Partners discussed and @Joe to propose some revisions
The document is a living doc so we'll continue to work on it over time
5b Task 10 Support for Three Zone Systems and TVPB (milestone)
Jeff working on caching strategies
Jeff checked in latest examples so I'm going to start on QA/QC/verification
My plan is to:
make sure it runs for me and all the settings/inputs make sense
do some simple test runs, trace results, visually check results for reasonable transit paths
do a full run and check results against Marin TM2
Can we compare detailed logging between the two setups?
Results will be a little different, but the general distributions - taps chosen, mode share, district-to-district flows, etc. should be similar
Let's be clear eyed about adding new dependencies
Unlike ct-ramp, which does generic calculations and then individual level calculations, we're instead just doing all calculations for a user defined set of market segments for consistency and ease-of-use
Questions remain about how/when to pre-compute versus on-demand/cache and we're still working toward doing full runs to help us answer these questions
More task design would have been good
Let's make sure the documentation includes:
highlights all the heuristics, assumptions, etc.
Gives good guidance to new users to avoid frustration
Assumptions around network coding requirements
Also include a discussion of the tradeoffs and benefits:
more accurate transit walk times
station choice / capacity restraint for BART
multipath transit routing to better consider tradeoffs of walking time versus ride time
basically maz to maz transit assignment
greater level of spatial detail / precision for integration with land use modeling
slower but worth it in some cases and hopefully will get faster over time through improvements
Once software in place, we could actually do a comparison of with and without, assuming we build the two data sets required