{process.name}_process_model_graphs/process_model_graph_{process.cp_count}.json')
def test_run(self, process):
stream = read_csv("demo/Detail_Supplier_IW-Frozen.csv")
process.run(stream)
assert process.initialized
assert path.isfile(
f"output/metrics/{process.name}_process_model_graphs/process_model_graph_{process.cp_count}.json"
)
def test_pipeline():
event_stream_test = read_csv(f"demo/sudden_trace_noise20_500_cb.csv")
cdesf = CDESF(
name="sudden_trace_noise20_500_cb",
time_horizon=259200,
lambda_=0.05,
beta=0.2,
epsilon=0.6,
mu=4,
stream_speed=100,
gen_metrics=True,
gen_plot=False,
)
cdesf.run(event_stream_test)
def test_attributed_pipeline():
event_stream_test = read_csv(
f"demo/sudden_trace_noise20_500_cb_attributed.csv")
cdesf = CDESF(
name="sudden_trace_noise20_500_cb_attributed",
time_horizon=259200,
lambda_=0.05,
beta=0.2,
epsilon=0.6,
mu=4,
stream_speed=100,
gen_metrics=True,
gen_plot=False,
additional_attributes=["first", "second", "country"],
)
cdesf.run(event_stream_test)
def test_pipeline():
path = 'demo'
filename = 'sudden_trace_noise20_500_cb.csv'
event_stream_test = read_csv(f'{path}/{filename}')
cdesf = CDESF(name='sudden_trace_noise20_500_cb',
time_horizon=259200,
lambda_=0.05,
beta=0.2,
epsilon=0.6,
mu=4,
stream_speed=100,
n_features=2,
gen_metrics=True,
gen_plot=False)
cdesf.run(event_stream_test)
def test_read_csv():
path = 'demo'
filename = 'Detail_Supplier_IW-Frozen.csv'
event_stream_test = read_csv(f'{path}/{filename}')
assert isinstance(event_stream_test, np.ndarray)
assert len(event_stream_test) == 5000
first_event = event_stream_test[0]
assert first_event[0] == 'Case 496'
assert first_event[1] == 'Process_Creation'
assert isinstance(first_event[2], datetime)
assert first_event[2] == datetime(2010, 9, 21, 9, 0, 13)
second_event = event_stream_test[1]
assert second_event[0] == 'Case 12186'
assert second_event[1] == 'Process_Creation'
assert isinstance(second_event[2], datetime)
assert second_event[2] == datetime(2010, 9, 21, 9, 0, 21)
second_to_last_event = event_stream_test[4998]
assert second_to_last_event[0] == 'Case 8848'
assert second_to_last_event[1] == 'ME_Fabrication_Checker'
assert isinstance(second_to_last_event[2], datetime)
assert second_to_last_event[2] == datetime(2011, 4, 27, 8, 0, 59)
last_event = event_stream_test[4999]
assert last_event[0] == 'Case 10634'
assert last_event[1] == 'ME_Fabrication_Checker'
assert isinstance(last_event[2], datetime)
assert last_event[2] == datetime(2011, 4, 27, 9, 0, 0)
event_stream = pd.read_csv(f'{path}/{filename}',
usecols=['case', 'activity', 'timestamp'],
parse_dates=['timestamp'],
infer_datetime_format=True,
memory_map=True)
event_stream['activity'].replace(' ', '_', regex=True, inplace=True)
assert np.all(event_stream.values == event_stream_test)