Project Meeting 2017.07.07

Revisions for support for logsums

Working through this plan
The mode choice availability rules are not implemented so the logsums will be funny
We'd like to implement the availability rules now as expressions instead of orca calculated columns
We don’t yet have support for chunked simple_simulate, but we will probably want to implement it now that we are running mode_choice simple to calculate compute_logsums on person_count*tazcount choosers
Revised work location with logsums model is working, but still untidy, no efficiency, not much testing, no pep8
The in-progress logsums branch is here
We still need to update the documentation and run it on the full model setup as well

We're on budget and have spent around 50% of the budget
We should finish up the logsums task in a few weeks and then we have two smaller tasks remaining
So we should be done with this phase of work in a few months
I'll share the phase 3 amendment scope and cost estimate with the group so they can work on accelerating funding
@Clint discuss outreach to interested MPOs next time

@Wu discussed results of the updated performance tests
For most tested methods, the improved software is 30-80 times faster
Previously, two model components ("stop location" and "trip mode choices") were not able to complete most of the test scenarios that mimic SANDAG settings, except one trip mode choice scenario, due to the extreme runtime. With the software improvements, all tests finished within reasonable runtime
See Wu's memo for more details
@Wu update/create new tests page
With the remaining budget, we'll look into speeding up get_taps_mazs

Stefan confirmed this week that the results of the parcel buffering in ActivitySim are the same as in Soundcast
He is pretty close to issuing a pull request once he cleans things up a bit
He expects a bit of back and forth before the request is actually accepted, but thinks it is time for feedback