Project Meeting 2025.06.10

Agenda

• Input File Validation Update (Andrew Kay)
• Telecommute Design (Sijia Wang)

Technical

Input File Validation Update

• Progress Update:
  • Additional configurations to Settings Checker
  • Update core Settings to toggle Settings Checker on and off
  • Ability to check multiple spec/coefficient files in checker (e.g. School Escort model)
  • Introduce SettingsCheckError
  • Allow attempting to load SPEC files from Annotators and Preprocessors defined at top level
  • Set up evaluation for segmented coefficients using omnibus spec (PTYPE)
  • Set up evaluation for templated coefficients
  • Created pattern for including settings to check from Extensions modules in SANDAG example
  • Improved logging
  • Opened PR from Fork
• Things for the future:
  • Trouble with some specific model components (need to revisit to integrate into the checker)
    • trip_scheduling_choice: The default YAML file (trip_scheduling_choice.yaml) is missing
    • trip_scheduling: The required trip_scheduling_coefficients file is missing
  • Pydantic should be handling issues related to handling "missing" if NONE is returned (what fields are required), not for the validation checker to handle
• David H will review the PR

Telecommute Design Update

• Follow up to poll about choices of options
• Option B was the first choice of all who voted with one exception for whom it was the second choice
• Sijia will update the memo to indicate the final choice and discuss next steps that would be needed for software implementation and downstream model components
• Sijia additional analysis:
  • All options try to answer 3 questions: does person work today, does person have out-of-home work, does person have in-home work activity?
  • Difference is in the order.
  • With Option B, there may be some conflicts with out-of-home and in-home work because they are not decided jointly and may be conflicts with CDAP escorting. (Option D represent these as joint choices)
  • May need to impose constraints on choices. Greater level of effort for Option B than previously indicated.
  • CDAP would not be re-estimated. If person has out of home activities chosen next. Would children being in school or not impact out of home activities of adults (workers)? Would need to re-estimate telecommute arrangement model or impose constraints.
  • Joe and Joel still comfortable with B. Suggest providing context in describing the limitations.
  • Marty M chose Option C first. Felt that treatment was more consistent. B does not require redefining CDAP. C would require re-estimation of CDAP which would decide whether person would work that day before making the in-home out of home choice. In Option B, out-of-home mandatory is the elemental choice.
  • Redefining CDAP choices (required in Option C) would require careful changes to downstream impacts wherever it appears in a UEC (Work X Out of Home = Mandatory Out of Home). Would require adopting the re-estimated CDAP.
  • Option B would also require mods to downstream models
• Feng X wondering whether HH surveys need to ask additional questions about in-home work activity durations. Joe, Joel thinks privacy and response rates would hamper this effort. He shared recent results from MWCOG model.