Project Meeting 2022.01.27 - ActivitySim/activitysim GitHub Wiki

Agenda

Vehicle type model estimation results

• For option 4 (which predicts age, fuel type, and body type all at once), there are two model estimations that we can choose from. The first has a strong nesting structure, and the second trimmed some of the coefficients that might not make sense.
  • Went through estimation coefficients to understand what parameters were included
  • There was discussion about the difficulty of implementing the nesting structure in ActivitySim.
• Data was compiled to include variables like number of models, range, mpg, etc, and has the years 1991 - 2021 already included. Future year implementations would have to add years and projections for those years to the input data file. Average, inflation adjusted new car prices were also included as input data for price sensitivity.
  • Input prices can be adjusted to reflect subsidies and other policy related impacts. Another option would be to adjust the utility expression in the config file to apply different price reductions based on household attributes for instance.
• Estimation was performed on the year 2017

Vehicle Allocation estimation results

• Alternatives would choose between the vehicles owned by that household and also non-household vehicles
• There is no determination for shared rides as to whether the person is a driver or a passenger. The shared ride will just use the vehicle type allocated to the tour. Driver/passenger choices are taken into account when creating trip tables for assignment where the number of trips is divided by the occupancy.

Admin

• Clint is moving to another term, so the project management task will be going to CS