Description
We need to investigate whether SANDAG ABM resident workplace simulation shadow pricing is using the correct chooser segmentation column during workplace resimulation.
The current resident workplace location model segments choosers by occupation:
CHOOSER_SEGMENT_COLUMN_NAME: occupation
However, ActivitySim simulation-based shadow pricing can default the workplace segment field to income_segment unless SEGMENT_TO_NAME is explicitly set in shadow_pricing.yaml. Further details can be found in ActivitySim/activitysim#1082.
Proposed test change
Add the following to src/asim/configs/resident/shadow_pricing.yaml:
SEGMENT_TO_NAME:
school: school_segment
workplace: occupation
Create scatter plots comparing modeled workers by workplace zone against jobs/employment from the land-use file for both existing setup and test change scenarios.
- x-axis: land-use employment by zone
total_emp_revised
- y-axis: modeled resident workers assigned to workplace zone (remove external workers)
- one point per workplace zone
- include 45-degree reference line
Additional proposed statistics that might be helpful:
- number of workplace zones included
- total modeled workers
- total land-use employment target
- correlation between modeled workers and land-use employment
A review of the activitysim run logs for shadow pricing details would also be helpful.
Description
We need to investigate whether SANDAG ABM resident workplace simulation shadow pricing is using the correct chooser segmentation column during workplace resimulation.
The current resident workplace location model segments choosers by
occupation:CHOOSER_SEGMENT_COLUMN_NAME: occupationHowever, ActivitySim simulation-based shadow pricing can default the workplace segment field to
income_segmentunlessSEGMENT_TO_NAMEis explicitly set inshadow_pricing.yaml. Further details can be found in ActivitySim/activitysim#1082.Proposed test change
Add the following to
src/asim/configs/resident/shadow_pricing.yaml:Create scatter plots comparing modeled workers by workplace zone against jobs/employment from the land-use file for both existing setup and test change scenarios.
total_emp_revisedAdditional proposed statistics that might be helpful:
A review of the activitysim run logs for shadow pricing details would also be helpful.