Project Meeting 2018.01.09

SANDAG multiple zone system prototype

• Project completed and final test results posted
• Made some additional speed-ups due to sampling improvements, as described in #167

Documentation and release

• Finished outstanding documentation updates, merged develop into master, updated issues and milestones
• Posted a new release to pypi
• Updated the Progress Report
• Drafted initial Contribution Review Questions for review
• Elizabeth suggested we look into Python 3 compatibility now #206
• We plan to soon address issue #208 as well

TRB meeting

• Work program for 1st half of 2018
  • Bill should be forwarding the contract to RSG any day
  • Contract will be for 12 months, but we plan to work ahead of schedule so we can add funds for additional work in the second half of the year
  • Trip models will be more difficult to implement than the joint travel models so maybe we should do them first
• Improvements to the repository administration:
  • We're getting significant interest from additional potential partners
  • The related PopulationSim, benefits calculator, and buffering tool (@stefancoe) packages are all close to done
  • So, to continue to formalize the ActivitySim effort, we will:
    • Adopt github.com/activitysim as our official home, similar to visioneval, tlumip, astropy, etc. and make the PMC members the organization owners
    • Transfer activitysim/activitysim to activitysim/activitysim (the old site will redirect)
    • Redirect ActivitySim.org to github.com/activitysim
    • Create a very lightweight website for the project that is hosted on GitHub. See VisionEval's site and source repo for example.
    • Add a page describing affiliated packages, much like astropy does, with links to popsim, bca, etc. JC: Making clear that these are not yet formally part of the activitysim project/governance, though long term vision to unite all these projects/packages.
    • Update the Governance document as needed
    • @Stefan to refactor the buffering model to be an affiliated package
• Initial thoughts around who might first adopt the new platform
  • ARC is a good candidate since it has a very similar model design to the example and uses one zone system (TAZs)
  • MTC and SANDAG both use multiple zone systems so likely to wait until this feature is added
  • PSRC and SFCTA are DaySim model designs and have multiple zone systems as well
  • An agency without an ABM is also a good candidate
  • Maybe we should migrate the current example to ARC's data instead of MTC's
• Response to agencies interested in joining the effort:
  • We assume new agencies will join the consortium and contribute financially each year like the rest of the agencies
  • This gets them into the prioritization discussion about how to spend the annual budget
  • In order to develop an activitysim-based model in a region, the agency should issue a separate RFP with a task for "localization" of the framework, which includes modifying/adding/removing features and updating utilities, data, reporting, etc. as needed for their local needs
  • We're planning a beta release around next TRB, so localization could happen after that