Skip to content

Project Meeting 2024.12.10

rcopperman edited this page Dec 10, 2024 · 1 revision

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)
Clone this wiki locally