Merged develop2 to develop and need to work on documentation/user guide
Continued SANDAG performance testing discussion
Its really good the 3 zone system code is working for a full population run at SANDAG
But runtimes are too slow
Now running 22 hours, including 16 hours for destination choice models
The destination choice models can be faster and this is good for everyone
Why are they slower than the existing model?
Its good for the project to keep populating the multiregion timing sheet by submodel to identify performance issues and then make improvements
RSG also added de-duplication for the TVPB to the accessibility calculator so this will help with SANDAG runtimes
SANDAG trip purpose and destination failure rate is low so that's not the issue
We can investigate and fix the slow trip destination performance with the MTC example too
SANDAG will engage RSG separately to help understand performance issues and so this may pull Jeff away for a bit, but
that's ok since performance tuning is important to everyone
Each phase of work should have an administration TO, performance TO, and maintenance TO, plus whatever else is desired
We should think about designing a better accessibility calculator - i.e. the disaggregate accessibility calculator discussed during the last round of scoping