SPUR Wish #3 - Yesterday's Transit

[ondrae]

Short description:

A dashboard to easily see the status of the Bay Area's public transit, yesterday.

Long Description:

SPUR presented at hack night tonight and had three wishes. Number three was for this dashboard. They showed a horrendous looking example and said they dream about having something like that. We can do better.

We should talk to her further about what data would be useful for SPUR.

To start though, we could have:

• How many minutes of transit delay happened yesterday region wide.
• Avg speed of transit across the region
• How many transit trips happened yesterday
• How much co2 was avoided yesterday
• Lots more

All of these examples can be figured out with just the GTFS Realtime data.

We could start with just that first one. How many minutes delay was there. Here's how we could answer that:

• [ ] Get the GTFS Realtime data for one system
• [ ] Figure out how to differences between realtime and the scheduled times for one line
• [ ] Do that for all the lines in that system
• [ ] Research GTFS Realtime for all the of the region's transit agencies (I doubt they all have it).
• [ ] Use the algorithm on all of the transit agencies.
• [ ] Display the total minutes of delay in the Bay.

  Project Needs:

• Project owner - hopefully SPUR
• Transit Data Researchers - go find the data
• Programmers to write the algorithms
• Designers to make a presentable dashboard

  Status:

Still just in discussion here in this issue.

[stevepepple]

Related to transit feeds and "what happened yesterday", there are a couple of data sources and projects I'm aware of:

Aggregation and Normalization of Transit Feed, e.g. GTFS: https://transit.land/

Real-time Database of Muni and other NextBus-enabled Services: https://www.firebase.com/docs/open-data/transit.html

[irees]

I can calculate the # of service hours for transit on a given day, using public static GTFS feeds, as well as for each route: average frequency, average speed. Also split by bus/rail, agency. I'll try and cook something up this weekend.

[ondrae]

Thats great. We could get the expected stats for each day that way. Any chance of calculating the real value of what actually happened that day though? Does GTFS Realtime allow for that?

The static values are a great start though.

[irees]

I'll start with static schedule data; RT data might be a good next step for looking at schedule adherence and delays.

[stevepepple]

A project like Car Free A to Z would help multi-modal megaregion transit users: http://www.carfreeatoz.com/planner

They use OpenTripPlanner, OpenStreetMaps, and GTFS, which means it could be ported to another city.

[ondrae]

Some interesting charts to show delays. https://mbtaviz.github.io/

[luqmaan]

A first step towards this idea could be storing historical GTFS-RT data.

For example, we have 9 months of GTFS-RT Vehicle Positions stored on GitHub for Austin: https://github.com/scascketta/capmetrics.

Then you can show a dashboard for any day, instead of just yesterday's transit. You can also do other types of analysis (e.g. which route is faster: https://github.com/open-austin/project-ideas/issues/3).

[ondrae]

Exactly @luqmaan thanks for the link. I'll check out that project. I was hoping there was a service out there that already had the historical data.

@invisiblefunnel do know of any services or agencies that track historical data? Is doing it our damn selves the way?