Input checker framework for each user of ActivitySim to have a place to script out in a repeatable way, which verifies all data coming into ActivitySim.
Additional functionality to create additional summaries.
Much of the work has been done with SEMCOG to integrate all practical checks. With SEMCOG’s work, able to ensure all changes and additions run well and pass as to be expected.
In terms of log output, much cleaner and easier from a user perspective.
Addition to test_input_checker.py, which tests passing data frame, etc.
Next steps: Finish continuous integration testing and work with SEMCOG to clean up their example and answer any questions. From there, the output log file will be complete.
Model Development Update from RSG, WSP, SANDAG, and MTC
Up to the consortium as to whether or not the discussed enhancements will be included within Phase 9.
Timeline regarding the pull request depends on how stable these enhancements will be. David H.’s estimate is approximately a few weeks to a few months – depends on the amount of testing effort.
SANDAG’s preference would be to include enhancements within the same branch, as opposed to maintaining 2 branches.
A cyclical predicament: Without testing, no obvious issues. However, upon testing and making slight changes or running under different conditions, it is at this point in which issues/errors butt their heads. Although a testing effort will not be the ultimate solution, still a necessary aspect to allocate resources towards.
Going Forward
Next Meeting: Circle back regarding testing (and testing budget) in Phase 9. Open pull request back into ActivitySim for review? Decisions to be made include:
What should be included in CI testing?
Should improvements be generalized, or should we wait for Phase 9? Enhancements, thus far, have been made to keep SANDAG moving.
What additions should be kept for, for example, external models that aren’t part of the ActivitySim code base?