Polars: Parallelizing creating a new dataframe while using groupby of another dataframe - python-polars

Question

I have an artifacts dataframe which I need to groupby some columns. This sub-dataframe is then passed to a function where I create a new dataframe.
for sub_df in merged_dataset.groupby(['PRODUCT','PROGRAM','TESTSUITE','MODULE', 'BASECLASS', 'SUBCLASS']):
flow_summary_df = sub_df.select(['PRODUCT','PROGRAM','TESTSUITE','MODULE', 'BASECLASS', 'SUBCLASS', 'VERSION', 'RELEASE_DATE', 'Status'])
current_test_events_dataset = add_test_events(flow_summary_df)
Here is the add_test_events code:
def add_test_events_new(flow_summary_df):
previous_test_status = None
known_versions = flow_summary_df.VERSION.to_list()
test_events = []
test_event_versions = []
test_events_data = {}
for i, version in enumerate(known_versions):
if i == 0:
test_events.append('Introduced')
test_event_versions.append(version)
if flow_summary_df.Status[i] == 'BINNING':
test_events.append('Binning Enabled')
test_event_versions.append(version)
previous_test_status = flow_summary_df.Status[i]
continue
if flow_summary_df.Status[i] == 'BYPASSED' and previous_test_status != 'BYPASSED':
test_events.append('Removed')
test_event_versions.append(version)
previous_test_status = 'BYPASSED'
elif flow_summary_df.Status[i] == 'BINNING' and previous_test_status != 'BINNING':
test_events.append('Binning Enabled')
test_event_versions.append(version)
previous_test_status = 'BINNING'
elif flow_summary_df.Status[i] == 'NON-BINNING' and previous_test_status == 'BINNING':
test_events.append('Binning Disabled')
test_event_versions.append(version)
previous_test_status = 'NON-BINNING'
elif flow_summary_df.Status[i] != 'BYPASSED' and previous_test_status == 'BYPASSED':
test_events.append('Re-Introduced')
test_event_versions.append(version)
if flow_summary_df.Status[i] == 'BINNING':
test_events.append('Binning Enabled')
for column in ['PRODUCT','PROGRAM','TESTSUITE','MODULE', 'BASECLASS', 'SUBCLASS', 'VERSION', 'RELEASE_DATE', 'TEST_EVENT']:
if column == 'TEST_EVENT':
test_events_data[column] = test_events
elif column == 'VERSION':
test_events_data[column] = test_event_versions
else:
if flow_summary_df[column][0] is None:
test_events_data[column] = [''] * len(test_events)
else:
test_events_data[column] = [flow_summary_df[column][0]] * len(test_events)
test_events_df = pl.DataFrame(test_events_data)
return test_events_df
It just feels like this is not the polars way to do things.
thx