Michelle to email group for feedback on the proposed approaches below. (Done)
Alternatives for the vehicle type choice model will be specified in the model settings and will be generated programmatically by combining the supplied body type, fuel types, and age. Data for each of these models will be joined to the alternatives from the supplied vehicle_type_data.csv file. This allows the user to have one data file for many scenario years and additional vehicle types that may not want to always be included in every scenario. If an alternative does not have a match from the data file, an error will be thrown and the run will stop.
The vehicle allocation model will only be run once before tour mode choice as opposed to every time tour mode choice is called. This will significantly decrease run time and complexity. For logsum calculations in upstream models, user supplied default values for auto operating characteristics will be used.
User supplied default values for auto operating characteristics will also be used in mode choice when a non-household vehicle is chosen by the vehicle allocation model.
SANDAG to conduct code review of latest visualization commits from WSP, or report to the group if they are no able to do so and the task will be assigned to someone else. (Michelle to add this as an issue, to track.)
WSP to look at conda forge to better consolidate dependencies.
Notes
WSP Visualization Task Update
WSP cleaned up loose ends from previous request, such as updated documentation (see presentation for list of items).
SANDAG is building visualizations out of their actual model outputs for standard things like VMT, mode share, etc. They are also including two use cases: trips to/from employment centers and mobility hubs, looking at mode share, travel times, demographics, etc. They may be interested in presenting their visualization work after presenting internally to their planners and leadership in the coming weeks.
RSG Vehicle Type Model Update
Future Year Scenarios
Noted differences between vehicle year and model age (which is model year - vehicle year). You could set the Scenario Year in the settings file to 2017 to use existing data; but in theory, you would want to project vehicle type input data for your scenario year.
It was suggested that someone develop a few future year scenario inputs (high EVs, low EVs, for example) instead of having individuals generating their own each time, to have something available out of the box. This is not part of anyone's current scope of work. SEMCOG and SANDAG are the likely first ones to use this component in application and could potentially present on or provide their future year assumptions.
Estimation of Vehicle Type
Discuss which terms were found to be significant and insignificant.
Noted that the vehicle(s) that a household already owns is an explanatory variable.
Vehicle Allocation Model, needs to be run for each occupancy level