Project Meeting 2023.05.09 - ActivitySim/activitysim GitHub Wiki

Agenda

  • Road Map Update - Discussion: Software + Behavior (WSP)

Meeting Notes

Road Map Update - Discussion: Software + Behavior

  • ActivitySim 2.0 (Level 3 example model)
    • Not proposing a completely novel approach, we would want to borrow from successful implementations.
    • Proposed big leap would be the consistency in space/time for activities, vehicles, etc.
      • A first step towards that could be to output itineraries
      • Vehicle allocation could be a good first step towards the big leap to test level of effort / implementation requirements.
      • There was a general question about the quantity of schedule conflicts that we would be proposing re-simulating (unknown at this time).
    • Path forward could be fundamental incrementalism or broader refactoring
      • Data model and LOS manager could be a good proof of concept / incremental step. Both could be stand alone things.
    • Voiced support for the following things:
      • Model specification GUI / data model
      • vehicle allocation, accessibility-based sampling
    • Voiced concerns for the following things:
      • Do we have the data to support the proposed big leap for a Level 3 model? In-home activity data is difficult to collect.
      • Would the consortium have the funds to support development and maintenance of all three levels of ActivitySim?
      • At a conceptual level, can we sketch out the level of computational requirements? Do we have the means to implement? Is there an incremental approach possible?
      • Would the proposed framework be able to handle disaggregate establishments? CMAP and Ohio have or plan to have this.
    • Additional questions:
      • How much detail can you provide on a proposed Level 3 framework?
      • How would/could an LOS manager efficiently interact with trajectories and DTAs?
        • Trajectories record node-to-node movements instead of skimming zone-to-zone (something to replace skimming a ridiculous number of zones, especially at the MAZ level)
    • Need to make a decision to understand how to decide on Level 3 framework.
      • Do we just put everything on the table and see how things shake out? What’s the process? Will the roadmap provide this type of direction?
      • Proposed to enumerate the benefits and then have a discussion.
  • ActivitySimple
    • Concerns
      • Don’t overly constrain with a max number of nodes, for example. Still needs to be customizable.
  • Roadmap will set a direction but will be routinely updated and actively adjusting the course as we move along and learn new things.