Project Meeting 2021.12.02 - ActivitySim/activitysim GitHub Wiki
Agenda
Update on the visualizer development
Model summaries are created in output folder and simwrapper reads the summaries to create plots
SimWrapper can be installed via pip
To create dashboard, add the "summarize" step at the end of the model list in settings.yaml
Summary csvs are created using python expressions in summarize.csv
These expressions have access to all of the data in an activitysim run including skims
Segmentations by demographic group, e.g. income, are under development
The local machine creating the dashboard acts as a file server is created to run the dashboard on the web as a background process.
Hosting on the cloud should be no problem
Request for "best practice" recommendations to handle creation and sharing (internally and externally) of dashboards
Berlin has a dedicated server that holds all model runs and acts as the dashboard server
Request to add more description text options for plots
Request for comparison between scenarios
Previous request for filtering / brushing is currently under development
Recent infrastructure bill mentions Travel Demand Data and Modeling, including web-based evaluation tools (section 11205)
Some more testing is still needed
Goal is to wrap up phase 1 development by the end of the year
Vehicle type model development
Model has been added to mtc example
New model, vehicle_choice, was created and a new "canonical table" called vehicles was added
There is one single vehicle_choice.yaml, and to change between option 2 or option 4, the user would change the spec and coefficient csv options to the one they want in the vehicle_choice.yaml
Would like some automated code to create the spec file
Need to confirm that model specification has been finalized
Combinatorial effects are treated dynamically in the CDAP model, could we take that implementation to improve the vehicle type model?
Will consider alternative options to make specification more extensible before giving go-ahead for code review
Suggest using model option 4 for model calibration and validation
How will regional differences be implemented and are they important?