Agenda
Note: Boston (CTPS) agency update has been moved to 7/14
- Admin
- Meeting Time Change Proposal
- Agreement review status and deadline
- Task Order Progress Check-Ins
- ActivitySim Application & Analysis Guide
- Explicit Telecommute Implementation
- EET Implementation and Improvement (Poisson & Other Fixes)
- EET Updated Random Number Generation
- EET Investigation of Inconsistencies & EET Output Analysis
- Model Calibration Partial Automation Design
- Telecommute Frequency Model Update and Tests (next week)
Notes
Admin Items
Meeting Schedule Changes
Joe proposed two structural changes to the consortium meeting cadence:
-
Alternating engineering meeting times. Following the successful pilot on the Tuesday all-hands series, engineering team meetings would also alternate between 11 AM and 2 PM Pacific to improve access for Australia-based participants. This will require separate recurring Zoom series for each time slot — attendees should expect a batch of new calendar invites.
-
Converting the "all-hands" meeting slot to a partners-only meeting. The "all-hands" meeting has been largely indistinguishable from the executive team meeting. Converting it to a partners-only discussion every three weeks could eliminate the need for a separate monthly partners-only meeting and a separate monthly Australia team meeting, reducing the total number of meetings in a typical month meetings from roughly 10 to 8. No objections were raised. Joe will send a follow-up email giving partners a final opportunity to flag concerns before implementing the changes.
Partner Agreements
Agreements have been circulating for approximately three weeks. Over 9 of the anticipated 14 agencies have indicated consent or are routing internally. Joe strongly prefers to keep agreement language unchanged to avoid a full recirculation. Any suggested changes should be submitted now for potential incorporation in next year's cycle. Agencies with multi-year agreements (e.g., Met Council through 2029, SANDAG through 2030) do not need to re-execute at this time.
Task Updates
ActivitySim Application Analysis Guide (Joe Flood)
Joe Flood confirmed he is working on an outline for the land use change scenario. A key open question — whether to use a smaller, clipped SANDAG dataset rather than the full model — was discussed and provisionally approved. The smaller dataset (from the SANDAG ABM3 example repo) would allow readers to download and run scenarios themselves more easily. It was agreed that all guides in the series should use the same example dataset, and that the telecommute and network change scenarios could potentially be represented in the smaller dataset as well. Sumit sought clarification that this is not the 100-zone CI/CD test dataset but a larger intermediate one. Application analysis guide presentation moved to June 16
Explicit Telecommute Implementation (Sijia Wang)
Sijia has been setting up the software for two new model components and drafting unit tests. The next step is building the estimation data bundle using SANDAG survey data. Bhargava Sana confirmed he is on the critical path and will provide a previously processed survey file within one to two days. Sijia plans to present draft estimation results at the May 26 check-in, with the caveat that results may still be preliminary.
EET Implementation & Poisson Sampling (Jan Zill / Outer Loop)
Joe sent a scope amendment request to partners the prior day asking for consent to proceed with a modest expansion of Outer Loop's scope to incorporate Poisson sampling and related fixes. Jan confirmed the implementation is complete and ready to fold in within approximately a few days of receiving formal go-ahead. Partners were asked to raise any objections by the following day. The inconsistency investigation task depends on this implementation being in place first.
Random Number Generation (RNG) Testing & Implementation (RSG / David Hensle)
David and Jan have done profiling on ActivitySim's RNG pipeline and found meaningful opportunities to improve runtime for explicit error terms. The profiling work will help isolate specific pain points in the code — both in the RNG itself and in the plumbing that routes generated numbers into choice tables.
The group agreed that RNG work should be completed before beginning the EET output analysis, to avoid having to redo analysis after code changes. Target for RNG implementation and reporting: June 2. RNG progress will be a standing item at engineering team meetings in the interim.
EET Output Analysis & Inconsistency Investigation (RSG + Outer Loop)
The group discussed which scenarios to rerun for the analysis (original SANDAG runs, potentially a Victoria-based model). Jan noted that if the group drops from two scenarios to one, it may be possible to include a non-SANDAG model within existing budget. Joe noted the scenarios have produced meaningfully different results in prior analysis (e.g., employment vs. transit service tests affected destination choice differently), and suggested the topic be discussed further at an engineering meeting. No decisions were made today.
Revised schedule: EET output analysis and investigation moved to June 23.
Model Calibration Automation (RSG / David Hensle)
RSG has completed the design memo and is circulating it internally for review. It is expected to go out to the broader agency group by end of week. Models in scope are auto ownership, tour mode choice, and workplace location choice. David will share additional notebooks for the destination choice models to help partners review and provide feedback. The non-mandatory tour destination notebook is functionally representative of the workplace location model.
Telecommute Frequency Model Update (RSG / David Hensle)
RSG has received and inspected the estimation data bundle from SANDAG and is actively working through the estimation. David expects to present results at the May 19 engineering meeting.
