Discussed Ohio DOT insights related to CT-RAMP 2 design:
Includes Taxi mode with utility = ivt_time + constant_fare + cost_per_mile * distance
No wait time
TNC mode applies modifiers to taxi mode
Auto ownership uses an accessibility measure by TNC mode as well. The accessibility by TNC coefficient was asserted.
@Ben add notes above to Adding TNCs. This is now done.
Discussed lognormal distributed wait times and the need for data driven model design
In addition to wait time variability, there is variability in cost and variability by time-of-day
We only want defensible, data driven, empirically observed model designs
Can we setup the model software structure without observed data or should we wait until after we get the data?
We have two goals for activitysim:
A great platform / technology
A recommended example / good starting point for a model design
So, it was decided to:
Use just mean wait times for now
Not add TNC to transit since the mode share is super tiny, it means changes in logsums and calibration updates, and it isn't worth the additional effort, expression complexity, and runtime.
@SFCTA to look into new TNC data for better insights related to wait times and costs
@MTC to review Portland data used for estimating TM1.5 wait time distributions
The exercise of adding TNC illustrates that is straightforward to add TNC to transit if desired