Project Meeting 2024.12.10 - ActivitySim/activitysim GitHub Wiki
Agenda
Telecommuting Update - Software changes
Technical
Telecommuting Update - software changes
Review of existing model design
Work from Home
Work Location - only runs for workers not working exclusively from home
Telecommute frequency - simulates number of days a worker telecommutes
CDAP - simulates day pattern of worker. Frequent telecommuters are less likely to have mandatory pattern (out of home work tour); only mandatory pattern will have out of home work tour
Very little behavioral difference in model between telecommuters working from home on travel day and non-workers on travel day
But they do have behavioral differences
Ideally have three groups: Commuters; telecommuters; not working
Proposed changes
Revise the CDAP configuration
Mandatory pattern = working on the day, regardless of working from home or working outside the home
Non-mandatory pattern or home pattern = not working on the day
Simulate whether telecommuters are telecommuting that day based on telecommuting frequency information
If telecommuting, overwrite the work location to be home location; make sure subsequent tour generation is correctly anchored
There is concern that its changing definition of tour and may trigger, for example, running mode choice model for a non-travel tour; Concern that it introduces non-intuitive representation of behavior
Question about whether there is even a simpler solution
Implications
Run time and memory usage
Will increase choosers in mandatory tour scheduling model; increase will be moderate since with explicit chunking the burden should be alleviated
SANDAG model result
Would do light calibration
Positive: Allows for modelers to explicitly identify telecommuters and tweak behavioral representation
Next Steps
Implement software changes
Run model
See how far off model is compared to existing SANDAG model to inform calibration
Suggested next meeting: January 14th (Tuesday after TRB)