Fixed a coefficients lookup bug for university tour mode choice that was introduced when separating data from coefficients for estimation. The meaning of school in TM1 was overloaded - sometimes it meant school (both k-12 and univ) and sometimes it meant just k-12. We should eventually clean this up throughout the system.
Our test system should catch this kind of stuff
We're using a very small sample / set of zones for testing so some issues can slip through. Maybe we should fix this?
@Stefan/Joe to help with community questions like #326.
This version is much stronger than the previous version
Still lots to talk about, such as agreeing on the principles (i.e. what is ActivitySim), what the test system is, and supporting use in multiple regions
@Everyone review principles ahead of next call
Task 10 Support for Three Zone Systems and TVPB (milestone)
Three zones and TVPB working through tour mode choice for the simple test example
Running slower than one/two zone systems but this is expected at this time
We discussed some improvements to the design/specification and Jeff is working on those - namely syncing up the accessibility calculator and the tour/trip mode choice model to use the same underlying three zone / TVPB engine so there is no (or less) duplication
Plan to get three zone / TVPB working through trip models next week
Will turn our attention to performance / caching / multiprocessing needs after that but need the better example soon
@Lisa and I coordinating on the TM2 tour mode choice data prep
Plan to use the Marin County version of TM2 since it is the most stable
Lisa and I will reach out to Marin for an official data request
Maybe test TM2 accessibility calculator too since it also changed from TM1 and is before tour mode choice in the model stream
May need to hold / slow development until better example ready