Upcoming Exec Team & Product / Community Team Meeting Discussion Items
| Date |
Topic |
| May 19 |
Telecommute frequency model update |
| May 26 |
Explicit telecommute implementation |
| June 2 |
RNG test/implementation |
| June 9 |
Auto-Calibration Partial Implementation |
| June 16 |
Application analysis guide (moved up from original date) |
| June 23 |
EET output analysis & inconsistency investigation |
Agenda
Note: Boston (CTPS) agency update has been moved to 7/14
Notes
Admin Items
Meeting Schedule Changes
Joe proposed two structural changes to the consortium meeting cadence:
Alternating engineering meeting times. Following the successful pilot on the Tuesday all-hands series, engineering team meetings would also alternate between 11 AM and 2 PM Pacific to improve access for Australia-based participants. This will require separate recurring Zoom series for each time slot — attendees should expect a batch of new calendar invites.
Converting the "all-hands" meeting slot to a partners-only meeting. The "all-hands" meeting has been largely indistinguishable from the executive team meeting. Converting it to a partners-only discussion every three weeks could eliminate the need for a separate monthly partners-only meeting and a separate monthly Australia team meeting, reducing the total number of meetings in a typical month meetings from roughly 10 to 8. No objections were raised. Joe will send a follow-up email giving partners a final opportunity to flag concerns before implementing the changes.
Partner Agreements
Agreements have been circulating for approximately three weeks. Over 9 of the anticipated 14 agencies have indicated consent or are routing internally. Joe strongly prefers to keep agreement language unchanged to avoid a full recirculation. Any suggested changes should be submitted now for potential incorporation in next year's cycle. Agencies with multi-year agreements (e.g., Met Council through 2029, SANDAG through 2030) do not need to re-execute at this time.
Task Updates
ActivitySim Application Analysis Guide (Joe Flood)
Joe Flood confirmed he is working on an outline for the land use change scenario. A key open question — whether to use a smaller, clipped SANDAG dataset rather than the full model — was discussed and provisionally approved. The smaller dataset (from the SANDAG ABM3 example repo) would allow readers to download and run scenarios themselves more easily. It was agreed that all guides in the series should use the same example dataset, and that the telecommute and network change scenarios could potentially be represented in the smaller dataset as well. Sumit sought clarification that this is not the 100-zone CI/CD test dataset but a larger intermediate one. Application analysis guide presentation moved to June 16
Explicit Telecommute Implementation (Sijia Wang)
Sijia has been setting up the software for two new model components and drafting unit tests. The next step is building the estimation data bundle using SANDAG survey data. Bhargava Sana confirmed he is on the critical path and will provide a previously processed survey file within one to two days. Sijia plans to present draft estimation results at the May 26 check-in, with the caveat that results may still be preliminary.
EET Implementation & Poisson Sampling (Jan Zill / Outer Loop)
Joe sent a scope amendment request to partners the prior day asking for consent to proceed with a modest expansion of Outer Loop's scope to incorporate Poisson sampling and related fixes. Jan confirmed the implementation is complete and ready to fold in within approximately a few days of receiving formal go-ahead. Partners were asked to raise any objections by the following day. The inconsistency investigation task depends on this implementation being in place first.
Random Number Generation (RNG) Testing & Implementation (RSG / David Hensle)
David and Jan have done profiling on ActivitySim's RNG pipeline and found meaningful opportunities to improve runtime for explicit error terms. The profiling work will help isolate specific pain points in the code — both in the RNG itself and in the plumbing that routes generated numbers into choice tables.
The group agreed that RNG work should be completed before beginning the EET output analysis, to avoid having to redo analysis after code changes. Target for RNG implementation and reporting: June 2. RNG progress will be a standing item at engineering team meetings in the interim.
EET Output Analysis & Inconsistency Investigation (RSG + Outer Loop)
The group discussed which scenarios to rerun for the analysis (original SANDAG runs, potentially a Victoria-based model). Jan noted that if the group drops from two scenarios to one, it may be possible to include a non-SANDAG model within existing budget. Joe noted the scenarios have produced meaningfully different results in prior analysis (e.g., employment vs. transit service tests affected destination choice differently), and suggested the topic be discussed further at an engineering meeting. No decisions were made today.
Revised schedule: EET output analysis and investigation moved to June 23.
Model Calibration Automation (RSG / David Hensle)
RSG has completed the design memo and is circulating it internally for review. It is expected to go out to the broader agency group by end of week. Models in scope are auto ownership, tour mode choice, and workplace location choice. David will share additional notebooks for the destination choice models to help partners review and provide feedback. The non-mandatory tour destination notebook is functionally representative of the workplace location model.
Telecommute Frequency Model Update (RSG / David Hensle)
RSG has received and inspected the estimation data bundle from SANDAG and is actively working through the estimation. David expects to present results at the May 19 engineering meeting.
Upcoming Exec Team & Product / Community Team Meeting Discussion Items