Guidance Needed for Public Transit Assignment Using Custom TNTP-Based Network in AequilibraE

19 views
Skip to first unread message

Justine Kojo

unread,
Mar 29, 2025, 5:49:25 PMMar 29
to AequilibraE

Hi AequilibraE team,

I'm currently working on implementing public transit assignment using the AequilibraE library, specifically aiming to replicate the functionality shown in the public example here: Public transport assignment with skimming

However, I am using locally modified TNTP network data — specifically the Sioux Falls dataset — which I have adapted to include additional attributes like travel_time, link_type, and mode. Here are the files I'm working with:

I've successfully loaded the data and created a project, but I keep running into issues when trying to use create_graph() for the transit assignment. I believe the main challenges are:

  1. Understanding the exact minimum required structure and attributes for nodes, links, and modes in a transit assignment context.

  2. Correctly inserting and registering custom link_types and modes to prevent IntegrityError or caching issues.

  3. Adapting TNTP-format data to match the expectations of AequilibraE's transit assignment framework.

Would you be able to advise on:

  • The exact structure required for the network to be used in Transit.create_graph()?

  • Whether TNTP data is viable for public transit modeling, and if so, what the necessary transformations are?

  • Whether there is a working example with openly accessible data that closely resembles this workflow?

Any insights or guidance would be greatly appreciated. I’ve invested a few weeks into this and would love to get things up and running properly!

Best regards,
Justine Kojo

Pedro Camargo

unread,
Mar 30, 2025, 1:45:23 AMMar 30
to AequilibraE
Hi Justine,

Thanks for reaching out.

AequilibraE is not currently set out to perform public transport assignment (or skimming) from generic graphs, although that is theoretically possible.

Setting it up would require quite a bit of data wrangling to get everything into place, though.

I suspect, however, that you are not going down the write path on this, however.

The TNTP format is designed to handle regular networks and does NOT include the temporal elements found in transit networks (i.e. trips). The best data source to build transit networks from is GTFS, and we always recommend users stsart from that.


A question: Wouldn't an all-or-nothing assignment with skimming do the trick for you?  Wht transit elements are you using that this route wouldn't be suitable?

Cheers,
Pedro




---- On Sat, 29 Mar 2025 23:46:34 +1000 Justine Kojo <awum...@gmail.com> wrote ---

--
You received this message because you are subscribed to the Google Groups "AequilibraE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to aequilibrae...@googlegroups.com.


Justine Kojo

unread,
Mar 30, 2025, 3:59:45 AMMar 30
to Pedro Camargo, AequilibraE

Hi Pedro,

Thanks so much for your quick and insightful response.

You’re absolutely right—my current aim is more on the strategic planning side of things. I’m looking to evaluate different transit integration scenarios (like BRT vs. paratransit) using modified TNTP-style data (e.g., Sioux Falls) that I’ve enriched with attributes relevant for transit modeling (headways, travel times, modes, etc.).

I now understand that AequilibraE’s public transit functionality is geared toward GTFS-based workflows, which makes sense given the importance of temporal dynamics. That said, my current use case is less about modeling individual passenger trips and more about estimating aggregate flows, costs, and service impacts—so even an all-or-nothing assignment with skimming, if structured well, could serve my purposes for now.

If it’s feasible within the framework, I’d love any guidance on:

• How best to adapt a TNTP-style network (with added transit-specific fields) for such a workflow;

• Whether using the project-based interface or working with generic graphs is more robust for these kinds of strategic assignments;

• Any critical fields or schema requirements the system expects (e.g., minimum requirements for link types, modes, connectors, etc.).

I’m happy to make adjustments if it means staying within AequilibraE’s strengths. Thanks again for your time and for the excellent toolset.

Best regards,

Justine




Pedro Camargo

unread,
Mar 30, 2025, 4:56:15 PMMar 30
to AequilibraE
Hi Justine,

* I am not sure why you would insist on the TNTP format, but any changes yoiu make to it would have to have corresponding changes for you to load into an AequilibraE graph. I don't think it is worth it

* As a consequence of the point before, working with an AequilibraE project is substantially easier. I am still not sure what exactly is the methodology you are applying

* The documentation is the place to start from: https://www.aequilibrae.com/latest/python/aequilibrae_project.html


Cheers,
Pedro




---- On Sun, 30 Mar 2025 17:58:43 +1000 Justine Kojo <awum...@gmail.com> wrote ---

James Bunch

unread,
Mar 31, 2025, 3:33:38 PMMar 31
to Justine Kojo, Pedro Camargo, AequilibraE
I'm confused about whether it is even possible to use the  TNTP-style network format which is anode to bnode link (or edge) based to represent transit networks without first converting any transit network to a legs based format (see the old UTPP Process) that connects each boarding node to each alightinig node along a transit route.  An example, is a service plan that uses alternating stop patterns in a CBD, or one with a trunk portion of ROW with multiple branches that include closed door suburban services to a CBD, Express services, and local services all traveling on the same physical road way.  Even if you use average headways for each line you can't represent these well with out converting to a legs based structure and also accounting for combined headways on the trunk portions.  This is not easy in a simple graph representation such as TNTP.

To view this discussion visit https://groups.google.com/d/msgid/aequilibrae/CAMnW_NWsKyaFJk9jKfkrX%2Buc3Eo3vp%2BMCn1anOjFrWGvp5vFiA%40mail.gmail.com.


--
James (Jim) Allday Bunch, JABunch Transportation Consulting
411 Penwood Road, Silver Spring Maryland, 20901 

Justine Kojo

unread,
Apr 2, 2025, 9:47:24 AMApr 2
to Pedro Camargo, AequilibraE

Hi Pedro and team,

Your suggestions and feedback are noted. Thank you for your guidance and support.

Best regards,

Justine Kojo



Reply all
Reply to author
Forward
0 new messages