Check segments

[tiffanychu90]

segment speeds

• Do a comprehensive review of segments, especially look at how loop_or_inlining segments are, and whether improvements could be made Research Request - Plug in gaps in stop segments #811
• Exploratory notebooks: 16_stop_combos_for_segments and 17_debug_empty_stop_segments
• Several issues identified:
  • Multiple stop_sequences for different stop_ids present in a shape --> select 1 trip with most stops to cut segments for that shape
  • Gaps remain in between segments --> especially on curves and turns, the straight line distance underestimates (hypotenuse vs sum of 2 straight edges). add an extra check where the proposed end point is compared to original, and we tack on the leftover distance when we're subsetting the shape for the portion between 2 stops. marginal improvements, and making segments slightly longer won't hurt
  • Difficulty in segmenting loops near origin --> tried an extra super_project check to then try subsetting again using a combination of shape_geometry and altered/monotonically cast shape_geometry. this produced some improvements in problem segments, but also created more errors in otherwise correctly cut segments. forgo this approach because we don't want to create more issues than we're solving. incorrect segments are 1-2%.

[github-actions]

nbviewer URLs for impacted notebooks:

• rt_segment_speeds/16_stop_combos_for_segments.ipynb
• rt_segment_speeds/17_debug_empty_stop_segments.ipynb

[github-actions]

nbviewer URLs for impacted notebooks:

• rt_segment_speeds/16_stop_combos_for_segments.ipynb
• rt_segment_speeds/17_debug_empty_stop_segments.ipynb

[github-actions]

nbviewer URLs for impacted notebooks:

• rt_segment_speeds/16_stop_combos_for_segments.ipynb
• rt_segment_speeds/17_debug_empty_stop_segments.ipynb

[github-actions]

nbviewer URLs for impacted notebooks:

• rt_segment_speeds/16_stop_combos_for_segments.ipynb
• rt_segment_speeds/17_debug_empty_stop_segments.ipynb