Duplicate transit segments cause join explosion

[yueshuaing]

Background
tm2py exports two set of transit segment-level outputs:

• boardings_by_segment_{time_period}.geojson

• transit_segment_{time_period}.csv (more variables)

Transit segments can repeat within a line (loops/round trips). e.g., line 10_573_201_AM_d0_s490 traverses the same segment (INODE=266386, JNODE=294084) twice. The second pass is indicated by a “-2” suffix. The two passes have different boardings.

Issue
Because of duplicate segments, it is not reliable to merge the two outputs based on LINE_ID, instead we should merge based on the unique longer segment name (line in the screenshots)

tm2py-utils/tm2py_utils/summary/acceptance/simulated.py

Lines 434 to 446 in 3871f75

	in_file = self.scenario_dir / f"output_summaries/boardings_by_segment_{time_period}.geojson"
	logging.info(f"Reading {in_file}")
	gdf = gpd.read_file(in_file)
	logging.debug(f"gdf:\n{gdf}")
	gdf["first_row_in_line"] = gdf.groupby("LINE_ID").cumcount() == 0
	df = pd.DataFrame(gdf.drop(columns=["geometry"]))
	# Add am boards
	a_df = self.am_segment_boardings_df
	df = pd.merge(df, a_df, how="left", on=["LINE_ID", "INODE", "JNODE"])

Plan
We could add the unique segment name to "boardings_by_segment_{time_period}.geojson, and also add the boarding variable to it so we don't need extra steps to merge boarding from transit_segment_{time_period}.csv. This can be done in tm2py, post-processor